Pluggable fetchers

Flakes are now fetched using an extensible mechanism. Also lots of
other flake cleanups.

flake.lock

@@ -2,9 +2,9 @@
     "inputs": {
         "nixpkgs": {
             "inputs": {},
-            "narHash": "sha256-UcPmWgmf7Xgr4Ta8YjLuuxqzLmIYSk+uL2gPy/5bqmk=",
+            "narHash": "sha256-HkMF+s/yqNOOxqZGp+rscaC8LPtOGc50nEAjLFsnJpg=",
-            "originalUrl": "nixpkgs/release-19.09",
+            "originalUrl": "flake:nixpkgs/release-19.09",
-            "url": "github:edolstra/nixpkgs/44603b4103dbce2c9c18e6cc0df51a74f5eb8975"
+            "url": "git+file:///home/eelco/Dev/nixpkgs-flake?ref=release-19.09&rev=e7223c602152ee4544b05157fc9d88a3feed22c2"
         }
     },
     "version": 3

flake.nix

@@ -213,7 +213,8 @@
             name = "nix-tarball";
             version = builtins.readFile ./.version;
             versionSuffix = if officialRelease then "" else
-              "pre${builtins.substring 0 8 self.lastModified}_${self.shortRev}";
+              "pre${builtins.substring 0 8 self.lastModified}" +
+              (if self ? shortRev then "_${self.shortRev}" else "");
             src = self;
             inherit officialRelease;
 

src/libexpr/eval.hh

@@ -20,10 +20,6 @@ class EvalState;
 struct StorePath;
 enum RepairFlag : bool;
 
-namespace flake {
-struct FlakeRegistry;
-}
-
 
 typedef void (* PrimOpFun) (EvalState & state, const Pos & pos, Value * * args, Value & v);
 
@@ -67,8 +63,6 @@ typedef std::list<SearchPathElem> SearchPath;
 /* Initialise the Boehm GC, if applicable. */
 void initGC();
 
-typedef std::vector<std::pair<std::string, std::string>> RegistryOverrides;
-
 
 class EvalState
 {
@@ -95,8 +89,6 @@ public:
 
     const ref<Store> store;
 
-    RegistryOverrides registryOverrides;
-
 
 private:
     SrcToStore srcToStore;
@@ -224,8 +216,6 @@ public:
        path.  Nothing is copied to the store. */
     Path coerceToPath(const Pos & pos, Value & v, PathSet & context);
 
-    void addRegistryOverrides(RegistryOverrides overrides) { registryOverrides = overrides; }
-
 public:
 
     /* The base environment, containing the builtin functions and
@@ -328,16 +318,6 @@ private:
     friend struct ExprOpConcatLists;
     friend struct ExprSelect;
     friend void prim_getAttr(EvalState & state, const Pos & pos, Value * * args, Value & v);
-
-public:
-
-    const std::vector<std::shared_ptr<flake::FlakeRegistry>> getFlakeRegistries();
-
-    std::shared_ptr<flake::FlakeRegistry> getGlobalFlakeRegistry();
-
-private:
-    std::shared_ptr<flake::FlakeRegistry> _globalFlakeRegistry;
-    std::once_flag _globalFlakeRegistryInit;
 };
 
 
@@ -388,15 +368,6 @@ struct EvalSettings : Config
 
     Setting<bool> traceFunctionCalls{this, false, "trace-function-calls",
         "Emit log messages for each function entry and exit at the 'vomit' log level (-vvvv)."};
-
-    Setting<std::string> flakeRegistry{this, "https://github.com/NixOS/flake-registry/raw/master/flake-registry.json", "flake-registry",
-        "Path or URI of the global flake registry."};
-
-    Setting<bool> allowDirty{this, true, "allow-dirty",
-        "Whether to allow dirty Git/Mercurial trees."};
-
-    Setting<bool> warnDirty{this, true, "warn-dirty",
-        "Whether to warn about dirty Git/Mercurial trees."};
 };
 
 extern EvalSettings evalSettings;

src/libexpr/flake/flake.cc

@@ -2,16 +2,12 @@
 #include "lockfile.hh"
 #include "primops.hh"
 #include "eval-inline.hh"
-#include "primops/fetchGit.hh"
+#include "store-api.hh"
-#include "download.hh"
+#include "fetchers/fetchers.hh"
-#include "args.hh"
 
 #include <iostream>
-#include <queue>
-#include <regex>
 #include <ctime>
 #include <iomanip>
-#include <nlohmann/json.hpp>
 
 namespace nix {
 
@@ -19,105 +15,6 @@ using namespace flake;
 
 namespace flake {
 
-/* Read a registry. */
-std::shared_ptr<FlakeRegistry> readRegistry(const Path & path)
-{
-    auto registry = std::make_shared<FlakeRegistry>();
-
-    if (!pathExists(path))
-        return std::make_shared<FlakeRegistry>();
-
-    auto json = nlohmann::json::parse(readFile(path));
-
-    auto version = json.value("version", 0);
-    if (version != 1)
-        throw Error("flake registry '%s' has unsupported version %d", path, version);
-
-    auto flakes = json["flakes"];
-    for (auto i = flakes.begin(); i != flakes.end(); ++i) {
-        // FIXME: remove 'uri' soon.
-        auto url = i->value("url", i->value("uri", ""));
-        if (url.empty())
-            throw Error("flake registry '%s' lacks a 'url' attribute for entry '%s'",
-                path, i.key());
-        registry->entries.emplace(i.key(), url);
-    }
-
-    return registry;
-}
-
-/* Write a registry to a file. */
-void writeRegistry(const FlakeRegistry & registry, const Path & path)
-{
-    nlohmann::json json;
-    json["version"] = 1;
-    for (auto elem : registry.entries)
-        json["flakes"][elem.first.to_string()] = { {"url", elem.second.to_string()} };
-    createDirs(dirOf(path));
-    writeFile(path, json.dump(4)); // The '4' is the number of spaces used in the indentation in the json file.
-}
-
-Path getUserRegistryPath()
-{
-    return getHome() + "/.config/nix/registry.json";
-}
-
-std::shared_ptr<FlakeRegistry> getUserRegistry()
-{
-    return readRegistry(getUserRegistryPath());
-}
-
-std::shared_ptr<FlakeRegistry> getFlagRegistry(RegistryOverrides registryOverrides)
-{
-    auto flagRegistry = std::make_shared<FlakeRegistry>();
-    for (auto const & x : registryOverrides) {
-        flagRegistry->entries.insert_or_assign(FlakeRef(x.first), FlakeRef(x.second));
-    }
-    return flagRegistry;
-}
-
-static FlakeRef lookupFlake(EvalState & state, const FlakeRef & flakeRef, const Registries & registries,
-    std::vector<FlakeRef> pastSearches = {});
-
-FlakeRef updateFlakeRef(EvalState & state, const FlakeRef & newRef, const Registries & registries, std::vector<FlakeRef> pastSearches)
-{
-    std::string errorMsg = "found cycle in flake registries: ";
-    for (FlakeRef oldRef : pastSearches) {
-        errorMsg += oldRef.to_string();
-        if (oldRef == newRef)
-            throw Error(errorMsg);
-        errorMsg += " - ";
-    }
-    pastSearches.push_back(newRef);
-    return lookupFlake(state, newRef, registries, pastSearches);
-}
-
-static FlakeRef lookupFlake(EvalState & state, const FlakeRef & flakeRef, const Registries & registries,
-    std::vector<FlakeRef> pastSearches)
-{
-    for (std::shared_ptr<FlakeRegistry> registry : registries) {
-        auto i = registry->entries.find(flakeRef);
-        if (i != registry->entries.end()) {
-            auto newRef = i->second;
-            return updateFlakeRef(state, newRef, registries, pastSearches);
-        }
-
-        auto j = registry->entries.find(flakeRef.baseRef());
-        if (j != registry->entries.end()) {
-            auto newRef = j->second;
-            newRef.ref = flakeRef.ref;
-            newRef.rev = flakeRef.rev;
-            newRef.subdir = flakeRef.subdir;
-            return updateFlakeRef(state, newRef, registries, pastSearches);
-        }
-    }
-
-    if (!flakeRef.isDirect())
-        throw Error("could not resolve flake reference '%s'", flakeRef);
-
-    return flakeRef;
-}
-
 /* If 'allowLookup' is true, then resolve 'flakeRef' using the
    registries. */
 static FlakeRef maybeLookupFlake(
@@ -127,7 +24,7 @@ static FlakeRef maybeLookupFlake(
 {
     if (!flakeRef.isDirect()) {
         if (allowLookup)
-            return lookupFlake(state, flakeRef, state.getFlakeRegistries());
+            return flakeRef.resolve(state.store);
         else
             throw Error("'%s' is an indirect flake reference, but registry lookups are not allowed", flakeRef);
     } else
@@ -140,6 +37,7 @@ static FlakeRef lookupInRefMap(
     const RefMap & refMap,
     const FlakeRef & flakeRef)
 {
+#if 0
     // FIXME: inefficient.
     for (auto & i : refMap) {
         if (flakeRef.contains(i.first)) {
@@ -148,45 +46,11 @@ static FlakeRef lookupInRefMap(
             return i.second;
         }
     }
+#endif
 
     return flakeRef;
 }
 
-static SourceInfo fetchInput(EvalState & state, const FlakeRef & resolvedRef)
-{
-    assert(resolvedRef.isDirect());
-
-    auto doGit = [&](const GitInfo & gitInfo) {
-        FlakeRef ref(resolvedRef.baseRef());
-        ref.ref = gitInfo.ref;
-        ref.rev = gitInfo.rev;
-        SourceInfo info(ref);
-        info.storePath = gitInfo.storePath;
-        info.revCount = gitInfo.revCount;
-        info.narHash = state.store->queryPathInfo(state.store->parseStorePath(info.storePath))->narHash;
-        info.lastModified = gitInfo.lastModified;
-        return info;
-    };
-
-    // This only downloads one revision of the repo, not the entire history.
-    if (auto refData = std::get_if<FlakeRef::IsGitHub>(&resolvedRef.data)) {
-        return doGit(exportGitHub(state.store, refData->owner, refData->repo, resolvedRef.ref, resolvedRef.rev));
-    }
-
-    // This downloads the entire git history.
-    else if (auto refData = std::get_if<FlakeRef::IsGit>(&resolvedRef.data)) {
-        return doGit(exportGit(state.store, refData->uri, resolvedRef.ref, resolvedRef.rev, "source"));
-    }
-
-    else if (auto refData = std::get_if<FlakeRef::IsPath>(&resolvedRef.data)) {
-        if (!pathExists(refData->path + "/.git"))
-            throw Error("flake '%s' does not reference a Git repository", refData->path);
-        return doGit(exportGit(state.store, refData->path, resolvedRef.ref, resolvedRef.rev, "source"));
-    }
-
-    else abort();
-}
-
 static void expectType(EvalState & state, ValueType type,
     Value & value, const Pos & pos)
 {
@@ -204,34 +68,38 @@ static Flake getFlake(EvalState & state, const FlakeRef & originalRef,
         maybeLookupFlake(state,
             lookupInRefMap(refMap, originalRef), allowLookup));
 
-    SourceInfo sourceInfo = fetchInput(state, flakeRef);
+    auto [sourceInfo, resolvedInput] = flakeRef.input->fetchTree(state.store);
-    debug("got flake source '%s' with flakeref %s", sourceInfo.storePath, sourceInfo.resolvedRef.to_string());
 
-    FlakeRef resolvedRef = sourceInfo.resolvedRef;
+    FlakeRef resolvedRef(resolvedInput, flakeRef.subdir);
+
+    debug("got flake source '%s' from flake URL '%s'",
+        state.store->printStorePath(sourceInfo.storePath), resolvedRef);
 
     refMap.push_back({originalRef, resolvedRef});
     refMap.push_back({flakeRef, resolvedRef});
 
-    state.store->parseStorePath(sourceInfo.storePath);
-
     if (state.allowedPaths)
-        state.allowedPaths->insert(state.store->toRealPath(sourceInfo.storePath));
+        state.allowedPaths->insert(sourceInfo.actualPath);
 
     // Guard against symlink attacks.
-    Path flakeFile = canonPath(sourceInfo.storePath + "/" + resolvedRef.subdir + "/flake.nix");
+    auto flakeFile = canonPath(sourceInfo.actualPath + "/" + resolvedRef.subdir + "/flake.nix");
-    Path realFlakeFile = state.store->toRealPath(flakeFile);
+    if (!isInDir(flakeFile, sourceInfo.actualPath))
-    if (!isInDir(realFlakeFile, state.store->toRealPath(sourceInfo.storePath)))
+        throw Error("'flake.nix' file of flake '%s' escapes from '%s'",
-        throw Error("'flake.nix' file of flake '%s' escapes from '%s'", resolvedRef, sourceInfo.storePath);
+            resolvedRef, state.store->printStorePath(sourceInfo.storePath));
 
-    Flake flake(originalRef, sourceInfo);
+    Flake flake {
+        .originalRef = originalRef,
+        .resolvedRef = resolvedRef,
+        .sourceInfo = std::make_shared<fetchers::Tree>(std::move(sourceInfo))
+    };
 
-    if (!pathExists(realFlakeFile))
+    if (!pathExists(flakeFile))
         throw Error("source tree referenced by '%s' does not contain a '%s/flake.nix' file", resolvedRef, resolvedRef.subdir);
 
     Value vInfo;
-    state.evalFile(realFlakeFile, vInfo, true); // FIXME: symlink attack
+    state.evalFile(flakeFile, vInfo, true); // FIXME: symlink attack
 
-    expectType(state, tAttrs, vInfo, Pos(state.symbols.create(realFlakeFile), 0, 0));
+    expectType(state, tAttrs, vInfo, Pos(state.symbols.create(flakeFile), 0, 0));
 
     auto sEdition = state.symbols.create("edition");
     auto sEpoch = state.symbols.create("epoch"); // FIXME: remove soon
@@ -266,12 +134,12 @@ static Flake getFlake(EvalState & state, const FlakeRef & originalRef,
         for (Attr inputAttr : *(*(**inputs).value).attrs) {
             expectType(state, tAttrs, *inputAttr.value, *inputAttr.pos);
 
-            FlakeInput input(FlakeRef(inputAttr.name));
+            FlakeInput input(parseFlakeRef(inputAttr.name));
 
             for (Attr attr : *(inputAttr.value->attrs)) {
                 if (attr.name == sUrl || attr.name == sUri) {
                     expectType(state, tString, *attr.value, *attr.pos);
-                    input.ref = std::string(attr.value->string.s);
+                    input.ref = parseFlakeRef(attr.value->string.s);
                 } else if (attr.name == sFlake) {
                     expectType(state, tBool, *attr.value, *attr.pos);
                     input.isFlake = attr.value->boolean;
@@ -293,7 +161,7 @@ static Flake getFlake(EvalState & state, const FlakeRef & originalRef,
         if (flake.vOutputs->lambda.fun->matchAttrs) {
             for (auto & formal : flake.vOutputs->lambda.fun->formals->formals) {
                 if (formal.name != state.sSelf)
-                    flake.inputs.emplace(formal.name, FlakeInput(FlakeRef(formal.name)));
+                    flake.inputs.emplace(formal.name, FlakeInput(parseFlakeRef(formal.name)));
             }
         }
 
@@ -319,27 +187,30 @@ Flake getFlake(EvalState & state, const FlakeRef & originalRef, bool allowLookup
     return getFlake(state, originalRef, allowLookup, refMap);
 }
 
-static SourceInfo getNonFlake(EvalState & state, const FlakeRef & originalRef,
+static std::pair<fetchers::Tree, FlakeRef> getNonFlake(
-    bool allowLookup, RefMap & refMap)
+    EvalState & state,
+    const FlakeRef & originalRef,
+    bool allowLookup,
+    RefMap & refMap)
 {
     auto flakeRef = lookupInRefMap(refMap,
         maybeLookupFlake(state,
             lookupInRefMap(refMap, originalRef), allowLookup));
 
-    auto sourceInfo = fetchInput(state, flakeRef);
+    auto [sourceInfo, resolvedInput] = flakeRef.input->fetchTree(state.store);
-    debug("got non-flake source '%s' with flakeref %s", sourceInfo.storePath, sourceInfo.resolvedRef.to_string());
 
-    FlakeRef resolvedRef = sourceInfo.resolvedRef;
+    FlakeRef resolvedRef(resolvedInput, flakeRef.subdir);
+
+    debug("got non-flake source '%s' with flakeref %s",
+        state.store->printStorePath(sourceInfo.storePath), resolvedRef);
 
     refMap.push_back({originalRef, resolvedRef});
     refMap.push_back({flakeRef, resolvedRef});
 
-    state.store->parseStorePath(sourceInfo.storePath);
-
     if (state.allowedPaths)
-        state.allowedPaths->insert(sourceInfo.storePath);
+        state.allowedPaths->insert(sourceInfo.actualPath);
 
-    return sourceInfo;
+    return std::make_pair(std::move(sourceInfo), resolvedRef);
 }
 
 bool allowedToWrite(HandleLockFile handle)
@@ -382,9 +253,9 @@ static std::pair<Flake, LockedInput> updateLocks(
     bool topRef)
 {
     LockedInput newEntry(
-        flake.sourceInfo.resolvedRef,
+        flake.resolvedRef,
         flake.originalRef,
-        flake.sourceInfo.narHash);
+        flake.sourceInfo->narHash);
 
     std::vector<std::function<void()>> postponed;
 
