Add git_reference class

3 files changed, 120 insertions, 40 deletions

diff --git a/libbpkg/manifest.cxx b/libbpkg/manifest.cxx
index a972793..d7cf7cf 100644
--- a/libbpkg/manifest.cxx
+++ b/libbpkg/manifest.cxx
@@ -1770,9 +1770,9 @@ namespace bpkg
       }
     case repository_type::git:
       {
-        if (!url_.fragment)
-          throw invalid_argument ("missing branch/tag for git repository");
-
+        // Verify the URL fragment.
+        //
+        git_reference r (url_.fragment);
         break;
       }
     }
@@ -1942,6 +1942,37 @@ namespace bpkg
     canonical_name_ += cp;
   }
 
+  // git_reference
+  //
+  git_reference::
+  git_reference (const optional<string>& frag)
+  {
+    if (frag)
+    {
+      const string& s (*frag);
+
+      size_t p (s.find ('@'));
+      if (p != string::npos)
+      {
+        if (p != 0)
+          branch = string (s, 0, p);
+
+        if (p + 1 != s.size ())
+          commit = string (s, p + 1);
+      }
+      else if (!s.empty ())
+        branch = s;
+    }
+
+    if (!branch && !commit)
+      throw invalid_argument (
+        "missing branch/tag or commit id for git repository");
+
+    if (commit && commit->size () != 40)
+      throw invalid_argument (
+        "git repository commit id must be 40 characters long");
+  }
+
   // repository_manifest
   //
   repository_manifest::
diff --git a/libbpkg/manifest.hxx b/libbpkg/manifest.hxx
index 8c20e76..9d19b2c 100644
--- a/libbpkg/manifest.hxx
+++ b/libbpkg/manifest.hxx
@@ -510,22 +510,24 @@ namespace bpkg
     // the URL matches the repository type. Throw std::invalid_argument if the
     // URL object is a relative local path.
     //
-    // @@ Note that the repository location string representation may differ
-    //    from the original URL in the presence of the trailing slash. This
-    //    may cause problems with some WEB servers that are sensitive to the
-    //    trailing slash presence/absence. For example:
+    // Note that the repository location string representation may differ from
+    // the original URL in the presence of the trailing slash. This may cause
+    // problems with some WEB servers that are sensitive to the trailing slash
+    // presence/absence. For example:
     //
-    //    $ git clone http://git.sv.gnu.org/r/config.git
-    //    warning: redirecting to http://git.savannah.gnu.org/r/config.git/
+    // $ git clone http://git.sv.gnu.org/r/config.git
+    //   warning: redirecting to http://git.savannah.gnu.org/r/config.git/
     //
-    //    Also note that we disregard the slash presence/absence on multiple
-    //    levels:
+    // Also note that we disregard the slash presence/absence on multiple
+    // levels:
     //
-    //    - reduce absent path to an empty one in
-    //      repository_url_traits::translate_scheme() (so a.com/ becomes a.com)
-    //    - use path::*string() rather than path::*representation() functions
-    //      in repository_url_traits::translate_*() functions
-    //    - may append slash in repository_location ctor
+    // - reduce absent path to an empty one in
+    //   repository_url_traits::translate_scheme() (so a.com/ becomes a.com)
+    //
+    // - use path::*string() rather than path::*representation() functions
+    //   in repository_url_traits::translate_*() functions
+    //
+    // - may append slash in repository_location ctor
     //
     explicit
     repository_location (repository_url, repository_type);
@@ -685,6 +687,28 @@ namespace bpkg
     return os << l.string ();
   }
 
+  // Branch and/or commit. At least one of them must be present. If both of
+  // them are present then the commit is expected to belong to the branch
+  // history. Note that the branch member can also denote a tag.
+  //
+  class LIBBPKG_EXPORT git_reference
+  {
+  public:
+    butl::optional<std::string> branch;
+    butl::optional<std::string> commit;
+
+  public:
+    // Parse the [<branch>][@<commit>] repository URL fragment representation.
+    //
+    explicit
+    git_reference (const butl::optional<std::string>&);
+
+    git_reference (butl::optional<std::string> b,
+                   butl::optional<std::string> c)
+        : branch (std::move (b)),
+          commit (std::move (c)) {}
+  };
+
   enum class repository_role
   {
     base,
diff --git a/tests/repository-location/driver.cxx b/tests/repository-location/driver.cxx
index 098ecab..e5a818a 100644
--- a/tests/repository-location/driver.cxx
+++ b/tests/repository-location/driver.cxx
@@ -150,7 +150,7 @@ namespace bpkg
     assert (bad_loc ("c:\\aaa\\bbb"));
 #endif
 
-    // No URL fragment.
+    // Invalid/absent URL fragment.
     //
 #ifndef _WIN32
     assert (bad_loc ("file://localhost/", repository_type::git));
@@ -158,6 +158,18 @@ namespace bpkg
     assert (bad_loc ("file://localhost/c:/", repository_type::git));
 #endif
 
