diff processor displays wrong line number (+1)

[Shunichi Goto <gotoh@…>]

The diff processor (in wiki/ticket, not in browser) displays wrong line number added by 1.

For example, I put a raw diff text and a processed one to compare. both has same content.

--- templates/changeset.cs      (revision 2143)
+++ templates/changeset.cs      (working copy)
@@ -106,7 +106,8 @@
  <dt class="author">Author:</dt>
  <dd class="author"><?cs var:changeset.author ?></dd>
  <dt class="message">Message:</dt>
- <dd class="message" id="searchable"><?cs var:changeset.message ?></dd>
+ <dd class="message" id="searchable"><?cs
+  alt:changeset.message ?>&nbsp;<?cs /alt ?></dd>
  <dt class="files">Files:</dt>
  <dd class="files">
   <ul><?cs each:item = changeset.changes ?>

templates/changeset.cs

@@ -106 +106 @@
  <dt class="author">Author:</dt>
  <dd class="author"><?cs var:changeset.author ?></dd>
  <dt class="message">Message:</dt>
- <dd class="message" id="searchable"><?cs var:changeset.message ?></dd>
+ <dd class="message" id="searchable"><?cs
+  alt:changeset.message ?>&nbsp;<?cs /alt ?></dd>
  <dt class="files">Files:</dt>
  <dd class="files">
   <ul><?cs each:item = changeset.changes ?>

[Shunichi Goto <gotoh@…>]

This is due to bug of PatchRenderer._diff_to_hdf() caused by the difference of 'line number' (1 base) and 'offset number' (0 base). Here is a patch:

patch.py

@@ -151 +151 @@
             if (command == ' ') != last_type:
                 last_type = command == ' '
                 blocks.append({'type': last_type and 'unmod' or 'mod',
-                               'base.offset': fromline, 'base.lines': [],
-                               'changed.offset': toline,'changed.lines': []})
+                               'base.offset': fromline-1, 'base.lines': [],
+                               'changed.offset': toline-1,'changed.lines': []})
             if command == ' ':
                 blocks[-1]['changed.lines'].append(line)
                 blocks[-1]['base.lines'].append(line)

[Christian Boos]

I'll try to add a unit test for that, too.

[Christopher Lenz]

I'll take this (sorry cboos). If you have a unit test ready, that'd be cool of course ;-)

[Christopher Lenz]

Fixed in [2264].