widgets/thread: add heuristics for broken quote levels

1 files changed, 40 insertions, 8 deletions

diff --git a/alot/widgets/thread.py b/alot/widgets/thread.py
index 78d974f2..2af22b07 100644
--- a/alot/widgets/thread.py
+++ b/alot/widgets/thread.py
@@ -192,10 +192,13 @@ class _TextPart(_MIMEPartWidget):
         fold_context = settings.get('thread_fold_context')
 
         blocks = self._split_quotes(text)
-        max_level = max((b[0] for b in blocks))
+        max_level = max((b['level'] for b in blocks))
 
         block_wgts = []
-        for level, lines in blocks:
+        for b in blocks:
+            level = b['level']
+            lines = b['data']
+
             if level == 0 or len(attrs_quote) < 1:
                 attr = attr_text
             else:
@@ -215,19 +218,48 @@ class _TextPart(_MIMEPartWidget):
         """
         lines = text.splitlines()
 
+        # first pass: assign a level to each line based on the number of quote
+        # characters at the beginning
         blocks = []
         for line in lines:
             level = 0
             m = re.match(self._QUOTE_REGEX, line)
             if m is not None:
-                g = m.group(0)
+                quote_prefix = m.group(0)
+                remainder    = line[len(quote_prefix):]
                 for c in self._QUOTE_CHARS:
-                    level += g.count(c)
-
-            if len(blocks) > 0 and blocks[-1][0] == level:
-                blocks[-1][1].append(line)
+                    level += quote_prefix.count(c)
             else:
-                blocks.append((level, [line]))
+                quote_prefix = ''
+                remainder    = line
+
+            if len(blocks) == 0 or blocks[-1]['level'] != level:
+                # start a new block
+                blocks.append({ 'level' : level, 'data' : [], 'data_empty' : True})
+
+            blocks[-1]['data'].append(line)
+            if len(remainder) > 0:
+                blocks[-1]['data_empty'] = False
+
+        # second pass: every quoted block where the remainder (line minus the
+        # leading quote characters) is empty gets merged into an adjacent quoted
+        # block, higher level preferred
+        for idx, b in enumerate(blocks):
+            if b['level'] > 0 and b['data_empty']:
+                b_prev = blocks[idx - 1] if idx > 0               else None
+                b_next = blocks[idx + 1] if idx + 1 < len(blocks) else None
+
+                merge_level = None
+                if (b_prev is not None and not b_prev['data_empty'] and
+                    b_prev['level'] > 0):
+                    merge_level = b_prev['level']
+
+                if (b_next is not None and not b_next['data_empty'] and
+                    (merge_level is None or merge_level < b_next['level'])):
+                    merge_level = b_next['level']
+
+                if merge_level is not None:
+                    b['level'] = merge_level
 
         return blocks