From d5e7f83da4297f69094438d6fd515166aaeffe94 Mon Sep 17 00:00:00 2001
From: Charles Haley <>
Date: Thu, 5 Jul 2012 17:44:36 +0200
Subject: [PATCH] Change tag browser to handle multi-character "characters"
 that some languages use when doing first letter categorization.

---
 src/calibre/gui2/tag_browser/model.py | 27 ++++++++++++++++-----------
 1 file changed, 16 insertions(+), 11 deletions(-)

diff --git a/src/calibre/gui2/tag_browser/model.py b/src/calibre/gui2/tag_browser/model.py
index 5b3996ac76..0bc602622c 100644
--- a/src/calibre/gui2/tag_browser/model.py
+++ b/src/calibre/gui2/tag_browser/model.py
@@ -17,7 +17,7 @@ from PyQt4.Qt import (QAbstractItemModel, QIcon, QVariant, QFont, Qt,
 from calibre.gui2 import NONE, gprefs, config, error_dialog
 from calibre.library.database2 import Tag
 from calibre.utils.config import tweaks
-from calibre.utils.icu import sort_key, lower, strcmp, span_contractions
+from calibre.utils.icu import sort_key, lower, strcmp, contractions
 from calibre.library.field_metadata import TagsIcons, category_icon_map
 from calibre.gui2.dialogs.confirm_delete import confirm
 from calibre.utils.formatter import EvalFormatter
@@ -259,6 +259,16 @@ class TagsModel(QAbstractItemModel): # {{{
                 self.hidden_categories.add(cat)
         db.prefs.set('tag_browser_hidden_categories', list(self.hidden_categories))
 
+        conts = contractions()
+        if len(conts) == 0:
+            self.do_contraction = False
+        else:
+            self.do_contraction = True
+            nconts = set()
+            for s in conts:
+                nconts.add(icu_upper(s))
+            self.contraction_set = frozenset(nconts)
+
         self.db = db
         self._run_rebuild()
         self.endResetModel()
@@ -379,10 +389,6 @@ class TagsModel(QAbstractItemModel): # {{{
             else:
                 collapse_model = 'partition'
                 collapse_template = tweaks['categories_collapsed_popularity_template']
-        if True or get_lang() in self.langs_no_span_contractions:
-            use_span_contractions = False
-        else:
-            use_span_contractions = True
 
         def get_name_components(name):
             components = [t.strip() for t in name.split('.') if t.strip()]
@@ -423,15 +429,14 @@ class TagsModel(QAbstractItemModel): # {{{
                     if not tag.sort:
                         c = ' '
                     else:
-                        if not use_span_contractions:
+                        if not self.do_contraction:
                             c = icu_upper(tag.sort)[0]
                         else:
                             v = icu_upper(tag.sort)
-                            le = span_contractions(v)
-                            if le > 0:
-                                c = v[:le]
-                            else:
-                                c = v[0]
+                            c = v[0]
+                            for s in self.contraction_set:
+                                if len(s) > len(c) and v.startswith(s):
+                                    c = s
                     if c not in chardict:
                         chardict[c] = [idx, idx]
                     else: