lix/src/libutil/serialise.hh

408 lines
9.5 KiB
C++
Raw Normal View History

#pragma once
2016-07-13 10:03:37 +00:00
#include <memory>
#include "types.hh"
2015-07-19 23:16:16 +00:00
#include "util.hh"
namespace nix {
/* Abstract destination of binary data. */
2015-07-19 23:16:16 +00:00
struct Sink
{
virtual ~Sink() { }
virtual void operator () (const unsigned char * data, size_t len) = 0;
virtual bool good() { return true; }
2016-05-04 13:46:25 +00:00
void operator () (const std::string & s)
{
(*this)((const unsigned char *) s.data(), s.size());
}
};
2020-08-07 19:09:26 +00:00
/* Just throws away data. */
struct NullSink : Sink
{
void operator () (const unsigned char * data, size_t len) override
{ }
};
/* A buffered abstract sink. Warning: a BufferedSink should not be
used from multiple threads concurrently. */
Allow content-addressable paths to have references This adds a command 'nix make-content-addressable' that rewrites the specified store paths into content-addressable paths. The advantage of such paths is that 1) they can be imported without signatures; 2) they can enable deduplication in cases where derivation changes do not cause output changes (apart from store path hashes). For example, $ nix make-content-addressable -r nixpkgs.cowsay rewrote '/nix/store/g1g31ah55xdia1jdqabv1imf6mcw0nb1-glibc-2.25-49' to '/nix/store/48jfj7bg78a8n4f2nhg269rgw1936vj4-glibc-2.25-49' ... rewrote '/nix/store/qbi6rzpk0bxjw8lw6azn2mc7ynnn455q-cowsay-3.03+dfsg1-16' to '/nix/store/iq6g2x4q62xp7y7493bibx0qn5w7xz67-cowsay-3.03+dfsg1-16' We can then copy the resulting closure to another store without signatures: $ nix copy --trusted-public-keys '' ---to ~/my-nix /nix/store/iq6g2x4q62xp7y7493bibx0qn5w7xz67-cowsay-3.03+dfsg1-16 In order to support self-references in content-addressable paths, these paths are hashed "modulo" self-references, meaning that self-references are zeroed out during hashing. Somewhat annoyingly, this means that the NAR hash stored in the Nix database is no longer necessarily equal to the output of "nix hash-path"; for content-addressable paths, you need to pass the --modulo flag: $ nix path-info --json /nix/store/iq6g2x4q62xp7y7493bibx0qn5w7xz67-cowsay-3.03+dfsg1-16 | jq -r .[].narHash sha256:0ri611gdilz2c9rsibqhsipbfs9vwcqvs811a52i2bnkhv7w9mgw $ nix hash-path --type sha256 --base32 /nix/store/iq6g2x4q62xp7y7493bibx0qn5w7xz67-cowsay-3.03+dfsg1-16 1ggznh07khq0hz6id09pqws3a8q9pn03ya3c03nwck1kwq8rclzs $ nix hash-path --type sha256 --base32 /nix/store/iq6g2x4q62xp7y7493bibx0qn5w7xz67-cowsay-3.03+dfsg1-16 --modulo iq6g2x4q62xp7y7493bibx0qn5w7xz67 0ri611gdilz2c9rsibqhsipbfs9vwcqvs811a52i2bnkhv7w9mgw
2018-03-29 22:56:13 +00:00
struct BufferedSink : virtual Sink
{
size_t bufSize, bufPos;
2016-07-13 10:03:37 +00:00
std::unique_ptr<unsigned char[]> buffer;
BufferedSink(size_t bufSize = 32 * 1024)
2016-07-13 10:03:37 +00:00
: bufSize(bufSize), bufPos(0), buffer(nullptr) { }
void operator () (const unsigned char * data, size_t len) override;
2015-07-19 23:16:16 +00:00
2016-05-04 13:46:25 +00:00
void operator () (const std::string & s)
{
Sink::operator()(s);
}
void flush();
2015-07-19 23:16:16 +00:00
virtual void write(const unsigned char * data, size_t len) = 0;
};
/* Abstract source of binary data. */
struct Source
{
virtual ~Source() { }
2015-07-19 23:16:16 +00:00
/* Store exactly len bytes in the buffer pointed to by data.