@@ -397,29 +268,29 @@ static std::pair<Flake, LockedInput> updateLocks(
             if (handleLockFile == AllPure || handleLockFile == TopRefUsesRegistries)
                 throw Error("cannot update flake input '%s' in pure mode", id);
 
-            auto warn = [&](const SourceInfo & sourceInfo) {
+            auto warn = [&](const FlakeRef & resolvedRef, const fetchers::Tree & sourceInfo) {
                 if (i == oldEntry.inputs.end())
                     printInfo("mapped flake input '%s' to '%s'",
-                        inputPath2, sourceInfo.resolvedRef);
+                        inputPath2, resolvedRef);
                 else
                     printMsg(lvlWarn, "updated flake input '%s' from '%s' to '%s'",
-                        inputPath2, i->second.originalRef, sourceInfo.resolvedRef);
+                        inputPath2, i->second.originalRef, resolvedRef);
             };
 
             if (input.isFlake) {
                 auto actualInput = getFlake(state, input.ref,
                     allowedToUseRegistries(handleLockFile, false), refMap);
-                warn(actualInput.sourceInfo);
+                warn(actualInput.resolvedRef, *actualInput.sourceInfo);
                 postponed.push_back([&, id{id}, inputPath2, actualInput]() {
                     newEntry.inputs.insert_or_assign(id,
                         updateLocks(refMap, inputPath2, state, actualInput, handleLockFile, {}, false).second);
                 });
             } else {
-                auto sourceInfo = getNonFlake(state, input.ref,
+                auto [sourceInfo, resolvedRef] = getNonFlake(state, input.ref,
                     allowedToUseRegistries(handleLockFile, false), refMap);
-                warn(sourceInfo);
+                warn(resolvedRef, sourceInfo);
                 newEntry.inputs.insert_or_assign(id,
-                    LockedInput(sourceInfo.resolvedRef, input.ref, sourceInfo.narHash));
+                    LockedInput(resolvedRef, input.ref, sourceInfo.narHash));
             }
         }
     }
@@ -444,8 +315,7 @@ ResolvedFlake resolveFlake(EvalState & state, const FlakeRef & topRef, HandleLoc
         // If recreateLockFile, start with an empty lockfile
         // FIXME: symlink attack
         oldLockFile = LockFile::read(
-            state.store->toRealPath(flake.sourceInfo.storePath)
+            flake.sourceInfo->actualPath + "/" + flake.resolvedRef.subdir + "/flake.lock");
-            + "/" + flake.sourceInfo.resolvedRef.subdir + "/flake.lock");
     }
 
     debug("old lock file: %s", oldLockFile);
@@ -459,19 +329,26 @@ ResolvedFlake resolveFlake(EvalState & state, const FlakeRef & topRef, HandleLoc
 
     if (!(lockFile == oldLockFile)) {
         if (allowedToWrite(handleLockFile)) {
-            if (auto refData = std::get_if<FlakeRef::IsPath>(&topRef.data)) {
+            if (auto sourcePath = topRef.input->getSourcePath()) {
-                if (lockFile.isDirty()) {
+                if (!lockFile.isImmutable()) {
-                    if (evalSettings.warnDirty)
+                    if (settings.warnDirty)
-                        warn("will not write lock file of flake '%s' because it has a dirty input", topRef);
+                        warn("will not write lock file of flake '%s' because it has a mutable input", topRef);
                 } else {
-                    lockFile.write(refData->path + (topRef.subdir == "" ? "" : "/" + topRef.subdir) + "/flake.lock");
+                    warn("updated lock file of flake '%s'", topRef);
 
+                    lockFile.write(*sourcePath + (topRef.subdir == "" ? "" : "/" + topRef.subdir) + "/flake.lock");
+
+                    // FIXME: rewriting the lockfile changed the
+                    // top-level repo, so we should re-read it.
+
+                    #if 0
                     // Hack: Make sure that flake.lock is visible to Git, so it ends up in the Nix store.
                     runProgram("git", true,
-                        { "-C", refData->path, "add",
+                        { "-C", *sourcePath, "add",
                           "--force",
                           "--intent-to-add",
                           (topRef.subdir == "" ? "" : topRef.subdir + "/") + "flake.lock" });
+                    #endif
                 }
             } else
                 warn("cannot write lock file of remote flake '%s'", topRef);
@@ -479,7 +356,7 @@ ResolvedFlake resolveFlake(EvalState & state, const FlakeRef & topRef, HandleLoc
             warn("using updated lock file without writing it to file");
     }
 
-    return ResolvedFlake(std::move(flake), std::move(lockFile));
+    return ResolvedFlake { .flake = std::move(flake), .lockFile = std::move(lockFile) };
 }
 
 void updateLockFile(EvalState & state, const FlakeRef & flakeRef, bool recreateLockFile)
@@ -487,17 +364,17 @@ void updateLockFile(EvalState & state, const FlakeRef & flakeRef, bool recreateL
     resolveFlake(state, flakeRef, recreateLockFile ? RecreateLockFile : UpdateLockFile);
 }
 
-static void emitSourceInfoAttrs(EvalState & state, const SourceInfo & sourceInfo, Value & vAttrs)
+static void emitSourceInfoAttrs(EvalState & state, const fetchers::Tree & sourceInfo, Value & vAttrs)
 {
-    auto & path = sourceInfo.storePath;
+    assert(state.store->isValidPath(sourceInfo.storePath));
-    assert(state.store->isValidPath(state.store->parseStorePath(path)));
+    auto pathS = state.store->printStorePath(sourceInfo.storePath);
-    mkString(*state.allocAttr(vAttrs, state.sOutPath), path, {path});
+    mkString(*state.allocAttr(vAttrs, state.sOutPath), pathS, {pathS});
 
-    if (sourceInfo.resolvedRef.rev) {
+    if (sourceInfo.rev) {
         mkString(*state.allocAttr(vAttrs, state.symbols.create("rev")),
-            sourceInfo.resolvedRef.rev->gitRev());
+            sourceInfo.rev->gitRev());
         mkString(*state.allocAttr(vAttrs, state.symbols.create("shortRev")),
-            sourceInfo.resolvedRef.rev->gitShortRev());
+            sourceInfo.rev->gitShortRev());
     }
 
     if (sourceInfo.revCount)
@@ -505,8 +382,7 @@ static void emitSourceInfoAttrs(EvalState & state, const SourceInfo & sourceInfo
 
     if (sourceInfo.lastModified)
         mkString(*state.allocAttr(vAttrs, state.symbols.create("lastModified")),
-            fmt("%s",
+            fmt("%s", std::put_time(std::gmtime(&*sourceInfo.lastModified), "%Y%m%d%H%M%S")));
-                std::put_time(std::gmtime(&*sourceInfo.lastModified), "%Y%m%d%H%M%S")));
 }
 
 struct LazyInput
@@ -522,19 +398,17 @@ static void prim_callFlake(EvalState & state, const Pos & pos, Value * * args, V
 {
     auto lazyInput = (LazyInput *) args[0]->attrs;
 
-    assert(lazyInput->lockedInput.ref.isImmutable());
-
     if (lazyInput->isFlake) {
         auto flake = getFlake(state, lazyInput->lockedInput.ref, false);
 
-        if (flake.sourceInfo.narHash != lazyInput->lockedInput.narHash)
+        if (flake.sourceInfo->narHash != lazyInput->lockedInput.narHash)
             throw Error("the content hash of flake '%s' doesn't match the hash recorded in the referring lockfile",
                 lazyInput->lockedInput.ref);
 
         callFlake(state, flake, lazyInput->lockedInput, v);
     } else {
         RefMap refMap;
-        auto sourceInfo = getNonFlake(state, lazyInput->lockedInput.ref, false, refMap);
+        auto [sourceInfo, resolvedRef] = getNonFlake(state, lazyInput->lockedInput.ref, false, refMap);
 
         if (sourceInfo.narHash != lazyInput->lockedInput.narHash)
             throw Error("the content hash of repository '%s' doesn't match the hash recorded in the referring lockfile",
@@ -542,10 +416,11 @@ static void prim_callFlake(EvalState & state, const Pos & pos, Value * * args, V
 
         state.mkAttrs(v, 8);
 
-        assert(state.store->isValidPath(state.store->parseStorePath(sourceInfo.storePath)));
+        assert(state.store->isValidPath(sourceInfo.storePath));
 
-        mkString(*state.allocAttr(v, state.sOutPath),
+        auto pathS = state.store->printStorePath(sourceInfo.storePath);
-            sourceInfo.storePath, {sourceInfo.storePath});
+
+        mkString(*state.allocAttr(v, state.sOutPath), pathS, {pathS});
 
         emitSourceInfoAttrs(state, sourceInfo, v);
 
@@ -580,7 +455,7 @@ void callFlake(EvalState & state,
 
     auto & vSourceInfo = *state.allocValue();
     state.mkAttrs(vSourceInfo, 8);
-    emitSourceInfoAttrs(state, flake.sourceInfo, vSourceInfo);
+    emitSourceInfoAttrs(state, *flake.sourceInfo, vSourceInfo);
     vSourceInfo.attrs->sort();
 
     vInputs.attrs->push_back(Attr(state.sSelf, &vRes));
@@ -614,70 +489,12 @@ void callFlake(EvalState & state,
 // This function is exposed to be used in nix files.
 static void prim_getFlake(EvalState & state, const Pos & pos, Value * * args, Value & v)
 {
-    callFlake(state, resolveFlake(state, state.forceStringNoCtx(*args[0], pos),
+    callFlake(state, resolveFlake(state, parseFlakeRef(state.forceStringNoCtx(*args[0], pos)),
             evalSettings.pureEval ? AllPure : UseUpdatedLockFile), v);
 }
 
 static RegisterPrimOp r2("getFlake", 1, prim_getFlake);
 
-void gitCloneFlake(FlakeRef flakeRef, EvalState & state, Registries registries, const Path & destDir)
-{
-    flakeRef = lookupFlake(state, flakeRef, registries);
-
-    std::string uri;
-
-    Strings args = {"clone"};
-
-    if (auto refData = std::get_if<FlakeRef::IsGitHub>(&flakeRef.data)) {
-        uri = "git@github.com:" + refData->owner + "/" + refData->repo + ".git";
-        args.push_back(uri);
-        if (flakeRef.ref) {
-            args.push_back("--branch");
-            args.push_back(*flakeRef.ref);
-        }
-    } else if (auto refData = std::get_if<FlakeRef::IsGit>(&flakeRef.data)) {
-        args.push_back(refData->uri);
-        if (flakeRef.ref) {
-            args.push_back("--branch");
-            args.push_back(*flakeRef.ref);
-        }
-    }
-
-    if (destDir != "")
-        args.push_back(destDir);
-
-    runProgram("git", true, args);
-}
-
-}
-
-std::shared_ptr<flake::FlakeRegistry> EvalState::getGlobalFlakeRegistry()
-{
-    std::call_once(_globalFlakeRegistryInit, [&]() {
-        auto path = evalSettings.flakeRegistry;
-
-        if (!hasPrefix(path, "/")) {
-            CachedDownloadRequest request(evalSettings.flakeRegistry);
-            request.name = "flake-registry.json";
-            request.gcRoot = true;
-            path = getDownloader()->downloadCached(store, request).path;
-        }
-
-        _globalFlakeRegistry = readRegistry(path);
-    });
-
-    return _globalFlakeRegistry;
-}
-
-// This always returns a vector with flakeReg, userReg, globalReg.
-// If one of them doesn't exist, the registry is left empty but does exist.
-const Registries EvalState::getFlakeRegistries()
-{
-    Registries registries;
-    registries.push_back(getFlagRegistry(registryOverrides));
-    registries.push_back(getUserRegistry());
-    registries.push_back(getGlobalFlakeRegistry());
-    return registries;
 }
 
 Fingerprint ResolvedFlake::getFingerprint() const
@@ -687,10 +504,12 @@ Fingerprint ResolvedFlake::getFingerprint() const
     // flake.sourceInfo.storePath for the fingerprint.
     return hashString(htSHA256,
         fmt("%s;%d;%d;%s",
-            flake.sourceInfo.storePath,
+            flake.sourceInfo->storePath.to_string(),
-            flake.sourceInfo.revCount.value_or(0),
+            flake.sourceInfo->revCount.value_or(0),
-            flake.sourceInfo.lastModified.value_or(0),
+            flake.sourceInfo->lastModified.value_or(0),
             lockFile));
 }
 
+Flake::~Flake() { }
+
 }

src/libexpr/flake/flake.hh

@@ -9,25 +9,10 @@ namespace nix {
 struct Value;
 class EvalState;
 
+namespace fetchers { struct Tree; }
+
 namespace flake {
 
-static const size_t FLAG_REGISTRY = 0;
-static const size_t USER_REGISTRY = 1;
-static const size_t GLOBAL_REGISTRY = 2;
-
-struct FlakeRegistry
-{
-    std::map<FlakeRef, FlakeRef> entries;
-};
-
-typedef std::vector<std::shared_ptr<FlakeRegistry>> Registries;
-
-std::shared_ptr<FlakeRegistry> readRegistry(const Path &);
-
-void writeRegistry(const FlakeRegistry &, const Path &);
-
-Path getUserRegistryPath();
-
 enum HandleLockFile : unsigned int
     { AllPure // Everything is handled 100% purely
     , TopRefUsesRegistries // The top FlakeRef uses the registries, apart from that, everything happens 100% purely
@@ -37,27 +22,6 @@ enum HandleLockFile : unsigned int
     , UseNewLockFile // `RecreateLockFile` without writing to file
     };
 
-struct SourceInfo
-{
-    // Immutable flakeref that this source tree was obtained from.
-    FlakeRef resolvedRef;
-
-    Path storePath;
-
-    // Number of ancestors of the most recent commit.
-    std::optional<uint64_t> revCount;
-
-    // NAR hash of the store path.
-    Hash narHash;
-
-    // A stable timestamp of this source tree. For Git and GitHub
-    // flakes, the commit date (not author date!) of the most recent
-    // commit.
-    std::optional<time_t> lastModified;
-
-    SourceInfo(const FlakeRef & resolvRef) : resolvedRef(resolvRef) {};
-};
-
 struct FlakeInput
 {
     FlakeRef ref;
@@ -68,14 +32,13 @@ struct FlakeInput
 struct Flake
 {
     FlakeRef originalRef;
+    FlakeRef resolvedRef;
     std::string description;
-    SourceInfo sourceInfo;
+    std::shared_ptr<const fetchers::Tree> sourceInfo;
     std::map<FlakeId, FlakeInput> inputs;
     Value * vOutputs; // FIXME: gc
     unsigned int edition;
-
+    ~Flake();
-    Flake(const FlakeRef & origRef, const SourceInfo & sourceInfo)
-        : originalRef(origRef), sourceInfo(sourceInfo) {};
 };
 
 Flake getFlake(EvalState & state, const FlakeRef & flakeRef, bool allowLookup);
@@ -88,9 +51,6 @@ struct ResolvedFlake
     Flake flake;
     LockFile lockFile;
 
-    ResolvedFlake(Flake && flake, LockFile && lockFile)
-        : flake(flake), lockFile(lockFile) {}
-
     Fingerprint getFingerprint() const;
 };
 
@@ -107,8 +67,6 @@ void callFlake(EvalState & state,
 
 void updateLockFile(EvalState &, const FlakeRef & flakeRef, bool recreateLockFile);
 
-void gitCloneFlake(FlakeRef flakeRef, EvalState &, Registries, const Path & destDir);
-
 }
 
 }

src/libexpr/flake/flakeref.cc

@@ -1,239 +1,33 @@
 #include "flakeref.hh"
 #include "store-api.hh"
-
+#include "fetchers/parse.hh"
-#include <regex>
+#include "fetchers/fetchers.hh"
+#include "fetchers/registry.hh"
+#include "fetchers/regex.hh"
 
 namespace nix {
 
-// A Git ref (i.e. branch or tag name).
+#if 0
-const static std::string refRegex = "[a-zA-Z0-9][a-zA-Z0-9_.-]*"; // FIXME: check
-
-// A Git revision (a SHA-1 commit hash).
-const static std::string revRegexS = "[0-9a-fA-F]{40}";
-std::regex revRegex(revRegexS, std::regex::ECMAScript);
-
-// A Git ref or revision.
-const static std::string revOrRefRegex = "(?:(" + revRegexS + ")|(" + refRegex + "))";
-
-// A rev ("e72daba8250068216d79d2aeef40d4d95aff6666"), or a ref
-// optionally followed by a rev (e.g. "master" or
-// "master/e72daba8250068216d79d2aeef40d4d95aff6666").
-const static std::string refAndOrRevRegex = "(?:(" + revRegexS + ")|(?:(" + refRegex + ")(?:/(" + revRegexS + "))?))";
-
-const static std::string flakeId = "[a-zA-Z][a-zA-Z0-9_-]*";
-
-// GitHub references.
-const static std::string ownerRegex = "[a-zA-Z][a-zA-Z0-9_-]*";
-const static std::string repoRegex = "[a-zA-Z][a-zA-Z0-9_-]*";
-
-// URI stuff.
-const static std::string schemeRegex = "[a-z+]+";
-const static std::string authorityRegex = "[a-zA-Z0-9._~-]*";
-const static std::string segmentRegex = "[a-zA-Z0-9._~-]+";
-const static std::string pathRegex = "/?" + segmentRegex + "(?:/" + segmentRegex + ")*";
-
 // 'dir' path elements cannot start with a '.'. We also reject
 // potentially dangerous characters like ';'.
 const static std::string subDirElemRegex = "(?:[a-zA-Z0-9_-]+[a-zA-Z0-9._-]*)";
 const static std::string subDirRegex = subDirElemRegex + "(?:/" + subDirElemRegex + ")*";
+#endif
 