+    assert (bad_loc ("https://www.example.com/test.git",
+                     repository_type::git));
+
+    assert (bad_loc ("https://www.example.com/test.git#",
+                     repository_type::git));
+
+    assert (bad_loc ("https://www.example.com/test.git#@",
+                     repository_type::git));
+
+    assert (bad_loc ("https://www.example.com/test.git#@123",
+                     repository_type::git));
+
     // Invalid version.
     //
     assert (bad_loc ("3/aaa/bbb"));
@@ -410,25 +422,25 @@ namespace bpkg
       assert (l.type () == repository_type::bpkg);
     }
     {
-      repository_location l (loc ("git://github.com/test#master",
+      repository_location l (loc ("git://example.com/test#master",
                                   repository_type::git));
-      assert (l.string () == "git://github.com/test#master");
-      assert (l.canonical_name () == "git:github.com/test");
+      assert (l.string () == "git://example.com/test#master");
+      assert (l.canonical_name () == "git:example.com/test");
       assert (l.proto () == proto::git);
       assert (l.type () == repository_type::git);
     }
     {
-      repository_location l (loc ("http://github.com/test.git#master",
+      repository_location l (loc ("http://example.com/test.git#master",
                                   repository_type::git));
-      assert (l.string () == "http://github.com/test.git#master");
-      assert (l.canonical_name () == "git:github.com/test");
+      assert (l.string () == "http://example.com/test.git#master");
+      assert (l.canonical_name () == "git:example.com/test");
       assert (l.proto () == proto::http);
     }
     {
-      repository_location l (loc ("https://github.com/test.git#master",
+      repository_location l (loc ("https://example.com/test.git#master",
                                   repository_type::git));
-      assert (l.string () == "https://github.com/test.git#master");
-      assert (l.canonical_name () == "git:github.com/test");
+      assert (l.string () == "https://example.com/test.git#master");
+      assert (l.canonical_name () == "git:example.com/test");
       assert (l.proto () == proto::https);
       assert (l.type () == repository_type::git);
     }
@@ -487,22 +499,22 @@ namespace bpkg
       assert (l.canonical_name () == "bpkg:www.cppget.org/qw/a/b");
     }
     {
-      repository_location l (loc ("https://git.github.com/test.git#master",
+      repository_location l (loc ("https://git.example.com/test.git#master",
                                   repository_type::git));
-      assert (l.string () == "https://git.github.com/test.git#master");
-      assert (l.canonical_name () == "git:github.com/test");
+      assert (l.string () == "https://git.example.com/test.git#master");
+      assert (l.canonical_name () == "git:example.com/test");
     }
     {
-      repository_location l (loc ("https://scm.github.com/test.git#master",
+      repository_location l (loc ("https://scm.example.com/test.git#master",
                                   repository_type::git));
-      assert (l.string () == "https://scm.github.com/test.git#master");
-      assert (l.canonical_name () == "git:github.com/test");
+      assert (l.string () == "https://scm.example.com/test.git#master");
+      assert (l.canonical_name () == "git:example.com/test");
     }
     {
-      repository_location l (loc ("https://www.github.com/test.git#master",
+      repository_location l (loc ("https://www.example.com/test.git#master",
                                   repository_type::git));
-      assert (l.string () == "https://www.github.com/test.git#master");
-      assert (l.canonical_name () == "git:github.com/test");
+      assert (l.string () == "https://www.example.com/test.git#master");
+      assert (l.canonical_name () == "git:example.com/test");
     }
     {
       repository_location l (loc ("http://cppget.org/qw//1/a//b/"));
@@ -711,14 +723,13 @@ namespace bpkg
               "http://cppget.org/foo/misc/stable");
     }
     {
-      repository_location l (
-        loc ("http://cppget.org/pkg/foo/1/misc/stable"));
+      repository_location l (loc ("http://cppget.org/pkg/foo/1/misc/stable"));
 
       assert (effective_url ("../..", l) ==
               "http://cppget.org/pkg/foo/misc/stable");
     }
     {
-      repository_location l (loc ("https://git.github.com/test.git#master",
+      repository_location l (loc ("https://git.example.com/test.git#master",
                                   repository_type::git));
       assert (effective_url ("../..", l) == "../..");
     }
@@ -726,8 +737,7 @@ namespace bpkg
       repository_location l (
         loc ("http://www.cppget.org/foo/bpkg/1/misc/stable"));
 
-      assert (effective_url ("../../..", l) ==
-              "http://cppget.org/foo/misc");
+      assert (effective_url ("../../..", l) == "http://cppget.org/foo/misc");
     }
     {
       repository_location l (loc ("http://pkg.cppget.org/foo/pkg/1/misc"));
@@ -741,6 +751,21 @@ namespace bpkg
               "http://cppget.org/foo/misc/abc");
     }
 
+    // Repository URL fragments.
+    //
+    {
+      loc ("https://www.example.com/test.git#master", repository_type::git);
+      loc ("https://www.example.com/test.git#master@", repository_type::git);
+
+      loc ("https://www.example.com/test.git#"
+           "@0a53e9ddeaddad63ad106860237bbf53411d11a7",
+           repository_type::git);
+
+      loc ("https://www.example.com/test.git#"
+           "master@0a53e9ddeaddad63ad106860237bbf53411d11a7",
+           repository_type::git);
+    }
+
     // repository_url
     //
     assert (repository_url ("git://example.com/test.git") ==