It blocks until all the requested data is available, or throws
an error if it is not going to be available. */
void operator () (unsigned char * data, size_t len);
/* Store up to len in the buffer pointed to by data, and
return the number of bytes stored. It blocks until at least
one byte is available. */
virtual size_t read(unsigned char * data, size_t len) = 0;
virtual bool good() { return true; }
std::string drain();
};
2020-08-27 12:51:50 +00:00
/* A buffered abstract source. Warning: a BufferedSource should not be
used from multiple threads concurrently. */
struct BufferedSource : Source
{
size_t bufSize, bufPosIn, bufPosOut;
2016-07-13 10:03:37 +00:00
std::unique_ptr<unsigned char[]> buffer;
BufferedSource(size_t bufSize = 32 * 1024)
2016-07-13 10:03:37 +00:00
: bufSize(bufSize), bufPosIn(0), bufPosOut(0), buffer(nullptr) { }
2015-07-19 23:16:16 +00:00
size_t read(unsigned char * data, size_t len) override;
2015-07-19 23:16:16 +00:00
bool hasData();
protected:
/* Underlying read call, to be overridden. */
virtual size_t readUnbuffered(unsigned char * data, size_t len) = 0;
};
/* A sink that writes data to a file descriptor. */
struct FdSink : BufferedSink
{
int fd;
2016-02-26 15:16:08 +00:00
bool warn = false;
size_t written = 0;
2016-02-26 15:16:08 +00:00
FdSink() : fd(-1) { }
FdSink(int fd) : fd(fd) { }
FdSink(FdSink&&) = default;
FdSink& operator=(FdSink && s)
{
flush();
fd = s.fd;
s.fd = -1;
warn = s.warn;
written = s.written;
return *this;
}
2011-12-16 15:45:42 +00:00
~FdSink();
2015-07-19 23:16:16 +00:00
void write(const unsigned char * data, size_t len) override;
bool good() override;
private:
bool _good = true;
};
/* A source that reads data from a file descriptor. */
struct FdSource : BufferedSource
{
int fd;
2016-02-26 15:16:08 +00:00
size_t read = 0;
FdSource() : fd(-1) { }
FdSource(int fd) : fd(fd) { }
FdSource(FdSource&&) = default;
FdSource& operator=(FdSource && s)
{
fd = s.fd;
s.fd = -1;
read = s.read;
return *this;
}
bool good() override;
protected:
size_t readUnbuffered(unsigned char * data, size_t len) override;
private:
bool _good = true;
};
/* A sink that writes data to a string. */
struct StringSink : Sink
{
2016-03-04 15:49:56 +00:00
ref<std::string> s;
StringSink() : s(make_ref<std::string>()) { };
explicit StringSink(const size_t reservedSize) : s(make_ref<std::string>()) {
s->reserve(reservedSize);
};
2016-03-04 15:49:56 +00:00
StringSink(ref<std::string> s) : s(s) { };
void operator () (const unsigned char * data, size_t len) override;
};
/* A source that reads data from a string. */
struct StringSource : Source
{
const string & s;
size_t pos;
StringSource(const string & _s) : s(_s), pos(0) { }
size_t read(unsigned char * data, size_t len) override;
};
/* A sink that writes all incoming data to two other sinks. */
struct TeeSink : Sink
{
Sink & sink1, & sink2;
TeeSink(Sink & sink1, Sink & sink2) : sink1(sink1), sink2(sink2) { }
virtual void operator () (const unsigned char * data, size_t len)
{
sink1(data, len);
sink2(data, len);
}
};
2020-07-13 15:30:42 +00:00
/* Adapter class of a Source that saves all data read to a sink. */
struct TeeSource : Source
{
Source & orig;
2020-07-13 15:30:42 +00:00
Sink & sink;
TeeSource(Source & orig, Sink & sink)
: orig(orig), sink(sink) { }
size_t read(unsigned char * data, size_t len)
{
size_t n = orig.read(data, len);
sink(data, n);
return n;
}
};
/* A reader that consumes the original Source until 'size'. */
struct SizedSource : Source
{
Source & orig;
size_t remain;
SizedSource(Source & orig, size_t size)
: orig(orig), remain(size) { }
size_t read(unsigned char * data, size_t len)
{
if (this->remain <= 0) {
throw EndOfFile("sized: unexpected end-of-file");
}
len = std::min(len, this->remain);
size_t n = this->orig.read(data, len);
this->remain -= n;
return n;
}
/* Consume the original source until no remain data is left to consume. */
size_t drainAll()
{
std::vector<unsigned char> buf(8192);
size_t sum = 0;
