Fix wrong |flexion removal

JackPotte · Sep 29, 2024 · 1b77866 · 1b77866
1 parent ab9abd0
commit 1b77866
Show file tree

Hide file tree

Showing 2 changed files with 19 additions and 17 deletions.
diff --git a/src/wiktionary/fr_wiktionary_format.py b/src/wiktionary/fr_wiktionary_format.py
@@ -393,10 +393,16 @@ def treat_page(page):
                         if debug_level > 1:
                             print(' Definition paragraph')
                         add_language_code = True  # Paragraphe avec code langue dans les modèles lexicaux
+
                         if language_code is None:
+                            if debug_level > 0:
+                                pywikibot.output("  find language_code...")
                             # TODO: gérer les {{S|étymologie}} en milieu d'article
                             language_code = page_content[
                                 end_position + 1 + len(section) + 1:page_content.find('}}')].replace('|flexion', '')
+                            if debug_level > 0:
+                                pywikibot.output("  language_code found: " + language_code)
+
                             # TODO: num=, genre=...
                             summary = summary + ' ajout du {{langue|' + language_code + '}} manquant'
                             page_content = '== {{langue|' + language_code + '}} ==\n' + final_page_content[
@@ -677,8 +683,7 @@ def treat_page(page):
                             final_page_content, page_content = next_template(final_page_content, page_content,
                                                                              current_template, 'nocat=1')
                     else:
-                        final_page_content, page_content = next_template(
-                            final_page_content, page_content)
+                        final_page_content, page_content = next_template(final_page_content, page_content)
 
                 elif p < limit8:
                     if debug_level > 0:
@@ -794,8 +799,7 @@ def treat_page(page):
                     page_content.find('}}') < page_content.find('{{') or page_content.find('{{') == -1):
                 if debug_level > 1:
                     print('    possible duplicated "lang=" in ' + current_template)
-                final_page_content, page_content = next_template(
-                    final_page_content, page_content)
+                final_page_content, page_content = next_template(final_page_content, page_content)
                 # TODO bug with nested templates:
                 # https://fr.wiktionary.org/w/index.php?title=Utilisateur:JackBot/test_unitaire&diff=prev&oldid=25811164
                 # regex = r'({{' + re.escape(current_template) + r')\|lang=' + language_code + '(\|[^{}]*({{(.*?)}}|.)*[^{}]*\|lang=' + language_code + ')'

diff --git a/src/wiktionary/fr_wiktionary_functions.py b/src/wiktionary/fr_wiktionary_functions.py
@@ -1100,13 +1100,11 @@ def add_language_code_with_named_parameter_to_template(
 
 def next_template(final_page_content, current_page_content, current_template=None, language_code=None):
     if language_code is None:
-        final_page_content = final_page_content + \
-            current_page_content[:current_page_content.find('}}')+2]
+        final_page_content = final_page_content + current_page_content[:current_page_content.find('}}')+2]
     else:
-        final_page_content = final_page_content + \
-            current_template + '|' + language_code + '}}'
-    current_page_content = current_page_content[current_page_content.find(
-        '}}')+2:]
+        final_page_content = final_page_content + current_template + '|' + language_code + '}}'
+
+    current_page_content = current_page_content[current_page_content.find('}}')+2:]
     return final_page_content, current_page_content
 
 
@@ -1683,7 +1681,9 @@ def format_sections(page_content, summary):
 
     # Normalize sections title casing
     for f in re.findall(r'{{S\|([^}]+)}}', page_content):
-        page_content = page_content.replace(f, f.lower())
+        # Do not touch false ISO codes like "gallo-italique de Sicile"
+        if ' ' not in f:
+            page_content = page_content.replace(f, f.lower())
 
     # Replace deprecated aliases with language parameter
     page_content = page_content.replace('{{S|adj|', '{{S|adjectif|')
@@ -2938,15 +2938,13 @@ def treat_conjugation(page_content, final_page_content, summary, current_templat
             else:
                 page_content = '|groupe=3' + page_content
 
-    if (page_content.find(language_code) != -1 and page_content.find(language_code) < page_content.find(
-            '}}')) or language_code == 'fr':
-        final_page_content, page_content = next_template(
-            final_page_content, page_content)
+    if (page_content.find(language_code) != -1 and page_content.find(language_code) < page_content.find('}}')) \
+            or language_code == 'fr':
+        final_page_content, page_content = next_template(final_page_content, page_content)
     else:
         if page_content.find('|nocat=1') != -1:
             page_content = page_content[:page_content.find('|nocat=1')] + page_content[
-                page_content.find('|nocat=1') + len(
-                    '|nocat=1'):]
+                page_content.find('|nocat=1') + len('|nocat=1'):]
         final_page_content = final_page_content + '|' + language_code + '}}'
         page_content = page_content[page_content.find('}}') + 2:]
     return page_content, final_page_content, summary