identify unmarked blobs and commits by line numbers

3 files changed, 36 insertions, 11 deletions

diff --git a/commands.py b/commands.py
index 3cf3df6..3f0e986 100644
--- a/commands.py
+++ b/commands.py
@@ -64,10 +64,16 @@ class ImportCommand(object):
 
 class BlobCommand(ImportCommand):
 
-    def __init__(self, mark, data):
+    def __init__(self, mark, data, lineno=0):
         ImportCommand.__init__(self, 'blob')
         self.mark = mark
         self.data = data
+        self.lineno = lineno
+        # Provide a unique id in case the mark is missing
+        if mark is None:
+            self.id = '@%d' % lineno
+        else:
+            self.id = ':' + mark
         self._binary = ['data']
 
 
@@ -80,7 +86,7 @@ class CheckpointCommand(ImportCommand):
 class CommitCommand(ImportCommand):
 
     def __init__(self, ref, mark, author, committer, message, parents,
-        file_iter):
+        file_iter, lineno=0):
         ImportCommand.__init__(self, 'commit')
         self.ref = ref
         self.mark = mark
@@ -89,7 +95,13 @@ class CommitCommand(ImportCommand):
         self.message = message
         self.parents = parents
         self.file_iter = file_iter
+        self.lineno = lineno
         self._binary = ['file_iter']
+        # Provide a unique id in case the mark is missing
+        if mark is None:
+            self.id = '@%d' % lineno
+        else:
+            self.id = ':' + mark
 
     def dump_str(self, names=None, child_lists=None, verbose=False):
         result = [ImportCommand.dump_str(self, names, verbose=verbose)]
diff --git a/parser.py b/parser.py
index 3e362bf..38abcea 100644
--- a/parser.py
+++ b/parser.py
@@ -323,12 +323,14 @@ class ImportParser(LineBasedParser):
 
     def _parse_blob(self):
         """Parse a blob command."""
+        lineno = self.lineno
         mark = self._get_mark_if_any()
         data = self._get_data('blob')
-        return commands.BlobCommand(mark, data)
+        return commands.BlobCommand(mark, data, lineno)
 
     def _parse_commit(self, ref):
         """Parse a commit command."""
+        lineno  = self.lineno
         mark = self._get_mark_if_any()
         author = self._get_user_info('commit', 'author', False)
         committer = self._get_user_info('commit', 'committer')
@@ -345,7 +347,7 @@ class ImportParser(LineBasedParser):
         else:
             parents = []
         return commands.CommitCommand(ref, mark, author, committer, message,
-            parents, self.iter_file_commands)
+            parents, self.iter_file_commands, lineno)
 
     def _parse_file_modify(self, info):
         """Parse a filemodify command within a commit.
diff --git a/processors/info_processor.py b/processors/info_processor.py
index a6958b9..b0d430e 100644
--- a/processors/info_processor.py
+++ b/processors/info_processor.py
@@ -67,10 +67,11 @@ class InfoProcessor(processor.ImportProcessor):
         self.named_branches = []
         # Blob usage tracking
         self.blobs = {}
-        for usage in ['new', 'used', 'multi', 'unknown']:
+        for usage in ['new', 'used', 'multi', 'unknown', 'unmarked']:
             self.blobs[usage] = set()
         # Head tracking: map of commit mark to ref
         self.heads = {}
+        self.last_ids = {}
 
     def post_process(self):
         # Dump statistics
@@ -153,7 +154,10 @@ class InfoProcessor(processor.ImportProcessor):
     def blob_handler(self, cmd):
         """Process a BlobCommand."""
         self.cmd_counts[cmd.name] += 1
-        self.blobs['new'].add(":" + cmd.mark)
+        if cmd.mark is None:
+            self.blobs['unmarked'].add(cmd.id)
+        else:
+            self.blobs['new'].add(cmd.id)
 
     def checkpoint_handler(self, cmd):
         """Process a CheckpointCommand."""
@@ -179,14 +183,21 @@ class InfoProcessor(processor.ImportProcessor):
                     else:
                         self.sha_blob_references = True
         # Track the heads
-        for parent in cmd.parents:
+        if cmd.mark is None:
+            last_id = self.last_ids.get(cmd.ref)
+            if last_id is not None:
+                parents = [last_id]
+            else:
+                parents = []
+        else:
+            parents = cmd.parents
+        for parent in parents:
             try:
                 del self.heads[parent]
             except KeyError:
-                print "hmm - didn't find parent %s" % parent
-                pass
-        self.heads[":" + cmd.mark] = cmd.ref
-            
+                note("hmm - didn't find parent %s" % parent)
+        self.heads[cmd.id] = cmd.ref
+        self.last_ids[cmd.ref] = cmd.id
 
     def reset_handler(self, cmd):
         """Process a ResetCommand."""