jenkins-bot merged this change.

View Change

[FEAT] cc: Remove empty sections

Bug: T140570
Change-Id: I0b5be6daaa4a827db6781ae5b90e381f682f31b5
---
M pywikibot/cosmetic_changes.py
M tests/cosmetic_changes_tests.py
2 files changed, 50 insertions(+), 2 deletions(-)

diff --git a/pywikibot/cosmetic_changes.py b/pywikibot/cosmetic_changes.py
index 0abd016..9c3a541 100755
--- a/pywikibot/cosmetic_changes.py
+++ b/pywikibot/cosmetic_changes.py
@@ -71,7 +71,8 @@
 import pywikibot
 
 from pywikibot import config, textlib
-from pywikibot.textlib import _MultiTemplateMatchBuilder, FILE_LINK_REGEX
+from pywikibot.textlib import (_MultiTemplateMatchBuilder, FILE_LINK_REGEX,
+                               _get_regexes)
 from pywikibot.tools import deprecated_args, first_lower, first_upper
 from pywikibot.tools import MediaWikiVersion
 
@@ -255,6 +256,7 @@
             self.replaceDeprecatedTemplates,
             # FIXME: fix bugs and re-enable
             # self.resolveHtmlEntities,
+            self.removeEmptySections,
             self.removeUselessSpaces,
             self.removeNonBreakingSpaceBeforePercent,
 
@@ -639,6 +641,30 @@
         text = pywikibot.html2unicode(text, ignore=ignore)
         return text
 
+    def removeEmptySections(self, text):
+        """Cleanup multiple empty sections."""
+        exceptions = ['comment', 'pre', 'source', 'nowiki', 'code',
+                      'startspace']
+
+        skippings = ['comment']
+        skip_regexes = _get_regexes(skippings, self.site)
+        skip_templates = {
+            'cs': ('Pahýl[ _]část',),  # stub section
+        }
+        if self.site.code in skip_templates:
+            for template in skip_templates[self.site.code]:
+                skip_regexes.append(
+                    re.compile(r'\{\{\s*' + template + r'\s*\}\}', re.I))
+        skip_regexes.append(re.compile(r'\s*'))
+
+        pattern = re.compile(r'\n(=+) *[^=]+? *\1(?:'
+                             + '|'.join(x.pattern for x in skip_regexes)
+                             + r')+(?=\1 *[^=]+? *\1)', re.I)
+        text = textlib.replaceExcept(text, pattern, r'\n',
+                                     exceptions=exceptions,
+                                     caseInsensitive=True)
+        return text
+
     def removeUselessSpaces(self, text):
         """Cleanup multiple or trailing spaces."""
         exceptions = ['comment', 'math', 'nowiki', 'pre', 'startspace',
diff --git a/tests/cosmetic_changes_tests.py b/tests/cosmetic_changes_tests.py
index 64c6a34..9695aae 100644
--- a/tests/cosmetic_changes_tests.py
+++ b/tests/cosmetic_changes_tests.py
@@ -1,7 +1,7 @@
 # -*- coding: utf-8 -*-
 """Test cosmetic_changes module."""
 #
-# (C) Pywikibot team, 2015-2017
+# (C) Pywikibot team, 2015-2018
 #
 # Distributed under the terms of the MIT license.
 #
@@ -56,6 +56,28 @@
             '&# # #0#>#x',
             self.cct.resolveHtmlEntities('&# # #0#>#x'))
 
+    def test_removeEmptySections(self):
+        """Test removeEmptySections method."""
+        # same level
+        self.assertEqual(
+            '\n==Bar==',
+            self.cct.removeEmptySections('\n== Foo ==\n\n==Bar=='))
+        # different level
+        self.assertEqual(
+            '\n===Foo===\n\n==Bar==',
+            self.cct.removeEmptySections('\n===Foo===\n\n==Bar=='))
+        self.assertEqual(
+            '\n==Foo==\n\n===Bar===',
+            self.cct.removeEmptySections('\n==Foo==\n\n===Bar==='))
+        # comment inside
+        self.assertEqual(
+            '\n==Bar==',
+            self.cct.removeEmptySections('\n==Foo==\n<!-- Baz -->\n==Bar=='))
+        # inside comment
+        self.assertEqual(
+            '<!--\n==Foo==\n\n==Bar==\n-->',
+            self.cct.removeEmptySections('<!--\n==Foo==\n\n==Bar==\n-->'))
+
     def test_removeUselessSpaces(self):
         """Test removeUselessSpaces method."""
         self.assertEqual('Foo bar',

To view, visit change 433914. To unsubscribe, visit settings.