nrabulinski/lix
0
0
Fork 0
forked from lix-project/lix
Code Pull requests Activity
lix/src/libexpr/parser.y

760 lines
23 KiB
Plaintext
Raw Normal View History

* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
%glr-parser
Fix Bison 2.4 warning
2019-10-09 20:57:37 +00:00
%define api.pure
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
%locations
Fix Bison 3.3 warning
2019-03-27 20:09:31 +00:00
%define parse.error verbose
* Hack for Bison 2.3 compatability.
2006-10-02 14:43:15 +00:00
%defines
* Removed URIs from the evaluator (NIX-66). They are now just another kind of notation for strings.
2006-10-11 21:59:33 +00:00
/* %no-lines */
Fix building against Bison 2.6
2013-03-14 17:31:08 +00:00
%parse-param { void * scanner }
%parse-param { nix::ParseData * data }
%lex-param { void * scanner }
%lex-param { nix::ParseData * data }
Use Bison 2.5
2012-04-13 12:28:26 +00:00
%expect 1
%expect-rr 1
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
* Optimise string constants by putting them in the symbol table.
2010-10-23 21:11:59 +00:00
%code requires {
Fix whitespace
2013-09-02 14:29:15 +00:00
* Optimise string constants by putting them in the symbol table.
2010-10-23 21:11:59 +00:00
#ifndef BISON_HEADER
#define BISON_HEADER
Fix whitespace
2013-09-02 14:29:15 +00:00
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
#include "util.hh"
Fix whitespace
2013-09-02 14:29:15 +00:00
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
#include "nixexpr.hh"
* Add a Nix expression search path feature. Paths between angle brackets, e.g. import <nixpkgs/pkgs/lib> are resolved by looking them up relative to the elements listed in the search path. This allows us to get rid of hacks like import "${builtins.getEnv "NIXPKGS_ALL"}/pkgs/lib" The search path can be specified through the ‘-I’ command-line flag and through the colon-separated ‘NIX_PATH’ environment variable, e.g., $ nix-build -I /etc/nixos ... If a file is not found in the search path, an error message is lazily thrown.
2011-08-06 16:05:24 +00:00
#include "eval.hh"
Add feature to disable URL literals E.g. $ nix-build '<nixpkgs>' -A hello --experimental-features no-url-literals error: URL literals are disabled, at /nix/store/vsjamkzh15r3c779q2711az826hqgvzr-nixpkgs-20.03pre194957.bef773ed53f/nixpkgs/pkgs/top-level/all-packages.nix:1236:11 Helps with implementing https://github.com/NixOS/rfcs/pull/45.
2019-11-26 18:48:34 +00:00
#include "globals.hh"
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
* Optimise string constants by putting them in the symbol table.
2010-10-23 21:11:59 +00:00
namespace nix {
Fix whitespace
2013-09-02 14:29:15 +00:00
struct ParseData
* Optimise string constants by putting them in the symbol table.
2010-10-23 21:11:59 +00:00
{
* Add a Nix expression search path feature. Paths between angle brackets, e.g. import <nixpkgs/pkgs/lib> are resolved by looking them up relative to the elements listed in the search path. This allows us to get rid of hacks like import "${builtins.getEnv "NIXPKGS_ALL"}/pkgs/lib" The search path can be specified through the ‘-I’ command-line flag and through the colon-separated ‘NIX_PATH’ environment variable, e.g., $ nix-build -I /etc/nixos ... If a file is not found in the search path, an error message is lazily thrown.
2011-08-06 16:05:24 +00:00
EvalState & state;
* Optimise string constants by putting them in the symbol table.
2010-10-23 21:11:59 +00:00
SymbolTable & symbols;
Expr * result;
Path basePath;
print LOC for stdin, string args
2020-05-21 04:18:26 +00:00
Symbol file;
FileOrigin origin;
more pos reporting
2020-05-11 23:34:57 +00:00
ErrorInfo error;
* Optimise string constants by putting them in the symbol table.
2010-10-23 21:11:59 +00:00
Symbol sLetBody;
* Add a Nix expression search path feature. Paths between angle brackets, e.g. import <nixpkgs/pkgs/lib> are resolved by looking them up relative to the elements listed in the search path. This allows us to get rid of hacks like import "${builtins.getEnv "NIXPKGS_ALL"}/pkgs/lib" The search path can be specified through the ‘-I’ command-line flag and through the colon-separated ‘NIX_PATH’ environment variable, e.g., $ nix-build -I /etc/nixos ... If a file is not found in the search path, an error message is lazily thrown.
2011-08-06 16:05:24 +00:00
ParseData(EvalState & state)
: state(state)
, symbols(state.symbols)
* Optimise string constants by putting them in the symbol table.
2010-10-23 21:11:59 +00:00
, sLetBody(symbols.create("<let-body>"))
{ };
};
Fix whitespace
2013-09-02 14:29:15 +00:00
* Optimise string constants by putting them in the symbol table.
2010-10-23 21:11:59 +00:00
}
#define YY_DECL int yylex \
(YYSTYPE * yylval_param, YYLTYPE * yylloc_param, yyscan_t yyscanner, nix::ParseData * data)
#endif
}
%{
* Remove unnecessary inclusions of aterm2.h.
2006-09-04 22:08:40 +00:00
#include "parser-tab.hh"
#include "lexer-tab.hh"
* Optimise string constants by putting them in the symbol table.
2010-10-23 21:11:59 +00:00
YY_DECL;
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
* Use a proper namespace. * Optimise header file usage a bit. * Compile the parser as C++.
2006-09-04 21:06:23 +00:00
using namespace nix;
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
* Use a proper namespace. * Optimise header file usage a bit. * Compile the parser as C++.
2006-09-04 21:06:23 +00:00
namespace nix {
Fix whitespace
2013-09-02 14:29:15 +00:00
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
* In the ‘?’ operator, allow attribute paths. For instance, you can write ‘attrs ? a.b’ to test whether ‘attrs’ has an attribute ‘a’ containing an attribute ‘b’. This is more convenient than ‘attrs ? a && attrs.a ? b’. Slight change in the semantics: it's no longer an error if the left-hand side of ‘?’ is not an attribute set. In that case it just returns false. So, ‘null ? foo’ no longer throws an error.
2011-07-06 10:58:17 +00:00
static void dupAttr(const AttrPath & attrPath, const Pos & pos, const Pos & prevPos)
* Check for duplicate attribute names / function arguments. `make check' now succeeds :-) * An attribute set such as `{ foo = { enable = true; }; foo.port = 23; }' now parses. It was previously rejected, but I'm too lazy to implement the check. (The only reason to reject it is that the reverse, `{ foo.port = 23; foo = { enable = true; }; }', is rejected, which is kind of ugly.)
2010-04-22 11:02:24 +00:00
{
new format for pos
2020-05-11 22:58:08 +00:00
throw ParseError(
ErrorInfo {
.hint = hintfmt("attribute '%1%' already defined at %2%",
showAttrPath(attrPath), prevPos),
.nixCode = NixCode { .errPos = pos },
});
* Check for duplicate attribute names / function arguments. `make check' now succeeds :-) * An attribute set such as `{ foo = { enable = true; }; foo.port = 23; }' now parses. It was previously rejected, but I'm too lazy to implement the check. (The only reason to reject it is that the reverse, `{ foo.port = 23; foo = { enable = true; }; }', is rejected, which is kind of ugly.)
2010-04-22 11:02:24 +00:00
}
Fix whitespace
2013-09-02 14:29:15 +00:00
* Check for duplicate attribute names / function arguments. `make check' now succeeds :-) * An attribute set such as `{ foo = { enable = true; }; foo.port = 23; }' now parses. It was previously rejected, but I'm too lazy to implement the check. (The only reason to reject it is that the reverse, `{ foo.port = 23; foo = { enable = true; }; }', is rejected, which is kind of ugly.)
2010-04-22 11:02:24 +00:00
* Store attribute positions in the AST and report duplicate attribute errors with position info. * For all positions, use the position of the first character of the first token, rather than the last character of the first token plus one.
2010-05-06 16:46:48 +00:00
static void dupAttr(Symbol attr, const Pos & pos, const Pos & prevPos)
* Check for duplicate attribute names / function arguments. `make check' now succeeds :-) * An attribute set such as `{ foo = { enable = true; }; foo.port = 23; }' now parses. It was previously rejected, but I'm too lazy to implement the check. (The only reason to reject it is that the reverse, `{ foo.port = 23; foo = { enable = true; }; }', is rejected, which is kind of ugly.)
2010-04-22 11:02:24 +00:00
{
new format for pos
2020-05-11 22:58:08 +00:00
throw ParseError(
ErrorInfo {
.hint = hintfmt("attribute '%1%' already defined at %2%",
attr, prevPos),
.nixCode = NixCode { .errPos = pos },
});
* Check for duplicate attribute names / function arguments. `make check' now succeeds :-) * An attribute set such as `{ foo = { enable = true; }; foo.port = 23; }' now parses. It was previously rejected, but I'm too lazy to implement the check. (The only reason to reject it is that the reverse, `{ foo.port = 23; foo = { enable = true; }; }', is rejected, which is kind of ugly.)
2010-04-22 11:02:24 +00:00
}
Fix whitespace
2013-09-02 14:29:15 +00:00
* Some syntactic sugar for attribute sets: allow {x.y.z = ...;} as a shorthand for {x = {y = {z = ...;};};}. This is especially useful for NixOS configuration files, e.g. { services = { sshd = { enable = true; port = 2022; }; }; } can now be written as { services.sshd.enable = true; services.sshd.port = 2022; } However, it is currently not permitted to write { services.sshd = {enable = true;}; services.sshd.port = 2022; } as this is considered a duplicate definition of `services.sshd'.
2009-05-15 12:35:23 +00:00
Don't use any syntactic sugar for dynamic attrs This doesn't change any functionality but moves some behavior out of the parser and into the evaluator in order to simplify the code. Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 23:56:26 +00:00
static void addAttr(ExprAttrs * attrs, AttrPath & attrPath,
* Use a symbol table to represent identifiers and attribute names efficiently. The symbol table ensures that there is only one copy of each symbol, thus allowing symbols to be compared efficiently using a pointer equality test.
2010-04-13 12:25:42 +00:00
Expr * e, const Pos & pos)
* Some syntactic sugar for attribute sets: allow {x.y.z = ...;} as a shorthand for {x = {y = {z = ...;};};}. This is especially useful for NixOS configuration files, e.g. { services = { sshd = { enable = true; port = 2022; }; }; } can now be written as { services.sshd.enable = true; services.sshd.port = 2022; } However, it is currently not permitted to write { services.sshd = {enable = true;}; services.sshd.port = 2022; } as this is considered a duplicate definition of `services.sshd'.
2009-05-15 12:35:23 +00:00
{
Don't use any syntactic sugar for dynamic attrs This doesn't change any functionality but moves some behavior out of the parser and into the evaluator in order to simplify the code. Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 23:56:26 +00:00
AttrPath::iterator i;
Fold dynamic binds handling into addAttr Since addAttr has to iterate through the AttrPath we pass it, it makes more sense to just iterate through the AttrNames in addAttr instead. As an added bonus, this allows attrsets where two dynamic attribute paths have the same static leading part (see added test case for an example that failed previously). Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 22:57:10 +00:00
// All attrpaths have at least one attr
Don't use any syntactic sugar for dynamic attrs This doesn't change any functionality but moves some behavior out of the parser and into the evaluator in order to simplify the code. Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 23:56:26 +00:00
assert(!attrPath.empty());
parser: Allow mixed nested and top-level attrpaths Fixes #2077.
2018-04-18 16:39:40 +00:00
// Checking attrPath validity.
