fetch/git: add support for removing arbitrary revs for shallow

In certain cases, it's valuable to be able to exert more control over what
history is removed, beyond srcrev+depth. As one example, you can remove most
of the upstream kernel history from a kernel repository, keeping predominently
the non-publically-accessible content. If the repository is private, the
history in that repo couldn't be restored via `git fetch --unshallow`, but
upstream history could be.

Example usage:

    # Remove only these revs, not at a particular depth
    BB_GIT_SHALLOW_DEPTH_pn-linux-foo = "0"
    BB_GIT_SHALLOW_REVS_pn-linux-foo = "v4.1"

Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

2 files changed, 82 insertions, 2 deletions

diff --git a/lib/bb/fetch2/git.py b/lib/bb/fetch2/git.py
index aa972c5cf..534c93d3c 100644
--- a/lib/bb/fetch2/git.py
+++ b/lib/bb/fetch2/git.py
@@ -196,6 +196,8 @@ class Git(FetchMethod):
             depth_default = 1
         ud.shallow_depths = collections.defaultdict(lambda: depth_default)
 
+        revs_default = d.getVar("BB_GIT_SHALLOW_REVS", True)
+        ud.shallow_revs = []
         ud.branches = {}
         for pos, name in enumerate(ud.names):
             branch = branches[pos]
@@ -213,7 +215,14 @@ class Git(FetchMethod):
                         raise bb.fetch2.FetchError("Invalid depth for BB_GIT_SHALLOW_DEPTH_%s: %s" % (name, shallow_depth))
                     ud.shallow_depths[name] = shallow_depth
 
+            revs = d.getVar("BB_GIT_SHALLOW_REVS_%s" % name)
+            if revs is not None:
+                ud.shallow_revs.extend(revs.split())
+            elif revs_default is not None:
+                ud.shallow_revs.extend(revs_default.split())
+
         if (ud.shallow and
+                not ud.shallow_revs and
                 all(ud.shallow_depths[n] == 0 for n in ud.names)):
             # Shallow disabled for this URL
             ud.shallow = False
@@ -261,6 +270,9 @@ class Git(FetchMethod):
             if ud.bareclone:
                 tarballname = "%s_bare" % tarballname
 
+            if ud.shallow_revs:
+                tarballname = "%s_%s" % (tarballname, "_".join(sorted(ud.shallow_revs)))
+
             for name, revision in sorted(ud.revisions.items()):
                 tarballname = "%s_%s" % (tarballname, ud.revisions[name][:7])
                 depth = ud.shallow_depths[name]
@@ -413,7 +425,11 @@ class Git(FetchMethod):
             runfetchcmd("%s update-ref %s %s" % (ud.basecmd, ref, revision), d, workdir=dest)
 
         # Map srcrev+depths to revisions
-        shallow_revisions = runfetchcmd("%s rev-parse %s" % (ud.basecmd, " ".join(to_parse)), d, workdir=dest).splitlines()
+        parsed_depths = runfetchcmd("%s rev-parse %s" % (ud.basecmd, " ".join(to_parse)), d, workdir=dest)
+
+        # Resolve specified revisions
+        parsed_revs = runfetchcmd("%s rev-parse %s" % (ud.basecmd, " ".join('"%s^{}"' % r for r in ud.shallow_revs)), d, workdir=dest)
+        shallow_revisions = parsed_depths.splitlines() + parsed_revs.splitlines()
 
