Convert fetchMercurial to a input type

This enables Mercurial flakes. It also fixes a bug in pure mode where
you could use a branch/tag name rather than a revision.

src/libexpr/primops/fetchMercurial.cc

@@ -1,175 +1,19 @@
 #include "primops.hh"
 #include "eval-inline.hh"
-#include "download.hh"
 #include "store-api.hh"
-#include "pathlocks.hh"
+#include "fetchers/fetchers.hh"
-
+#include "fetchers/parse.hh"
-#include <sys/time.h>
+#include "fetchers/regex.hh"
 
 #include <regex>
 
-#include <nlohmann/json.hpp>
-
-using namespace std::string_literals;
-
 namespace nix {
 
-struct HgInfo
-{
-    Path storePath;
-    std::string branch;
-    std::string rev;
-    uint64_t revCount = 0;
-};
-
-std::regex commitHashRegex("^[0-9a-fA-F]{40}$");
-
-HgInfo exportMercurial(ref<Store> store, const std::string & uri,
-    std::string rev, const std::string & name)
-{
-    if (rev == "" && hasPrefix(uri, "/") && pathExists(uri + "/.hg")) {
-
-        bool clean = runProgram("hg", true, { "status", "-R", uri, "--modified", "--added", "--removed" }) == "";
-
-        if (!clean) {
-
-            /* This is an unclean working tree. So copy all tracked
-               files. */
-
-            if (!settings.allowDirty)
-                throw Error("Mercurial tree '%s' is unclean", uri);
-
-            if (settings.warnDirty)
-                warn("Mercurial tree '%s' is unclean", uri);
-
-            HgInfo hgInfo;
-            hgInfo.rev = "0000000000000000000000000000000000000000";
-            hgInfo.branch = chomp(runProgram("hg", true, { "branch", "-R", uri }));
-
-            auto files = tokenizeString<std::set<std::string>>(
-                runProgram("hg", true, { "status", "-R", uri, "--clean", "--modified", "--added", "--no-status", "--print0" }), "\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);
-            };
-
-            hgInfo.storePath = store->printStorePath(store->addToStore("source", uri, true, htSHA256, filter));
-
-            return hgInfo;
-        }
-    }
-
-    if (rev == "") rev = "default";
-
-    Path cacheDir = fmt("%s/nix/hg/%s", getCacheDir(), hashString(htSHA256, uri).to_string(Base32, false));
-
-    Path stampFile = fmt("%s/.hg/%s.stamp", cacheDir, hashString(htSHA512, rev).to_string(Base32, false));
-
-    /* If we haven't pulled this repo less than ‘tarball-ttl’ seconds,
-       do so now. */
-    time_t now = time(0);
-    struct stat st;
-    if (stat(stampFile.c_str(), &st) != 0 ||
-        (uint64_t) st.st_mtime + settings.tarballTtl <= (uint64_t) now)
-    {
-        /* Except that if this is a commit hash that we already have,
-           we don't have to pull again. */
-        if (!(std::regex_match(rev, commitHashRegex)
-                && pathExists(cacheDir)
-                && runProgram(
-                    RunOptions("hg", { "log", "-R", cacheDir, "-r", rev, "--template", "1" })
-                    .killStderr(true)).second == "1"))
-        {
-            Activity act(*logger, lvlTalkative, actUnknown, fmt("fetching Mercurial repository '%s'", uri));
-
-            if (pathExists(cacheDir)) {
-                try {
-                    runProgram("hg", true, { "pull", "-R", cacheDir, "--", uri });
-                }
-                catch (ExecError & e) {
-                    string transJournal = cacheDir + "/.hg/store/journal";
-                    /* hg throws "abandoned transaction" error only if this file exists */
-                    if (pathExists(transJournal)) {
-                        runProgram("hg", true, { "recover", "-R", cacheDir });
-                        runProgram("hg", true, { "pull", "-R", cacheDir, "--", uri });
-                    } else {
-                        throw ExecError(e.status, fmt("'hg pull' %s", statusToString(e.status)));
-                    }
-                }
-            } else {
-                createDirs(dirOf(cacheDir));
-                runProgram("hg", true, { "clone", "--noupdate", "--", uri, cacheDir });
-            }
-        }
-
-        writeFile(stampFile, "");
-    }
-
-    auto tokens = tokenizeString<std::vector<std::string>>(
-        runProgram("hg", true, { "log", "-R", cacheDir, "-r", rev, "--template", "{node} {rev} {branch}" }));
-    assert(tokens.size() == 3);
-
-    HgInfo hgInfo;
-    hgInfo.rev = tokens[0];
-    hgInfo.revCount = std::stoull(tokens[1]);
-    hgInfo.branch = tokens[2];
-
-    std::string storeLinkName = hashString(htSHA512, name + std::string("\0"s) + hgInfo.rev).to_string(Base32, false);
-    Path storeLink = fmt("%s/.hg/%s.link", cacheDir, storeLinkName);
-
-    try {
-        auto json = nlohmann::json::parse(readFile(storeLink));
-
-        assert(json["name"] == name && json["rev"] == hgInfo.rev);
-
-        hgInfo.storePath = json["storePath"];
-
-        if (store->isValidPath(store->parseStorePath(hgInfo.storePath))) {
-            printTalkative("using cached Mercurial store path '%s'", hgInfo.storePath);
-            return hgInfo;
-        }
-
-    } catch (SysError & e) {
-        if (e.errNo != ENOENT) throw;
-    }
-
-    Path tmpDir = createTempDir();
-    AutoDelete delTmpDir(tmpDir, true);
-
-    runProgram("hg", true, { "archive", "-R", cacheDir, "-r", rev, tmpDir });
-
-    deletePath(tmpDir + "/.hg_archival.txt");
-
-    hgInfo.storePath = store->printStorePath(store->addToStore(name, tmpDir));
-
-    nlohmann::json json;
-    json["storePath"] = hgInfo.storePath;
-    json["uri"] = uri;
-    json["name"] = name;
-    json["branch"] = hgInfo.branch;
-    json["rev"] = hgInfo.rev;
-    json["revCount"] = hgInfo.revCount;
-
-    writeFile(storeLink, json.dump());
-
-    return hgInfo;
-}
-
 static void prim_fetchMercurial(EvalState & state, const Pos & pos, Value * * args, Value & v)
 {
     std::string url;
-    std::string rev;
+    std::optional<Hash> rev;
+    std::optional<std::string> ref;
     std::string name = "source";
     PathSet context;
 