 
-FlakeRef::FlakeRef(const std::string & uri_, bool allowRelative)
-{
-    // FIXME: could combine this into one regex.
-
-    static std::regex flakeRegex(
-        "(?:flake:)?(" + flakeId + ")(?:/(?:" + refAndOrRevRegex + "))?",
-        std::regex::ECMAScript);
-
-    static std::regex githubRegex(
-        "github:(" + ownerRegex + ")/(" + repoRegex + ")(?:/" + revOrRefRegex + ")?",
-        std::regex::ECMAScript);
-
-    static std::regex uriRegex(
-        "((" + schemeRegex + "):" +
-        "(?://(" + authorityRegex + "))?" +
-        "(" + pathRegex + "))",
-        std::regex::ECMAScript);
-
-    static std::regex refRegex2(refRegex, std::regex::ECMAScript);
-
-    static std::regex subDirRegex2(subDirRegex, std::regex::ECMAScript);
-
-    auto [uri2, params] = splitUriAndParams(uri_);
-    std::string uri(uri2);
-
-    auto handleSubdir = [&](const std::string & name, const std::string & value) {
-        if (name == "dir") {
-            if (value != "" && !std::regex_match(value, subDirRegex2))
-                throw BadFlakeRef("flake '%s' has invalid subdirectory '%s'", uri, value);
-            subdir = value;
-            return true;
-        } else
-            return false;
-    };
-
-    auto handleGitParams = [&](const std::string & name, const std::string & value) {
-        if (name == "rev") {
-            if (!std::regex_match(value, revRegex))
-                throw BadFlakeRef("invalid Git revision '%s'", value);
-            rev = Hash(value, htSHA1);
-        } else if (name == "ref") {
-            if (!std::regex_match(value, refRegex2))
-                throw BadFlakeRef("invalid Git ref '%s'", value);
-            ref = value;
-        } else if (handleSubdir(name, value))
-            ;
-        else return false;
-        return true;
-    };
-
-    std::smatch match;
-    if (std::regex_match(uri, match, flakeRegex)) {
-        IsId d;
-        d.id = match[1];
-        if (match[2].matched)
-            rev = Hash(match[2], htSHA1);
-        else if (match[3].matched) {
-            ref = match[3];
-            if (match[4].matched)
-                rev = Hash(match[4], htSHA1);
-        }
-        data = d;
-    }
-
-    else if (std::regex_match(uri, match, githubRegex)) {
-        IsGitHub d;
-        d.owner = match[1];
-        d.repo = match[2];
-        if (match[3].matched)
-            rev = Hash(match[3], htSHA1);
-        else if (match[4].matched) {
-            ref = match[4];
-        }
-        for (auto & param : params) {
-            if (handleSubdir(param.first, param.second))
-                ;
-            else
-                throw BadFlakeRef("invalid Git flakeref parameter '%s', in '%s'", param.first, uri);
-        }
-        data = d;
-    }
-
-    else if (std::regex_match(uri, match, uriRegex)) {
-        auto & scheme = match[2];
-        if (scheme == "git" ||
-            scheme == "git+http" ||
-            scheme == "git+https" ||
-            scheme == "git+ssh" ||
-            scheme == "git+file" ||
-            scheme == "file")
-        {
-            IsGit d;
-            d.uri = match[1];
-            for (auto & param : params) {
-                if (handleGitParams(param.first, param.second))
-                    ;
-                else
-                    // FIXME: should probably pass through unknown parameters
-                    throw BadFlakeRef("invalid Git flakeref parameter '%s', in '%s'", param.first, uri);
-            }
-            if (rev && !ref)
-                throw BadFlakeRef("flake URI '%s' lacks a Git ref", uri);
-            data = d;
-        } else
-            throw BadFlakeRef("unsupported URI scheme '%s' in flake reference '%s'", scheme, uri);
-    }
-
-    else if ((hasPrefix(uri, "/") || (allowRelative && (hasPrefix(uri, "./") || hasPrefix(uri, "../") || uri == ".")))
-        && uri.find(':') == std::string::npos)
-    {
-        IsPath d;
-        if (allowRelative) {
-            d.path = absPath(uri);
-            try {
-                if (!S_ISDIR(lstat(d.path).st_mode))
-                    throw MissingFlake("path '%s' is not a flake (sub)directory", d.path);
-            } catch (SysError & e) {
-                if (e.errNo == ENOENT || e.errNo == EISDIR)
-                    throw MissingFlake("flake '%s' does not exist", d.path);
-                throw;
-            }
-            while (true) {
-                if (pathExists(d.path + "/.git")) break;
-                subdir = std::string(baseNameOf(d.path)) + (subdir.empty() ? "" : "/" + subdir);
-                d.path = dirOf(d.path);
-                if (d.path == "/")
-                    throw MissingFlake("path '%s' is not a flake (because it does not reference a Git repository)", uri);
-            }
-        } else
-            d.path = canonPath(uri);
-        data = d;
-        for (auto & param : params) {
-            if (handleGitParams(param.first, param.second))
-                ;
-            else
-                throw BadFlakeRef("invalid Git flakeref parameter '%s', in '%s'", param.first, uri);
-        }
-    }
-
-    else
-        throw BadFlakeRef("'%s' is not a valid flake reference", uri);
-}
-
 std::string FlakeRef::to_string() const
 {
-    std::string string;
+    return input->to_string();
-    bool first = true;
-
-    auto addParam =
-        [&](const std::string & name, std::string value) {
-            string += first ? '?' : '&';
-            first = false;
-            string += name;
-            string += '=';
-            string += value; // FIXME: escaping
-        };
-
-    if (auto refData = std::get_if<FlakeRef::IsId>(&data)) {
-        string = refData->id;
-        if (ref) string += '/' + *ref;
-        if (rev) string += '/' + rev->gitRev();
 }
 
-    else if (auto refData = std::get_if<FlakeRef::IsPath>(&data)) {
+bool FlakeRef::isDirect() const
-        string = refData->path;
+{
-        if (ref) addParam("ref", *ref);
+    return input->isDirect();
-        if (rev) addParam("rev", rev->gitRev());
-        if (subdir != "") addParam("dir", subdir);
 }
 
-    else if (auto refData = std::get_if<FlakeRef::IsGitHub>(&data)) {
+bool FlakeRef::isImmutable() const
-        assert(!(ref && rev));
+{
-        string = "github:" + refData->owner + "/" + refData->repo;
+    return input->isImmutable();
-        if (ref) { string += '/'; string += *ref; }
-        if (rev) { string += '/'; string += rev->gitRev(); }
-        if (subdir != "") addParam("dir", subdir);
-    }
-
-    else if (auto refData = std::get_if<FlakeRef::IsGit>(&data)) {
-        assert(!rev || ref);
-        string = refData->uri;
-
-        if (ref) {
-            addParam("ref", *ref);
-            if (rev)
-                addParam("rev", rev->gitRev());
-        }
-
-        if (subdir != "") addParam("dir", subdir);
-    }
-
-    else abort();
-
-    assert(FlakeRef(string) == *this);
-
-    return string;
 }
 
 std::ostream & operator << (std::ostream & str, const FlakeRef & flakeRef)
@@ -242,42 +36,130 @@ std::ostream & operator << (std::ostream & str, const FlakeRef & flakeRef)
     return str;
 }
 
-bool FlakeRef::isImmutable() const
+bool FlakeRef::operator==(const FlakeRef & other) const
 {
-    return (bool) rev;
+    return *input == *other.input && subdir == other.subdir;
 }
 
-FlakeRef FlakeRef::baseRef() const // Removes the ref and rev from a FlakeRef.
+FlakeRef FlakeRef::resolve(ref<Store> store) const
 {
-    FlakeRef result(*this);
+    return FlakeRef(lookupInRegistries(store, input), subdir);
-    result.ref = std::nullopt;
-    result.rev = std::nullopt;
-    return result;
 }
 
-bool FlakeRef::contains(const FlakeRef & other) const
+FlakeRef parseFlakeRef(
+    const std::string & url, const std::optional<Path> & baseDir)
 {
-    if (!(data == other.data))
+    auto [flakeRef, fragment] = parseFlakeRefWithFragment(url, baseDir);
-        return false;
+    if (fragment != "")
-
+        throw Error("unexpected fragment '%s' in flake reference '%s'", fragment, url);
-    if (ref && ref != other.ref)
+    return flakeRef;
-        return false;
-
-    if (rev && rev != other.rev)
-        return false;
-
-    if (subdir != other.subdir)
-        return false;
-
-    return true;
 }
 
-std::optional<FlakeRef> parseFlakeRef(
+std::optional<FlakeRef> maybeParseFlakeRef(
-    const std::string & uri, bool allowRelative)
+    const std::string & url, const std::optional<Path> & baseDir)
 {
     try {
-        return FlakeRef(uri, allowRelative);
+        return parseFlakeRef(url, baseDir);
-    } catch (BadFlakeRef & e) {
+    } catch (Error &) {
+        return {};
+    }
+}
+
+std::pair<FlakeRef, std::string> parseFlakeRefWithFragment(
+    const std::string & url, const std::optional<Path> & baseDir)
+{
+    using namespace fetchers;
+
+    static std::regex pathUrlRegex(
+        "(" + pathRegex + "/?)"
+        + "(?:\\?(" + queryRegex + "))?"
+        + "(?:#(" + queryRegex + "))?",
+        std::regex::ECMAScript);
+
+    static std::regex flakeRegex(
+        "((" + flakeId + ")(?:/(?:" + refAndOrRevRegex + "))?)"
+        + "(?:#(" + queryRegex + "))?",
+        std::regex::ECMAScript);
+
+    std::smatch match;
+
+    /* Check if 'url' is a flake ID. This is an abbreviated syntax for
+       'flake:<flake-id>?ref=<ref>&rev=<rev>'. */
+
+    if (std::regex_match(url, match, flakeRegex)) {
+        auto parsedURL = ParsedURL{
+            .url = url,
+            .base = "flake:" + std::string(match[1]),
+            .scheme = "flake",
+            .authority = "",
+            .path = match[1],
+            .fragment = percentDecode(std::string(match[6]))
+        };
+
+        return std::make_pair(
+            FlakeRef(inputFromURL(parsedURL), ""),
+            parsedURL.fragment);
+    }
+
+    /* Check if 'url' is a path (either absolute or relative to
+       'baseDir'). If so, search upward to the root of the repo
+       (i.e. the directory containing .git). */
+
+    else if (std::regex_match(url, match, pathUrlRegex)) {
+        std::string path = match[1];
+        if (!baseDir && !hasPrefix(path, "/"))
+            throw BadURL("flake reference '%s' is not an absolute path", url);
+        path = absPath(path, baseDir, true);
+
+        auto flakeRoot = path;
+        std::string subdir;
+
+        while (true) {
+            if (pathExists(flakeRoot + "/.git")) break;
+            subdir = std::string(baseNameOf(flakeRoot)) + (subdir.empty() ? "" : "/" + subdir);
+            flakeRoot = dirOf(flakeRoot);
+            if (flakeRoot == "/")
+                throw BadURL("path '%s' is not a flake (because it does not reference a Git repository)", path);
+        }
+
+        auto base = std::string("git+file://") + flakeRoot;
+
+        auto parsedURL = ParsedURL{
+            .url = base, // FIXME
+            .base = base,
+            .scheme = "git+file",
+            .authority = "",
+            .path = flakeRoot,
+            .query = decodeQuery(match[2]),
+            .fragment = percentDecode(std::string(match[3]))
+        };
+
+        if (subdir != "") {
+            if (parsedURL.query.count("subdir"))
+                throw Error("flake URL '%s' has an inconsistent 'subdir' parameter", url);
+            parsedURL.query.insert_or_assign("subdir", subdir);
+        }
+
+        return std::make_pair(
+            FlakeRef(inputFromURL(parsedURL), get(parsedURL.query, "subdir").value_or("")),
+            parsedURL.fragment);
+    }
+
+    else {
+        auto parsedURL = parseURL(url);
+        return std::make_pair(
+            FlakeRef(inputFromURL(parsedURL), get(parsedURL.query, "subdir").value_or("")),
+            parsedURL.fragment);
+    }
+}
+
+std::optional<std::pair<FlakeRef, std::string>> maybeParseFlakeRefWithFragment(
+    const std::string & url, const std::optional<Path> & baseDir)
+{
+    try {
+        return parseFlakeRefWithFragment(url, baseDir);
+    } catch (Error & e) {
+        printError("FOO: %s", e.what());
         return {};
     }
 }

src/libexpr/flake/flakeref.hh

@@ -7,194 +7,52 @@
 
 namespace nix {
 
-/* Flake references are a URI-like syntax to specify a flake.
+class Store;
 
-   Examples:
+namespace fetchers { struct Input; }
-
-   * <flake-id>(/rev-or-ref(/rev)?)?
-
-     Look up a flake by ID in the flake lock file or in the flake
-     registry. These must specify an actual location for the flake
-     using the formats listed below. Note that in pure evaluation
-     mode, the flake registry is empty.
-
-     Optionally, the rev or ref from the dereferenced flake can be
-     overriden. For example,
-
-       nixpkgs/19.09
-
-     uses the "19.09" branch of the nixpkgs' flake GitHub repository,
-     while
-
-       nixpkgs/98a2a5b5370c1e2092d09cb38b9dcff6d98a109f
-
-     uses the specified revision. For Git (rather than GitHub)
-     repositories, both the rev and ref must be given, e.g.
-
-       nixpkgs/19.09/98a2a5b5370c1e2092d09cb38b9dcff6d98a109f
-
-   * github:<owner>/<repo>(/<rev-or-ref>)?
-
-     A repository on GitHub. These differ from Git references in that
-     they're downloaded in a efficient way (via the tarball mechanism)
-     and that they support downloading a specific revision without
-     specifying a branch. <rev-or-ref> is either a commit hash ("rev")
-     or a branch or tag name ("ref"). The default is: "master" if none
-     is specified. Note that in pure evaluation mode, a commit hash
-     must be used.
-
-     Flakes fetched in this manner expose "rev" and "lastModified"
-     attributes, but not "revCount".
-
-     Examples:
-
-       github:edolstra/dwarffs
-       github:edolstra/dwarffs/unstable
-       github:edolstra/dwarffs/41c0c1bf292ea3ac3858ff393b49ca1123dbd553
-
-   * git+https://<server>/<path>(\?attr(&attr)*)?
-     git+ssh://<server>/<path>(\?attr(&attr)*)?
-     git://<server>/<path>(\?attr(&attr)*)?
-     file:///<path>(\?attr(&attr)*)?
-
-     where 'attr' is one of:
-       rev=<rev>
-       ref=<ref>
-
-     A Git repository fetched through https. The default for "ref" is
-     "master".
-
-     Examples:
-
-       git+https://example.org/my/repo.git
-       git+https://example.org/my/repo.git?ref=release-1.2.3
-       git+https://example.org/my/repo.git?rev=e72daba8250068216d79d2aeef40d4d95aff6666
-       git://github.com/edolstra/dwarffs.git?ref=flake&rev=2efca4bc9da70fb001b26c3dc858c6397d3c4817
-
-   * /path(\?attr(&attr)*)?
-
-     Like file://path, but if no "ref" or "rev" is specified, the
-     (possibly dirty) working tree will be used. Using a working tree
-     is not allowed in pure evaluation mode.
-
-     Examples:
-
-       /path/to/my/repo
-       /path/to/my/repo?ref=develop
-       /path/to/my/repo?rev=e72daba8250068216d79d2aeef40d4d95aff6666
-
-   * https://<server>/<path>.tar.xz(?hash=<sri-hash>)
-     file:///<path>.tar.xz(?hash=<sri-hash>)
-
-     A flake distributed as a tarball. In pure evaluation mode, an SRI
-     hash is mandatory. It exposes a "lastModified" attribute, being
-     the newest file inside the tarball.
-
-     Example:
-
-       https://releases.nixos.org/nixos/unstable/nixos-19.03pre167858.f2a1a4e93be/nixexprs.tar.xz
-       https://releases.nixos.org/nixos/unstable/nixos-19.03pre167858.f2a1a4e93be/nixexprs.tar.xz?hash=sha256-56bbc099995ea8581ead78f22832fee7dbcb0a0b6319293d8c2d0aef5379397c
-
-  Note: currently, there can be only one flake per Git repository, and
-  it must be at top-level. In the future, we may want to add a field
-  (e.g. "dir=<dir>") to specify a subdirectory inside the repository.
-*/
 
 typedef std::string FlakeId;
-typedef std::string FlakeUri;
 
 struct FlakeRef
 {
-    struct IsId
+    std::shared_ptr<const fetchers::Input> input;
-    {
-        FlakeId id;
-        bool operator<(const IsId & b) const { return id < b.id; };
-        bool operator==(const IsId & b) const { return id == b.id; };
-    };
 