// ===========================
Iterate over references
2019-05-28 21:05:08 +00:00
for (i = attrPath.begin(); i + 1 < attrPath.end(); i++) {
Fold dynamic binds handling into addAttr Since addAttr has to iterate through the AttrPath we pass it, it makes more sense to just iterate through the AttrNames in addAttr instead. As an added bonus, this allows attrsets where two dynamic attribute paths have the same static leading part (see added test case for an example that failed previously). Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 22:57:10 +00:00
if (i->symbol.set()) {
ExprAttrs::AttrDefs::iterator j = attrs->attrs.find(i->symbol);
if (j != attrs->attrs.end()) {
if (!j->second.inherited) {
ExprAttrs * attrs2 = dynamic_cast<ExprAttrs *>(j->second.e);
Don't use any syntactic sugar for dynamic attrs This doesn't change any functionality but moves some behavior out of the parser and into the evaluator in order to simplify the code. Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 23:56:26 +00:00
if (!attrs2) dupAttr(attrPath, pos, j->second.pos);
Fold dynamic binds handling into addAttr Since addAttr has to iterate through the AttrPath we pass it, it makes more sense to just iterate through the AttrNames in addAttr instead. As an added bonus, this allows attrsets where two dynamic attribute paths have the same static leading part (see added test case for an example that failed previously). Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 22:57:10 +00:00
attrs = attrs2;
} else
Don't use any syntactic sugar for dynamic attrs This doesn't change any functionality but moves some behavior out of the parser and into the evaluator in order to simplify the code. Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 23:56:26 +00:00
dupAttr(attrPath, pos, j->second.pos);
Fold dynamic binds handling into addAttr Since addAttr has to iterate through the AttrPath we pass it, it makes more sense to just iterate through the AttrNames in addAttr instead. As an added bonus, this allows attrsets where two dynamic attribute paths have the same static leading part (see added test case for an example that failed previously). Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 22:57:10 +00:00
} else {
* Finished the ATerm-less parser.
2010-04-12 23:33:23 +00:00
ExprAttrs * nested = new ExprAttrs;
Fold dynamic binds handling into addAttr Since addAttr has to iterate through the AttrPath we pass it, it makes more sense to just iterate through the AttrNames in addAttr instead. As an added bonus, this allows attrsets where two dynamic attribute paths have the same static leading part (see added test case for an example that failed previously). Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 22:57:10 +00:00
attrs->attrs[i->symbol] = ExprAttrs::AttrDef(nested, pos);
* Finished the ATerm-less parser.
2010-04-12 23:33:23 +00:00
attrs = nested;
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
}
Fold dynamic binds handling into addAttr Since addAttr has to iterate through the AttrPath we pass it, it makes more sense to just iterate through the AttrNames in addAttr instead. As an added bonus, this allows attrsets where two dynamic attribute paths have the same static leading part (see added test case for an example that failed previously). Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 22:57:10 +00:00
} else {
ExprAttrs *nested = new ExprAttrs;
attrs->dynamicAttrs.push_back(ExprAttrs::DynamicAttrDef(i->expr, nested, pos));
attrs = nested;
* Some syntactic sugar for attribute sets: allow {x.y.z = ...;} as a shorthand for {x = {y = {z = ...;};};}. This is especially useful for NixOS configuration files, e.g. { services = { sshd = { enable = true; port = 2022; }; }; } can now be written as { services.sshd.enable = true; services.sshd.port = 2022; } However, it is currently not permitted to write { services.sshd = {enable = true;}; services.sshd.port = 2022; } as this is considered a duplicate definition of `services.sshd'.
2009-05-15 12:35:23 +00:00
}
* Check for duplicate attributes in fixAttrs, rather than doing a separate traversal after parsing. Likewise, check for duplicate pattern variables right away.
2009-05-14 14:29:45 +00:00
}
parser: Allow mixed nested and top-level attrpaths Fixes #2077.
2018-04-18 16:39:40 +00:00
// Expr insertion.
// ==========================
Iterate over references
2019-05-28 21:05:08 +00:00
if (i->symbol.set()) {
Fold dynamic binds handling into addAttr Since addAttr has to iterate through the AttrPath we pass it, it makes more sense to just iterate through the AttrNames in addAttr instead. As an added bonus, this allows attrsets where two dynamic attribute paths have the same static leading part (see added test case for an example that failed previously). Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 22:57:10 +00:00
ExprAttrs::AttrDefs::iterator j = attrs->attrs.find(i->symbol);
if (j != attrs->attrs.end()) {
parser: Allow mixed nested and top-level attrpaths Fixes #2077.
2018-04-18 16:39:40 +00:00
// This attr path is already defined. However, if both
// e and the expr pointed by the attr path are two attribute sets,
// we want to merge them.
// Otherwise, throw an error.
Iterate over references
2019-05-28 21:05:08 +00:00
auto ae = dynamic_cast<ExprAttrs *>(e);
auto jAttrs = dynamic_cast<ExprAttrs *>(j->second.e);
parser: Allow mixed nested and top-level attrpaths Fixes #2077.
2018-04-18 16:39:40 +00:00
if (jAttrs && ae) {
Iterate over references
2019-05-28 21:05:08 +00:00
for (auto & ad : ae->attrs) {
auto j2 = jAttrs->attrs.find(ad.first);
parser: Allow mixed nested and top-level attrpaths Fixes #2077.
2018-04-18 16:39:40 +00:00
if (j2 != jAttrs->attrs.end()) // Attr already defined in iAttrs, error.
dupAttr(ad.first, j2->second.pos, ad.second.pos);
jAttrs->attrs[ad.first] = ad.second;
}
} else {
dupAttr(attrPath, pos, j->second.pos);
}
Fold dynamic binds handling into addAttr Since addAttr has to iterate through the AttrPath we pass it, it makes more sense to just iterate through the AttrNames in addAttr instead. As an added bonus, this allows attrsets where two dynamic attribute paths have the same static leading part (see added test case for an example that failed previously). Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 22:57:10 +00:00
} else {
parser: Allow mixed nested and top-level attrpaths Fixes #2077.
2018-04-18 16:39:40 +00:00
// This attr path is not defined. Let's create it.
Fold dynamic binds handling into addAttr Since addAttr has to iterate through the AttrPath we pass it, it makes more sense to just iterate through the AttrNames in addAttr instead. As an added bonus, this allows attrsets where two dynamic attribute paths have the same static leading part (see added test case for an example that failed previously). Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 22:57:10 +00:00
attrs->attrs[i->symbol] = ExprAttrs::AttrDef(e, pos);
e->setName(i->symbol);
}
} else {
attrs->dynamicAttrs.push_back(ExprAttrs::DynamicAttrDef(i->expr, e, pos));
}
* Check for duplicate attributes in fixAttrs, rather than doing a separate traversal after parsing. Likewise, check for duplicate pattern variables right away.
2009-05-14 14:29:45 +00:00
}
* Check for duplicate attribute names / function arguments. `make check' now succeeds :-) * An attribute set such as `{ foo = { enable = true; }; foo.port = 23; }' now parses. It was previously rejected, but I'm too lazy to implement the check. (The only reason to reject it is that the reverse, `{ foo.port = 23; foo = { enable = true; }; }', is rejected, which is kind of ugly.)
2010-04-22 11:02:24 +00:00
static void addFormal(const Pos & pos, Formals * formals, const Formal & formal)
* Check for duplicate attributes in fixAttrs, rather than doing a separate traversal after parsing. Likewise, check for duplicate pattern variables right away.
2009-05-14 14:29:45 +00:00
{
OCD performance fix: {find,count}+insert => insert
2019-10-09 13:51:52 +00:00
if (!formals->argNames.insert(formal.name).second)
new format for pos
2020-05-11 22:58:08 +00:00
throw ParseError(
ErrorInfo {
.hint = hintfmt("duplicate formal function argument '%1%'",
formal.name),
.nixCode = NixCode { .errPos = pos },
});
* Check for duplicate attribute names / function arguments. `make check' now succeeds :-) * An attribute set such as `{ foo = { enable = true; }; foo.port = 23; }' now parses. It was previously rejected, but I'm too lazy to implement the check. (The only reason to reject it is that the reverse, `{ foo.port = 23; foo = { enable = true; }; }', is rejected, which is kind of ugly.)
2010-04-22 11:02:24 +00:00
formals->formals.push_front(formal);
* Check for duplicate attributes in fixAttrs, rather than doing a separate traversal after parsing. Likewise, check for duplicate pattern variables right away.
2009-05-14 14:29:45 +00:00
}
Show position info in string concatenation / addition errors
2014-04-04 20:19:33 +00:00
static Expr * stripIndentation(const Pos & pos, SymbolTable & symbols, vector<Expr *> & es)
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
{
* Optimise string constants by putting them in the symbol table.
2010-10-23 21:11:59 +00:00
if (es.empty()) return new ExprString(symbols.create(""));
Fix whitespace
2013-09-02 14:29:15 +00:00
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
/* Figure out the minimum indentation. Note that by design
whitespace-only final lines are not taken into account. (So
the " " in "\n ''" is ignored, but the " " in "\n foo''" is.) */
bool atStartOfLine = true; /* = seen only whitespace in the current line */
Fix some random -Wconversion warnings
2018-05-02 11:56:34 +00:00
size_t minIndent = 1000000;
size_t curIndent = 0;
OCD: foreach -> C++11 ranged for
2015-07-17 17:24:28 +00:00
for (auto & i : es) {
ExprIndStr * e = dynamic_cast<ExprIndStr *>(i);
* Indented strings.
2010-04-12 22:03:27 +00:00
if (!e) {
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
/* Anti-quotations end the current start-of-line whitespace. */
if (atStartOfLine) {
atStartOfLine = false;
if (curIndent < minIndent) minIndent = curIndent;
}
continue;
}
Fix some random -Wconversion warnings
2018-05-02 11:56:34 +00:00
for (size_t j = 0; j < e->s.size(); ++j) {
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
if (atStartOfLine) {
* Indented strings.
2010-04-12 22:03:27 +00:00
if (e->s[j] == ' ')
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
curIndent++;
* Indented strings.
2010-04-12 22:03:27 +00:00
else if (e->s[j] == '\n') {
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
/* Empty line, doesn't influence minimum
indentation. */
curIndent = 0;
} else {
atStartOfLine = false;
if (curIndent < minIndent) minIndent = curIndent;
}
* Indented strings.
2010-04-12 22:03:27 +00:00
} else if (e->s[j] == '\n') {
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
atStartOfLine = true;
curIndent = 0;
}
}
}
/* Strip spaces from each line. */
* Indented strings.
2010-04-12 22:03:27 +00:00
vector<Expr *> * es2 = new vector<Expr *>;
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
atStartOfLine = true;
Fix some random -Wconversion warnings
2018-05-02 11:56:34 +00:00
size_t curDropped = 0;
size_t n = es.size();
* Indented strings.
2010-04-12 22:03:27 +00:00
for (vector<Expr *>::iterator i = es.begin(); i != es.end(); ++i, --n) {
ExprIndStr * e = dynamic_cast<ExprIndStr *>(*i);
if (!e) {
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
atStartOfLine = false;
curDropped = 0;
* Indented strings.
2010-04-12 22:03:27 +00:00
es2->push_back(*i);
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
continue;
}
Fix whitespace
2013-09-02 14:29:15 +00:00
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
string s2;
Fix some random -Wconversion warnings
2018-05-02 11:56:34 +00:00
for (size_t j = 0; j < e->s.size(); ++j) {
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
if (atStartOfLine) {
* Indented strings.
2010-04-12 22:03:27 +00:00
if (e->s[j] == ' ') {
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
if (curDropped++ >= minIndent)
* Indented strings.
2010-04-12 22:03:27 +00:00
s2 += e->s[j];
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
}
* Indented strings.
2010-04-12 22:03:27 +00:00
else if (e->s[j] == '\n') {
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
curDropped = 0;
* Indented strings.
2010-04-12 22:03:27 +00:00
s2 += e->s[j];
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
} else {
atStartOfLine = false;
curDropped = 0;
* Indented strings.
2010-04-12 22:03:27 +00:00
s2 += e->s[j];
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
}
} else {
* Indented strings.
2010-04-12 22:03:27 +00:00
s2 += e->s[j];
if (e->s[j] == '\n') atStartOfLine = true;
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
}
}
/* Remove the last line if it is empty and consists only of
spaces. */
if (n == 1) {
* Fix a few "comparison is always false/true due to limited range of data type" warnings on 64-bit platforms. The one in parser.y is likely to be a real bug.