@@ -183,8 +27,15 @@ static void prim_fetchMercurial(EvalState & state, const Pos & pos, Value * * ar
             string n(attr.name);
             if (n == "url")
                 url = state.coerceToString(*attr.pos, *attr.value, context, false, false);
-            else if (n == "rev")
+            else if (n == "rev") {
-                rev = state.forceStringNoCtx(*attr.value, *attr.pos);
+                // Ugly: unlike fetchGit, here the "rev" attribute can
+                // be both a revision or a branch/tag name.
+                auto value = state.forceStringNoCtx(*attr.value, *attr.pos);
+                if (std::regex_match(value, fetchers::revRegex))
+                    rev = Hash(value, htSHA1);
+                else
+                    ref = value;
+            }
             else if (n == "name")
                 name = state.forceStringNoCtx(*attr.value, *attr.pos);
             else
@@ -201,21 +52,36 @@ static void prim_fetchMercurial(EvalState & state, const Pos & pos, Value * * ar
     // whitelist. Ah well.
     state.checkURI(url);
 
-    if (evalSettings.pureEval && rev == "")
+    if (evalSettings.pureEval && !rev)
         throw Error("in pure evaluation mode, 'fetchMercurial' requires a Mercurial revision");
 
-    auto hgInfo = exportMercurial(state.store, url, rev, name);
+    auto parsedUrl = fetchers::parseURL(
+        url.find("://") != std::string::npos
+        ? "hg+" + url
+        : "hg+file://" + url);
+    if (rev) parsedUrl.query.insert_or_assign("rev", rev->gitRev());
+    if (ref) parsedUrl.query.insert_or_assign("ref", *ref);
+    // FIXME: use name
+    auto input = inputFromURL(parsedUrl);
+
+    auto [tree, input2] = input->fetchTree(state.store);
 
     state.mkAttrs(v, 8);
-    mkString(*state.allocAttr(v, state.sOutPath), hgInfo.storePath, PathSet({hgInfo.storePath}));
+    auto storePath = state.store->printStorePath(tree.storePath);
-    mkString(*state.allocAttr(v, state.symbols.create("branch")), hgInfo.branch);
+    mkString(*state.allocAttr(v, state.sOutPath), storePath, PathSet({storePath}));
-    mkString(*state.allocAttr(v, state.symbols.create("rev")), hgInfo.rev);
+    if (input2->getRef())
-    mkString(*state.allocAttr(v, state.symbols.create("shortRev")), std::string(hgInfo.rev, 0, 12));
+        mkString(*state.allocAttr(v, state.symbols.create("branch")), *input2->getRef());
-    mkInt(*state.allocAttr(v, state.symbols.create("revCount")), hgInfo.revCount);
+    // Backward compatibility: set 'rev' to
+    // 0000000000000000000000000000000000000000 for a dirty tree.
+    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")), std::string(rev2.gitRev(), 0, 12));
+    if (tree.revCount)
+        mkInt(*state.allocAttr(v, state.symbols.create("revCount")), *tree.revCount);
     v.attrs->sort();
 