-    struct IsGitHub {
+    Path subdir;
-        std::string owner, repo;
+
-        bool operator<(const IsGitHub & b) const {
+    bool operator==(const FlakeRef & other) const;
-            return std::make_tuple(owner, repo) < std::make_tuple(b.owner, b.repo);
+
+    FlakeRef(const std::shared_ptr<const fetchers::Input> & input, const Path & subdir)
+        : input(input), subdir(subdir)
+    {
+        assert(input);
     }
-        bool operator==(const IsGitHub & b) const {
-            return owner == b.owner && repo == b.repo;
-        }
-    };
-
-    // Git, Tarball
-    struct IsGit
-    {
-        std::string uri;
-        bool operator<(const IsGit & b) const { return uri < b.uri; }
-        bool operator==(const IsGit & b) const { return uri == b.uri; }
-    };
-
-    struct IsPath
-    {
-        Path path;
-        bool operator<(const IsPath & b) const { return path < b.path; }
-        bool operator==(const IsPath & b) const { return path == b.path; }
-    };
-
-    // Git, Tarball
-
-    std::variant<IsId, IsGitHub, IsGit, IsPath> data;
-
-    std::optional<std::string> ref;
-    std::optional<Hash> rev;
-    Path subdir = ""; // This is a relative path pointing at the flake.nix file's directory, relative to the git root.
-
-    bool operator<(const FlakeRef & flakeRef) const
-    {
-        return std::make_tuple(data, ref, rev, subdir) <
-            std::make_tuple(flakeRef.data, flakeRef.ref, flakeRef.rev, subdir);
-    }
-
-    bool operator==(const FlakeRef & flakeRef) const
-    {
-        return std::make_tuple(data, ref, rev, subdir) ==
-            std::make_tuple(flakeRef.data, flakeRef.ref, flakeRef.rev, flakeRef.subdir);
-    }
-
-    // Parse a flake URI.
-    FlakeRef(const std::string & uri, bool allowRelative = false);
 
     // FIXME: change to operator <<.
     std::string to_string() const;
 
     /* Check whether this is a "direct" flake reference, that is, not
        a flake ID, which requires a lookup in the flake registry. */
-    bool isDirect() const
+    bool isDirect() const;
-    {
-        return !std::get_if<FlakeRef::IsId>(&data);
-    }
 
     /* Check whether this is an "immutable" flake reference, that is,
        one that contains a commit hash or content hash. */
     bool isImmutable() const;
 
-    FlakeRef baseRef() const;
+    FlakeRef resolve(ref<Store> store) const;
-
-    bool isDirty() const
-    {
-        return std::get_if<FlakeRef::IsPath>(&data)
-            && rev == Hash(rev->type);
-    }
-
-    /* Return true if 'other' is not less specific than 'this'. For
-       example, 'nixpkgs' contains 'nixpkgs/release-19.03', and both
-       'nixpkgs' and 'nixpkgs/release-19.03' contain
-       'nixpkgs/release-19.03/<hash>'. */
-    bool contains(const FlakeRef & other) const;
 };
 
 std::ostream & operator << (std::ostream & str, const FlakeRef & flakeRef);
 
-MakeError(BadFlakeRef, Error);
+FlakeRef parseFlakeRef(
-MakeError(MissingFlake, BadFlakeRef);
+    const std::string & url, const std::optional<Path> & baseDir = {});
 
-std::optional<FlakeRef> parseFlakeRef(
+std::optional<FlakeRef> maybeParseFlake(
-    const std::string & uri, bool allowRelative = false);
+    const std::string & url, const std::optional<Path> & baseDir = {});
+
+std::pair<FlakeRef, std::string> parseFlakeRefWithFragment(
+    const std::string & url, const std::optional<Path> & baseDir = {});
+
+std::optional<std::pair<FlakeRef, std::string>> maybeParseFlakeRefWithFragment(
+    const std::string & url, const std::optional<Path> & baseDir = {});
 
 }

src/libexpr/flake/lockfile.cc

@@ -7,8 +7,8 @@ namespace nix::flake {
 
 LockedInput::LockedInput(const nlohmann::json & json)
     : LockedInputs(json)
-    , ref(json.value("url", json.value("uri", "")))
+    , ref(parseFlakeRef(json.value("url", json.value("uri", ""))))
-    , originalRef(json.value("originalUrl", json.value("originalUri", "")))
+    , originalRef(parseFlakeRef(json.value("originalUrl", json.value("originalUri", ""))))
     , narHash(Hash((std::string) json["narHash"]))
 {
     if (!ref.isImmutable())
@@ -47,12 +47,12 @@ nlohmann::json LockedInputs::toJson() const
     return json;
 }
 
-bool LockedInputs::isDirty() const
+bool LockedInputs::isImmutable() const
 {
     for (auto & i : inputs)
-        if (i.second.ref.isDirty() || i.second.isDirty()) return true;
+        if (!i.second.ref.isImmutable() || !i.second.isImmutable()) return false;
 
-    return false;
+    return true;
 }
 
 nlohmann::json LockFile::toJson() const

src/libexpr/flake/lockfile.hh

@@ -22,9 +22,7 @@ struct LockedInputs
 
     nlohmann::json toJson() const;
 
-    /* A lock file is dirty if it contains a dirty flakeref
+    bool isImmutable() const;
-       (i.e. reference to a dirty working tree). */
-    bool isDirty() const;
 };
 
 /* Lock file information about a flake input. */
@@ -35,9 +33,7 @@ struct LockedInput : LockedInputs
 
     LockedInput(const FlakeRef & ref, const FlakeRef & originalRef, const Hash & narHash)
         : ref(ref), originalRef(originalRef), narHash(narHash)
-    {
+    { }
-        assert(ref.isImmutable());
-    };
 
     LockedInput(const nlohmann::json & json);
 

src/libexpr/primops/fetchGit.cc

@@ -1,332 +1,12 @@
-#include "fetchGit.hh"
 #include "primops.hh"
 #include "eval-inline.hh"
-#include "download.hh"
 #include "store-api.hh"
-#include "pathlocks.hh"
 #include "hash.hh"
-#include "tarfile.hh"
+#include "fetchers/fetchers.hh"
-
+#include "fetchers/parse.hh"
-#include <sys/time.h>
-
-#include <regex>
-
-#include <nlohmann/json.hpp>
-
-using namespace std::string_literals;
 
 namespace nix {
 
-extern std::regex revRegex;
-
-static Path getCacheInfoPathFor(const std::string & name, const Hash & rev)
-{
-    Path cacheDir = getCacheDir() + "/nix/git-revs";
-    std::string linkName =
-        name == "source"
-        ? rev.gitRev()
-        : hashString(htSHA512, name + std::string("\0"s) + rev.gitRev()).to_string(Base32, false);
-    return cacheDir + "/" + linkName + ".link";
-}
-
-static void cacheGitInfo(const std::string & name, const GitInfo & gitInfo)
-{
-    nlohmann::json json;
-    json["storePath"] = gitInfo.storePath;
-    json["name"] = name;
-    json["rev"] = gitInfo.rev.gitRev();
-    if (gitInfo.revCount)
-        json["revCount"] = *gitInfo.revCount;
-    json["lastModified"] = gitInfo.lastModified;
-
-    auto cacheInfoPath = getCacheInfoPathFor(name, gitInfo.rev);
-    createDirs(dirOf(cacheInfoPath));
-    writeFile(cacheInfoPath, json.dump());
-}
-
-static std::optional<GitInfo> lookupGitInfo(
-    ref<Store> store,
-    const std::string & name,
-    const Hash & rev)
-{
-    try {
-        auto json = nlohmann::json::parse(readFile(getCacheInfoPathFor(name, rev)));
-
-        assert(json["name"] == name && Hash((std::string) json["rev"], htSHA1) == rev);
-
-        Path storePath = json["storePath"];
-
-        if (store->isValidPath(store->parseStorePath(storePath))) {
-            GitInfo gitInfo;
-            gitInfo.storePath = storePath;
-            gitInfo.rev = rev;
-            if (json.find("revCount") != json.end())
-                gitInfo.revCount = json["revCount"];
-            gitInfo.lastModified = json["lastModified"];
-            return gitInfo;
-        }
-
-    } catch (SysError & e) {
-        if (e.errNo != ENOENT) throw;
-    }
-
-    return {};
-}
-
-GitInfo exportGit(ref<Store> store, std::string uri,
-    std::optional<std::string> ref,
-    std::optional<Hash> rev,
-    const std::string & name)
-{
-    assert(!rev || rev->type == htSHA1);
-
-    if (rev) {
-        if (auto gitInfo = lookupGitInfo(store, name, *rev)) {
-            // If this gitInfo was produced by exportGitHub, then it won't
-            // have a revCount. So we have to do a full clone.
-            if (gitInfo->revCount) {
-                gitInfo->ref = ref;
-                return *gitInfo;
-            }
-        }
-    }
-
-    if (hasPrefix(uri, "git+")) uri = std::string(uri, 4);
-
-    bool isLocal = hasPrefix(uri, "/") && pathExists(uri + "/.git");
-
-    // If this is a local directory (but not a file:// URI) and no ref
-    // or revision is given, then allow the use of an unclean working
-    // tree.
-    if (!ref && !rev && isLocal) {
-        bool clean = false;
-
-        /* Check whether this repo has any commits. There are
-           probably better ways to do this. */
-        bool haveCommits = !readDirectory(uri + "/.git/refs/heads").empty();
-
-        try {
-            if (haveCommits) {
-                runProgram("git", true, { "-C", uri, "diff-index", "--quiet", "HEAD", "--" });
-                clean = true;
-            }
-        } catch (ExecError & e) {
-            if (!WIFEXITED(e.status) || WEXITSTATUS(e.status) != 1) throw;
-        }
-
-        if (!clean) {
-
-            /* This is an unclean working tree. So copy all tracked files. */
-
-            if (!evalSettings.allowDirty)
-                throw Error("Git tree '%s' is dirty", uri);
-
-            if (evalSettings.warnDirty)
-                warn("Git tree '%s' is dirty", uri);
-
-            GitInfo gitInfo;
-            gitInfo.ref = "HEAD";
-
-            auto files = tokenizeString<std::set<std::string>>(
-                runProgram("git", true, { "-C", uri, "ls-files", "-z" }), "\0"s);
-
-            PathFilter filter = [&](const Path & p) -> bool {
-                assert(hasPrefix(p, uri));
-                std::string file(p, uri.size() + 1);
-
-                auto st = lstat(p);
-
-                if (S_ISDIR(st.st_mode)) {
-                    auto prefix = file + "/";
-                    auto i = files.lower_bound(prefix);
-                    return i != files.end() && hasPrefix(*i, prefix);
-                }
-
-                return files.count(file);
-            };
-
-            gitInfo.storePath = store->printStorePath(store->addToStore("source", uri, true, htSHA256, filter));
-            gitInfo.revCount = haveCommits ? std::stoull(runProgram("git", true, { "-C", uri, "rev-list", "--count", "HEAD" })) : 0;
-            // FIXME: maybe we should use the timestamp of the last
-            // modified dirty file?
-            gitInfo.lastModified = haveCommits ? std::stoull(runProgram("git", true, { "-C", uri, "log", "-1", "--format=%ct", "HEAD" })) : 0;
-
-            return gitInfo;
-        }
-    }
-
-    if (!ref) ref = isLocal ? "HEAD" : "master";
-
-    // Don't clone file:// URIs (but otherwise treat them the same as
-    // remote URIs, i.e. don't use the working tree or HEAD).
-    static bool forceHttp = getEnv("_NIX_FORCE_HTTP") == "1"; // for testing
-    if (!forceHttp && hasPrefix(uri, "file://")) {
-        uri = std::string(uri, 7);
-        isLocal = true;
-    }
-
-    Path repoDir;
-
-    if (isLocal) {
-
-        if (!rev)
-            rev = Hash(chomp(runProgram("git", true, { "-C", uri, "rev-parse", *ref })), htSHA1);
-
-        repoDir = uri;
-
-    } else {
-
-        Path cacheDir = getCacheDir() + "/nix/gitv3/" + hashString(htSHA256, uri).to_string(Base32, false);
-        repoDir = cacheDir;
-
-        if (!pathExists(cacheDir)) {
-            createDirs(dirOf(cacheDir));
-            runProgram("git", true, { "init", "--bare", repoDir });
-        }
-
-        Path localRefFile =
-            ref->compare(0, 5, "refs/") == 0
-            ? cacheDir + "/" + *ref
-            : cacheDir + "/refs/heads/" + *ref;
-
-        bool doFetch;
-        time_t now = time(0);
-
-        /* If a rev was specified, we need to fetch if it's not in the
-           repo. */
-        if (rev) {
-            try {
-                runProgram("git", true, { "-C", repoDir, "cat-file", "-e", rev->gitRev() });
-                doFetch = false;
-            } catch (ExecError & e) {
-                if (WIFEXITED(e.status)) {
-                    doFetch = true;
-                } else {
-                    throw;
-                }
-            }
-        } else {
-            /* If the local ref is older than ‘tarball-ttl’ seconds, do a
-               git fetch to update the local ref to the remote ref. */
-            struct stat st;
-            doFetch = stat(localRefFile.c_str(), &st) != 0 ||
-                (uint64_t) st.st_mtime + settings.tarballTtl <= (uint64_t) now;
-        }
-
-        if (doFetch) {
-            Activity act(*logger, lvlTalkative, actUnknown, fmt("fetching Git repository '%s'", uri));
-
-            // FIXME: git stderr messes up our progress indicator, so
-            // we're using --quiet for now. Should process its stderr.
-            try {
-                runProgram("git", true, { "-C", repoDir, "fetch", "--quiet", "--force", "--", uri, fmt("%s:%s", *ref, *ref) });
-            } catch (Error & e) {
-                if (!pathExists(localRefFile)) throw;
-                warn("could not update local clone of Git repository '%s'; continuing with the most recent version", uri);
-            }
-
-            struct timeval times[2];
-            times[0].tv_sec = now;
-            times[0].tv_usec = 0;
-            times[1].tv_sec = now;
-            times[1].tv_usec = 0;
-
-            utimes(localRefFile.c_str(), times);
-        }
-
-        if (!rev)
-            rev = Hash(chomp(readFile(localRefFile)), htSHA1);
-    }
-
-    if (auto gitInfo = lookupGitInfo(store, name, *rev)) {
-        if (gitInfo->revCount) {
-            gitInfo->ref = ref;
-            return *gitInfo;
-        }
-    }
-
-    // FIXME: check whether rev is an ancestor of ref.
-    GitInfo gitInfo;
-    gitInfo.ref = *ref;
-    gitInfo.rev = *rev;
-
-    printTalkative("using revision %s of repo '%s'", gitInfo.rev, uri);
-
-    // FIXME: should pipe this, or find some better way to extract a
-    // revision.
-    auto source = sinkToSource([&](Sink & sink) {
-        RunOptions gitOptions("git", { "-C", repoDir, "archive", gitInfo.rev.gitRev() });
-        gitOptions.standardOut = &sink;
-        runProgram2(gitOptions);
-    });
-
-    Path tmpDir = createTempDir();
-    AutoDelete delTmpDir(tmpDir, true);
-
-    unpackTarfile(*source, tmpDir);
-
-    gitInfo.storePath = store->printStorePath(store->addToStore(name, tmpDir));
-
-    gitInfo.revCount = std::stoull(runProgram("git", true, { "-C", repoDir, "rev-list", "--count", gitInfo.rev.gitRev() }));
-    gitInfo.lastModified = std::stoull(runProgram("git", true, { "-C", repoDir, "log", "-1", "--format=%ct", gitInfo.rev.gitRev() }));
-
-    cacheGitInfo(name, gitInfo);
-
-    return gitInfo;
-}
-
-GitInfo exportGitHub(
-    ref<Store> store,
-    const std::string & owner,
-    const std::string & repo,
-    std::optional<std::string> ref,
-    std::optional<Hash> rev)
-{
-    if (rev) {
-        if (auto gitInfo = lookupGitInfo(store, "source", *rev))
-            return *gitInfo;
-    }
-
-    if (!rev) {
-        auto url = fmt("https://api.github.com/repos/%s/%s/commits/%s",
-            owner, repo, ref ? *ref : "master");
-        CachedDownloadRequest request(url);
-        request.ttl = rev ? 1000000000 : settings.tarballTtl;
-        auto result = getDownloader()->downloadCached(store, request);
-        auto json = nlohmann::json::parse(readFile(result.path));
-        rev = Hash(json["sha"], htSHA1);
-    }
-
-    // FIXME: use regular /archive URLs instead? api.github.com
-    // might have stricter rate limits.
-
-    auto url = fmt("https://api.github.com/repos/%s/%s/tarball/%s",
-        owner, repo, rev->to_string(Base16, false));
-
-    std::string accessToken = settings.githubAccessToken.get();
-    if (accessToken != "")
-        url += "?access_token=" + accessToken;
-
-    CachedDownloadRequest request(url);
-    request.unpack = true;
-    request.name = "source";
-    request.ttl = 1000000000;
-    request.getLastModified = true;
-    auto result = getDownloader()->downloadCached(store, request);
-
-    assert(result.lastModified);
-
-    GitInfo gitInfo;
-    gitInfo.storePath = result.storePath;
-    gitInfo.rev = *rev;
-    gitInfo.lastModified = *result.lastModified;
-
-    // FIXME: this can overwrite a cache file that contains a revCount.
-    cacheGitInfo("source", gitInfo);
-
-    return gitInfo;
-}
-
 static void prim_fetchGit(EvalState & state, const Pos & pos, Value * * args, Value & v)
 {
     std::string url;
@@ -368,18 +48,31 @@ static void prim_fetchGit(EvalState & state, const Pos & pos, Value * * args, Va
     if (evalSettings.pureEval && !rev)
         throw Error("in pure evaluation mode, 'fetchGit' requires a Git revision");
 