2009-04-16 12:03:17 +00:00
string::size_type p = s2.find_last_of('\n');
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
if (p != string::npos && s2.find_first_not_of(' ', p + 1) == string::npos)
s2 = string(s2, 0, p + 1);
}
* Indented strings.
2010-04-12 22:03:27 +00:00
* Optimise string constants by putting them in the symbol table.
2010-10-23 21:11:59 +00:00
es2->push_back(new ExprString(symbols.create(s2)));
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
}
Ensure proper type checking/coercion of "${expr}" Now we only rewrite "${expr}" to expr if expr is a string literal.
2013-10-16 22:57:24 +00:00
/* If this is a single string, then don't do a concatenation. */
Show position info in string concatenation / addition errors
2014-04-04 20:19:33 +00:00
return es2->size() == 1 && dynamic_cast<ExprString *>((*es2)[0]) ? (*es2)[0] : new ExprConcatStrings(pos, true, es2);
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
}
Include position info in function application This allows error messages like: error: the anonymous function at `/etc/nixos/configuration.nix:1:1' called without required argument `foo', at `/nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs/lib/modules.nix:77:59'
2014-04-04 15:53:52 +00:00
static inline Pos makeCurPos(const YYLTYPE & loc, ParseData * data)
* When something goes wrong in the evaluation of a Nix expression, print a nice backtrace of the stack, rather than vomiting a gigantic (and useless) aterm on the screen. Example: error: while evaluating file `.../pkgs/system/test.nix': while evaluating attribute `subversion' at `.../pkgs/system/all-packages-generic.nix', line 533: while evaluating function at `.../pkgs/applications/version-management/subversion/default.nix', line 1: assertion failed at `.../pkgs/applications/version-management/subversion/default.nix', line 13 Since the Nix expression language is lazy, the trace may be misleading. The purpose is to provide a hint as to the location of the problem.
2004-04-05 22:27:41 +00:00
{
print LOC for stdin, string args
2020-05-21 04:18:26 +00:00
return Pos(data->origin, data->file, loc.first_line, loc.first_column);
* When something goes wrong in the evaluation of a Nix expression, print a nice backtrace of the stack, rather than vomiting a gigantic (and useless) aterm on the screen. Example: error: while evaluating file `.../pkgs/system/test.nix': while evaluating attribute `subversion' at `.../pkgs/system/all-packages-generic.nix', line 533: while evaluating function at `.../pkgs/applications/version-management/subversion/default.nix', line 1: assertion failed at `.../pkgs/applications/version-management/subversion/default.nix', line 13 Since the Nix expression language is lazy, the trace may be misleading. The purpose is to provide a hint as to the location of the problem.
2004-04-05 22:27:41 +00:00
}
* Store attribute positions in the AST and report duplicate attribute errors with position info. * For all positions, use the position of the first character of the first token, rather than the last character of the first token plus one.
2010-05-06 16:46:48 +00:00
#define CUR_POS makeCurPos(*yylocp, data)
* Override YYMALLOC and YYFREE so that we can call AT[un]protectMemory on the Bison parse stack. Otherwise, a garbage collect during parsing could lead to a crash.
2006-02-13 13:09:23 +00:00
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
}
* Hack for Bison 2.3 compatability.
2006-10-02 14:43:15 +00:00
void yyerror(YYLTYPE * loc, yyscan_t scanner, ParseData * data, const char * error)
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
{
more pos reporting
2020-05-11 23:34:57 +00:00
data->error = ErrorInfo {
.hint = hintfmt(error),
.nixCode = NixCode { .errPos = makeCurPos(*loc, data) }
};
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
}
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
%}
%union {
Fix whitespace
2013-09-02 14:29:15 +00:00
// !!! We're probably leaking stuff here.
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
nix::Expr * e;
nix::ExprList * list;
nix::ExprAttrs * attrs;
nix::Formals * formals;
nix::Formal * formal;
Store Nix integers as longs So on 64-bit systems, integers are now 64-bit. Fixes #158.
2013-08-19 10:35:03 +00:00
nix::NixInt n;
First hit at providing support for floats in the language.
2016-01-04 23:40:40 +00:00
nix::NixFloat nf;
Fix some clang warnings
2014-01-21 17:29:55 +00:00
const char * id; // !!! -> Symbol
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
char * path;
char * uri;
Don't use any syntactic sugar for dynamic attrs This doesn't change any functionality but moves some behavior out of the parser and into the evaluator in order to simplify the code. Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 23:56:26 +00:00
std::vector<nix::AttrName> * attrNames;
* More missing constructs.
2010-04-12 21:21:24 +00:00
std::vector<nix::Expr *> * string_parts;
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
}
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
%type <e> start expr expr_function expr_if expr_op
%type <e> expr_app expr_select expr_simple
%type <list> expr_list
%type <attrs> binds
* Added an experimental feature suggested by Andres: ellipses ("...") in attribute set pattern matches. This allows defining a function that takes *at least* the listed attributes, while ignoring additional attributes. For instance, {stdenv, fetchurl, fuse, ...}: stdenv.mkDerivation { ... }; defines a function that requires an attribute set that contains the specified attributes but ignores others. The main advantage is that we can then write in all-packages.nix aefs = import ../bla/aefs pkgs; instead of aefs = import ../bla/aefs { inherit stdenv fetchurl fuse; }; This saves a lot of typing (not to mention not having to update all-packages.nix with purely mechanical changes). It saves as much typing as the "args: with args;" style, but has the advantage that the function arguments are properly declared (not implicit in what the body of the "with" uses).
2008-08-14 14:00:44 +00:00
%type <formals> formals
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
%type <formal> formal
Don't use any syntactic sugar for dynamic attrs This doesn't change any functionality but moves some behavior out of the parser and into the evaluator in order to simplify the code. Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 23:56:26 +00:00
%type <attrNames> attrs attrpath
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
%type <string_parts> string_parts_interpolated ind_string_parts
%type <e> string_parts string_attr
* Allow a default value in attribute selection by writing x.y.z or default (as originally proposed in https://mail.cs.uu.nl/pipermail/nix-dev/2009-September/002989.html). For instance, an expression like stdenv.lib.attrByPath ["features" "ckSched"] false args can now be written as args.features.ckSched or false
2011-07-13 12:19:57 +00:00
%type <id> attr
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
%token <id> ID ATTRPATH
* More missing constructs.
2010-04-12 21:21:24 +00:00
%token <e> STR IND_STR
* Don't use ATerms to represent integers in the lexer.
2010-04-12 10:38:18 +00:00
%token <n> INT
First hit at providing support for floats in the language.
2016-01-04 23:40:40 +00:00
%token <nf> FLOAT
Allow the leading component of a path to be a ~
2015-02-19 13:05:16 +00:00
%token <path> PATH HPATH SPATH
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
%token <uri> URI
* Allow a default value in attribute selection by writing x.y.z or default (as originally proposed in https://mail.cs.uu.nl/pipermail/nix-dev/2009-September/002989.html). For instance, an expression like stdenv.lib.attrByPath ["features" "ckSched"] false args can now be written as args.features.ckSched or false
2011-07-13 12:19:57 +00:00
%token IF THEN ELSE ASSERT WITH LET IN REC INHERIT EQ NEQ AND OR IMPL OR_KW
* String interpolation. Expressions like "--with-freetype2-library=" + freetype + "/lib" can now be written as "--with-freetype2-library=${freetype}/lib" An arbitrary expression can be enclosed within ${...}, not just identifiers. * Escaping in string literals: \n, \r, \t interpreted as in C, any other character following \ is interpreted as-is. * Newlines are now allowed in string literals.
2006-05-01 14:01:47 +00:00
%token DOLLAR_CURLY /* == ${ */
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
%token IND_STRING_OPEN IND_STRING_CLOSE
* Added an experimental feature suggested by Andres: ellipses ("...") in attribute set pattern matches. This allows defining a function that takes *at least* the listed attributes, while ignoring additional attributes. For instance, {stdenv, fetchurl, fuse, ...}: stdenv.mkDerivation { ... }; defines a function that requires an attribute set that contains the specified attributes but ignores others. The main advantage is that we can then write in all-packages.nix aefs = import ../bla/aefs pkgs; instead of aefs = import ../bla/aefs { inherit stdenv fetchurl fuse; }; This saves a lot of typing (not to mention not having to update all-packages.nix with purely mechanical changes). It saves as much typing as the "args: with args;" style, but has the advantage that the function arguments are properly declared (not implicit in what the body of the "with" uses).
2008-08-14 14:00:44 +00:00
%token ELLIPSIS
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
parser.y: right-associativity of ->
2018-07-23 07:28:48 +00:00
%right IMPL
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
%left OR
%left AND
%nonassoc EQ NEQ
parser.y: fix assoc of -> and < > <= >= The parser allowed senseless `a > b > c` but disallowed `a -> b -> c` which seems valid It might be a typo
2018-07-21 15:24:51 +00:00
%nonassoc '<' '>' LEQ GEQ
* An attribute set update operator (//). E.g., {x=1; y=2; z=3;} // {y=4;} => {x=1; y=4; z=3;}
2004-02-04 16:49:51 +00:00
%right UPDATE
Add a unary integer negation operator This allows saying "-1" instead of "builtins.sub 0 1".
2013-08-02 15:35:59 +00:00
%left NOT
Add integer ‘-’, ‘*’ and ‘/’ operators
2013-08-02 16:03:02 +00:00
%left '+' '-'
%left '*' '/'
* List concatenation must be right-associative for efficiency.
2005-09-14 11:41:59 +00:00
%right CONCAT
* Added an operator `?' to test for attribute existence, e.g., `attrs ? x' yields true iff `attrs' has an attribute named `x'.
2004-03-28 21:15:01 +00:00
%nonassoc '?'
Add a unary integer negation operator This allows saying "-1" instead of "builtins.sub 0 1".
2013-08-02 15:35:59 +00:00
%nonassoc NEGATE
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
%%
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
start: expr { data->result = $1; };
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
expr: expr_function;
expr_function
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
: ID ':' expr_function
* Check for duplicate attribute names / function arguments. `make check' now succeeds :-) * An attribute set such as `{ foo = { enable = true; }; foo.port = 23; }' now parses. It was previously rejected, but I'm too lazy to implement the check. (The only reason to reject it is that the reverse, `{ foo.port = 23; foo = { enable = true; }; }', is rejected, which is kind of ugly.)
2010-04-22 11:02:24 +00:00
{ $$ = new ExprLambda(CUR_POS, data->symbols.create($1), false, 0, $3); }
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
| '{' formals '}' ':' expr_function
* Use a symbol table to represent identifiers and attribute names efficiently. The symbol table ensures that there is only one copy of each symbol, thus allowing symbols to be compared efficiently using a pointer equality test.
2010-04-13 12:25:42 +00:00
{ $$ = new ExprLambda(CUR_POS, data->symbols.create(""), true, $2, $5); }
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
| '{' formals '}' '@' ID ':' expr_function
* Use a symbol table to represent identifiers and attribute names efficiently. The symbol table ensures that there is only one copy of each symbol, thus allowing symbols to be compared efficiently using a pointer equality test.
2010-04-13 12:25:42 +00:00
{ $$ = new ExprLambda(CUR_POS, data->symbols.create($5), true, $2, $7); }
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
| ID '@' '{' formals '}' ':' expr_function
* Use a symbol table to represent identifiers and attribute names efficiently. The symbol table ensures that there is only one copy of each symbol, thus allowing symbols to be compared efficiently using a pointer equality test.
2010-04-13 12:25:42 +00:00
{ $$ = new ExprLambda(CUR_POS, data->symbols.create($1), true, $4, $7); }
* More missing constructs.