     if (state.allowedPaths)
-        state.allowedPaths->insert(state.store->toRealPath(hgInfo.storePath));
+        state.allowedPaths->insert(tree.actualPath);
 }
 
 static RegisterPrimOp r("fetchMercurial", 1, prim_fetchMercurial);

src/libstore/fetchers/mercurial.cc

@@ -0,0 +1,274 @@
+#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 {
+
+struct MercurialInput : Input
+{
+    ParsedURL url;
+    std::optional<std::string> ref;
+    std::optional<Hash> rev;
+
+    MercurialInput(const ParsedURL & url) : url(url)
+    {
+        type = "hg";
+    }
+
+    bool operator ==(const Input & other) const override
+    {
+        auto other2 = dynamic_cast<const MercurialInput *>(&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();
+    }
+
+    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<MercurialInput>(*this);
+
+        if (ref) res->ref = ref;
+        if (rev) res->rev = rev;
+
+        return res;
+    }
+
+    std::optional<Path> getSourcePath() const
+    {
+        if (url.scheme == "hg+file" && !ref && !rev)
+            return url.path;
+        return {};
+    }
+
+    std::pair<bool, std::string> getActualUrl() const
+    {
+        bool isLocal = url.scheme == "hg+file";
+        return {isLocal, isLocal ? url.path : std::string(url.base, 3)};
+    }
+
+    std::pair<Tree, std::shared_ptr<const Input>> fetchTreeInternal(nix::ref<Store> store) const override
+    {
+        auto name = "source";
+
+        auto input = std::make_shared<MercurialInput>(*this);
+
+        auto [isLocal, actualUrl] = getActualUrl();
+
+        // FIXME: return lastModified.
+
+        if (!input->ref && !input->rev && isLocal && pathExists(actualUrl + "/.hg")) {
+
+            bool clean = runProgram("hg", true, { "status", "-R", actualUrl, "--modified", "--added", "--removed" }) == "";
+
+            if (!clean) {
+
+                /* This is an unclean working tree. So copy all tracked
+                   files. */
+
+                if (!settings.allowDirty)
+                    throw Error("Mercurial tree '%s' is unclean", actualUrl);
+
+                if (settings.warnDirty)
+                    warn("Mercurial tree '%s' is unclean", actualUrl);
+
+                input->ref = chomp(runProgram("hg", true, { "branch", "-R", actualUrl }));
+
+                auto files = tokenizeString<std::set<std::string>>(
+                    runProgram("hg", true, { "status", "-R", actualUrl, "--clean", "--modified", "--added", "--no-status", "--print0" }), "\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);
+
+                return {Tree {
+                    .actualPath = store->printStorePath(storePath),
+                    .storePath = std::move(storePath),
+                }, input};
+            }
+        }
+
+        if (!input->ref) input->ref = "default";
+
+        Path cacheDir = fmt("%s/nix/hg/%s", getCacheDir(), hashString(htSHA256, actualUrl).to_string(Base32, false));
+
+        assert(input->rev || input->ref);
+        auto revOrRef = input->rev ? input->rev->gitRev() : *input->ref;
+
+        Path stampFile = fmt("%s/.hg/%s.stamp", cacheDir, hashString(htSHA512, revOrRef).to_string(Base32, false));
+
+        /* If we haven't pulled this repo less than ‘tarball-ttl’ seconds,
+           do so now. */
+        time_t now = time(0);
+        struct stat st;
+        if (stat(stampFile.c_str(), &st) != 0 ||
+            (uint64_t) st.st_mtime + settings.tarballTtl <= (uint64_t) now)
+        {
+            /* Except that if this is a commit hash that we already have,
+               we don't have to pull again. */
+            if (!(input->rev
+                    && pathExists(cacheDir)
+                    && runProgram(
+                        RunOptions("hg", { "log", "-R", cacheDir, "-r", input->rev->gitRev(), "--template", "1" })
+                        .killStderr(true)).second == "1"))
+            {
+                Activity act(*logger, lvlTalkative, actUnknown, fmt("fetching Mercurial repository '%s'", actualUrl));
+
+                if (pathExists(cacheDir)) {
+                    try {
+                        runProgram("hg", true, { "pull", "-R", cacheDir, "--", actualUrl });
+                    }
+                    catch (ExecError & e) {
+                        string transJournal = cacheDir + "/.hg/store/journal";
+                        /* hg throws "abandoned transaction" error only if this file exists */