-    auto gitInfo = exportGit(state.store, url, ref, rev, name);
+    auto parsedUrl = fetchers::parseURL(
+        url.find("://") != std::string::npos
+        ? "git+" + url
+        : "git+file://" + url);
+    if (ref) parsedUrl.query.insert_or_assign("ref", *ref);
+    if (rev) parsedUrl.query.insert_or_assign("rev", rev->gitRev());
+    // FIXME: use name
+    auto input = inputFromURL(parsedUrl);
+
+    auto tree = input->fetchTree(state.store).first;
 
     state.mkAttrs(v, 8);
-    mkString(*state.allocAttr(v, state.sOutPath), gitInfo.storePath, PathSet({gitInfo.storePath}));
+    auto storePath = state.store->printStorePath(tree.storePath);
-    mkString(*state.allocAttr(v, state.symbols.create("rev")), gitInfo.rev.gitRev());
+    mkString(*state.allocAttr(v, state.sOutPath), storePath, PathSet({storePath}));
-    mkString(*state.allocAttr(v, state.symbols.create("shortRev")), gitInfo.rev.gitShortRev());
+    // Backward compatibility: set 'rev' to
-    assert(gitInfo.revCount);
+    // 0000000000000000000000000000000000000000 for a dirty tree.
-    mkInt(*state.allocAttr(v, state.symbols.create("revCount")), *gitInfo.revCount);
+    auto rev2 = tree.rev.value_or(Hash(htSHA1));
+    mkString(*state.allocAttr(v, state.symbols.create("rev")), rev2.gitRev());
+    mkString(*state.allocAttr(v, state.symbols.create("shortRev")), rev2.gitShortRev());
+    assert(tree.revCount);
+    mkInt(*state.allocAttr(v, state.symbols.create("revCount")), *tree.revCount);
     v.attrs->sort();
 
     if (state.allowedPaths)
-        state.allowedPaths->insert(state.store->toRealPath(gitInfo.storePath));
+        state.allowedPaths->insert(tree.actualPath);
 }
 
 static RegisterPrimOp r("fetchGit", 1, prim_fetchGit);

src/libexpr/primops/fetchGit.hh

@@ -1,32 +0,0 @@
-#pragma once
-
-#include "store-api.hh"
-
-#include <regex>
-
-namespace nix {
-
-struct GitInfo
-{
-    Path storePath;
-    std::optional<std::string> ref;
-    Hash rev{htSHA1};
-    std::optional<uint64_t> revCount;
-    time_t lastModified;
-};
-
-GitInfo exportGit(
-    ref<Store> store,
-    std::string uri,
-    std::optional<std::string> ref,
-    std::optional<Hash> rev,
-    const std::string & name);
-
-GitInfo exportGitHub(
-    ref<Store> store,
-    const std::string & owner,
-    const std::string & repo,
-    std::optional<std::string> ref,
-    std::optional<Hash> rev);
-
-}

src/libexpr/primops/fetchMercurial.cc

@@ -36,10 +36,10 @@ HgInfo exportMercurial(ref<Store> store, const std::string & uri,
             /* This is an unclean working tree. So copy all tracked
                files. */
 
-            if (!evalSettings.allowDirty)
+            if (!settings.allowDirty)
                 throw Error("Mercurial tree '%s' is unclean", uri);
 
-            if (evalSettings.warnDirty)
+            if (settings.warnDirty)
                 warn("Mercurial tree '%s' is unclean", uri);
 
             HgInfo hgInfo;

src/libstore/fetchers/fetchers.cc

@@ -0,0 +1,56 @@
+#include "fetchers.hh"
+#include "parse.hh"
+#include "store-api.hh"
+
+namespace nix::fetchers {
+
+std::unique_ptr<std::vector<std::unique_ptr<InputScheme>>> inputSchemes = nullptr;
+
+void registerInputScheme(std::unique_ptr<InputScheme> && inputScheme)
+{
+    if (!inputSchemes) inputSchemes = std::make_unique<std::vector<std::unique_ptr<InputScheme>>>();
+    inputSchemes->push_back(std::move(inputScheme));
+}
+
+std::unique_ptr<Input> inputFromURL(const ParsedURL & url)
+{
+    for (auto & inputScheme : *inputSchemes) {
+        auto res = inputScheme->inputFromURL(url);
+        if (res) return res;
+    }
+    throw Error("input '%s' is unsupported", url.url);
+}
+
+std::unique_ptr<Input> inputFromURL(const std::string & url)
+{
+    return inputFromURL(parseURL(url));
+}
+
+std::pair<Tree, std::shared_ptr<const Input>> Input::fetchTree(ref<Store> store) const
+{
+    auto [tree, input] = fetchTreeInternal(store);
+
+    if (tree.actualPath == "")
+        tree.actualPath = store->toRealPath(store->printStorePath(tree.storePath));
+
+    if (!tree.narHash)
+        tree.narHash = store->queryPathInfo(tree.storePath)->narHash;
+
+    if (input->narHash)
+        assert(input->narHash == tree.narHash);
+
+    return {std::move(tree), input};
+}
+
+std::shared_ptr<const Input> Input::applyOverrides(
+    std::optional<std::string> ref,
+    std::optional<Hash> rev) const
+{
+    if (ref)
+        throw Error("don't know how to apply '%s' to '%s'", *ref, to_string());
+    if (rev)
+        throw Error("don't know how to apply '%s' to '%s'", rev->to_string(Base16, false), to_string());
+    return shared_from_this();
+}
+
+}

src/libstore/fetchers/fetchers.hh

@@ -0,0 +1,75 @@
+#pragma once
+
+#include "types.hh"
+#include "hash.hh"
+#include "path.hh"
+
+#include <memory>
+
+namespace nix { class Store; }
+
+namespace nix::fetchers {
+
+struct Input;
+
+struct Tree
+{
+    Path actualPath;
+    StorePath storePath;
+    Hash narHash;
+    std::optional<Hash> rev;
+    std::optional<uint64_t> revCount;
+    std::optional<time_t> lastModified;
+};
+
+struct Input : std::enable_shared_from_this<Input>
+{
+    std::string type;
+    std::optional<Hash> narHash;
+
+    virtual bool operator ==(const Input & other) const { return false; }
+
+    virtual bool isDirect() const { return true; }
+
+    virtual bool isImmutable() const { return (bool) narHash; }
+
+    virtual bool contains(const Input & other) const { return false; }
+
+    virtual std::optional<std::string> getRef() const { return {}; }
+
+    virtual std::optional<Hash> getRev() const { return {}; }
+
+    virtual std::string to_string() const = 0;
+
+    std::pair<Tree, std::shared_ptr<const Input>> fetchTree(ref<Store> store) const;
+
+    virtual std::shared_ptr<const Input> applyOverrides(
+        std::optional<std::string> ref,
+        std::optional<Hash> rev) const;
+
+    virtual std::optional<Path> getSourcePath() const { return {}; }
+
+    virtual void clone(const Path & destDir) const
+    {
+        throw Error("do not know how to clone input '%s'", to_string());
+    }
+
+private:
+
+    virtual std::pair<Tree, std::shared_ptr<const Input>> fetchTreeInternal(ref<Store> store) const = 0;
+};
+
+struct ParsedURL;
+
+struct InputScheme
+{
+    virtual std::unique_ptr<Input> inputFromURL(const ParsedURL & url) = 0;
+};
+
+std::unique_ptr<Input> inputFromURL(const ParsedURL & url);
+
+std::unique_ptr<Input> inputFromURL(const std::string & url);
+
+void registerInputScheme(std::unique_ptr<InputScheme> && fetcher);
+
+}

src/libstore/fetchers/git.cc

@@ -0,0 +1,382 @@
+#include "fetchers.hh"
+#include "parse.hh"
+#include "globals.hh"
+#include "tarfile.hh"
+#include "store-api.hh"
+#include "regex.hh"
+
+#include <sys/time.h>
+
+#include <nlohmann/json.hpp>
+
+using namespace std::string_literals;
+
+namespace nix::fetchers {
+
+static Path getCacheInfoPathFor(const std::string & name, const Hash & rev)
+{
+    Path cacheDir = getCacheDir() + "/nix/git-revs-v2";
+    std::string linkName =
+        name == "source"
+        ? rev.gitRev()
+        : hashString(htSHA512, name + std::string("\0"s) + rev.gitRev()).to_string(Base32, false);
+    return cacheDir + "/" + linkName + ".link";
+}
+
+static void cacheGitInfo(Store & store, const std::string & name, const Tree & tree)
+{
+    nlohmann::json json;
+    json["storePath"] = store.printStorePath(tree.storePath);
+    json["name"] = name;
+    json["rev"] = tree.rev->gitRev();
+    json["revCount"] = *tree.revCount;
+    json["lastModified"] = *tree.lastModified;
+
+    auto cacheInfoPath = getCacheInfoPathFor(name, *tree.rev);
+    createDirs(dirOf(cacheInfoPath));
+    writeFile(cacheInfoPath, json.dump());
+}
+
+static std::optional<Tree> lookupGitInfo(
+    ref<Store> store,
+    const std::string & name,
+    const Hash & rev)
+{
+    try {
+        auto json = nlohmann::json::parse(readFile(getCacheInfoPathFor(name, rev)));
+
+        assert(json["name"] == name && Hash((std::string) json["rev"], htSHA1) == rev);
+
+        auto storePath = store->parseStorePath((std::string) json["storePath"]);
+
+        if (store->isValidPath(storePath)) {
+            Tree tree{
+                .actualPath = store->toRealPath(store->printStorePath(storePath)),
+                .storePath = std::move(storePath),
+                .rev = rev,
+                .revCount = json["revCount"],
+                .lastModified = json["lastModified"],
+            };
+            return tree;
+        }
+
+    } catch (SysError & e) {
+        if (e.errNo != ENOENT) throw;
+    }
+
+    return {};
+}
+
+struct GitInput : Input
+{
+    ParsedURL url;
+    std::optional<std::string> ref;
+    std::optional<Hash> rev;
+
+    GitInput(const ParsedURL & url) : url(url)
+    {
+        type = "git";
+    }
+
+    bool operator ==(const Input & other) const override
+    {
+        auto other2 = dynamic_cast<const GitInput *>(&other);
+        return
+            other2
+            && url.url == other2->url.url
+            && rev == other2->rev
+            && ref == other2->ref;
+    }
+
+    bool isImmutable() const override
+    {
+        return (bool) rev;
+    }
+
+    std::optional<std::string> getRef() const override { return ref; }
+
+    std::optional<Hash> getRev() const override { return rev; }
+
+    std::string to_string() const override
+    {
+        ParsedURL url2(url);
+        if (rev) url2.query.insert_or_assign("rev", rev->gitRev());
+        if (ref) url2.query.insert_or_assign("ref", *ref);
+        return url2.to_string();
+    }
+
+    void clone(const Path & destDir) const override
+    {
+        auto [isLocal, actualUrl] = getActualUrl();
+
+        Strings args = {"clone"};
+
+        args.push_back(actualUrl);
+
+        if (ref) {
+            args.push_back("--branch");
+            args.push_back(*ref);
+        }
+
+        if (rev) throw Error("cloning a specific revision is not implemented");
+
+        args.push_back(destDir);
+
+        runProgram("git", true, args);
+    }
+
+    std::shared_ptr<const Input> applyOverrides(
+        std::optional<std::string> ref,
+        std::optional<Hash> rev) const override
+    {
+        if (!ref && !rev) return shared_from_this();
+
+        auto res = std::make_shared<GitInput>(*this);
+
+        if (ref) res->ref = ref;
+        if (rev) res->rev = rev;
+
+        if (!res->ref && res->rev)
+            throw Error("Git input '%s' has a commit hash but no branch/tag name", res->to_string());
+
+        return res;
+    }
+
+    std::optional<Path> getSourcePath() const
+    {
+        if (url.scheme == "git+file" && !ref && !rev)
+            return url.path;
+        return {};
+    }
+
+    std::pair<bool, std::string> getActualUrl() const
+    {
+        // Don't clone git+file:// URIs (but otherwise treat them the
+        // same as remote URIs, i.e. don't use the working tree or
+        // HEAD).
+        static bool forceHttp = getEnv("_NIX_FORCE_HTTP") == "1"; // for testing
+        bool isLocal = url.scheme == "git+file" && !forceHttp;
+        return {isLocal, isLocal ? url.path : std::string(url.base, 4)};
+    }
+
+    std::pair<Tree, std::shared_ptr<const Input>> fetchTreeInternal(nix::ref<Store> store) const override
+    {
+        auto name = "source";
+
+        auto input = std::make_shared<GitInput>(*this);
+
+        assert(!rev || rev->type == htSHA1);
+
+        if (rev) {
+            if (auto tree = lookupGitInfo(store, name, *rev))
+                return {std::move(*tree), input};
+        }
+
+        auto [isLocal, actualUrl] = getActualUrl();
+
+        // If this is a local directory and no ref or revision is
+        // given, then allow the use of an unclean working tree.
+        if (!input->ref && !input->rev && isLocal) {
+            bool clean = false;
+
+            /* Check whether this repo has any commits. There are
+               probably better ways to do this. */
+            bool haveCommits = !readDirectory(actualUrl + "/.git/refs/heads").empty();
+
+            try {
+                if (haveCommits) {
+                    runProgram("git", true, { "-C", actualUrl, "diff-index", "--quiet", "HEAD", "--" });
+                    clean = true;
+                }
+            } catch (ExecError & e) {
+                if (!WIFEXITED(e.status) || WEXITSTATUS(e.status) != 1) throw;
+            }
+
+            if (!clean) {
+
+                /* This is an unclean working tree. So copy all tracked files. */
+
+                if (!settings.allowDirty)
+                    throw Error("Git tree '%s' is dirty", actualUrl);
+
+                if (settings.warnDirty)
+                    warn("Git tree '%s' is dirty", actualUrl);
+
+                auto files = tokenizeString<std::set<std::string>>(
+                    runProgram("git", true, { "-C", actualUrl, "ls-files", "-z" }), "\0"s);
+
+                PathFilter filter = [&](const Path & p) -> bool {
+                    assert(hasPrefix(p, actualUrl));
+                    std::string file(p, actualUrl.size() + 1);
+
+                    auto st = lstat(p);
+
+                    if (S_ISDIR(st.st_mode)) {
+                        auto prefix = file + "/";
+                        auto i = files.lower_bound(prefix);
+                        return i != files.end() && hasPrefix(*i, prefix);
+                    }
+
+                    return files.count(file);
+                };
+
+                auto storePath = store->addToStore("source", actualUrl, true, htSHA256, filter);
+
+                auto tree = Tree {
+                    .actualPath = store->printStorePath(storePath),
+                    .storePath = std::move(storePath),
+                    .revCount = haveCommits ? std::stoull(runProgram("git", true, { "-C", actualUrl, "rev-list", "--count", "HEAD" })) : 0,
+                    // FIXME: maybe we should use the timestamp of the last
+                    // modified dirty file?
+                    .lastModified = haveCommits ? std::stoull(runProgram("git", true, { "-C", actualUrl, "log", "-1", "--format=%ct", "HEAD" })) : 0,
+                };
+
+                return {std::move(tree), input};
+            }
+        }
+
+        if (!input->ref) input->ref = isLocal ? "HEAD" : "master";
+
+        Path repoDir;
+
+        if (isLocal) {
+
+            if (!input->rev)
+                input->rev = Hash(chomp(runProgram("git", true, { "-C", actualUrl, "rev-parse", *input->ref })), htSHA1);
+
+            repoDir = actualUrl;
+
+        } else {
+
+            Path cacheDir = getCacheDir() + "/nix/gitv3/" + hashString(htSHA256, actualUrl).to_string(Base32, false);
+            repoDir = cacheDir;
+
+            if (!pathExists(cacheDir)) {
+                createDirs(dirOf(cacheDir));
+                runProgram("git", true, { "init", "--bare", repoDir });
+            }
+
+            Path localRefFile =
+                input->ref->compare(0, 5, "refs/") == 0
+                ? cacheDir + "/" + *input->ref
+                : cacheDir + "/refs/heads/" + *input->ref;
+
+            bool doFetch;
+            time_t now = time(0);
+
+            /* If a rev was specified, we need to fetch if it's not in the
+               repo. */
+            if (input->rev) {
+                try {
+                    runProgram("git", true, { "-C", repoDir, "cat-file", "-e", input->rev->gitRev() });
+                    doFetch = false;
+                } catch (ExecError & e) {
+                    if (WIFEXITED(e.status)) {
+                        doFetch = true;
+                    } else {
+                        throw;
+                    }
+                }
+            } else {
+                /* If the local ref is older than ‘tarball-ttl’ seconds, do a
+                   git fetch to update the local ref to the remote ref. */
+                struct stat st;
+                doFetch = stat(localRefFile.c_str(), &st) != 0 ||
+                    (uint64_t) st.st_mtime + settings.tarballTtl <= (uint64_t) now;
+            }
+
+            if (doFetch) {
+                Activity act(*logger, lvlTalkative, actUnknown, fmt("fetching Git repository '%s'", actualUrl));
+
+                // FIXME: git stderr messes up our progress indicator, so
+                // we're using --quiet for now. Should process its stderr.
+                try {
+                    runProgram("git", true, { "-C", repoDir, "fetch", "--quiet", "--force", "--", actualUrl, fmt("%s:%s", *input->ref, *input->ref) });
+                } catch (Error & e) {
+                    if (!pathExists(localRefFile)) throw;
+                    warn("could not update local clone of Git repository '%s'; continuing with the most recent version", actualUrl);
+                }
+
+                struct timeval times[2];
+                times[0].tv_sec = now;
+                times[0].tv_usec = 0;
+                times[1].tv_sec = now;
+                times[1].tv_usec = 0;
+
+                utimes(localRefFile.c_str(), times);
+            }
+
+            if (!input->rev)
+                input->rev = Hash(chomp(readFile(localRefFile)), htSHA1);
+        }
+
+        if (auto tree = lookupGitInfo(store, name, *input->rev))
+            return {std::move(*tree), input};
+
+        // FIXME: check whether rev is an ancestor of ref.
+
+        printTalkative("using revision %s of repo '%s'", input->rev->gitRev(), actualUrl);
+
+        // FIXME: should pipe this, or find some better way to extract a
+        // revision.
+        auto source = sinkToSource([&](Sink & sink) {
+            RunOptions gitOptions("git", { "-C", repoDir, "archive", input->rev->gitRev() });
+            gitOptions.standardOut = &sink;
+            runProgram2(gitOptions);
+        });
+
+        Path tmpDir = createTempDir();
+        AutoDelete delTmpDir(tmpDir, true);
+
+        unpackTarfile(*source, tmpDir);
+
+        auto storePath = store->addToStore(name, tmpDir);
+        auto revCount = std::stoull(runProgram("git", true, { "-C", repoDir, "rev-list", "--count", input->rev->gitRev() }));
+        auto lastModified = std::stoull(runProgram("git", true, { "-C", repoDir, "log", "-1", "--format=%ct", input->rev->gitRev() }));
+
+        auto tree = Tree {
+            .actualPath = store->toRealPath(store->printStorePath(storePath)),
+            .storePath = std::move(storePath),
+            .rev = input->rev,
+            .revCount = revCount,
+            .lastModified = lastModified,
+        };
+
+        cacheGitInfo(*store, name, tree);
+
+        return {std::move(tree), input};
+    }
+};
+
+struct GitInputScheme : InputScheme
+{
+    std::unique_ptr<Input> inputFromURL(const ParsedURL & url) override
+    {
+        if (url.scheme != "git" &&
+            url.scheme != "git+http" &&
+            url.scheme != "git+https" &&
+            url.scheme != "git+ssh" &&
+            url.scheme != "git+file") return nullptr;
+
+        auto input = std::make_unique<GitInput>(url);
+
+        for (auto &[name, value] : url.query) {
+            if (name == "rev") {
+                if (!std::regex_match(value, revRegex))
+                    throw BadURL("Git URL '%s' contains an invalid commit hash", url.url);
+                input->rev = Hash(value, htSHA1);
+            }
+            else if (name == "ref") {
+                if (!std::regex_match(value, refRegex))
+                    throw BadURL("Git URL '%s' contains an invalid branch/tag name", url.url);
+                input->ref = value;
+            }
+        }
+
+        return input;
+    }
+};
+
+static auto r1 = OnStartup([] { registerInputScheme(std::make_unique<GitInputScheme>()); });
+
+}