2010-04-12 21:21:24 +00:00
| ASSERT expr ';' expr_function
{ $$ = new ExprAssert(CUR_POS, $2, $4); }
* New language feature: with expressions. The expression `with E1; E2' evaluates to E2 with all bindings in the attribute set E1 substituted. E.g., with {x = 123;}; x evaluates to 123. That is, the attribute set E1 is in scope in E2. This is particularly useful when importing files containing lots definitions. E.g., instead of let { inherit (import ./foo.nix) a b c d e f; body = ... a ... f ...; } we can now say with import ./foo.nix; ... a ... f ... I.e., we don't have to say what variables should be brought into scope.
2004-10-25 16:54:56 +00:00
| WITH expr ';' expr_function
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
{ $$ = new ExprWith(CUR_POS, $2, $4); }
* Finally, a real "let" syntax: `let x = ...; ... z = ...; in ...'.
2006-10-02 15:52:44 +00:00
| LET binds IN expr_function
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
{ if (!$2->dynamicAttrs.empty())
new format for pos
2020-05-11 22:58:08 +00:00
throw ParseError(
ErrorInfo {
.hint = hintfmt("dynamic attributes not allowed in let"),
.nixCode = NixCode { .errPos = CUR_POS },
});
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
$$ = new ExprLet($2, $4);
}
* Resolve an ambiguity between ifs and attribute selection, e.g., `if b then x else y.z'.
2004-02-19 13:11:12 +00:00
| expr_if
;
expr_if
Attach pos to if expression errors
2020-04-09 07:45:15 +00:00
: IF expr THEN expr ELSE expr { $$ = new ExprIf(CUR_POS, $2, $4, $6); }
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
| expr_op
;
expr_op
Add a unary integer negation operator This allows saying "-1" instead of "builtins.sub 0 1".
2013-08-02 15:35:59 +00:00
: '!' expr_op %prec NOT { $$ = new ExprOpNot($2); }
Remove ExprBuiltin It's slower than ExprVar since it doesn't compute a static displacement. Since we're not using the throw primop in the implementation of <...> anymore, it's also not really needed.
2014-05-26 15:14:28 +00:00
| '-' expr_op %prec NEGATE { $$ = new ExprApp(CUR_POS, new ExprApp(new ExprVar(data->symbols.create("__sub")), new ExprInt(0)), $2); }
* More missing constructs.
2010-04-12 21:21:24 +00:00
| expr_op EQ expr_op { $$ = new ExprOpEq($1, $3); }
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
| expr_op NEQ expr_op { $$ = new ExprOpNEq($1, $3); }
Remove ExprBuiltin It's slower than ExprVar since it doesn't compute a static displacement. Since we're not using the throw primop in the implementation of <...> anymore, it's also not really needed.
2014-05-26 15:14:28 +00:00
| expr_op '<' expr_op { $$ = new ExprApp(CUR_POS, new ExprApp(new ExprVar(data->symbols.create("__lessThan")), $1), $3); }
| expr_op LEQ expr_op { $$ = new ExprOpNot(new ExprApp(CUR_POS, new ExprApp(new ExprVar(data->symbols.create("__lessThan")), $3), $1)); }
| expr_op '>' expr_op { $$ = new ExprApp(CUR_POS, new ExprApp(new ExprVar(data->symbols.create("__lessThan")), $3), $1); }
| expr_op GEQ expr_op { $$ = new ExprOpNot(new ExprApp(CUR_POS, new ExprApp(new ExprVar(data->symbols.create("__lessThan")), $1), $3)); }
Show position info in Boolean operations
2014-04-04 20:43:52 +00:00
| expr_op AND expr_op { $$ = new ExprOpAnd(CUR_POS, $1, $3); }
| expr_op OR expr_op { $$ = new ExprOpOr(CUR_POS, $1, $3); }
| expr_op IMPL expr_op { $$ = new ExprOpImpl(CUR_POS, $1, $3); }
| expr_op UPDATE expr_op { $$ = new ExprOpUpdate(CUR_POS, $1, $3); }
Don't use any syntactic sugar for dynamic attrs This doesn't change any functionality but moves some behavior out of the parser and into the evaluator in order to simplify the code. Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 23:56:26 +00:00
| expr_op '?' attrpath { $$ = new ExprOpHasAttr($1, *$3); }
* More missing constructs.
2010-04-12 21:21:24 +00:00
| expr_op '+' expr_op
Get rid of some unnecessary ExprConcatStrings nodes in dynamic attrs This gives a ~18% speedup in NixOS evaluation (after converting most calls to hasAttr/getAttr to dynamic attrs).
2014-10-04 23:04:58 +00:00
{ $$ = new ExprConcatStrings(CUR_POS, false, new vector<Expr *>({$1, $3})); }
Remove ExprBuiltin It's slower than ExprVar since it doesn't compute a static displacement. Since we're not using the throw primop in the implementation of <...> anymore, it's also not really needed.
2014-05-26 15:14:28 +00:00
| expr_op '-' expr_op { $$ = new ExprApp(CUR_POS, new ExprApp(new ExprVar(data->symbols.create("__sub")), $1), $3); }
| expr_op '*' expr_op { $$ = new ExprApp(CUR_POS, new ExprApp(new ExprVar(data->symbols.create("__mul")), $1), $3); }
| expr_op '/' expr_op { $$ = new ExprApp(CUR_POS, new ExprApp(new ExprVar(data->symbols.create("__div")), $1), $3); }
Show position info in Boolean operations
2014-04-04 20:43:52 +00:00
| expr_op CONCAT expr_op { $$ = new ExprOpConcatLists(CUR_POS, $1, $3); }
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
| expr_app
;
expr_app
: expr_app expr_select
Include position info in function application This allows error messages like: error: the anonymous function at `/etc/nixos/configuration.nix:1:1' called without required argument `foo', at `/nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs/lib/modules.nix:77:59'
2014-04-04 15:53:52 +00:00
{ $$ = new ExprApp(CUR_POS, $1, $2); }
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
| expr_select { $$ = $1; }
;
expr_select
* Change the right-hand side of the ‘.’ operator from an attribute to an attribute path. This is a refactoring to support default values.
2011-07-06 12:28:57 +00:00
: expr_simple '.' attrpath
Show position info in attribute selection errors
2014-04-04 20:52:14 +00:00
{ $$ = new ExprSelect(CUR_POS, $1, *$3, 0); }
* Allow a default value in attribute selection by writing x.y.z or default (as originally proposed in https://mail.cs.uu.nl/pipermail/nix-dev/2009-September/002989.html). For instance, an expression like stdenv.lib.attrByPath ["features" "ckSched"] false args can now be written as args.features.ckSched or false
2011-07-13 12:19:57 +00:00
| expr_simple '.' attrpath OR_KW expr_select
Show position info in attribute selection errors
2014-04-04 20:52:14 +00:00
{ $$ = new ExprSelect(CUR_POS, $1, *$3, $5); }
* Allow a default value in attribute selection by writing x.y.z or default (as originally proposed in https://mail.cs.uu.nl/pipermail/nix-dev/2009-September/002989.html). For instance, an expression like stdenv.lib.attrByPath ["features" "ckSched"] false args can now be written as args.features.ckSched or false
2011-07-13 12:19:57 +00:00
| /* Backwards compatibility: because Nixpkgs has a rarely used
Revert "Get rid of unicode quotes (#1140)" This reverts commit f78126bfd6b6c8477fcdbc09b2f98772dbe9a1e7. There really is no need for such a massive change...
2016-11-25 23:37:43 +00:00
function named ‘or’, allow stuff like ‘map or [...]’. */
* Allow a default value in attribute selection by writing x.y.z or default (as originally proposed in https://mail.cs.uu.nl/pipermail/nix-dev/2009-September/002989.html). For instance, an expression like stdenv.lib.attrByPath ["features" "ckSched"] false args can now be written as args.features.ckSched or false
2011-07-13 12:19:57 +00:00
expr_simple OR_KW
Include position info in function application This allows error messages like: error: the anonymous function at `/etc/nixos/configuration.nix:1:1' called without required argument `foo', at `/nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs/lib/modules.nix:77:59'
2014-04-04 15:53:52 +00:00
{ $$ = new ExprApp(CUR_POS, $1, new ExprVar(CUR_POS, data->symbols.create("or"))); }
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
| expr_simple { $$ = $1; }
;
expr_simple
Add a symbol __curPos that expands to the current source location I.e. an attribute set { file = <string>; line = <int>; column = <int>; }.
2013-11-18 19:14:54 +00:00
: ID {
if (strcmp($1, "__curPos") == 0)
$$ = new ExprPos(CUR_POS);
else
$$ = new ExprVar(CUR_POS, data->symbols.create($1));
}
* More missing constructs.
2010-04-12 21:21:24 +00:00
| INT { $$ = new ExprInt($1); }
First hit at providing support for floats in the language.
2016-01-04 23:40:40 +00:00
| FLOAT { $$ = new ExprFloat($1); }
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
| '"' string_parts '"' { $$ = $2; }
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
| IND_STRING_OPEN ind_string_parts IND_STRING_CLOSE {
Show position info in string concatenation / addition errors
2014-04-04 20:19:33 +00:00
$$ = stripIndentation(CUR_POS, data->symbols, *$2);
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
}
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
| PATH { $$ = new ExprPath(absPath($1, data->basePath)); }
Figure out the user's home directory if $HOME is not set
2017-05-05 14:40:12 +00:00
| HPATH { $$ = new ExprPath(getHome() + string{$1 + 1}); }
* Add a Nix expression search path feature. Paths between angle brackets, e.g. import <nixpkgs/pkgs/lib> are resolved by looking them up relative to the elements listed in the search path. This allows us to get rid of hacks like import "${builtins.getEnv "NIXPKGS_ALL"}/pkgs/lib" The search path can be specified through the ‘-I’ command-line flag and through the colon-separated ‘NIX_PATH’ environment variable, e.g., $ nix-build -I /etc/nixos ... If a file is not found in the search path, an error message is lazily thrown.
2011-08-06 16:05:24 +00:00
| SPATH {
string path($1 + 1, strlen($1) - 2);
Make the Nix search path declarative Nix search path lookups like <nixpkgs> are now desugared to ‘findFile nixPath <nixpkgs>’, where ‘findFile’ is a new primop. Thus you can override the search path simply by saying let nixPath = [ { prefix = "nixpkgs"; path = "/my-nixpkgs"; } ]; in ... <nixpkgs> ... In conjunction with ‘scopedImport’ (commit c273c15cb13bb86420dda1e5341a4e19517532b5), the Nix search path can be propagated across imports, e.g. let overrides = { nixPath = [ ... ] ++ builtins.nixPath; import = fn: scopedImport overrides fn; scopedImport = attrs: fn: scopedImport (overrides // attrs) fn; builtins = builtins // overrides; }; in scopedImport overrides ./nixos
2014-05-26 15:02:22 +00:00
$$ = new ExprApp(CUR_POS,
new ExprApp(new ExprVar(data->symbols.create("__findFile")),
Rename nixPath to __nixPath The name ‘nixPath’ breaks existing code.
2014-07-30 09:27:34 +00:00
new ExprVar(data->symbols.create("__nixPath"))),
Make the Nix search path declarative Nix search path lookups like <nixpkgs> are now desugared to ‘findFile nixPath <nixpkgs>’, where ‘findFile’ is a new primop. Thus you can override the search path simply by saying let nixPath = [ { prefix = "nixpkgs"; path = "/my-nixpkgs"; } ]; in ... <nixpkgs> ... In conjunction with ‘scopedImport’ (commit c273c15cb13bb86420dda1e5341a4e19517532b5), the Nix search path can be propagated across imports, e.g. let overrides = { nixPath = [ ... ] ++ builtins.nixPath; import = fn: scopedImport overrides fn; scopedImport = attrs: fn: scopedImport (overrides // attrs) fn; builtins = builtins // overrides; }; in scopedImport overrides ./nixos
2014-05-26 15:02:22 +00:00
new ExprString(data->symbols.create(path)));
* Add a Nix expression search path feature. Paths between angle brackets, e.g. import <nixpkgs/pkgs/lib> are resolved by looking them up relative to the elements listed in the search path. This allows us to get rid of hacks like import "${builtins.getEnv "NIXPKGS_ALL"}/pkgs/lib" The search path can be specified through the ‘-I’ command-line flag and through the colon-separated ‘NIX_PATH’ environment variable, e.g., $ nix-build -I /etc/nixos ... If a file is not found in the search path, an error message is lazily thrown.
