Always ls-tree before trying to cat-file

This provides much better performance in cases where most of the
repos do not have a .morph included.

2 files changed, 35 insertions, 36 deletions

diff --git a/morphlib/morphologyfactory.py b/morphlib/morphologyfactory.py
index c03d541e..ab56afde 100644
--- a/morphlib/morphologyfactory.py
+++ b/morphlib/morphologyfactory.py
@@ -44,10 +44,31 @@ class MorphologyFactory(object):
         self._rrc = remote_repo_cache
 
     def get_morphology(self, reponame, sha1, filename):
-        try:
-            text = self._cat_text(reponame, sha1, filename)
-        except:
-            text = self._autodetect_text(reponame, sha1, filename)
+
+        text = None
+        if self._lrc.has_repo(reponame):
+            repo = self._lrc.get_repo(reponame)
+            file_list = repo.ls_tree(sha1)
+            if filename in file_list:
+                text = repo.cat(sha1, filename)
+        elif self._rrc is not None:
+            file_list = self._rrc.ls_tree(reponame, sha1)
+            if filename in file_list:
+                text = self._rrc.cat_file(reponame, sha1, filename)
+        else:
+            raise NotcachedError(reponame)
+
+        if text is None:
+            bs = morphlib.buildsystem.detect_build_system(file_list)
+            if bs is None:
+                raise AutodetectError(reponame, sha1)
+            # TODO consider changing how morphs are located to be by morph
+            #      name rather than filename, it would save creating a
+            #      filename only to strip it back to its morph name again
+            #      and would allow future changes like morphologies being
+            #      stored as git metadata instead of as a file in the repo
+            morph_name = filename[:-len('.morph')]
+            text = bs.get_morphology_text(morph_name)
 
         try:
             morphology = morphlib.morph2.Morphology(text)
@@ -62,37 +83,6 @@ class MorphologyFactory(object):
 
         return morphology
 
-    def _cat_text(self, reponame, sha1, filename):
-        if self._lrc.has_repo(reponame):
-            repo = self._lrc.get_repo(reponame)
-            return repo.cat(sha1, filename)
-        elif self._rrc is not None:
-            return self._rrc.cat_file(reponame, sha1, filename)
-        else:
-            raise NotcachedError(reponame)
-
-    def _autodetect_text(self, reponame, sha1, filename):
-        if self._lrc.has_repo(reponame):
-            repo = self._lrc.get_repo(reponame)
-            file_list = repo.ls_tree(sha1)
-        elif self._rrc is not None:
-            file_list = self._rrc.ls_tree(reponame, sha1)
-        else:
-            raise NotcachedError(reponame)
-
-        bs = morphlib.buildsystem.detect_build_system(file_list)
-        if bs is None:
-            raise AutodetectError(reponame, sha1)
-        # TODO consider changing how morphs are located to be by morph
-        #      name rather than filename, it would save creating a
-        #      filename only to strip it back to its morph name again
-        #      and would allow future changes like morphologies being
-        #      stored as git metadata instead of as a file in the repo
-        assert filename.endswith('.morph')
-        morph_name = filename[:-len('.morph')]
-        morph_text = bs.get_morphology_text(morph_name)
-        return morph_text
-
     def _check_and_tweak_system(self, morphology, reponame, sha1, filename):
         '''Check and tweak a system morphology.'''
 
diff --git a/morphlib/morphologyfactory_tests.py b/morphlib/morphologyfactory_tests.py
index 9d6900d7..8b69f1de 100644
--- a/morphlib/morphologyfactory_tests.py
+++ b/morphlib/morphologyfactory_tests.py
@@ -90,7 +90,7 @@ class FakeLocalRepo(object):
         return 'text'
 
     def ls_tree(self, sha1):
-        return []
+        return self.morphologies.keys()
 
 class FakeLocalRepoCache(object):
 
@@ -119,6 +119,9 @@ class MorphologyFactoryTests(unittest.TestCase):
     def noremotefile(self, *args):
         raise CatFileError('reponame', 'ref', 'filename')
 
+    def localmorph(self, *args):
+        return ['foo.morph']
+
     def nolocalmorph(self, *args):
         if args[-1].endswith('.morph'):
             raise IOError('File not found')
@@ -127,6 +130,9 @@ class MorphologyFactoryTests(unittest.TestCase):
     def autotoolsbuildsystem(self, *args):
         return ['configure.in']
 
+    def remotemorph(self, *args):
+        return ['foo.morph']
+
     def noremotemorph(self, *args):
         if args[-1].endswith('.morph'):
             raise CatFileError('reponame', 'ref', 'filename')
@@ -136,11 +142,13 @@ class MorphologyFactoryTests(unittest.TestCase):
         return False
 
     def test_gets_morph_from_local_repo(self):
+        self.lr.ls_tree = self.localmorph
         morph = self.mf.get_morphology('reponame', 'sha1',
                                        'foo.morph')
         self.assertEqual('local-foo', morph['name'])
 
     def test_gets_morph_from_remote_repo(self):
+        self.rrc.ls_tree = self.remotemorph
         self.lrc.has_repo = self.doesnothaverepo
         morph = self.mf.get_morphology('reponame', 'sha1',
                                        'foo.morph')
@@ -174,6 +182,7 @@ class MorphologyFactoryTests(unittest.TestCase):
                           'reponame', 'sha1', 'unreached.morph')
 
     def test_looks_locally_with_no_remote(self):
+        self.lr.ls_tree = self.localmorph
         morph = self.lmf.get_morphology('reponame', 'sha1',
                                         'foo.morph')
         self.assertEqual('local-foo', morph['name'])