src/libstore/fetchers/github.cc

@@ -0,0 +1,183 @@
+#include "fetchers.hh"
+#include "download.hh"
+#include "globals.hh"
+#include "parse.hh"
+#include "regex.hh"
+#include "store-api.hh"
+
+#include <nlohmann/json.hpp>
+
+namespace nix::fetchers {
+
+std::regex ownerRegex("[a-zA-Z][a-zA-Z0-9_-]*", std::regex::ECMAScript);
+std::regex repoRegex("[a-zA-Z][a-zA-Z0-9_-]*", std::regex::ECMAScript);
+
+struct GitHubInput : Input
+{
+    std::string owner;
+    std::string repo;
+    std::optional<std::string> ref;
+    std::optional<Hash> rev;
+
+    bool operator ==(const Input & other) const override
+    {
+        auto other2 = dynamic_cast<const GitHubInput *>(&other);
+        return
+            other2
+            && owner == other2->owner
+            && repo == other2->repo
+            && rev == other2->rev
+            && ref == other2->ref;
+    }
+
+    bool isImmutable() const override
+    {
+        return (bool) rev;
+    }
+
+    std::optional<std::string> getRef() const override { return ref; }
+
+    std::optional<Hash> getRev() const override { return rev; }
+
+    std::string to_string() const override
+    {
+        auto s = fmt("github:%s/%s", owner, repo);
+        assert(!(ref && rev));
+        if (ref) s += "/" + *ref;
+        if (rev) s += "/" + rev->to_string(Base16, false);
+        return s;
+    }
+
+    void clone(const Path & destDir) const override
+    {
+        std::shared_ptr<const Input> input = inputFromURL(fmt("git+ssh://git@github.com/%s/%s.git", owner, repo));
+        input = input->applyOverrides(ref.value_or("master"), rev);
+        input->clone(destDir);
+    }
+
+    std::pair<Tree, std::shared_ptr<const Input>> fetchTreeInternal(nix::ref<Store> store) const override
+    {
+        auto rev = this->rev;
+
+    #if 0
+        if (rev) {
+            if (auto gitInfo = lookupGitInfo(store, "source", *rev))
+                return *gitInfo;
+        }
+    #endif
+
+        if (!rev) {
+            auto url = fmt("https://api.github.com/repos/%s/%s/commits/%s",
+                owner, repo, ref ? *ref : "master");
+            CachedDownloadRequest request(url);
+            request.ttl = rev ? 1000000000 : settings.tarballTtl;
+            auto result = getDownloader()->downloadCached(store, request);
+            auto json = nlohmann::json::parse(readFile(result.path));
+            rev = Hash(json["sha"], htSHA1);
+            debug("HEAD revision for '%s' is %s", url, rev->gitRev());
+        }
+
+        // FIXME: use regular /archive URLs instead? api.github.com
+        // might have stricter rate limits.
+
+        auto url = fmt("https://api.github.com/repos/%s/%s/tarball/%s",
+            owner, repo, rev->to_string(Base16, false));
+
+        std::string accessToken = settings.githubAccessToken.get();
+        if (accessToken != "")
+            url += "?access_token=" + accessToken;
+
+        CachedDownloadRequest request(url);
+        request.unpack = true;
+        request.name = "source";
+        request.ttl = 1000000000;
+        request.getLastModified = true;
+        auto dresult = getDownloader()->downloadCached(store, request);
+
+        assert(dresult.lastModified);
+
+        Tree result{
+            .actualPath = dresult.path,
+            .storePath = store->parseStorePath(dresult.storePath),
+            .rev = *rev,
+            .lastModified = *dresult.lastModified
+        };
+
+    #if 0
+        // FIXME: this can overwrite a cache file that contains a revCount.
+        cacheGitInfo("source", gitInfo);
+    #endif
+
+        auto input = std::make_shared<GitHubInput>(*this);
+        input->ref = {};
+        input->rev = *rev;
+
+        return {std::move(result), input};
+    }
+
+    std::shared_ptr<const Input> applyOverrides(
+        std::optional<std::string> ref,
+        std::optional<Hash> rev) const override
+    {
+        if (!ref && !rev) return shared_from_this();
+
+        auto res = std::make_shared<GitHubInput>(*this);
+
+        if (ref) res->ref = ref;
+        if (rev) res->rev = rev;
+
+        return res;
+    }
+};
+
+struct GitHubInputScheme : InputScheme
+{
+    std::unique_ptr<Input> inputFromURL(const ParsedURL & url) override
+    {
+        if (url.scheme != "github") return nullptr;
+
+        auto path = tokenizeString<std::vector<std::string>>(url.path, "/");
+        auto input = std::make_unique<GitHubInput>();
+        input->type = "github";
+
+        if (path.size() == 2) {
+        } else if (path.size() == 3) {
+            if (std::regex_match(path[2], revRegex))
+                input->rev = Hash(path[2], htSHA1);
+            else if (std::regex_match(path[2], refRegex))
+                input->ref = path[2];
+            else
+                throw BadURL("in GitHub URL '%s', '%s' is not a commit hash or branch/tag name", url.url, path[2]);
+        } else
+            throw BadURL("GitHub URL '%s' is invalid", url.url);
+
+        for (auto &[name, value] : url.query) {
+            if (name == "rev") {
+                if (!std::regex_match(value, revRegex))
+                    throw BadURL("GitHub URL '%s' contains an invalid commit hash", url.url);
+                if (input->rev)
+                    throw BadURL("GitHub URL '%s' contains multiple commit hashes", url.url);
+                input->rev = Hash(value, htSHA1);
+            }
+            else if (name == "ref") {
+                if (!std::regex_match(value, refRegex))
+                    throw BadURL("GitHub URL '%s' contains an invalid branch/tag name", url.url);
+                if (input->ref)
+                    throw BadURL("GitHub URL '%s' contains multiple branch/tag names", url.url);
+                input->ref = value;
+            }
+        }
+
+        if (input->ref && input->rev)
+            throw BadURL("GitHub URL '%s' contains both a commit hash and a branch/tag name", url.url);
+
+        input->owner = path[0];
+        input->repo = path[1];
+
+        return input;
+    }
+};
+
+static auto r1 = OnStartup([] { registerInputScheme(std::make_unique<GitHubInputScheme>()); });
+
+}

src/libstore/fetchers/indirect.cc

@@ -0,0 +1,114 @@
+#include "fetchers.hh"
+#include "parse.hh"
+#include "regex.hh"
+
+namespace nix::fetchers {
+
+std::regex flakeRegex("[a-zA-Z][a-zA-Z0-9_-]*", std::regex::ECMAScript);
+
+struct IndirectInput : Input
+{
+    std::string id;
+    std::optional<Hash> rev;
+    std::optional<std::string> ref;
+
+    bool operator ==(const Input & other) const override
+    {
+        auto other2 = dynamic_cast<const IndirectInput *>(&other);
+        return
+            other2
+            && id == other2->id
+            && rev == other2->rev
+            && ref == other2->ref;
+    }
+
+    bool isDirect() const override
+    {
+        return false;
+    }
+
+    std::optional<std::string> getRef() const override { return ref; }
+
+    std::optional<Hash> getRev() const override { return rev; }
+
+    bool contains(const Input & other) const override
+    {
+        auto other2 = dynamic_cast<const IndirectInput *>(&other);
+        return
+            other2
+            && id == other2->id
+            && (!ref || ref == other2->ref)
+            && (!rev || rev == other2->rev);
+    }
+
+    std::string to_string() const override
+    {
+        ParsedURL url;
+        url.scheme = "flake";
+        url.path = id;
+        if (ref) { url.path += '/'; url.path += *ref; };
+        if (rev) { url.path += '/'; url.path += rev->gitRev(); };
+        return url.to_string();
+    }
+
+    std::shared_ptr<const Input> applyOverrides(
+        std::optional<std::string> ref,
+        std::optional<Hash> rev) const override
+    {
+        if (!ref && !rev) return shared_from_this();
+
+        auto res = std::make_shared<IndirectInput>(*this);
+
+        if (ref) res->ref = ref;
+        if (rev) res->rev = rev;
+
+        return res;
+    }
+
+    std::pair<Tree, std::shared_ptr<const Input>> fetchTreeInternal(nix::ref<Store> store) const override
+    {
+        throw Error("indirect input '%s' cannot be fetched directly", to_string());
+    }
+};
+
+struct IndirectInputScheme : InputScheme
+{
+    std::unique_ptr<Input> inputFromURL(const ParsedURL & url) override
+    {
+        if (url.scheme != "flake") return nullptr;
+
+        auto path = tokenizeString<std::vector<std::string>>(url.path, "/");
+        auto input = std::make_unique<IndirectInput>();
+        input->type = "indirect";
+
+        if (path.size() == 1) {
+        } else if (path.size() == 2) {
+            if (std::regex_match(path[1], revRegex))
+                input->rev = Hash(path[1], htSHA1);
+            else if (std::regex_match(path[1], refRegex))
+                input->ref = path[1];
+            else
+                throw BadURL("in flake URL '%s', '%s' is not a commit hash or branch/tag name", url.url, path[1]);
+        } else if (path.size() == 3) {
+            if (!std::regex_match(path[1], refRegex))
+                throw BadURL("in flake URL '%s', '%s' is not a branch/tag name", url.url, path[1]);
+            input->ref = path[1];
+            if (!std::regex_match(path[2], revRegex))
+                throw BadURL("in flake URL '%s', '%s' is not a commit hash", url.url, path[2]);
+            input->rev = Hash(path[2], htSHA1);
+        } else
+            throw BadURL("GitHub URL '%s' is invalid", url.url);
+
+        // FIXME: forbid query params?
+
+        input->id = path[0];
+        if (!std::regex_match(input->id, flakeRegex))
+            throw BadURL("'%s' is not a valid flake ID", input->id);
+
+        return input;
+    }
+};
+
+static auto r1 = OnStartup([] { registerInputScheme(std::make_unique<IndirectInputScheme>()); });
+
+}

src/libstore/fetchers/parse.cc

@@ -0,0 +1,129 @@
+#include "parse.hh"
+#include "util.hh"
+#include "regex.hh"
+
+namespace nix::fetchers {
+
+std::regex refRegex(refRegexS, std::regex::ECMAScript);
+std::regex revRegex(revRegexS, std::regex::ECMAScript);
+
+ParsedURL parseURL(const std::string & url)
+{
+    static std::regex uriRegex(
+        "(((" + schemeRegex + "):"
+        + "(//(" + authorityRegex + "))?"
+        + "(" + pathRegex + "))"
+        + "(?:\\?(" + queryRegex + "))?"
+        + "(?:#(" + queryRegex + "))?"
+        + ")",
+        std::regex::ECMAScript);
+
+    std::smatch match;
+
+    if (std::regex_match(url, match, uriRegex)) {
+        auto & base = match[2];
+        std::string scheme = match[3];
+        auto authority = match[4].matched
+            ? std::optional<std::string>(match[5]) : std::nullopt;
+        std::string path = match[6];
+        auto & query = match[7];
+        auto & fragment = match[8];
+
+        auto isFile = scheme.find("file") != std::string::npos;
+
+        if (authority && *authority != "" && isFile)
+            throw Error("file:// URL '%s' has unexpected authority '%s'",
+                url, *authority);
+
+        if (isFile && path.empty())
+            path = "/";
+
+        return ParsedURL{
+            .url = url,
+            .base = base,
+            .scheme = scheme,
+            .authority = authority,
+            .path = path,
+            .query = decodeQuery(query),
+            .fragment = percentDecode(std::string(fragment))
+        };
+    }
+
+    else
+        throw BadURL("'%s' is not a valid URL", url);
+}
+
+std::string percentDecode(std::string_view in)
+{
+    std::string decoded;
+    for (size_t i = 0; i < in.size(); ) {
+        if (in[i] == '%') {
+            if (i + 2 >= in.size())
+                throw BadURL("invalid URI parameter '%s'", in);
+            try {
+                decoded += std::stoul(std::string(in, i + 1, 2), 0, 16);
+                i += 3;
+            } catch (...) {
+                throw BadURL("invalid URI parameter '%s'", in);
+            }
+        } else
+            decoded += in[i++];
+    }
+    return decoded;
+}
+
+std::map<std::string, std::string> decodeQuery(const std::string & query)
+{
+    std::map<std::string, std::string> result;
+
+    for (auto s : tokenizeString<Strings>(query, "&")) {
+        auto e = s.find('=');
+        if (e != std::string::npos)
+            result.emplace(
+                s.substr(0, e),
+                percentDecode(std::string_view(s).substr(e + 1)));
+    }
+
+    return result;
+}
+
+std::string percentEncode(std::string_view s)
+{
+    std::string res;
+    for (auto & c : s)
+        if ((c >= 'a' && c <= 'z')
+            || (c >= 'A' && c <= 'Z')
+            || (c >= '0' && c <= '9')
+            || strchr("-._~!$&'()*+,;=:@", c))
+            res += c;
+        else
+            res += fmt("%%%02x", (unsigned int) c);
+    return res;
+}
+
+std::string encodeQuery(const std::map<std::string, std::string> & ss)
+{
+    std::string res;
+    bool first = true;
+    for (auto & [name, value] : ss) {
+        if (!first) res += '&';
+        first = false;
+        res += percentEncode(name);
+        res += '=';
+        res += percentEncode(value);
+    }
+    return res;
+}
+
+std::string ParsedURL::to_string() const
+{
+    return
+        scheme
+        + ":"
+        + (authority ? "//" + *authority : "")
+        + path
+        + (query.empty() ? "" : "?" + encodeQuery(query))
+        + (fragment.empty() ? "" : "#" + percentEncode(fragment));
+}
+
+}

src/libstore/fetchers/parse.hh

@@ -0,0 +1,28 @@
+#pragma once
+
+#include "types.hh"
+
+namespace nix::fetchers {
+
+struct ParsedURL
+{
+    std::string url;
+    std::string base; // URL without query/fragment
+    std::string scheme;
+    std::optional<std::string> authority;
+    std::string path;
+    std::map<std::string, std::string> query;
+    std::string fragment;
+
+    std::string to_string() const;
+};
+
+MakeError(BadURL, Error);
+
+std::string percentDecode(std::string_view in);
+
+std::map<std::string, std::string> decodeQuery(const std::string & query);
+
+ParsedURL parseURL(const std::string & url);
+
+}