2011-08-06 16:05:24 +00:00
}
Add feature to disable URL literals E.g. $ nix-build '<nixpkgs>' -A hello --experimental-features no-url-literals error: URL literals are disabled, at /nix/store/vsjamkzh15r3c779q2711az826hqgvzr-nixpkgs-20.03pre194957.bef773ed53f/nixpkgs/pkgs/top-level/all-packages.nix:1236:11 Helps with implementing https://github.com/NixOS/rfcs/pull/45.
2019-11-26 18:48:34 +00:00
| URI {
static bool noURLLiterals = settings.isExperimentalFeatureEnabled("no-url-literals");
if (noURLLiterals)
add pos to errorinfo, remove from hints
2020-05-09 00:18:28 +00:00
throw ParseError(
ErrorInfo {
.hint = hintfmt("URL literals are disabled"),
.nixCode = NixCode { .errPos = CUR_POS }
});
Add feature to disable URL literals E.g. $ nix-build '<nixpkgs>' -A hello --experimental-features no-url-literals error: URL literals are disabled, at /nix/store/vsjamkzh15r3c779q2711az826hqgvzr-nixpkgs-20.03pre194957.bef773ed53f/nixpkgs/pkgs/top-level/all-packages.nix:1236:11 Helps with implementing https://github.com/NixOS/rfcs/pull/45.
2019-11-26 18:48:34 +00:00
$$ = new ExprString(data->symbols.create($1));
}
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
| '(' expr ')' { $$ = $2; }
* Added syntactic sugar to the construction of attribute sets to `inherit' variables from the surrounding lexical scope. E.g., {stdenv, libfoo}: derivation { builder = ./bla; inherit stdenv libfoo; xyzzy = 1; } is equivalent to {stdenv, libfoo}: derivation { builder = ./bla; stdenv = stdenv; libfoo = libfoo; xyzzy = 1; } Note that for mutually recursive attribute set definitions (`rec {...}'), this also works, that is, `rec {inherit x;}' is equivalent to `let {fresh = x; body = rec {x = fresh;};}', *not* `rec {x = x}'.
2004-02-02 21:39:33 +00:00
/* Let expressions `let {..., body = ...}' are just desugared
* More missing constructs.
2010-04-12 21:21:24 +00:00
into `(rec {..., body = ...}).body'. */
* Added syntactic sugar to the construction of attribute sets to `inherit' variables from the surrounding lexical scope. E.g., {stdenv, libfoo}: derivation { builder = ./bla; inherit stdenv libfoo; xyzzy = 1; } is equivalent to {stdenv, libfoo}: derivation { builder = ./bla; stdenv = stdenv; libfoo = libfoo; xyzzy = 1; } Note that for mutually recursive attribute set definitions (`rec {...}'), this also works, that is, `rec {inherit x;}' is equivalent to `let {fresh = x; body = rec {x = fresh;};}', *not* `rec {x = x}'.
2004-02-02 21:39:33 +00:00
| LET '{' binds '}'
Show position info in attribute selection errors
2014-04-04 20:52:14 +00:00
{ $3->recursive = true; $$ = new ExprSelect(noPos, $3, data->symbols.create("body")); }
* Added syntactic sugar to the construction of attribute sets to `inherit' variables from the surrounding lexical scope. E.g., {stdenv, libfoo}: derivation { builder = ./bla; inherit stdenv libfoo; xyzzy = 1; } is equivalent to {stdenv, libfoo}: derivation { builder = ./bla; stdenv = stdenv; libfoo = libfoo; xyzzy = 1; } Note that for mutually recursive attribute set definitions (`rec {...}'), this also works, that is, `rec {inherit x;}' is equivalent to `let {fresh = x; body = rec {x = fresh;};}', *not* `rec {x = x}'.
2004-02-02 21:39:33 +00:00
| REC '{' binds '}'
* Finished the ATerm-less parser.
2010-04-12 23:33:23 +00:00
{ $3->recursive = true; $$ = $3; }
* Added syntactic sugar to the construction of attribute sets to `inherit' variables from the surrounding lexical scope. E.g., {stdenv, libfoo}: derivation { builder = ./bla; inherit stdenv libfoo; xyzzy = 1; } is equivalent to {stdenv, libfoo}: derivation { builder = ./bla; stdenv = stdenv; libfoo = libfoo; xyzzy = 1; } Note that for mutually recursive attribute set definitions (`rec {...}'), this also works, that is, `rec {inherit x;}' is equivalent to `let {fresh = x; body = rec {x = fresh;};}', *not* `rec {x = x}'.
2004-02-02 21:39:33 +00:00
| '{' binds '}'
* Finished the ATerm-less parser.
2010-04-12 23:33:23 +00:00
{ $$ = $2; }
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
| '[' expr_list ']' { $$ = $2; }
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
;
* String interpolation. Expressions like "--with-freetype2-library=" + freetype + "/lib" can now be written as "--with-freetype2-library=${freetype}/lib" An arbitrary expression can be enclosed within ${...}, not just identifiers. * Escaping in string literals: \n, \r, \t interpreted as in C, any other character following \ is interpreted as-is. * Newlines are now allowed in string literals.
2006-05-01 14:01:47 +00:00
string_parts
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
: STR
Show position info in string concatenation / addition errors
2014-04-04 20:19:33 +00:00
| string_parts_interpolated { $$ = new ExprConcatStrings(CUR_POS, true, $1); }
Fix building against Bison 3.0.2
2014-01-21 14:34:04 +00:00
| { $$ = new ExprString(data->symbols.create("")); }
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
;
string_parts_interpolated
: string_parts_interpolated STR { $$ = $1; $1->push_back($2); }
Fix the hack that resets the scanner state.
2015-07-02 16:39:02 +00:00
| string_parts_interpolated DOLLAR_CURLY expr '}' { $$ = $1; $1->push_back($3); }
| DOLLAR_CURLY expr '}' { $$ = new vector<Expr *>; $$->push_back($2); }
| STR DOLLAR_CURLY expr '}' {
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
$$ = new vector<Expr *>;
$$->push_back($1);
$$->push_back($3);
}
* String interpolation. Expressions like "--with-freetype2-library=" + freetype + "/lib" can now be written as "--with-freetype2-library=${freetype}/lib" An arbitrary expression can be enclosed within ${...}, not just identifiers. * Escaping in string literals: \n, \r, \t interpreted as in C, any other character following \ is interpreted as-is. * Newlines are now allowed in string literals.
2006-05-01 14:01:47 +00:00
;
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
ind_string_parts
* Indented strings.
2010-04-12 22:03:27 +00:00
: ind_string_parts IND_STR { $$ = $1; $1->push_back($2); }
Fix the hack that resets the scanner state.
2015-07-02 16:39:02 +00:00
| ind_string_parts DOLLAR_CURLY expr '}' { $$ = $1; $1->push_back($3); }
* Indented strings.
2010-04-12 22:03:27 +00:00
| { $$ = new vector<Expr *>; }
* Added a new kind of multi-line string literal delimited by two single quotes. Example (from NixOS): job = '' start on network-interfaces start script rm -f /var/run/opengl-driver ${if videoDriver == "nvidia" then "ln -sf ${nvidiaDrivers} /var/run/opengl-driver" else if cfg.driSupport then "ln -sf ${mesa} /var/run/opengl-driver" else "" } rm -f /var/log/slim.log end script ''; This style has two big advantages: - \, ' and " aren't special, only '' and ${. So you get a lot less escaping in shell scripts / configuration files in Nixpkgs/NixOS. The delimiter '' is rare in scripts (and can usually be written as ""). ${ is also fairly rare. Other delimiters such as <<...>>, {{...}} and <|...|> were also considered but this one appears to have the fewest drawbacks (thanks Martin). - Indentation is intelligently stripped so that multi-line strings can follow the nesting structure of the containing Nix expression. E.g. in the example above 6 spaces are stripped from the start of each line. This prevents unnecessary indentation in generated files (which sometimes even breaks things). See tests/lang/eval-okay-ind-string.nix for some examples.
2007-11-30 16:48:45 +00:00
;
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
binds
Fold dynamic binds handling into addAttr Since addAttr has to iterate through the AttrPath we pass it, it makes more sense to just iterate through the AttrNames in addAttr instead. As an added bonus, this allows attrsets where two dynamic attribute paths have the same static leading part (see added test case for an example that failed previously). Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 22:57:10 +00:00
: binds attrpath '=' expr ';' { $$ = $1; addAttr($$, *$2, $4, makeCurPos(@2, data)); }
* Allow a default value in attribute selection by writing x.y.z or default (as originally proposed in https://mail.cs.uu.nl/pipermail/nix-dev/2009-September/002989.html). For instance, an expression like stdenv.lib.attrByPath ["features" "ckSched"] false args can now be written as args.features.ckSched or false
2011-07-13 12:19:57 +00:00
| binds INHERIT attrs ';'
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
{ $$ = $1;
OCD: foreach -> C++11 ranged for
2015-07-17 17:24:28 +00:00
for (auto & i : *$3) {
if ($$->attrs.find(i.symbol) != $$->attrs.end())
dupAttr(i.symbol, makeCurPos(@3, data), $$->attrs[i.symbol].pos);
* Store position info for inherited attributes.
2010-05-07 12:43:57 +00:00
Pos pos = makeCurPos(@3, data);
OCD: foreach -> C++11 ranged for
2015-07-17 17:24:28 +00:00
$$->attrs[i.symbol] = ExprAttrs::AttrDef(new ExprVar(CUR_POS, i.symbol), pos, true);
* Check for duplicate attribute names / function arguments. `make check' now succeeds :-) * An attribute set such as `{ foo = { enable = true; }; foo.port = 23; }' now parses. It was previously rejected, but I'm too lazy to implement the check. (The only reason to reject it is that the reverse, `{ foo.port = 23; foo = { enable = true; }; }', is rejected, which is kind of ugly.)
2010-04-22 11:02:24 +00:00
}
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
}
* Allow a default value in attribute selection by writing x.y.z or default (as originally proposed in https://mail.cs.uu.nl/pipermail/nix-dev/2009-September/002989.html). For instance, an expression like stdenv.lib.attrByPath ["features" "ckSched"] false args can now be written as args.features.ckSched or false
2011-07-13 12:19:57 +00:00
| binds INHERIT '(' expr ')' attrs ';'
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
{ $$ = $1;
/* !!! Should ensure sharing of the expression in $4. */
OCD: foreach -> C++11 ranged for
2015-07-17 17:24:28 +00:00
for (auto & i : *$6) {
if ($$->attrs.find(i.symbol) != $$->attrs.end())
dupAttr(i.symbol, makeCurPos(@6, data), $$->attrs[i.symbol].pos);
$$->attrs[i.symbol] = ExprAttrs::AttrDef(new ExprSelect(CUR_POS, $4, i.symbol), makeCurPos(@6, data));
* Allow a default value in attribute selection by writing x.y.z or default (as originally proposed in https://mail.cs.uu.nl/pipermail/nix-dev/2009-September/002989.html). For instance, an expression like stdenv.lib.attrByPath ["features" "ckSched"] false args can now be written as args.features.ckSched or false
2011-07-13 12:19:57 +00:00
}
}
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
| { $$ = new ExprAttrs; }
* Extended the `inherit' syntax to optionally select attributes from other attribute sets, rather than the current scope. E.g., {inherit (pkgs) gcc binutils;} is equivalent to {gcc = pkgs.gcc; binutils = pkgs.binutils;} I am not so happy about the syntax.