+                        if (pathExists(transJournal)) {
+                            runProgram("hg", true, { "recover", "-R", cacheDir });
+                            runProgram("hg", true, { "pull", "-R", cacheDir, "--", actualUrl });
+                        } else {
+                            throw ExecError(e.status, fmt("'hg pull' %s", statusToString(e.status)));
+                        }
+                    }
+                } else {
+                    createDirs(dirOf(cacheDir));
+                    runProgram("hg", true, { "clone", "--noupdate", "--", actualUrl, cacheDir });
+                }
+            }
+
+            writeFile(stampFile, "");
+        }
+
+        auto tokens = tokenizeString<std::vector<std::string>>(
+            runProgram("hg", true, { "log", "-R", cacheDir, "-r", revOrRef, "--template", "{node} {rev} {branch}" }));
+        assert(tokens.size() == 3);
+
+        input->rev = Hash(tokens[0], htSHA1);
+        auto revCount = std::stoull(tokens[1]);
+        input->ref = tokens[2];
+
+        std::string storeLinkName = hashString(htSHA512, name + std::string("\0"s) + input->rev->gitRev()).to_string(Base32, false);
+        Path storeLink = fmt("%s/.hg/%s.link", cacheDir, storeLinkName);
+
+        try {
+            auto json = nlohmann::json::parse(readFile(storeLink));
+
+            assert(json["name"] == name && json["rev"] == input->rev->gitRev());
+
+            auto storePath = store->parseStorePath((std::string) json["storePath"]);
+
+            if (store->isValidPath(storePath)) {
+                printTalkative("using cached Mercurial store path '%s'", store->printStorePath(storePath));
+                return {Tree {
+                    .actualPath = store->printStorePath(storePath),
+                    .storePath = std::move(storePath),
+                    .rev = input->rev,
+                    .revCount = revCount,
+                }, input};
+            }
+
+        } catch (SysError & e) {
+            if (e.errNo != ENOENT) throw;
+        }
+
+        Path tmpDir = createTempDir();
+        AutoDelete delTmpDir(tmpDir, true);
+
+        runProgram("hg", true, { "archive", "-R", cacheDir, "-r", input->rev->gitRev(), tmpDir });
+
+        deletePath(tmpDir + "/.hg_archival.txt");
+
+        auto storePath = store->addToStore(name, tmpDir);
+
+        nlohmann::json json;
+        json["storePath"] = store->printStorePath(storePath);
+        json["uri"] = actualUrl;
+        json["name"] = name;
+        json["branch"] = *input->ref;
+        json["rev"] = input->rev->gitRev();
+        json["revCount"] = revCount;
+
+        writeFile(storeLink, json.dump());
+
+        return {Tree {
+            .actualPath = store->printStorePath(storePath),
+            .storePath = std::move(storePath),
+            .rev = input->rev,
+            .revCount = revCount,
+        }, input};
+    }
+};
+
+struct MercurialInputScheme : InputScheme
+{
+    std::unique_ptr<Input> inputFromURL(const ParsedURL & url) override
+    {
+        if (url.scheme != "hg+http" &&
+            url.scheme != "hg+https" &&
+            url.scheme != "hg+ssh" &&
+            url.scheme != "hg+file") return nullptr;
+
+        auto input = std::make_unique<MercurialInput>(url);
+
+        for (auto &[name, value] : url.query) {
+            if (name == "rev") {
+                if (!std::regex_match(value, revRegex))
+                    throw BadURL("Mercurial 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("Mercurial URL '%s' contains an invalid branch/tag name", url.url);
+                input->ref = value;
+            }
+        }
+
+        return input;
+    }
+};
+
+static auto r1 = OnStartup([] { registerInputScheme(std::make_unique<MercurialInputScheme>()); });
+
+}

tests/fetchMercurial.sh

@@ -82,8 +82,8 @@ path2=$(nix eval --impure --raw --expr "(builtins.fetchMercurial $repo).outPath"
 
 [[ $(nix eval --impure --raw --expr "(builtins.fetchMercurial $repo).rev") = 0000000000000000000000000000000000000000 ]]
 
-# ... unless we're using an explicit rev.
+# ... unless we're using an explicit ref.
-path3=$(nix eval --raw --expr "(builtins.fetchMercurial { url = $repo; rev = \"default\"; }).outPath")
+path3=$(nix eval --impure --raw --expr "(builtins.fetchMercurial { url = $repo; rev = \"default\"; }).outPath")
 [[ $path = $path3 ]]
 
 # Committing should not affect the store path.