src/libstore/fetchers/regex.hh

@@ -0,0 +1,32 @@
+#pragma once
+
+#include <regex>
+
+namespace nix::fetchers {
+
+// URI stuff.
+const static std::string pctEncoded = "%[0-9a-fA-F][0-9a-fA-F]";
+const static std::string schemeRegex = "[a-z+]+";
+const static std::string authorityRegex =
+    "(?:(?:[a-z])*@)?"
+    "[a-zA-Z0-9._~-]*";
+const static std::string segmentRegex = "[a-zA-Z0-9._~-]+";
+const static std::string pathRegex = "(?:/?" + segmentRegex + "(?:/" + segmentRegex + ")*|/?)";
+const static std::string pcharRegex =
+    "(?:[a-zA-Z0-9-._~!$&'()*+,;=:@ ]|" + pctEncoded + ")";
+const static std::string queryRegex = "(?:" + pcharRegex + "|[/?])*";
+
+// A Git ref (i.e. branch or tag name).
+const static std::string refRegexS = "[a-zA-Z0-9][a-zA-Z0-9_.-]*"; // FIXME: check
+extern std::regex refRegex;
+
+// A Git revision (a SHA-1 commit hash).
+const static std::string revRegexS = "[0-9a-fA-F]{40}";
+extern std::regex revRegex;
+
+// A ref or revision, or a ref followed by a revision.
+const static std::string refAndOrRevRegex = "(?:(" + revRegexS + ")|(?:(" + refRegexS + ")(?:/(" + revRegexS + "))?))";
+
+const static std::string flakeId = "[a-zA-Z][a-zA-Z0-9_-]*";
+
+}

src/libstore/fetchers/registry.cc

@@ -0,0 +1,145 @@
+#include "registry.hh"
+#include "util.hh"
+#include "fetchers.hh"
+#include "globals.hh"
+#include "download.hh"
+
+#include <nlohmann/json.hpp>
+
+namespace nix::fetchers {
+
+std::shared_ptr<Registry> Registry::read(
+    const Path & path, RegistryType type)
+{
+    auto registry = std::make_shared<Registry>();
+    registry->type = type;
+
+    if (!pathExists(path))
+        return std::make_shared<Registry>();
+
+    auto json = nlohmann::json::parse(readFile(path));
+
+    auto version = json.value("version", 0);
+    if (version != 1)
+        throw Error("flake registry '%s' has unsupported version %d", path, version);
+
+    auto flakes = json["flakes"];
+    for (auto i = flakes.begin(); i != flakes.end(); ++i) {
+        // FIXME: remove 'uri' soon.
+        auto url = i->value("url", i->value("uri", ""));
+        if (url.empty())
+            throw Error("flake registry '%s' lacks a 'url' attribute for entry '%s'",
+                path, i.key());
+        registry->entries.push_back(
+            {inputFromURL(i.key()), inputFromURL(url)});
+    }
+
+    return registry;
+}
+
+void Registry::write(const Path & path)
+{
+    nlohmann::json json;
+    json["version"] = 1;
+    for (auto & elem : entries)
+        json["flakes"][elem.first->to_string()] = { {"url", elem.second->to_string()} };
+    createDirs(dirOf(path));
+    writeFile(path, json.dump(4));
+}
+
+void Registry::add(
+    const std::shared_ptr<const Input> & from,
+    const std::shared_ptr<const Input> & to)
+{
+    entries.emplace_back(from, to);
+}
+
+void Registry::remove(const std::shared_ptr<const Input> & input)
+{
+    // FIXME: use C++20 std::erase.
+    for (auto i = entries.begin(); i != entries.end(); )
+        if (*i->first == *input)
+            i = entries.erase(i);
+        else
+            ++i;
+}
+
+Path getUserRegistryPath()
+{
+    return getHome() + "/.config/nix/registry.json";
+}
+
+std::shared_ptr<Registry> getUserRegistry()
+{
+    return Registry::read(getUserRegistryPath(), Registry::User);
+}
+
+#if 0
+std::shared_ptr<Registry> getFlagRegistry(RegistryOverrides registryOverrides)
+{
+    auto flagRegistry = std::make_shared<Registry>();
+    for (auto const & x : registryOverrides)
+        flagRegistry->entries.insert_or_assign(
+            parseFlakeRef2(x.first),
+            parseFlakeRef2(x.second));
+    return flagRegistry;
+}
+#endif
+
+static std::shared_ptr<Registry> getGlobalRegistry(ref<Store> store)
+{
+    static auto reg = [&]() {
+        auto path = settings.flakeRegistry;
+
+        if (!hasPrefix(path, "/")) {
+            CachedDownloadRequest request(path);
+            request.name = "flake-registry.json";
+            request.gcRoot = true;
+            path = getDownloader()->downloadCached(store, request).path;
+        }
+
+        return Registry::read(path, Registry::Global);
+    }();
+
+    return reg;
+}
+
+Registries getRegistries(ref<Store> store)
+{
+    Registries registries;
+    //registries.push_back(getFlagRegistry(registryOverrides));
+    registries.push_back(getUserRegistry());
+    registries.push_back(getGlobalRegistry(store));
+    return registries;
+}
+
+std::shared_ptr<const Input> lookupInRegistries(
+    ref<Store> store,
+    std::shared_ptr<const Input> input)
+{
+    int n = 0;
+
+ restart:
+
+    n++;
+    if (n > 100) throw Error("cycle detected in flake registr for '%s'", input);
+
+    for (auto & registry : getRegistries(store)) {
+        // FIXME: O(n)
+        for (auto & entry : registry->entries) {
+            if (entry.first->contains(*input)) {
+                input = entry.second->applyOverrides(
+                    !entry.first->getRef() && input->getRef() ? input->getRef() : std::optional<std::string>(),
+                    !entry.first->getRev() && input->getRev() ? input->getRev() : std::optional<Hash>());
+                goto restart;
+            }
+        }
+    }
+
+    if (!input->isDirect())
+        throw Error("cannot find flake '%s' in the flake registries", input->to_string());
+
+    return input;
+}
+
+}

src/libstore/fetchers/registry.hh

@@ -0,0 +1,47 @@
+#pragma once
+
+#include "types.hh"
+
+namespace nix { class Store; }
+
+namespace nix::fetchers {
+
+struct Input;
+
+struct Registry
+{
+    enum RegistryType {
+        Flag = 0,
+        User = 1,
+        Global = 2,
+    };
+
+    RegistryType type;
+
+    std::vector<std::pair<std::shared_ptr<const Input>, std::shared_ptr<const Input>>> entries;
+
+    static std::shared_ptr<Registry> read(
+        const Path & path, RegistryType type);
+
+    void write(const Path & path);
+
+    void add(
+        const std::shared_ptr<const Input> & from,
+        const std::shared_ptr<const Input> & to);
+
+    void remove(const std::shared_ptr<const Input> & input);
+};
+
+typedef std::vector<std::shared_ptr<Registry>> Registries;
+
+std::shared_ptr<Registry> getUserRegistry();
+
+Path getUserRegistryPath();
+
+Registries getRegistries(ref<Store> store);
+
+std::shared_ptr<const Input> lookupInRegistries(
+    ref<Store> store,
+    std::shared_ptr<const Input> input);
+
+}

src/libstore/globals.hh

@@ -365,6 +365,15 @@ public:
     bool isExperimentalFeatureEnabled(const std::string & name);
 
     void requireExperimentalFeature(const std::string & name);
+
+    Setting<std::string> flakeRegistry{this, "https://github.com/NixOS/flake-registry/raw/master/flake-registry.json", "flake-registry",
+        "Path or URI of the global flake registry."};
+
+    Setting<bool> allowDirty{this, true, "allow-dirty",
+        "Whether to allow dirty Git/Mercurial trees."};
+
+    Setting<bool> warnDirty{this, true, "warn-dirty",
+        "Whether to warn about dirty Git/Mercurial trees."};
 };
 
 

src/libstore/local.mk

@@ -4,7 +4,7 @@ libstore_NAME = libnixstore
 
 libstore_DIR := $(d)
 
-libstore_SOURCES := $(wildcard $(d)/*.cc $(d)/builtins/*.cc)
+libstore_SOURCES := $(wildcard $(d)/*.cc $(d)/builtins/*.cc $(d)/fetchers/*.cc)
 
 libstore_LIBS = libutil libnixrust
 

src/libstore/store-api.cc

@@ -6,6 +6,7 @@
 #include "thread-pool.hh"
 #include "json.hh"
 #include "derivations.hh"
+#include "fetchers/parse.hh"
 
 #include <future>
 
@@ -864,27 +865,7 @@ std::pair<std::string, Store::Params> splitUriAndParams(const std::string & uri_
     Store::Params params;
     auto q = uri.find('?');
     if (q != std::string::npos) {
-        for (auto s : tokenizeString<Strings>(uri.substr(q + 1), "&")) {
+        params = fetchers::decodeQuery(uri.substr(q + 1));
-            auto e = s.find('=');
-            if (e != std::string::npos) {
-                auto value = s.substr(e + 1);
-                std::string decoded;
-                for (size_t i = 0; i < value.size(); ) {
-                    if (value[i] == '%') {
-                        if (i + 2 >= value.size())
-                            throw Error("invalid URI parameter '%s'", value);
-                        try {
-                            decoded += std::stoul(std::string(value, i + 1, 2), 0, 16);
-                            i += 3;
-                        } catch (...) {
-                            throw Error("invalid URI parameter '%s'", value);
-                        }
-                    } else
-                        decoded += value[i++];
-                }
-                params[s.substr(0, e)] = decoded;
-            }
-        }
         uri = uri_.substr(0, q);
     }
     return {uri, params};

src/libutil/types.hh

@@ -157,4 +157,12 @@ typedef list<Path> Paths;
 typedef set<Path> PathSet;
 
 
+/* Helper class to run code at startup. */
+template<typename T>
+struct OnStartup
+{
+    OnStartup(T && t) { t(); }
+};
+
+
 }

src/libutil/util.cc

@@ -98,7 +98,7 @@ void replaceEnv(std::map<std::string, std::string> newEnv)
 }
 
 
-Path absPath(Path path, std::optional<Path> dir)
+Path absPath(Path path, std::optional<Path> dir, bool resolveSymlinks)
 {
     if (path[0] != '/') {
         if (!dir) {
@@ -119,7 +119,7 @@ Path absPath(Path path, std::optional<Path> dir)
         }
         path = *dir + "/" + path;
     }
-    return canonPath(path);
+    return canonPath(path, resolveSymlinks);
 }
 
 

src/libutil/util.hh

@@ -46,7 +46,9 @@ void clearEnv();
 /* Return an absolutized path, resolving paths relative to the
    specified directory, or the current directory otherwise.  The path
    is also canonicalised. */
-Path absPath(Path path, std::optional<Path> dir = {});
+Path absPath(Path path,
+    std::optional<Path> dir = {},
+    bool resolveSymlinks = false);
 
 /* Canonicalise a path by removing all `.' or `..' components and
    double or trailing slashes.  Optionally resolves all symlink

src/nix/flake.cc

@@ -9,6 +9,8 @@
 #include "store-api.hh"
 #include "derivations.hh"
 #include "attr-path.hh"
+#include "fetchers/fetchers.hh"
+#include "fetchers/registry.hh"
 
 #include <nlohmann/json.hpp>
 #include <queue>
@@ -30,10 +32,7 @@ public:
 
     FlakeRef getFlakeRef()
     {
-        if (flakeUrl.find('/') != std::string::npos || flakeUrl == ".")
+        return parseFlakeRef(flakeUrl, absPath(".")); //FIXME
-            return FlakeRef(flakeUrl, true);
-        else
-            return FlakeRef(flakeUrl);
     }
 
     Flake getFlake()
@@ -57,63 +56,54 @@ struct CmdFlakeList : EvalCommand
 
     void run(nix::ref<nix::Store> store) override
     {
-        auto registries = getEvalState()->getFlakeRegistries();
+        using namespace fetchers;
+
+        auto registries = getRegistries(store);
 
         stopProgressBar();
 
-        for (auto & entry : registries[FLAG_REGISTRY]->entries)
+        for (auto & registry : registries) {
-            std::cout << entry.first.to_string() << " flags " << entry.second.to_string() << "\n";
+            for (auto & entry : registry->entries) {
-
+                // FIXME: format nicely
-        for (auto & entry : registries[USER_REGISTRY]->entries)
+                std::cout << fmt("%s %s %s\n",
-            std::cout << entry.first.to_string() << " user " << entry.second.to_string() << "\n";
+                    registry->type == Registry::Flag ? "flags " :
-
+                    registry->type == Registry::User ? "user  " :
-        for (auto & entry : registries[GLOBAL_REGISTRY]->entries)
+                    "global",
-            std::cout << entry.first.to_string() << " global " << entry.second.to_string() << "\n";
+                    entry.first->to_string(),
+                    entry.second->to_string());
+            }
+        }
     }
 };
 
-static void printSourceInfo(const SourceInfo & sourceInfo)
+static void printFlakeInfo(const Store & store, const Flake & flake)
 {
-    std::cout << fmt("URL:           %s\n", sourceInfo.resolvedRef.to_string());
+    std::cout << fmt("URL:           %s\n", flake.resolvedRef.input->to_string());
-    if (sourceInfo.resolvedRef.ref)
-        std::cout << fmt("Branch:        %s\n",*sourceInfo.resolvedRef.ref);
-    if (sourceInfo.resolvedRef.rev)
-        std::cout << fmt("Revision:      %s\n", sourceInfo.resolvedRef.rev->to_string(Base16, false));
-    if (sourceInfo.revCount)
-        std::cout << fmt("Revisions:     %s\n", *sourceInfo.revCount);
-    if (sourceInfo.lastModified)
-        std::cout << fmt("Last modified: %s\n",
-            std::put_time(std::localtime(&*sourceInfo.lastModified), "%F %T"));
-    std::cout << fmt("Path:          %s\n", sourceInfo.storePath);
-}
-
-static void sourceInfoToJson(const SourceInfo & sourceInfo, nlohmann::json & j)
-{
-    j["url"] = sourceInfo.resolvedRef.to_string();
-    if (sourceInfo.resolvedRef.ref)
-        j["branch"] = *sourceInfo.resolvedRef.ref;
-    if (sourceInfo.resolvedRef.rev)
-        j["revision"] = sourceInfo.resolvedRef.rev->to_string(Base16, false);
-    if (sourceInfo.revCount)
-        j["revCount"] = *sourceInfo.revCount;
-    if (sourceInfo.lastModified)
-        j["lastModified"] = *sourceInfo.lastModified;
-    j["path"] = sourceInfo.storePath;
-}
-
-static void printFlakeInfo(const Flake & flake)
-{
-    std::cout << fmt("Description:   %s\n", flake.description);
     std::cout << fmt("Edition:       %s\n", flake.edition);
-    printSourceInfo(flake.sourceInfo);
+    std::cout << fmt("Description:   %s\n", flake.description);
+    std::cout << fmt("Path:          %s\n", store.printStorePath(flake.sourceInfo->storePath));
+    if (flake.sourceInfo->rev)
+        std::cout << fmt("Revision:      %s\n", flake.sourceInfo->rev->to_string(Base16, false));
+    if (flake.sourceInfo->revCount)
+        std::cout << fmt("Revisions:     %s\n", *flake.sourceInfo->revCount);
+    if (flake.sourceInfo->lastModified)
+        std::cout << fmt("Last modified: %s\n",
+            std::put_time(std::localtime(&*flake.sourceInfo->lastModified), "%F %T"));
 }
 
-static nlohmann::json flakeToJson(const Flake & flake)
+static nlohmann::json flakeToJson(const Store & store, const Flake & flake)
 {
     nlohmann::json j;
     j["description"] = flake.description;
     j["edition"] = flake.edition;
-    sourceInfoToJson(flake.sourceInfo, j);
+    j["url"] = flake.resolvedRef.input->to_string();
+    if (flake.sourceInfo->rev)
+        j["revision"] = flake.sourceInfo->rev->to_string(Base16, false);
+    if (flake.sourceInfo->revCount)
+        j["revCount"] = *flake.sourceInfo->revCount;
+    if (flake.sourceInfo->lastModified)
+        j["lastModified"] = *flake.sourceInfo->lastModified;
+    j["path"] = store.printStorePath(flake.sourceInfo->storePath);
     return j;
 }
 
@@ -140,7 +130,7 @@ struct CmdFlakeDeps : FlakeCommand
             todo.pop();
 
             for (auto & info : resFlake.flakeDeps) {
-                printFlakeInfo(info.second.flake);
+                printFlakeInfo(*store, info.second.flake);
                 todo.push(info.second);
             }
         }
@@ -161,10 +151,12 @@ struct CmdFlakeUpdate : FlakeCommand
 
         auto flakeRef = getFlakeRef();
 
+#if 0
         if (std::get_if<FlakeRef::IsPath>(&flakeRef.data))
             updateLockFile(*evalState, flakeRef, true);
         else
             throw Error("cannot update lockfile of flake '%s'", flakeRef);
+#endif
     }
 };
 
@@ -195,7 +187,7 @@ struct CmdFlakeInfo : FlakeCommand, MixJSON
             auto state = getEvalState();
             auto flake = resolveFlake();
 
-            auto json = flakeToJson(flake.flake);
+            auto json = flakeToJson(*store, flake.flake);
 
             auto vFlake = state->allocValue();
             flake::callFlake(*state, flake, *vFlake);
@@ -222,7 +214,7 @@ struct CmdFlakeInfo : FlakeCommand, MixJSON
         } else {
             auto flake = getFlake();
             stopProgressBar();
-            printFlakeInfo(flake);
+            printFlakeInfo(*store, flake);
         }
     }
 };
@@ -495,8 +487,7 @@ struct CmdFlakeCheck : FlakeCommand, MixJSON
 
 struct CmdFlakeAdd : MixEvalArgs, Command
 {
-    FlakeUri alias;
+    std::string fromUrl, toUrl;
-    FlakeUri url;
 
     std::string description() override
     {
@@ -505,24 +496,24 @@ struct CmdFlakeAdd : MixEvalArgs, Command
 