2004-02-04 17:23:26 +00:00
;
* Allow a default value in attribute selection by writing x.y.z or default (as originally proposed in https://mail.cs.uu.nl/pipermail/nix-dev/2009-September/002989.html). For instance, an expression like stdenv.lib.attrByPath ["features" "ckSched"] false args can now be written as args.features.ckSched or false
2011-07-13 12:19:57 +00:00
attrs
Don't use any syntactic sugar for dynamic attrs This doesn't change any functionality but moves some behavior out of the parser and into the evaluator in order to simplify the code. Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 23:56:26 +00:00
: attrs attr { $$ = $1; $1->push_back(AttrName(data->symbols.create($2))); }
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
| attrs string_attr
{ $$ = $1;
Get rid of some unnecessary ExprConcatStrings nodes in dynamic attrs This gives a ~18% speedup in NixOS evaluation (after converting most calls to hasAttr/getAttr to dynamic attrs).
2014-10-04 23:04:58 +00:00
ExprString * str = dynamic_cast<ExprString *>($2);
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
if (str) {
Don't use any syntactic sugar for dynamic attrs This doesn't change any functionality but moves some behavior out of the parser and into the evaluator in order to simplify the code. Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 23:56:26 +00:00
$$->push_back(AttrName(str->s));
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
delete str;
} else
new format for pos
2020-05-11 22:58:08 +00:00
throw ParseError(
ErrorInfo {
.hint = hintfmt("dynamic attributes not allowed in inherit"),
.nixCode = NixCode { .errPos = makeCurPos(@2, data) },
});
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
}
Don't use any syntactic sugar for dynamic attrs This doesn't change any functionality but moves some behavior out of the parser and into the evaluator in order to simplify the code. Signed-off-by: Shea Levy <shea@shealevy.com>
2013-12-31 23:56:26 +00:00
| { $$ = new AttrPath; }
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
;
* Some syntactic sugar for attribute sets: allow {x.y.z = ...;} as a shorthand for {x = {y = {z = ...;};};}. This is especially useful for NixOS configuration files, e.g. { services = { sshd = { enable = true; port = 2022; }; }; } can now be written as { services.sshd.enable = true; services.sshd.port = 2022; } However, it is currently not permitted to write { services.sshd = {enable = true;}; services.sshd.port = 2022; } as this is considered a duplicate definition of `services.sshd'.
2009-05-15 12:35:23 +00:00
attrpath
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
: attrpath '.' attr { $$ = $1; $1->push_back(AttrName(data->symbols.create($3))); }
| attrpath '.' string_attr
{ $$ = $1;
Get rid of some unnecessary ExprConcatStrings nodes in dynamic attrs This gives a ~18% speedup in NixOS evaluation (after converting most calls to hasAttr/getAttr to dynamic attrs).
2014-10-04 23:04:58 +00:00
ExprString * str = dynamic_cast<ExprString *>($3);
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
if (str) {
$$->push_back(AttrName(str->s));
delete str;
} else
Get rid of some unnecessary ExprConcatStrings nodes in dynamic attrs This gives a ~18% speedup in NixOS evaluation (after converting most calls to hasAttr/getAttr to dynamic attrs).
2014-10-04 23:04:58 +00:00
$$->push_back(AttrName($3));
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
}
| attr { $$ = new vector<AttrName>; $$->push_back(AttrName(data->symbols.create($1))); }
| string_attr
{ $$ = new vector<AttrName>;
ExprString *str = dynamic_cast<ExprString *>($1);
if (str) {
$$->push_back(AttrName(str->s));
delete str;
} else
Get rid of some unnecessary ExprConcatStrings nodes in dynamic attrs This gives a ~18% speedup in NixOS evaluation (after converting most calls to hasAttr/getAttr to dynamic attrs).
2014-10-04 23:04:58 +00:00
$$->push_back(AttrName($1));
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
}
* Allow a default value in attribute selection by writing x.y.z or default (as originally proposed in https://mail.cs.uu.nl/pipermail/nix-dev/2009-September/002989.html). For instance, an expression like stdenv.lib.attrByPath ["features" "ckSched"] false args can now be written as args.features.ckSched or false
2011-07-13 12:19:57 +00:00
;
attr
: ID { $$ = $1; }
| OR_KW { $$ = "or"; }
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
;
string_attr
: '"' string_parts '"' { $$ = $2; }
Get rid of some unnecessary ExprConcatStrings nodes in dynamic attrs This gives a ~18% speedup in NixOS evaluation (after converting most calls to hasAttr/getAttr to dynamic attrs).
2014-10-04 23:04:58 +00:00
| DOLLAR_CURLY expr '}' { $$ = $2; }
* Some syntactic sugar for attribute sets: allow {x.y.z = ...;} as a shorthand for {x = {y = {z = ...;};};}. This is especially useful for NixOS configuration files, e.g. { services = { sshd = { enable = true; port = 2022; }; }; } can now be written as { services.sshd.enable = true; services.sshd.port = 2022; } However, it is currently not permitted to write { services.sshd = {enable = true;}; services.sshd.port = 2022; } as this is considered a duplicate definition of `services.sshd'.
2009-05-15 12:35:23 +00:00
;
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
expr_list
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
: expr_list expr_select { $$ = $1; $1->elems.push_back($2); /* !!! dangerous */ }
| { $$ = new ExprList; }
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
;
formals
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
: formal ',' formals
* Check for duplicate attribute names / function arguments. `make check' now succeeds :-) * An attribute set such as `{ foo = { enable = true; }; foo.port = 23; }' now parses. It was previously rejected, but I'm too lazy to implement the check. (The only reason to reject it is that the reverse, `{ foo.port = 23; foo = { enable = true; }; }', is rejected, which is kind of ugly.)
2010-04-22 11:02:24 +00:00
{ $$ = $3; addFormal(CUR_POS, $$, *$1); }
* Added an experimental feature suggested by Andres: ellipses ("...") in attribute set pattern matches. This allows defining a function that takes *at least* the listed attributes, while ignoring additional attributes. For instance, {stdenv, fetchurl, fuse, ...}: stdenv.mkDerivation { ... }; defines a function that requires an attribute set that contains the specified attributes but ignores others. The main advantage is that we can then write in all-packages.nix aefs = import ../bla/aefs pkgs; instead of aefs = import ../bla/aefs { inherit stdenv fetchurl fuse; }; This saves a lot of typing (not to mention not having to update all-packages.nix with purely mechanical changes). It saves as much typing as the "args: with args;" style, but has the advantage that the function arguments are properly declared (not implicit in what the body of the "with" uses).
2008-08-14 14:00:44 +00:00
| formal
* Check for duplicate attribute names / function arguments. `make check' now succeeds :-) * An attribute set such as `{ foo = { enable = true; }; foo.port = 23; }' now parses. It was previously rejected, but I'm too lazy to implement the check. (The only reason to reject it is that the reverse, `{ foo.port = 23; foo = { enable = true; }; }', is rejected, which is kind of ugly.)
2010-04-22 11:02:24 +00:00
{ $$ = new Formals; addFormal(CUR_POS, $$, *$1); $$->ellipsis = false; }
* Added an experimental feature suggested by Andres: ellipses ("...") in attribute set pattern matches. This allows defining a function that takes *at least* the listed attributes, while ignoring additional attributes. For instance, {stdenv, fetchurl, fuse, ...}: stdenv.mkDerivation { ... }; defines a function that requires an attribute set that contains the specified attributes but ignores others. The main advantage is that we can then write in all-packages.nix aefs = import ../bla/aefs pkgs; instead of aefs = import ../bla/aefs { inherit stdenv fetchurl fuse; }; This saves a lot of typing (not to mention not having to update all-packages.nix with purely mechanical changes). It saves as much typing as the "args: with args;" style, but has the advantage that the function arguments are properly declared (not implicit in what the body of the "with" uses).
2008-08-14 14:00:44 +00:00
|
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
{ $$ = new Formals; $$->ellipsis = false; }
* Added an experimental feature suggested by Andres: ellipses ("...") in attribute set pattern matches. This allows defining a function that takes *at least* the listed attributes, while ignoring additional attributes. For instance, {stdenv, fetchurl, fuse, ...}: stdenv.mkDerivation { ... }; defines a function that requires an attribute set that contains the specified attributes but ignores others. The main advantage is that we can then write in all-packages.nix aefs = import ../bla/aefs pkgs; instead of aefs = import ../bla/aefs { inherit stdenv fetchurl fuse; }; This saves a lot of typing (not to mention not having to update all-packages.nix with purely mechanical changes). It saves as much typing as the "args: with args;" style, but has the advantage that the function arguments are properly declared (not implicit in what the body of the "with" uses).
2008-08-14 14:00:44 +00:00
| ELLIPSIS
* Don't use ATerms for the abstract syntax trees anymore. Not finished yet.
2010-04-12 18:30:11 +00:00
{ $$ = new Formals; $$->ellipsis = true; }
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
;
formal
Make function arguments retain position info This allows querying the location of function arguments. E.g. builtins.unsafeGetAttrPos "x" (builtins.functionArgs ({ x }: null)) => { column = 57; file = "/home/infinisil/src/nix/inst/test.nix"; line = 1; }
2020-04-02 03:03:58 +00:00
: ID { $$ = new Formal(CUR_POS, data->symbols.create($1), 0); }
| ID '?' expr { $$ = new Formal(CUR_POS, data->symbols.create($1), $3); }
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
;
Fix whitespace
2013-09-02 14:29:15 +00:00
* Replaced the SDF parser by a substantially faster Bison/Flex parser (roughly 80x faster). The absolutely latest version of Bison (1.875c) is required for reentrant GLR support, as well as a recent version of Flex (say, 2.5.31). Note that most Unix distributions ship with the prehistoric Flex 2.5.4, which doesn't support reentrancy.
2004-01-30 15:21:42 +00:00
%%
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <unistd.h>
Support Git repos in the Nix path E.g. $ nix-build -I nixpkgs=git://github.com/NixOS/nixpkgs '<nixpkgs>' -A hello This is not extremely useful yet because you can't specify a branch/revision.
2016-04-29 19:04:40 +00:00
#include "eval.hh"
datatransfer.{cc,hh} -> filetransfer.{cc,hh}
2020-04-06 21:57:28 +00:00
#include "filetransfer.hh"
Backport libfetchers from the flakes branch This provides a pluggable mechanism for defining new fetchers. It adds a builtin function 'fetchTree' that generalizes existing fetchers like 'fetchGit', 'fetchMercurial' and 'fetchTarball'. 'fetchTree' takes a set of attributes, e.g. fetchTree { type = "git"; url = "https://example.org/repo.git"; ref = "some-branch"; rev = "abcdef..."; } The existing fetchers are just wrappers around this. Note that the input attributes to fetchTree are the same as flake input specifications and flake lock file entries. All fetchers share a common cache stored in ~/.cache/nix/fetcher-cache-v1.sqlite. This replaces the ad hoc caching mechanisms in fetchGit and download.cc (e.g. ~/.cache/nix/{tarballs,git-revs*}). This also adds support for Git worktrees (c169ea59049f861aaba429f48b828d0820b74d1d).
2020-03-30 14:04:18 +00:00
#include "fetchers.hh"
Support Git repos in the Nix path E.g. $ nix-build -I nixpkgs=git://github.com/NixOS/nixpkgs '<nixpkgs>' -A hello This is not extremely useful yet because you can't specify a branch/revision.
2016-04-29 19:04:40 +00:00
#include "store-api.hh"
* Use a symbol table to represent identifiers and attribute names efficiently. The symbol table ensures that there is only one copy of each symbol, thus allowing symbols to be compared efficiently using a pointer equality test.
2010-04-13 12:25:42 +00:00
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
namespace nix {
Fix whitespace
2013-09-02 14:29:15 +00:00
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
print LOC for stdin, string args
2020-05-21 04:18:26 +00:00
Expr * EvalState::parse(const char * text, FileOrigin origin,
Add some support code for nix-repl
2013-09-02 16:34:04 +00:00
const Path & path, const Path & basePath, StaticEnv & staticEnv)
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
{
yyscan_t scanner;
* Add a Nix expression search path feature. Paths between angle brackets, e.g. import <nixpkgs/pkgs/lib> are resolved by looking them up relative to the elements listed in the search path. This allows us to get rid of hacks like import "${builtins.getEnv "NIXPKGS_ALL"}/pkgs/lib" The search path can be specified through the ‘-I’ command-line flag and through the colon-separated ‘NIX_PATH’ environment variable, e.g., $ nix-build -I /etc/nixos ... If a file is not found in the search path, an error message is lazily thrown.