         # Apply extra ref wildcards
         all_refs = runfetchcmd('%s for-each-ref "--format=%%(refname)"' % ud.basecmd,
diff --git a/lib/bb/tests/fetch.py b/lib/bb/tests/fetch.py
index 73f7b3f78..343ae8fe5 100644
--- a/lib/bb/tests/fetch.py
+++ b/lib/bb/tests/fetch.py
@@ -1259,6 +1259,33 @@ class GitShallowTest(FetcherTest):
         self.add_empty_file('c')
         self.add_empty_file('d')
         self.git('checkout master', cwd=self.srcdir)
+        self.git('tag v0.0 a_branch', cwd=self.srcdir)
+        self.add_empty_file('e')
+        self.git('merge --no-ff --no-edit a_branch', cwd=self.srcdir)
+        self.add_empty_file('f')
+        self.assertRevCount(7, cwd=self.srcdir)
+
+        uri = self.d.getVar('SRC_URI', True).split()[0]
+        uri = '%s;branch=master,a_branch;name=master,a_branch' % uri
+
+        self.d.setVar('BB_GIT_SHALLOW_DEPTH', '0')
+        self.d.setVar('BB_GIT_SHALLOW_REVS', 'v0.0')
+        self.d.setVar('SRCREV_master', '${AUTOREV}')
+        self.d.setVar('SRCREV_a_branch', '${AUTOREV}')
+
+        self.fetch_shallow(uri)
+
+        self.assertRevCount(5)
+        self.assertRefs(['master', 'origin/master', 'origin/a_branch'])
+
+    def test_shallow_multi_one_uri_depths(self):
+        # Create initial git repo
+        self.add_empty_file('a')
+        self.add_empty_file('b')
+        self.git('checkout -b a_branch', cwd=self.srcdir)
+        self.add_empty_file('c')
+        self.add_empty_file('d')
+        self.git('checkout master', cwd=self.srcdir)
         self.add_empty_file('e')
         self.git('merge --no-ff --no-edit a_branch', cwd=self.srcdir)
         self.add_empty_file('f')
@@ -1375,6 +1402,38 @@ class GitShallowTest(FetcherTest):
         self.d.setVar('BB_GIT_SHALLOW_EXTRA_REFS', 'refs/tags/*')
         self.fetch()
 
+    def test_shallow_remove_revs(self):
+        # Create initial git repo
+        self.add_empty_file('a')
+        self.add_empty_file('b')
+        self.git('checkout -b a_branch', cwd=self.srcdir)
+        self.add_empty_file('c')
+        self.add_empty_file('d')
+        self.git('checkout master', cwd=self.srcdir)
+        self.git('tag v0.0 a_branch', cwd=self.srcdir)
+        self.add_empty_file('e')
+        self.git('merge --no-ff --no-edit a_branch', cwd=self.srcdir)
+        self.git('branch -d a_branch', cwd=self.srcdir)
+        self.add_empty_file('f')
+        self.assertRevCount(7, cwd=self.srcdir)
+
+        self.d.setVar('BB_GIT_SHALLOW_DEPTH', '0')
+        self.d.setVar('BB_GIT_SHALLOW_REVS', 'v0.0')
+
+        self.fetch_shallow()
+
+        self.assertRevCount(5)
+
+    def test_shallow_invalid_revs(self):
+        self.add_empty_file('a')
+        self.add_empty_file('b')
+
+        self.d.setVar('BB_GIT_SHALLOW_DEPTH', '0')
+        self.d.setVar('BB_GIT_SHALLOW_REVS', 'v0.0')
+
+        with self.assertRaises(bb.fetch2.FetchError):
+            self.fetch()
+
     if os.environ.get("BB_SKIP_NETTESTS") == "yes":
         print("Unset BB_SKIP_NETTESTS to run network tests")
     else:
@@ -1383,11 +1442,16 @@ class GitShallowTest(FetcherTest):
             self.git('config core.bare true', cwd=self.srcdir)
             self.git('fetch --tags', cwd=self.srcdir)
 
-            self.d.setVar('BB_GIT_SHALLOW_DEPTH', '100')
+            self.d.setVar('BB_GIT_SHALLOW_DEPTH', '0')
+            # Note that the 1.10.0 tag is annotated, so this also tests
+            # reference of an annotated vs unannotated tag
+            self.d.setVar('BB_GIT_SHALLOW_REVS', '1.10.0')
 
             self.fetch_shallow()
 
+            # Confirm that the history of 1.10.0 was removed
             orig_revs = len(self.git('rev-list master', cwd=self.srcdir).splitlines())
             revs = len(self.git('rev-list master').splitlines())
             self.assertNotEqual(orig_revs, revs)
             self.assertRefs(['master', 'origin/master'])
+            self.assertRevCount(orig_revs - 1758)