     CmdFlakeAdd()
     {
-        expectArg("alias", &alias);
+        expectArg("from-url", &fromUrl);
-        expectArg("flake-url", &url);
+        expectArg("to-url", &toUrl);
     }
 
     void run() override
     {
-        FlakeRef aliasRef(alias);
+        auto fromRef = parseFlakeRef(fromUrl);
-        Path userRegistryPath = getUserRegistryPath();
+        auto toRef = parseFlakeRef(toUrl);
-        auto userRegistry = readRegistry(userRegistryPath);
+        auto userRegistry = fetchers::getUserRegistry();
-        userRegistry->entries.erase(aliasRef);
+        userRegistry->remove(fromRef.input);
-        userRegistry->entries.insert_or_assign(aliasRef, FlakeRef(url));
+        userRegistry->add(fromRef.input, toRef.input);
-        writeRegistry(*userRegistry, userRegistryPath);
+        userRegistry->write(fetchers::getUserRegistryPath());
     }
 };
 
 struct CmdFlakeRemove : virtual Args, MixEvalArgs, Command
 {
-    FlakeUri alias;
+    std::string url;
 
     std::string description() override
     {
@@ -531,52 +522,38 @@ struct CmdFlakeRemove : virtual Args, MixEvalArgs, Command
 
     CmdFlakeRemove()
     {
-        expectArg("alias", &alias);
+        expectArg("url", &url);
     }
 
     void run() override
     {
-        Path userRegistryPath = getUserRegistryPath();
+        auto userRegistry = fetchers::getUserRegistry();
-        auto userRegistry = readRegistry(userRegistryPath);
+        userRegistry->remove(parseFlakeRef(url).input);
-        userRegistry->entries.erase(FlakeRef(alias));
+        userRegistry->write(fetchers::getUserRegistryPath());
-        writeRegistry(*userRegistry, userRegistryPath);
     }
 };
 
 struct CmdFlakePin : virtual Args, EvalCommand
 {
-    FlakeUri alias;
+    std::string url;
 
     std::string description() override
     {
-        return "pin flake require in user flake registry";
+        return "pin a flake to its current version in user flake registry";
     }
 
     CmdFlakePin()
     {
-        expectArg("alias", &alias);
+        expectArg("url", &url);
     }
 
     void run(nix::ref<nix::Store> store) override
     {
-        auto evalState = getEvalState();
+        auto ref = parseFlakeRef(url);
-
+        auto userRegistry = fetchers::getUserRegistry();
-        Path userRegistryPath = getUserRegistryPath();
+        userRegistry->remove(ref.input);
-        FlakeRegistry userRegistry = *readRegistry(userRegistryPath);
+        auto [tree, resolved] = ref.resolve(store).input->fetchTree(store);
-        auto it = userRegistry.entries.find(FlakeRef(alias));
+        userRegistry->add(ref.input, resolved);
-        if (it != userRegistry.entries.end()) {
-            it->second = getFlake(*evalState, it->second, true).sourceInfo.resolvedRef;
-            writeRegistry(userRegistry, userRegistryPath);
-        } else {
-            std::shared_ptr<FlakeRegistry> globalReg = evalState->getGlobalFlakeRegistry();
-            it = globalReg->entries.find(FlakeRef(alias));
-            if (it != globalReg->entries.end()) {
-                auto newRef = getFlake(*evalState, it->second, true).sourceInfo.resolvedRef;
-                userRegistry.entries.insert_or_assign(alias, newRef);
-                writeRegistry(userRegistry, userRegistryPath);
-            } else
-                throw Error("the flake alias '%s' does not exist in the user or global registry", alias);
-        }
     }
 };
 
@@ -616,15 +593,20 @@ struct CmdFlakeClone : FlakeCommand
 
     CmdFlakeClone()
     {
-        expectArg("dest-dir", &destDir, true);
+        mkFlag()
+            .shortName('f')
+            .longName("dest")
+            .label("path")
+            .description("destination path")
+            .dest(&destDir);
     }
 
     void run(nix::ref<nix::Store> store) override
     {
-        auto evalState = getEvalState();
+        if (destDir.empty())
+            throw Error("missing flag '--dest'");
 
-        Registries registries = evalState->getFlakeRegistries();
+        getFlakeRef().resolve(store).input->clone(destDir);
-        gitCloneFlake(getFlakeRef().to_string(), *evalState, registries, destDir);
     }
 };
 

src/nix/installables.cc

@@ -10,6 +10,7 @@
 #include "shared.hh"
 #include "flake/flake.hh"
 #include "flake/eval-cache.hh"
+#include "fetchers/parse.hh"
 
 #include <regex>
 #include <queue>
@@ -80,10 +81,8 @@ Strings SourceExprCommand::getDefaultFlakeAttrPathPrefixes()
 
 ref<EvalState> EvalCommand::getEvalState()
 {
-    if (!evalState) {
+    if (!evalState)
         evalState = std::make_shared<EvalState>(searchPath, getStore());
-        evalState->addRegistryOverrides(registryOverrides);
-    }
     return ref<EvalState>(evalState);
 }
 
@@ -243,6 +242,7 @@ void makeFlakeClosureGCRoot(Store & store,
     const FlakeRef & origFlakeRef,
     const flake::ResolvedFlake & resFlake)
 {
+#if 0
     if (std::get_if<FlakeRef::IsPath>(&origFlakeRef.data)) return;
 
     /* Get the store paths of all non-local flakes. */
@@ -285,6 +285,7 @@ void makeFlakeClosureGCRoot(Store & store,
     debug("writing GC root '%s' for flake closure of '%s'", symlink, origFlakeRef);
     replaceSymlink(store.printStorePath(closurePath), symlink);
     store.addIndirectRoot(symlink);
+#endif
 }
 
 std::vector<std::string> InstallableFlake::getActualAttrPaths()
@@ -334,7 +335,7 @@ std::tuple<std::string, FlakeRef, flake::EvalCache::Derivation> InstallableFlake
         auto drv = evalCache.getDerivation(fingerprint, attrPath);
         if (drv) {
             if (state->store->isValidPath(drv->drvPath))
-                return {attrPath, resFlake.flake.sourceInfo.resolvedRef, std::move(*drv)};
+                return {attrPath, resFlake.flake.resolvedRef, std::move(*drv)};
         }
 
         if (!vOutputs)
@@ -356,7 +357,7 @@ std::tuple<std::string, FlakeRef, flake::EvalCache::Derivation> InstallableFlake
 
             evalCache.addDerivation(fingerprint, attrPath, drv);
 
-            return {attrPath, resFlake.flake.sourceInfo.resolvedRef, std::move(drv)};
+            return {attrPath, resFlake.flake.resolvedRef, std::move(drv)};
         } catch (AttrPathNotFound & e) {
         }
     }
@@ -440,27 +441,23 @@ std::vector<std::shared_ptr<Installable>> SourceExprCommand::parseInstallables(
             if (hasPrefix(s, "nixpkgs.")) {
                 bool static warned;
                 warnOnce(warned, "the syntax 'nixpkgs.<attr>' is deprecated; use 'nixpkgs:<attr>' instead");
-                result.push_back(std::make_shared<InstallableFlake>(*this, FlakeRef("nixpkgs"),
+                result.push_back(std::make_shared<InstallableFlake>(*this, parseFlakeRef("flake:nixpkgs"),
-                        Strings{"legacyPackages." + settings.thisSystem.get() + "." + std::string(s, 8)}));
+                        Strings{"legacyPackages." + settings.thisSystem.get() + "." + std::string(s, 8)}, Strings{}));
             }
 
-            else if ((hash = s.rfind('#')) != std::string::npos)
-                result.push_back(std::make_shared<InstallableFlake>(
-                        *this,
-                        FlakeRef(std::string(s, 0, hash), true),
-                        std::string(s, hash + 1),
-                        getDefaultFlakeAttrPathPrefixes()));
-
             else {
-                try {
+                auto res = maybeParseFlakeRefWithFragment(s, absPath("."));
-                    auto flakeRef = FlakeRef(s, true);
+                if (res) {
+                    auto &[flakeRef, fragment] = *res;
                     result.push_back(std::make_shared<InstallableFlake>(
-                            *this, std::move(flakeRef), getDefaultFlakeAttrPaths()));
+                            *this, std::move(flakeRef),
-                } catch (...) {
+                            fragment == "" ? getDefaultFlakeAttrPaths() : Strings{fragment},
+                            getDefaultFlakeAttrPathPrefixes()));
+                } else {
                     if (s.find('/') != std::string::npos && (storePath = follow(s)))
                         result.push_back(std::make_shared<InstallableStorePath>(store, store->printStorePath(*storePath)));
                     else
-                        throw;
+                        throw Error("unrecognized argument '%s'", s);
                 }
             }
         }

src/nix/installables.hh

@@ -75,13 +75,9 @@ struct InstallableFlake : InstallableValue
     Strings attrPaths;
     Strings prefixes;
 
-    InstallableFlake(SourceExprCommand & cmd, FlakeRef && flakeRef, Strings attrPaths)
-        : InstallableValue(cmd), flakeRef(flakeRef), attrPaths(std::move(attrPaths))
-    { }
-
     InstallableFlake(SourceExprCommand & cmd, FlakeRef && flakeRef,
-        std::string attrPath, Strings && prefixes)
+        Strings && attrPaths, Strings && prefixes)
-        : InstallableValue(cmd), flakeRef(flakeRef), attrPaths{attrPath},
+        : InstallableValue(cmd), flakeRef(flakeRef), attrPaths(attrPaths),
           prefixes(prefixes)
     { }
 

src/nix/profile.cc

@@ -54,8 +54,8 @@ struct ProfileManifest
                 element.active = e["active"];
                 if (e.value("uri", "") != "") {
                     element.source = ProfileElementSource{
-                        FlakeRef(e["originalUri"]),
+                        parseFlakeRef(e["originalUri"]),
-                        FlakeRef(e["uri"]),
+                        parseFlakeRef(e["uri"]),
                         e["attrPath"]
                     };
                 }
@@ -336,7 +336,7 @@ struct CmdProfileUpgrade : virtual SourceExprCommand, MixDefaultProfile, MixProf
                 Activity act(*logger, lvlChatty, actUnknown,
                     fmt("checking '%s' for updates", element.source->attrPath));
 
-                InstallableFlake installable(*this, FlakeRef(element.source->originalRef), {element.source->attrPath});
+                InstallableFlake installable(*this, FlakeRef(element.source->originalRef), {element.source->attrPath}, {});
 
                 auto [attrPath, resolvedRef, drv] = installable.toDerivation();
 

tests/fetchGit.sh

@@ -71,6 +71,7 @@ echo bar > $repo/dir2/bar
 git -C $repo add dir1/foo
 git -C $repo rm hello
 
+unset _NIX_FORCE_HTTP
 path2=$(nix eval --impure --raw --expr "(builtins.fetchGit $repo).outPath")
 [ ! -e $path2/hello ]
 [ ! -e $path2/bar ]
@@ -107,9 +108,9 @@ path=$(nix eval --impure --raw --expr "(builtins.fetchGit file://$repo).outPath"
 git -C $repo checkout $rev2 -b dev
 echo dev > $repo/hello
 
-# File URI uses 'master' unless specified otherwise
+# File URI uses dirty tree unless specified otherwise
 path2=$(nix eval --impure --raw --expr "(builtins.fetchGit file://$repo).outPath")
-[[ $path = $path2 ]]
+[ $(cat $path2/hello) = dev ]
 
 # Using local path with branch other than 'master' should work when clean or dirty
 path3=$(nix eval --impure --raw --expr "(builtins.fetchGit $repo).outPath")

tests/flakes.sh

@@ -5,10 +5,8 @@ if [[ -z $(type -p git) ]]; then
     exit 99
 fi
 
-export _NIX_FORCE_HTTP=1
-
 clearStore
-rm -rf $TEST_HOME/.cache
+rm -rf $TEST_HOME/.cache $TEST_HOME/.config
 
 registry=$TEST_ROOT/registry.json
 
@@ -91,20 +89,20 @@ git -C $nonFlakeDir commit -m 'Initial'
 cat > $registry <<EOF
 {
     "flakes": {
-        "flake1": {
+        "flake:flake1": {
-            "url": "file://$flake1Dir"
+            "url": "git+file://$flake1Dir"
         },
-        "flake2": {
+        "flake:flake2": {
-            "url": "file://$flake2Dir"
+            "url": "git+file://$flake2Dir"
         },
-        "flake3": {
+        "flake:flake3": {
-            "url": "file://$flake3Dir"
+            "url": "git+file://$flake3Dir"
         },
-        "flake4": {
+        "flake:flake4": {
-            "url": "flake3"
+            "url": "flake:flake3"
         },
-        "nixpkgs": {
+        "flake:nixpkgs": {
-            "url": "flake1"
+            "url": "flake:flake1"
         }
     },
     "version": 1
@@ -137,10 +135,10 @@ nix build -o $TEST_ROOT/result --flake-registry $registry flake1
 [[ -e $TEST_ROOT/result/hello ]]
 
 nix build -o $TEST_ROOT/result --flake-registry $registry $flake1Dir
-nix build -o $TEST_ROOT/result --flake-registry $registry file://$flake1Dir
+nix build -o $TEST_ROOT/result --flake-registry $registry git+file://$flake1Dir
 
-# CHeck that store symlinks inside a flake are not interpreted as flakes.
+# Check that store symlinks inside a flake are not interpreted as flakes.
-nix build -o $flake1Dir/result --flake-registry $registry file://$flake1Dir
+nix build -o $flake1Dir/result --flake-registry $registry git+file://$flake1Dir
 nix path-info $flake1Dir/result
 
 # Building a flake with an unlocked dependency should fail in pure mode.
@@ -152,6 +150,7 @@ nix build -o $TEST_ROOT/result --flake-registry $registry flake2#bar --impure
 # Test automatic lock file generation.
 nix build -o $TEST_ROOT/result --flake-registry $registry $flake2Dir#bar
 [[ -e $flake2Dir/flake.lock ]]
+git -C $flake2Dir add flake.lock
 git -C $flake2Dir commit flake.lock -m 'Add flake.lock'
 
 # Rerunning the build should not change the lockfile.
@@ -170,10 +169,11 @@ nix build -o $TEST_ROOT/result --flake-registry $registry flake2#bar
 
 # Or without a registry.
 # FIXME: shouldn't need '--flake-registry /no-registry'?
-nix build -o $TEST_ROOT/result --flake-registry /no-registry file://$flake2Dir#bar --tarball-ttl 0
+nix build -o $TEST_ROOT/result --flake-registry /no-registry git+file://$flake2Dir#bar --tarball-ttl 0
 
 # Test whether indirect dependencies work.
 nix build -o $TEST_ROOT/result --flake-registry $registry $flake3Dir#xyzzy
+git -C $flake3Dir add flake.lock
 
 # Add dependency to flake3.
 rm $flake3Dir/flake.nix
@@ -196,9 +196,10 @@ git -C $flake3Dir commit -m 'Update flake.nix'
 
 # Check whether `nix build` works with an incomplete lockfile
 nix build -o $TEST_ROOT/result --flake-registry $registry $flake3Dir#"sth sth"
+nix build -o $TEST_ROOT/result --flake-registry $registry $flake3Dir#"sth%20sth"
 
 # Check whether it saved the lockfile
-[[ ! (-z $(git -C $flake3Dir diff master)) ]]
+(! [[ -z $(git -C $flake3Dir diff master) ]])
 
 git -C $flake3Dir add flake.lock
 
@@ -217,12 +218,12 @@ mv $registry.tmp $registry
 # Test whether flakes are registered as GC roots for offline use.
 # FIXME: use tarballs rather than git.
 rm -rf $TEST_HOME/.cache
-nix build -o $TEST_ROOT/result --flake-registry file://$registry file://$flake2Dir#bar
+_NIX_FORCE_HTTP=1 nix build -o $TEST_ROOT/result --flake-registry file://$registry git+file://$flake2Dir#bar
 mv $flake1Dir $flake1Dir.tmp
 mv $flake2Dir $flake2Dir.tmp
 nix-store --gc
-nix build -o $TEST_ROOT/result --flake-registry file://$registry file://$flake2Dir#bar
+_NIX_FORCE_HTTP=1 nix build -o $TEST_ROOT/result --flake-registry file://$registry git+file://$flake2Dir#bar
-nix build -o $TEST_ROOT/result --flake-registry file://$registry file://$flake2Dir#bar --tarball-ttl 0
+_NIX_FORCE_HTTP=1 nix build -o $TEST_ROOT/result --flake-registry file://$registry git+file://$flake2Dir#bar --tarball-ttl 0
 mv $flake1Dir.tmp $flake1Dir
 mv $flake2Dir.tmp $flake2Dir
 
@@ -237,7 +238,7 @@ cat > $flake3Dir/flake.nix <<EOF
     flake1 = {};
     flake2 = {};
     nonFlake = {
-      url = "$nonFlakeDir";
+      url = git+file://$nonFlakeDir;
       flake = false;
     };
   };
@@ -348,7 +349,8 @@ git -C $flake7Dir add flake.nix
 nix flake --flake-registry $registry check $flake7Dir
 
 rm -rf $TEST_ROOT/flake1-v2
-nix flake clone --flake-registry $registry flake1 $TEST_ROOT/flake1-v2
+nix flake clone --flake-registry $registry flake1 --dest $TEST_ROOT/flake1-v2
+[ -e $TEST_ROOT/flake1-v2/flake.nix ]
 
 # More 'nix flake check' tests.
 cat > $flake3Dir/flake.nix <<EOF