Strip leading and trailing whitespace in Links to prevent wrong fam_hashes (when receiving redfam from db) since MySQL drops it

This commit is contained in:
2015-09-19 21:29:12 +02:00
parent 4e21b6696a
commit 7d6cd8bb30

View File

@@ -210,13 +210,15 @@ class RedFamParser( RedFam ):
# Parse content of heading for generating section links later # Parse content of heading for generating section links later
match = type( self ).__sectionhead_pat.search( heading ) match = type( self ).__sectionhead_pat.search( heading )
if match: if match:
self._heading = match.group(2).lstrip() self._heading = match.group(2).strip()
else: else:
raise RedFamHeadingError( heading ) raise RedFamHeadingError( heading )
# We get the pages in first [0] element iterating over # We get the pages in first [0] element iterating over
# wikilink_pat.findall( line ) # wikilink_pat.findall( line )
self._articlesList = [ link[0] for link # Strip leading and trailing whitespace in Links to prevent wrong
# fam_hashes (when receiving redfam from db) since MySQL drops it
self._articlesList = [ link.strip() for link
in wikilink_pat.findall( self._heading ) ] in wikilink_pat.findall( self._heading ) ]
# Catch sections with more then 8 articles, print error # Catch sections with more then 8 articles, print error
@@ -405,7 +407,7 @@ class RedFamWorker( RedFam ):
if 'article' in key and mysql_data[ key ]: if 'article' in key and mysql_data[ key ]:
articlesList.append( mysql_data[ key ] ) articlesList.append( mysql_data[ key ] )
super().__init__( articlesList, mysql_data[ 'beginning' ], super().__init__( articlesList, mysql_data[ 'beginning' ],
mysql_data[ 'ending' ], mysql_data[ 'red_page_id' ], mysql_data[ 'ending' ], mysql_data[ 'red_page_id' ],
mysql_data[ 'status' ], mysql_data[ 'fam_hash' ], mysql_data[ 'status' ], mysql_data[ 'fam_hash' ],
mysql_data[ 'heading' ] ) mysql_data[ 'heading' ] )