[Pkg-bazaar-commits] ./bzr/unstable r125: - check progress indicator for file texts

[mbp at sourcefrog.net]

------------------------------------------------------------
revno: 125
committer: mbp at sourcefrog.net
timestamp: Tue 2005-03-29 10:32:52 +1000
message:
  - check progress indicator for file texts
  - avoid re-reading file texts that have already been checked
modified:
  bzrlib/check.py
-------------- next part --------------
=== modified file 'bzrlib/check.py'
--- a/bzrlib/check.py	2005-03-29 00:23:20 +0000
+++ b/bzrlib/check.py	2005-03-29 00:32:52 +0000
@@ -47,6 +47,8 @@
     history = branch.revision_history()
     revno = 0
     revcount = len(history)
+
+    checked_texts = {}
     
     for rid in history:
         revno += 1
@@ -65,51 +67,59 @@
         ## TODO: Check all the required fields are present on the revision.
 
         inv = branch.get_inventory(rev.inventory_id)
-        check_inventory(branch, inv, rid)
+        seen_ids = Set()
+        seen_names = Set()
+
+        p('revision %d/%d file ids' % (revno, revcount))
+        for file_id in inv:
+            if file_id in seen_ids:
+                bailout('duplicated file_id {%s} in inventory for revision {%s}'
+                        % (file_id, revid))
+            seen_ids.add(file_id)
+
+        i = 0
+        len_inv = len(inv)
+        for file_id in inv:
+            i += 1
+            if (i % 100) == 0:
+                p('revision %d/%d file text %d/%d' % (revno, revcount, i, len_inv))
+
+            ie = inv[file_id]
+
+            if ie.parent_id != None:
+                if ie.parent_id not in seen_ids:
+                    bailout('missing parent {%s} in inventory for revision {%s}'
+                            % (ie.parent_id, revid))
+
+            if ie.kind == 'file':
+                if ie.text_id in checked_texts:
+                    fp = checked_texts[ie.text_id]
+                else:
+                    if not ie.text_id in branch.text_store:
+                        bailout('text {%s} not in text_store' % ie.text_id)
+
+                    tf = branch.text_store[ie.text_id]
+                    fp = osutils.fingerprint_file(tf)
+                    checked_texts[ie.text_id] = fp
+
+                if ie.text_size != fp['size']:
+                    bailout('text {%s} wrong size' % ie.text_id)
+                if ie.text_sha1 != fp['sha1']:
+                    bailout('text {%s} wrong sha1' % ie.text_id)
+            elif ie.kind == 'directory':
+                if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
+                    bailout('directory {%s} has text in revision {%s}'
+                            % (file_id, revid))
+
+        p('revision %d/%d file paths' % (revno, revcount))
+        for path, ie in inv.iter_entries():
+            if path in seen_names:
+                bailout('duplicated path %r in inventory for revision {%s}' % (path, revid))
+            seen_names.add(path)
+
 
     p('done')
     if progress:
         print 
 
 
-
-def check_inventory(branch, inv, revid):
-    seen_ids = Set()
-    seen_names = Set()
-
-    for file_id in inv:
-        if file_id in seen_ids:
-            bailout('duplicated file_id {%s} in inventory for revision {%s}'
-                    % (file_id, revid))
-        seen_ids.add(file_id)
-        
-
-    for file_id in inv:
-        ie = inv[file_id]
-        
-        if ie.parent_id != None:
-            if ie.parent_id not in seen_ids:
-                bailout('missing parent {%s} in inventory for revision {%s}'
-                        % (ie.parent_id, revid))
-
-        if ie.kind == 'file':
-            if not ie.text_id in branch.text_store:
-                bailout('text {%s} not in text_store' % ie.text_id)
-
-            tf = branch.text_store[ie.text_id]
-            fp = osutils.fingerprint_file(tf)
-            if ie.text_size != fp['size']:
-                bailout('text {%s} wrong size' % ie.text_id)
-            if ie.text_sha1 != fp['sha1']:
-                bailout('text {%s} wrong sha1' % ie.text_id)
-        elif ie.kind == 'directory':
-            if ie.text_sha1 != None or ie.text_size != None or ie.text_id != None:
-                bailout('directory {%s} has text in revision {%s}'
-                        % (file_id, revid))
-                
-            
-    for path, ie in inv.iter_entries():
-        if path in seen_names:
-            bailout('duplicated path %r in inventory for revision {%s}' % (path, revid))
-        seen_names.add(path)
-