while (this->remain > 0) {
size_t n = read(buf.data(), buf.size());
sum += n;
}
return sum;
}
};
/* A sink that that just counts the number of bytes given to it */
struct LengthSink : Sink
{
uint64_t length = 0;
virtual void operator () (const unsigned char * _, size_t len)
{
length += len;
}
};
/* Convert a function into a sink. */
struct LambdaSink : Sink
{
typedef std::function<void(const unsigned char *, size_t)> lambda_t;
lambda_t lambda;
LambdaSink(const lambda_t & lambda) : lambda(lambda) { }
virtual void operator () (const unsigned char * data, size_t len)
{
lambda(data, len);
}
};
/* Convert a function into a source. */
struct LambdaSource : Source
{
typedef std::function<size_t(unsigned char *, size_t)> lambda_t;
lambda_t lambda;
LambdaSource(const lambda_t & lambda) : lambda(lambda) { }
size_t read(unsigned char * data, size_t len) override
{
return lambda(data, len);
}
};
/* Chain two sources together so after the first is exhausted, the second is
used */
struct ChainSource : Source
{
Source & source1, & source2;
bool useSecond = false;
ChainSource(Source & s1, Source & s2)
: source1(s1), source2(s2)
{ }
size_t read(unsigned char * data, size_t len) override;
};
/* Convert a function that feeds data into a Sink into a Source. The
Source executes the function as a coroutine. */
std::unique_ptr<Source> sinkToSource(
std::function<void(Sink &)> fun,
std::function<void()> eof = []() {
throw EndOfFile("coroutine has finished");
});
void writePadding(size_t len, Sink & sink);
void writeString(const unsigned char * buf, size_t len, Sink & sink);
2015-07-19 23:16:16 +00:00
inline Sink & operator << (Sink & sink, uint64_t n)
{
unsigned char buf[8];
buf[0] = n & 0xff;
buf[1] = (n >> 8) & 0xff;
buf[2] = (n >> 16) & 0xff;
buf[3] = (n >> 24) & 0xff;
buf[4] = (n >> 32) & 0xff;
buf[5] = (n >> 40) & 0xff;
buf[6] = (n >> 48) & 0xff;
2018-05-02 11:56:34 +00:00
buf[7] = (unsigned char) (n >> 56) & 0xff;
2015-07-19 23:16:16 +00:00
sink(buf, sizeof(buf));
return sink;
}
Sink & operator << (Sink & sink, const string & s);
Sink & operator << (Sink & sink, const Strings & s);
Sink & operator << (Sink & sink, const StringSet & s);
MakeError(SerialisationError, Error);
template<typename T>
T readNum(Source & source)
{
unsigned char buf[8];
source(buf, sizeof(buf));
uint64_t n =
2020-07-30 11:10:49 +00:00
((uint64_t) buf[0]) |
((uint64_t) buf[1] << 8) |
((uint64_t) buf[2] << 16) |
((uint64_t) buf[3] << 24) |
((uint64_t) buf[4] << 32) |
((uint64_t) buf[5] << 40) |
((uint64_t) buf[6] << 48) |
((uint64_t) buf[7] << 56);
if (n > (uint64_t)std::numeric_limits<T>::max())
throw SerialisationError("serialised integer %d is too large for type '%s'", n, typeid(T).name());
2018-05-02 11:56:34 +00:00
return (T) n;
}
inline unsigned int readInt(Source & source)
{
return readNum<unsigned int>(source);
}
inline uint64_t readLongLong(Source & source)
{
return readNum<uint64_t>(source);
}
void readPadding(size_t len, Source & source);
size_t readString(unsigned char * buf, size_t max, Source & source);
string readString(Source & source, size_t max = std::numeric_limits<size_t>::max());
template<class T> T readStrings(Source & source);
Source & operator >> (Source & in, string & s);
template<typename T>
Source & operator >> (Source & in, T & n)
{
n = readNum<T>(in);
return in;
}
template<typename T>
Source & operator >> (Source & in, bool & b)
{
b = readNum<uint64_t>(in);
return in;
}
/* An adapter that converts a std::basic_istream into a source. */
struct StreamToSourceAdapter : Source
{
std::shared_ptr<std::basic_istream<char>> istream;
StreamToSourceAdapter(std::shared_ptr<std::basic_istream<char>> istream)
: istream(istream)
{ }
size_t read(unsigned char * data, size_t len) override
{
if (!istream->read((char *) data, len)) {
if (istream->eof()) {
if (istream->gcount() == 0)
throw EndOfFile("end of file");
} else
throw Error("I/O error in StreamToSourceAdapter");
}
return istream->gcount();
}
};
}