From a3deae358084e5a4e1a2c0303d58aa47ae99574b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Fr=C3=A9d=C3=A9ric=20P=C3=A9ters?= <fpeters@entrouvert.com>
Date: Fri, 3 Apr 2015 12:36:40 +0200
Subject: [PATCH] odf2legi: parse static list styles (#6899)

The styles.xml file is supposed to be static but it can actually change in some
situations; we cannot therefore rely on known style names for lists and have to
parse the various levels to discover their formats.
---
 odf2legi/odf2legi.py | 72 +++++++++++++++++++++++++++++++++-----------
 1 file changed, 54 insertions(+), 18 deletions(-)

diff --git a/odf2legi/odf2legi.py b/odf2legi/odf2legi.py
index 371dfa4..68d417b 100644
--- a/odf2legi/odf2legi.py
+++ b/odf2legi/odf2legi.py
@@ -457,28 +457,32 @@ def handle_paragraph(parent, elem):
                 t_emph._children, para._children = para.getchildren(), [t_emph]
 
 def handle_list(parent, elem):
-
     style = elem.attrib.get('{%s}style-name' % TEXT_NS)
     level = 1
+    style_props = None
+
+    ancestor = elem
+    ancestorstyle = None
+    while True:
+        try:
+            gdparent = ancestor.parent.parent
+        except AttributeError:
+            break
+        if gdparent.tag != '{%s}list' % TEXT_NS:
+            break
+        level += 1
+        ancestor = gdparent
+        if ancestorstyle is None:
+            ancestorstyle = ancestor.attrib.get('{%s}style-name' % TEXT_NS)
+
     if style is None:
-        # means our parent was also a list
-        style_props = None
-        ancestor = elem
-        while True:
-            try:
-                gdparent = ancestor.parent.parent
-            except AttributeError:
-                break
-            if gdparent.tag != '{%s}list' % TEXT_NS:
-                break
-            level += 1
-            ancestor = gdparent
-        ancestorstyle = ancestor.attrib.get('{%s}style-name' % TEXT_NS)
-        style_props = STYLES.get('LIST:' + ancestorstyle)
+        if ancestorstyle:
+            style_props = STYLES.get('LIST:' + ancestorstyle)
     else:
         if 'parent' in STYLES.get(style, {}):
             style = STYLES.get(style).get('parent')
         style_props = STYLES.get('LIST:' + style)
+
     num_format = None
     if style_props:
         num_format = style_props.get('levels', {}).get(level, {}).get('format')
@@ -788,11 +792,13 @@ def convert(input_filename, output_filename):
             content = z.read(zfile)
         elif zfile == 'meta.xml':
             metadata = z.read(zfile)
+        elif zfile == 'styles.xml':
+            styles = z.read(zfile)
 
         if content and metadata:
             break
 
-    legi = convert_to_legi_xml(content, metadata)
+    legi = convert_to_legi_xml(content, metadata, styles)
     if debug:
         print legi
 
@@ -814,6 +820,31 @@ def convert(input_filename, output_filename):
         legiz.writestr(zi, z.read(zfile))
     legiz.close()
 
+def parse_static_styles(content_tree):
+    '''
+    Parse styles from styles.xml
+    '''
+    global STYLES
+
+    for elem in content_tree.findall('{%s}styles/{%s}list-style' % (OFFICE_NS, TEXT_NS)):
+        style_name = 'LIST:%s' % elem.attrib.get('{%s}name' % STYLE_NS)
+        STYLES[style_name] = {'levels': {}}
+        for level in elem.findall('{%s}list-level-style-number' % TEXT_NS):
+            num_level = level.attrib.get('{%s}level' % TEXT_NS)
+            num_format = level.attrib.get('{%s}num-format' % STYLE_NS)
+            STYLES[style_name]['levels'][int(num_level)] = {
+                'format': num_format,
+            }
+        for level in elem.findall('{%s}list-level-style-bullet' % TEXT_NS):
+            num_level = level.attrib.get('{%s}level' % TEXT_NS)
+            bullet_char = level.attrib.get('{%s}bullet-char' % TEXT_NS)
+            if not STYLES[style_name]['levels'].get(int(num_level)):
+                STYLES[style_name]['levels'][int(num_level)] = {}
+            STYLES[style_name]['levels'][int(num_level)].update({
+                'bullet': bullet_char,
+            })
+
+
 def parse_automatic_styles(content_tree):
     '''
     Parse styles created automatically and populate the global styles
@@ -870,7 +901,8 @@ def parse_automatic_styles(content_tree):
     # parse automatic list styles
     for elem in content_tree.findall('{%s}automatic-styles/{%s}list-style' % (OFFICE_NS, TEXT_NS)):
         style_name = 'LIST:%s' % elem.attrib.get('{%s}name' % STYLE_NS)
-        STYLES[style_name] = {'levels': {}}
+        if not style_name in STYLES:
+            STYLES[style_name] = {'levels': {}}
         for level in elem.findall('{%s}list-level-style-number' % TEXT_NS):
             num_level = level.attrib.get('{%s}level' % TEXT_NS)
             num_format = level.attrib.get('{%s}num-format' % STYLE_NS)
@@ -887,7 +919,7 @@ def parse_automatic_styles(content_tree):
             })
 
 
-def convert_to_legi_xml(content, metadata = None):
+def convert_to_legi_xml(content, metadata=None, styles=None):
     '''
     Convert a content.xml/metadata.xml pair from an odt file
     to the legi XML format.
@@ -907,6 +939,10 @@ def convert_to_legi_xml(content, metadata = None):
         metadata_element = ET.SubElement(legi, 'metadata')
         create_metadata(metadata_element, metadata_tree, content_tree)
 
+    if styles:
+        styles_tree = ET.ElementTree(ET.fromstring(styles))
+        parse_static_styles(styles_tree)
+
     current_top = [legi]
     current_legi = []
     speech = None