2011-08-06 16:05:24 +00:00
ParseData data(*this);
print LOC for stdin, string args
2020-05-21 04:18:26 +00:00
data.origin = origin;
switch (origin) {
case foFile:
data.file = data.symbols.create(path);
break;
case foStdin:
case foString:
data.file = data.symbols.create(text);
break;
default:
throw Error("invalid FileOrigin in parse");
}
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
data.basePath = basePath;
yylex_init(&scanner);
yy_scan_string(text, scanner);
int res = yyparse(scanner, &data);
yylex_destroy(scanner);
Fix whitespace
2013-09-02 14:29:15 +00:00
Revert "Throw a specific error for incomplete parse errors." This reverts commit 6498adb002bcf7e715afe46c23b8635d4592c156. We don't actually use IncompleteParseError in 'nix repl'.
2018-05-11 09:39:25 +00:00
if (res) throw ParseError(data.error);
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
Show the exact position of undefined variables In particular, undefined variable errors in a "with" previously didn't show *any* position information, so this should help a lot in those cases.
2013-10-08 12:40:51 +00:00
data.result->bindVars(staticEnv);
Fix whitespace
2013-09-02 14:29:15 +00:00
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
return data.result;
}
Get rid of the parse tree cache Since we already cache files in normal form (fileEvalCache), caching parse trees is redundant. Note that getting rid of this cache doesn't actually save much memory at the moment, because parse trees are currently not freed / GC'ed.
2013-09-03 10:56:33 +00:00
Path resolveExprPath(Path path)
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
{
assert(path[0] == '/');
bugfix: Adding depth limit to resolveExprPath There is no termination condition for evaluation of cyclical expression paths which can lead to infinite loops. This addresses one spot in the parser in a similar fashion as utils.cc/canonPath does. This issue can be reproduced by something like: ``` ln -s a b ln -s b a nix-instantiate -E 'import ./a' ```
2019-12-13 13:44:14 +00:00
unsigned int followCount = 0, maxFollow = 1024;
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
/* If `path' is a symlink, follow it. This is so that relative
path references work. */
struct stat st;
* Handle multiple indirect symlinks when loading a Nix expression.
2007-01-15 14:50:25 +00:00
while (true) {
bugfix: Adding depth limit to resolveExprPath There is no termination condition for evaluation of cyclical expression paths which can lead to infinite loops. This addresses one spot in the parser in a similar fashion as utils.cc/canonPath does. This issue can be reproduced by something like: ``` ln -s a b ln -s b a nix-instantiate -E 'import ./a' ```
2019-12-13 13:44:14 +00:00
// Basic cycle/depth limit to avoid infinite loops.
if (++followCount >= maxFollow)
Tweak error message
2019-12-15 09:44:53 +00:00
throw Error("too many symbolic links encountered while traversing the path '%s'", path);
* Handle multiple indirect symlinks when loading a Nix expression.
2007-01-15 14:50:25 +00:00
if (lstat(path.c_str(), &st))
Tweak error message
2019-12-15 09:44:53 +00:00
throw SysError("getting status of '%s'", path);
* Handle multiple indirect symlinks when loading a Nix expression.
2007-01-15 14:50:25 +00:00
if (!S_ISLNK(st.st_mode)) break;
path = absPath(readLink(path), dirOf(path));
}
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
/* If `path' refers to a directory, append `/default.nix'. */
if (S_ISDIR(st.st_mode))
path = canonPath(path + "/default.nix");
Get rid of the parse tree cache Since we already cache files in normal form (fileEvalCache), caching parse trees is redundant. Note that getting rid of this cache doesn't actually save much memory at the moment, because parse trees are currently not freed / GC'ed.
2013-09-03 10:56:33 +00:00
return path;
}
* Refactoring: move parseExprFromFile() and parseExprFromString() into the EvalState class.
2011-08-06 13:02:55 +00:00
Get rid of the parse tree cache Since we already cache files in normal form (fileEvalCache), caching parse trees is redundant. Note that getting rid of this cache doesn't actually save much memory at the moment, because parse trees are currently not freed / GC'ed.
2013-09-03 10:56:33 +00:00
Expr * EvalState::parseExprFromFile(const Path & path)
{
Add primop ‘scopedImport’ ‘scopedImport’ works like ‘import’, except that it takes a set of attributes to be added to the lexical scope of the expression, essentially extending or overriding the builtin variables. For instance, the expression scopedImport { x = 1; } ./foo.nix where foo.nix contains ‘x’, will evaluate to 1. This has a few applications: * It allows getting rid of function argument specifications in package expressions. For instance, a package expression like: { stdenv, fetchurl, libfoo }: stdenv.mkDerivation { ... buildInputs = [ libfoo ]; } can now we written as just stdenv.mkDerivation { ... buildInputs = [ libfoo ]; } and imported in all-packages.nix as: bar = scopedImport pkgs ./bar.nix; So whereas we once had dependencies listed in three places (buildInputs, the function, and the call site), they now only need to appear in one place. * It allows overriding builtin functions. For instance, to trace all calls to ‘map’: let overrides = { map = f: xs: builtins.trace "map called!" (map f xs); # Ensure that our override gets propagated by calls to # import/scopedImport. import = fn: scopedImport overrides fn; scopedImport = attrs: fn: scopedImport (overrides // attrs) fn; # Also update ‘builtins’. builtins = builtins // overrides; }; in scopedImport overrides ./bla.nix * Similarly, it allows extending the set of builtin functions. For instance, during Nixpkgs/NixOS evaluation, the Nixpkgs library functions could be added to the default scope. There is a downside: calls to scopedImport are not memoized, unlike import. So importing a file multiple times leads to multiple parsings / evaluations. It would be possible to construct the AST only once, but that would require careful handling of variables/environments.
2014-05-26 11:46:11 +00:00
return parseExprFromFile(path, staticBaseEnv);
}
Expr * EvalState::parseExprFromFile(const Path & path, StaticEnv & staticEnv)
{
print LOC for stdin, string args
2020-05-21 04:18:26 +00:00
return parse(readFile(path).c_str(), foFile, path, dirOf(path), staticEnv);
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
}
parseExprFromString(): Use std::string_view
2020-03-11 15:36:36 +00:00
Expr * EvalState::parseExprFromString(std::string_view s, const Path & basePath, StaticEnv & staticEnv)
Add some support code for nix-repl
2013-09-02 16:34:04 +00:00
{
print LOC for stdin, string args
2020-05-21 04:18:26 +00:00
return parse(s.data(), foString, "", basePath, staticEnv);
Add some support code for nix-repl
2013-09-02 16:34:04 +00:00
}
parseExprFromString(): Use std::string_view
2020-03-11 15:36:36 +00:00
Expr * EvalState::parseExprFromString(std::string_view s, const Path & basePath)
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
{
Add some support code for nix-repl
2013-09-02 16:34:04 +00:00
return parseExprFromString(s, basePath, staticBaseEnv);
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
}
* Add a Nix expression search path feature. Paths between angle brackets, e.g. import <nixpkgs/pkgs/lib> are resolved by looking them up relative to the elements listed in the search path. This allows us to get rid of hacks like import "${builtins.getEnv "NIXPKGS_ALL"}/pkgs/lib" The search path can be specified through the ‘-I’ command-line flag and through the colon-separated ‘NIX_PATH’ environment variable, e.g., $ nix-build -I /etc/nixos ... If a file is not found in the search path, an error message is lazily thrown.
2011-08-06 16:05:24 +00:00
nix-build/nix-shell: Eliminate call to nix-instantiate / nix-store Note that this removes the need for a derivation symlink, so the --drv-path and --add-drv-link flags now do nothing.
2017-07-25 13:09:06 +00:00
Expr * EvalState::parseStdin()
{
//Activity act(*logger, lvlTalkative, format("parsing standard input"));
print LOC for stdin, string args
2020-05-21 04:18:26 +00:00
// return parseExprFromString(foStdin, drainFD(0), absPath("."));
return parse(drainFD(0).data(), foStdin, "", absPath("."), staticBaseEnv);
nix-build/nix-shell: Eliminate call to nix-instantiate / nix-store Note that this removes the need for a derivation symlink, so the --drv-path and --add-drv-link flags now do nothing.
2017-07-25 13:09:06 +00:00
}
Make the search path lazier with non-fatal errors Thus, -I / $NIX_PATH entries are now downloaded only when they are needed for evaluation. An error to download an entry is a non-fatal warning (just like non-existant paths). This does change the semantics of builtins.nixPath, which now returns the original, rather than resulting path. E.g., before we had [ { path = "/nix/store/hgm3yxf1lrrwa3z14zpqaj5p9vs0qklk-nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] but now [ { path = "https://nixos.org/channels/nixos-16.03/nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] Fixes #792.
2016-04-14 13:32:24 +00:00
void EvalState::addToSearchPath(const string & s)
* Add a Nix expression search path feature. Paths between angle brackets, e.g. import <nixpkgs/pkgs/lib> are resolved by looking them up relative to the elements listed in the search path. This allows us to get rid of hacks like import "${builtins.getEnv "NIXPKGS_ALL"}/pkgs/lib" The search path can be specified through the ‘-I’ command-line flag and through the colon-separated ‘NIX_PATH’ environment variable, e.g., $ nix-build -I /etc/nixos ... If a file is not found in the search path, an error message is lazily thrown.
2011-08-06 16:05:24 +00:00
{
* Allow redirections in search path entries. E.g. if you have a directory /home/eelco/src/stdenv-updates that you want to use as the directory for import such as with (import <nixpkgs> { }); then you can say $ nix-build -I nixpkgs=/home/eelco/src/stdenv-updates
2011-08-06 17:48:57 +00:00
size_t pos = s.find('=');
string prefix;
Path path;
if (pos == string::npos) {
path = s;
} else {
prefix = string(s, 0, pos);
path = string(s, pos + 1);
}
Fix whitespace
2013-09-02 14:29:15 +00:00
Make the search path lazier with non-fatal errors Thus, -I / $NIX_PATH entries are now downloaded only when they are needed for evaluation. An error to download an entry is a non-fatal warning (just like non-existant paths). This does change the semantics of builtins.nixPath, which now returns the original, rather than resulting path. E.g., before we had [ { path = "/nix/store/hgm3yxf1lrrwa3z14zpqaj5p9vs0qklk-nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] but now [ { path = "https://nixos.org/channels/nixos-16.03/nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] Fixes #792.
2016-04-14 13:32:24 +00:00
searchPath.emplace_back(prefix, path);
* Add a Nix expression search path feature. Paths between angle brackets, e.g. import <nixpkgs/pkgs/lib> are resolved by looking them up relative to the elements listed in the search path. This allows us to get rid of hacks like import "${builtins.getEnv "NIXPKGS_ALL"}/pkgs/lib" The search path can be specified through the ‘-I’ command-line flag and through the colon-separated ‘NIX_PATH’ environment variable, e.g., $ nix-build -I /etc/nixos ... If a file is not found in the search path, an error message is lazily thrown.
2011-08-06 16:05:24 +00:00
}
Path EvalState::findFile(const string & path)
Make the Nix search path declarative Nix search path lookups like <nixpkgs> are now desugared to ‘findFile nixPath <nixpkgs>’, where ‘findFile’ is a new primop. Thus you can override the search path simply by saying let nixPath = [ { prefix = "nixpkgs"; path = "/my-nixpkgs"; } ]; in ... <nixpkgs> ... In conjunction with ‘scopedImport’ (commit c273c15cb13bb86420dda1e5341a4e19517532b5), the Nix search path can be propagated across imports, e.g. let overrides = { nixPath = [ ... ] ++ builtins.nixPath; import = fn: scopedImport overrides fn; scopedImport = attrs: fn: scopedImport (overrides // attrs) fn; builtins = builtins // overrides; }; in scopedImport overrides ./nixos
2014-05-26 15:02:22 +00:00
{
return findFile(searchPath, path);
}
Show position info for failing <...> lookups
2015-01-07 12:43:55 +00:00
Path EvalState::findFile(SearchPath & searchPath, const string & path, const Pos & pos)
* Add a Nix expression search path feature. Paths between angle brackets, e.g. import <nixpkgs/pkgs/lib> are resolved by looking them up relative to the elements listed in the search path. This allows us to get rid of hacks like import "${builtins.getEnv "NIXPKGS_ALL"}/pkgs/lib" The search path can be specified through the ‘-I’ command-line flag and through the colon-separated ‘NIX_PATH’ environment variable, e.g., $ nix-build -I /etc/nixos ... If a file is not found in the search path, an error message is lazily thrown.
