Extract git reference parsing to a shared library

These utility functions can be shared between the git and github fetchers.

4 files changed, 70 insertions, 35 deletions

diff --git a/src/libfetchers/git-utils.cc b/src/libfetchers/git-utils.cc
new file mode 100644
index 000000000..060077098
--- /dev/null
+++ b/src/libfetchers/git-utils.cc
@@ -0,0 +1,25 @@
+#include "git-utils.hh"
+
+#include <regex>
+
+std::optional<std::string> parseListReferenceHeadRef(std::string_view line) {
+    const static std::regex head_ref_regex("^ref: ([^\\s]+)\\t+HEAD$");
+    std::match_results<std::string_view::const_iterator> match;
+    if (std::regex_match(line.cbegin(), line.cend(), match, head_ref_regex)) {
+        return match[1];
+    } else {
+        return std::nullopt;
+    }
+}
+
+std::optional<std::string> parseListReferenceForRev(std::string_view rev, std::string_view line) {
+    const static std::regex rev_regex("^([^\\t]+)\\t+(.*)$");
+    std::match_results<std::string_view::const_iterator> match;
+    if (!std::regex_match(line.cbegin(), line.cend(), match, rev_regex)) {
+        return std::nullopt;
+    }
+    if (rev != match[2].str()) {
+        return std::nullopt;
+    }
+    return match[1];
+}
diff --git a/src/libfetchers/git-utils.hh b/src/libfetchers/git-utils.hh
new file mode 100644
index 000000000..946a68a9e
--- /dev/null
+++ b/src/libfetchers/git-utils.hh
@@ -0,0 +1,23 @@
+#pragma once
+
+#include <string>
+#include <string_view>
+#include <optional>
+
+// Parses the HEAD ref as reported by `git ls-remote --symref`
+//
+// Returns the head branch name as reported by `git ls-remote --symref`, e.g., if
+// ls-remote returns the output below, "main" is returned based on the ref line.
+//
+//   ref: refs/heads/main       HEAD
+//
+// If the repository is in 'detached head' state (HEAD is pointing to a rev
+// instead of a branch), parseListReferenceForRev("HEAD") may be used instead.
+std::optional<std::string> parseListReferenceHeadRef(std::string_view line);
+
+// Parses a reference line from `git ls-remote --symref`, e.g.,
+// parseListReferenceForRev("refs/heads/master", line) will return 6926...
+// given the line below.
+//
+// 6926beab444c33fb57b21819b6642d032016bb1e	refs/heads/master
+std::optional<std::string> parseListReferenceForRev(std::string_view rev, std::string_view line);
diff --git a/src/libfetchers/git.cc b/src/libfetchers/git.cc
index 968cd642a..9d4348cf1 100644
--- a/src/libfetchers/git.cc
+++ b/src/libfetchers/git.cc
@@ -6,6 +6,7 @@
 #include "url-parts.hh"
 #include "pathlocks.hh"
 #include "util.hh"
+#include "git-utils.hh"
 
 #include "fetch-settings.hh"
 
@@ -69,27 +70,19 @@ std::optional<std::string> readHead(const Path & path)
         .args = {"ls-remote", "--symref", path},
     });
     if (exit_code != 0) {
-      return std::nullopt;
+        return std::nullopt;
     }
 
-    // Matches the common case when HEAD points to a branch, e.g.:
-    //   "ref: refs/heads/main  HEAD".
-    const static std::regex head_ref_regex("^ref:\\s*([^\\s]+)\\s*HEAD$");
-    // Matches when HEAD points directly at a commit, e.g.:
-    //   "71abcd...  HEAD".
-    const static std::regex head_rev_regex("^([^\\s]+)\\s*HEAD$");
-
-    for (const auto & line : tokenizeString<std::vector<std::string>>(output, "\n")) {
-        std::smatch match;
-        if (std::regex_match(line, match, head_ref_regex)) {
-            debug("resolved HEAD ref '%s' for repo '%s'", match[1], path);
-            return match[1];
-        } else if (std::regex_match(line, match, head_rev_regex)) {
-            debug("resolved HEAD ref '%s' for repo '%s'", match[1], path);
-            return match[1];
-        }
+    std::string_view line = output;
+    line = line.substr(0, line.find("\n"));
+    if (const auto ref = parseListReferenceHeadRef(line); ref) {
+        debug("resolved HEAD ref '%s' for repo '%s'", *ref, path);
+        return *ref;
+    }
+    if (const auto rev = parseListReferenceForRev("HEAD", line); rev) {
+        debug("resolved HEAD rev '%s' for repo '%s'", *rev, path);
+        return *rev;
     }
-
     return std::nullopt;
 }
 
diff --git a/src/libfetchers/github.cc b/src/libfetchers/github.cc
index 58b6e7c04..1bdf2759f 100644
--- a/src/libfetchers/github.cc
+++ b/src/libfetchers/github.cc
@@ -4,7 +4,7 @@
 #include "store-api.hh"
 #include "types.hh"
 #include "url-parts.hh"
-
+#include "git-utils.hh"
 #include "fetchers.hh"
 #include "fetch-settings.hh"
 
@@ -383,35 +383,29 @@ struct SourceHutInputScheme : GitArchiveInputScheme
             std::string line;
             getline(is, line);
 
-            auto ref_index = line.find("ref: ");
-            if (ref_index == std::string::npos) {
+            auto r = parseListReferenceHeadRef(line);
+            if (!r) {
                 throw BadURL("in '%d', couldn't resolve HEAD ref '%d'", input.to_string(), ref);
             }
-
-            ref_uri = line.substr(ref_index+5, line.length()-1);
-        } else
+            ref_uri = *r;
+        } else {
             ref_uri = fmt("refs/(heads|tags)/%s", ref);
+        }
 
         auto file = store->toRealPath(
             downloadFile(store, fmt("%s/info/refs", base_url), "source", false, headers).storePath);
         std::ifstream is(file);
 
         std::string line;
-        std::string id;
-        while(getline(is, line)) {
-            // Append $ to avoid partial name matches
-            std::regex pattern(fmt("%s$", ref_uri));
-
-            if (std::regex_search(line, pattern)) {
-                id = line.substr(0, line.find('\t'));
-                break;
-            }
+        std::optional<std::string> id;
+        while(!id && getline(is, line)) {
+            id = parseListReferenceForRev(ref_uri, line);
         }
 
-        if(id.empty())
+        if(!id)
             throw BadURL("in '%d', couldn't find ref '%d'", input.to_string(), ref);
 
-        auto rev = Hash::parseAny(id, htSHA1);
+        auto rev = Hash::parseAny(*id, htSHA1);
         debug("HEAD revision for '%s' is %s", fmt("%s/%s", base_url, ref), rev.gitRev());
         return rev;
     }