openlp-core team mailing list archive

[Jonathan Corwin <j@xxxxxxxxxxxx>]

Jonathan Corwin has proposed merging lp:~j-corwin/openlp/bug-742825 into lp:openlp.

Requested reviews:
  OpenLP Core (openlp-core)
Related bugs:
  Bug #742825 in OpenLP: "Search lyrics/titles and apostrophes "
  https://bugs.launchpad.net/openlp/+bug/742825

For more details, see:
https://code.launchpad.net/~j-corwin/openlp/bug-742825/+merge/63304

Replace apostrophes with an empty string instead of a space. This allows the searcher to find words containing "it's" and "Lord's" should they type "its" or "Lords" (and vice versa).

Also reinstate the setting of search_lyrics by clean_song. 

A Tools -> Re-index songs is necessary to update the search fields in the database.
-- 
https://code.launchpad.net/~j-corwin/openlp/bug-742825/+merge/63304
Your team OpenLP Core is requested to review the proposed merge of lp:~j-corwin/openlp/bug-742825 into lp:openlp.

=== modified file 'openlp/core/lib/mediamanageritem.py'
--- openlp/core/lib/mediamanageritem.py	2011-06-01 18:58:27 +0000
+++ openlp/core/lib/mediamanageritem.py	2011-06-02 21:33:26 +0000
@@ -92,7 +92,7 @@
         """
         QtGui.QWidget.__init__(self, parent)
         self.hide()
-        self.whitespace = re.compile(r'\W+', re.UNICODE)
+        self.whitespace = re.compile(r'[\W_]+', re.UNICODE)
         self.plugin = plugin
         visible_title = self.plugin.getString(StringContent.VisibleName)
         self.title = unicode(visible_title[u'title'])

=== modified file 'openlp/plugins/songs/lib/__init__.py'
--- openlp/plugins/songs/lib/__init__.py	2011-05-28 17:01:41 +0000
+++ openlp/plugins/songs/lib/__init__.py	2011-06-02 21:33:26 +0000
@@ -32,6 +32,9 @@
 from db import Author
 from ui import SongStrings
 
+WHITESPACE = re.compile(r'[\W_]+', re.UNICODE)
+APOSTROPHE = re.compile(u'[\'`’ʻ′]', re.UNICODE)
+
 class VerseType(object):
     """
     VerseType provides an enumeration for the tags that may be associated
@@ -246,6 +249,12 @@
         return None
     return filter(lambda item: item[1] == choice[0], encodings)[0][0]
 
+def clean_string(string):
+    """
+    Strips punctuation from the passed string to assist searching
+    """
+    return WHITESPACE.sub(u' ', APOSTROPHE.sub(u'', string)).lower()
+
 def clean_song(manager, song):
     """
     Cleans the search title, rebuilds the search lyrics, adds a default author
@@ -262,9 +271,8 @@
     if song.alternate_title is None:
         song.alternate_title = u''
     song.alternate_title = song.alternate_title.strip()
-    whitespace = re.compile(r'\W+', re.UNICODE)
-    song.search_title = (whitespace.sub(u' ', song.title).strip() + u'@' +
-        whitespace.sub(u' ', song.alternate_title).strip()).strip().lower()
+    song.search_title = clean_string(song.title) + u'@' + \
+        clean_string(song.alternate_title)
     # Only do this, if we the song is a 1.9.4 song (or older).
     if song.lyrics.find(u'<lyrics language="en">') != -1:
         # Remove the old "language" attribute from lyrics tag (prior to 1.9.5).
@@ -273,8 +281,8 @@
         song.lyrics = song.lyrics.replace(
             u'<lyrics language="en">', u'<lyrics>')
         verses = SongXML().get_verses(song.lyrics)
-        lyrics = u' '.join([whitespace.sub(u' ', verse[1]) for verse in verses])
-        song.search_lyrics = lyrics.lower()
+        song.search_lyrics = u' '.join([clean_string(verse[1])
+            for verse in verses])
         # We need a new and clean SongXML instance.
         sxml = SongXML()
         # Rebuild the song's verses, to remove any wrong verse names (for
@@ -316,6 +324,11 @@
             if order not in compare_order:
                 song.verse_order = u''
                 break
+    else:
+        verses = SongXML().get_verses(song.lyrics)
+        song.search_lyrics = u' '.join([clean_string(verse[1])
+            for verse in verses])
+
     # The song does not have any author, add one.
     if not song.authors:
         name = SongStrings.AuthorUnknown

=== modified file 'openlp/plugins/songs/lib/mediaitem.py'
--- openlp/plugins/songs/lib/mediaitem.py	2011-05-30 20:03:09 +0000
+++ openlp/plugins/songs/lib/mediaitem.py	2011-06-02 21:33:26 +0000
@@ -38,7 +38,8 @@
 from openlp.core.lib.ui import UiStrings
 from openlp.plugins.songs.forms import EditSongForm, SongMaintenanceForm, \
     SongImportForm, SongExportForm
-from openlp.plugins.songs.lib import OpenLyrics, SongXML, VerseType
+from openlp.plugins.songs.lib import OpenLyrics, SongXML, VerseType, \
+    clean_string
 from openlp.plugins.songs.lib.db import Author, Song
 from openlp.plugins.songs.lib.ui import SongStrings
 
@@ -181,13 +182,14 @@
         elif search_type == SongSearch.Titles:
             log.debug(u'Titles Search')
             search_results = self.plugin.manager.get_all_objects(Song,
-                Song.search_title.like(u'%' + self.whitespace.sub(u' ',
-                search_keywords.lower()) + u'%'))
+                Song.search_title.like(u'%' + clean_string(search_keywords) +
+                    u'%'))
             self.displayResultsSong(search_results)
         elif search_type == SongSearch.Lyrics:
             log.debug(u'Lyrics Search')
             search_results = self.plugin.manager.get_all_objects(Song,
-                Song.search_lyrics.like(u'%' + search_keywords.lower() + u'%'))
+                Song.search_lyrics.like(u'%' + clean_string(search_keywords) +
+                    u'%'))
             self.displayResultsSong(search_results)
         elif search_type == SongSearch.Authors:
             log.debug(u'Authors Search')
@@ -198,16 +200,16 @@
         elif search_type == SongSearch.Themes:
             log.debug(u'Theme Search')
             search_results = self.plugin.manager.get_all_objects(Song,
-                Song.theme_name.like(u'%' + self.whitespace.sub(u' ',
-                search_keywords) + u'%'))
+                Song.theme_name.like(u'%' + search_keywords + u'%'))
             self.displayResultsSong(search_results)
         self.check_search_result()
 
     def searchEntire(self, search_keywords):
         return self.plugin.manager.get_all_objects(Song,
-            or_(Song.search_title.like(u'%' + self.whitespace.sub(u' ',
-            search_keywords.lower()) + u'%'),
-            Song.search_lyrics.like(u'%' + search_keywords.lower() + u'%'),
+            or_(Song.search_title.like(u'%' + clean_string(search_keywords)
+                + u'%'),
+            Song.search_lyrics.like(u'%' + clean_string(search_keywords)
+                + u'%'),
             Song.comments.like(u'%' + search_keywords.lower() + u'%')))
 
     def onSongListLoad(self):