2011-08-06 16:05:24 +00:00
{
Allow URLs in the Nix search path E.g. to install "hello" from the latest Nixpkgs: $ nix-build '<nixpkgs>' -A hello -I nixpkgs=https://nixos.org/channels/nixpkgs-unstable/nixexprs.tar.xz Or to install a specific version of NixOS: $ nixos-rebuild switch -I nixpkgs=https://github.com/NixOS/nixpkgs/archive/63def04891a0abc328b1b0b3a78ec02c58f48583.tar.gz
2015-05-05 15:09:42 +00:00
for (auto & i : searchPath) {
Make the search path lazier with non-fatal errors Thus, -I / $NIX_PATH entries are now downloaded only when they are needed for evaluation. An error to download an entry is a non-fatal warning (just like non-existant paths). This does change the semantics of builtins.nixPath, which now returns the original, rather than resulting path. E.g., before we had [ { path = "/nix/store/hgm3yxf1lrrwa3z14zpqaj5p9vs0qklk-nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] but now [ { path = "https://nixos.org/channels/nixos-16.03/nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] Fixes #792.
2016-04-14 13:32:24 +00:00
std::string suffix;
Allow URLs in the Nix search path E.g. to install "hello" from the latest Nixpkgs: $ nix-build '<nixpkgs>' -A hello -I nixpkgs=https://nixos.org/channels/nixpkgs-unstable/nixexprs.tar.xz Or to install a specific version of NixOS: $ nixos-rebuild switch -I nixpkgs=https://github.com/NixOS/nixpkgs/archive/63def04891a0abc328b1b0b3a78ec02c58f48583.tar.gz
2015-05-05 15:09:42 +00:00
if (i.first.empty())
Make the search path lazier with non-fatal errors Thus, -I / $NIX_PATH entries are now downloaded only when they are needed for evaluation. An error to download an entry is a non-fatal warning (just like non-existant paths). This does change the semantics of builtins.nixPath, which now returns the original, rather than resulting path. E.g., before we had [ { path = "/nix/store/hgm3yxf1lrrwa3z14zpqaj5p9vs0qklk-nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] but now [ { path = "https://nixos.org/channels/nixos-16.03/nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] Fixes #792.
2016-04-14 13:32:24 +00:00
suffix = "/" + path;
* Allow redirections in search path entries. E.g. if you have a directory /home/eelco/src/stdenv-updates that you want to use as the directory for import such as with (import <nixpkgs> { }); then you can say $ nix-build -I nixpkgs=/home/eelco/src/stdenv-updates
2011-08-06 17:48:57 +00:00
else {
Make the search path lazier with non-fatal errors Thus, -I / $NIX_PATH entries are now downloaded only when they are needed for evaluation. An error to download an entry is a non-fatal warning (just like non-existant paths). This does change the semantics of builtins.nixPath, which now returns the original, rather than resulting path. E.g., before we had [ { path = "/nix/store/hgm3yxf1lrrwa3z14zpqaj5p9vs0qklk-nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] but now [ { path = "https://nixos.org/channels/nixos-16.03/nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] Fixes #792.
2016-04-14 13:32:24 +00:00
auto s = i.first.size();
if (path.compare(0, s, i.first) != 0 ||
(path.size() > s && path[s] != '/'))
* Allow redirections in search path entries. E.g. if you have a directory /home/eelco/src/stdenv-updates that you want to use as the directory for import such as with (import <nixpkgs> { }); then you can say $ nix-build -I nixpkgs=/home/eelco/src/stdenv-updates
2011-08-06 17:48:57 +00:00
continue;
Make the search path lazier with non-fatal errors Thus, -I / $NIX_PATH entries are now downloaded only when they are needed for evaluation. An error to download an entry is a non-fatal warning (just like non-existant paths). This does change the semantics of builtins.nixPath, which now returns the original, rather than resulting path. E.g., before we had [ { path = "/nix/store/hgm3yxf1lrrwa3z14zpqaj5p9vs0qklk-nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] but now [ { path = "https://nixos.org/channels/nixos-16.03/nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] Fixes #792.
2016-04-14 13:32:24 +00:00
suffix = path.size() == s ? "" : "/" + string(path, s);
* Allow redirections in search path entries. E.g. if you have a directory /home/eelco/src/stdenv-updates that you want to use as the directory for import such as with (import <nixpkgs> { }); then you can say $ nix-build -I nixpkgs=/home/eelco/src/stdenv-updates
2011-08-06 17:48:57 +00:00
}
Make the search path lazier with non-fatal errors Thus, -I / $NIX_PATH entries are now downloaded only when they are needed for evaluation. An error to download an entry is a non-fatal warning (just like non-existant paths). This does change the semantics of builtins.nixPath, which now returns the original, rather than resulting path. E.g., before we had [ { path = "/nix/store/hgm3yxf1lrrwa3z14zpqaj5p9vs0qklk-nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] but now [ { path = "https://nixos.org/channels/nixos-16.03/nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] Fixes #792.
2016-04-14 13:32:24 +00:00
auto r = resolveSearchPathElem(i);
if (!r.first) continue;
Path res = r.second + suffix;
* Add a Nix expression search path feature. Paths between angle brackets, e.g. import <nixpkgs/pkgs/lib> are resolved by looking them up relative to the elements listed in the search path. This allows us to get rid of hacks like import "${builtins.getEnv "NIXPKGS_ALL"}/pkgs/lib" The search path can be specified through the ‘-I’ command-line flag and through the colon-separated ‘NIX_PATH’ environment variable, e.g., $ nix-build -I /etc/nixos ... If a file is not found in the search path, an error message is lazily thrown.
2011-08-06 16:05:24 +00:00
if (pathExists(res)) return canonPath(res);
}
add pos to errorinfo, remove from hints
2020-05-09 00:18:28 +00:00
throw ThrownError(
ErrorInfo {
.hint = hintfmt("file '%1%' was not found in the Nix search path (add it using $NIX_PATH or -I)", path),
.nixCode = NixCode { .errPos = pos }
});
* Add a Nix expression search path feature. Paths between angle brackets, e.g. import <nixpkgs/pkgs/lib> are resolved by looking them up relative to the elements listed in the search path. This allows us to get rid of hacks like import "${builtins.getEnv "NIXPKGS_ALL"}/pkgs/lib" The search path can be specified through the ‘-I’ command-line flag and through the colon-separated ‘NIX_PATH’ environment variable, e.g., $ nix-build -I /etc/nixos ... If a file is not found in the search path, an error message is lazily thrown.
2011-08-06 16:05:24 +00:00
}
Make the search path lazier with non-fatal errors Thus, -I / $NIX_PATH entries are now downloaded only when they are needed for evaluation. An error to download an entry is a non-fatal warning (just like non-existant paths). This does change the semantics of builtins.nixPath, which now returns the original, rather than resulting path. E.g., before we had [ { path = "/nix/store/hgm3yxf1lrrwa3z14zpqaj5p9vs0qklk-nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] but now [ { path = "https://nixos.org/channels/nixos-16.03/nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] Fixes #792.
2016-04-14 13:32:24 +00:00
std::pair<bool, std::string> EvalState::resolveSearchPathElem(const SearchPathElem & elem)
{
auto i = searchPathResolved.find(elem.second);
if (i != searchPathResolved.end()) return i->second;
std::pair<bool, std::string> res;
if (isUri(elem.second)) {
try {
Backport libfetchers from the flakes branch This provides a pluggable mechanism for defining new fetchers. It adds a builtin function 'fetchTree' that generalizes existing fetchers like 'fetchGit', 'fetchMercurial' and 'fetchTarball'. 'fetchTree' takes a set of attributes, e.g. fetchTree { type = "git"; url = "https://example.org/repo.git"; ref = "some-branch"; rev = "abcdef..."; } The existing fetchers are just wrappers around this. Note that the input attributes to fetchTree are the same as flake input specifications and flake lock file entries. All fetchers share a common cache stored in ~/.cache/nix/fetcher-cache-v1.sqlite. This replaces the ad hoc caching mechanisms in fetchGit and download.cc (e.g. ~/.cache/nix/{tarballs,git-revs*}). This also adds support for Git worktrees (c169ea59049f861aaba429f48b828d0820b74d1d).
2020-03-30 14:04:18 +00:00
res = { true, store->toRealPath(fetchers::downloadTarball(
store, resolveUri(elem.second), "source", false).storePath) };
after flake rebase
2020-04-08 12:12:22 +00:00
} catch (FileTransferError & e) {
convert to logWarning format
2020-05-12 18:19:34 +00:00
logWarning(
ErrorInfo {
.name = "Entry download",
.hint = hintfmt("Nix search path entry '%1%' cannot be downloaded, ignoring", elem.second)
});
Make the search path lazier with non-fatal errors Thus, -I / $NIX_PATH entries are now downloaded only when they are needed for evaluation. An error to download an entry is a non-fatal warning (just like non-existant paths). This does change the semantics of builtins.nixPath, which now returns the original, rather than resulting path. E.g., before we had [ { path = "/nix/store/hgm3yxf1lrrwa3z14zpqaj5p9vs0qklk-nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] but now [ { path = "https://nixos.org/channels/nixos-16.03/nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] Fixes #792.
2016-04-14 13:32:24 +00:00
res = { false, "" };
}
} else {
auto path = absPath(elem.second);
if (pathExists(path))
res = { true, path };
else {
convert some printError calls to logError
2020-05-03 14:01:25 +00:00
logWarning(
ErrorInfo {
convert to logWarning format
2020-05-12 18:19:34 +00:00
.name = "Entry not found",
convert some printError calls to logError
2020-05-03 14:01:25 +00:00
.hint = hintfmt("warning: Nix search path entry '%1%' does not exist, ignoring", elem.second)
});
Make the search path lazier with non-fatal errors Thus, -I / $NIX_PATH entries are now downloaded only when they are needed for evaluation. An error to download an entry is a non-fatal warning (just like non-existant paths). This does change the semantics of builtins.nixPath, which now returns the original, rather than resulting path. E.g., before we had [ { path = "/nix/store/hgm3yxf1lrrwa3z14zpqaj5p9vs0qklk-nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] but now [ { path = "https://nixos.org/channels/nixos-16.03/nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] Fixes #792.
2016-04-14 13:32:24 +00:00
res = { false, "" };
}
}
Replace Unicode quotes in user-facing strings by ASCII Relevant RFC: NixOS/rfcs#4 $ ag -l | xargs sed -i -e "/\"/s/’/'/g;/\"/s/‘/'/g"
2017-07-30 11:27:57 +00:00
debug(format("resolved search path element '%s' to '%s'") % elem.second % res.second);
Make the search path lazier with non-fatal errors Thus, -I / $NIX_PATH entries are now downloaded only when they are needed for evaluation. An error to download an entry is a non-fatal warning (just like non-existant paths). This does change the semantics of builtins.nixPath, which now returns the original, rather than resulting path. E.g., before we had [ { path = "/nix/store/hgm3yxf1lrrwa3z14zpqaj5p9vs0qklk-nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] but now [ { path = "https://nixos.org/channels/nixos-16.03/nixexprs.tar.xz"; prefix = "nixpkgs"; } ... ] Fixes #792.
2016-04-14 13:32:24 +00:00
searchPathResolved[elem.second] = res;
return res;
}
* Compile the lexer as C++ code. Remove all the redundant C/C++ marshalling code.
2006-09-04 21:36:15 +00:00
}
Copy permalink