From d337debc923bd2d4d03d8ab2a87e49a247ff52f0 Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Mon, 12 May 2014 17:47:49 +0530
Subject: [PATCH] Edit book: Fix incorrect syntax highlighting on linux if the
 text contains non-BMP unicode characters.

---
 src/calibre/gui2/tweak_book/editor/syntax/base.py | 12 +++++++++++-
 src/calibre/gui2/tweak_book/editor/syntax/html.py |  1 +
 2 files changed, 12 insertions(+), 1 deletion(-)

diff --git a/src/calibre/gui2/tweak_book/editor/syntax/base.py b/src/calibre/gui2/tweak_book/editor/syntax/base.py
index 909f1ead34..191eb06eb2 100644
--- a/src/calibre/gui2/tweak_book/editor/syntax/base.py
+++ b/src/calibre/gui2/tweak_book/editor/syntax/base.py
@@ -6,6 +6,7 @@ from __future__ import (unicode_literals, division, absolute_import,
 __license__ = 'GPL v3'
 __copyright__ = '2013, Kovid Goyal <kovid at kovidgoyal.net>'
 
+import sys
 from collections import defaultdict
 
 from PyQt4.Qt import (
@@ -13,10 +14,14 @@ from PyQt4.Qt import (
 
 from ..themes import highlight_to_char_format
 from calibre.gui2.tweak_book.widgets import BusyCursor
+from calibre.utils.icu import utf16_length
+
+is_wide_build = sys.maxunicode >= 0x10ffff
 
 def run_loop(user_data, state_map, formats, text):
     state = user_data.state
     i = 0
+    fix_offsets = is_wide_build and utf16_length(text) != len(text)
     seen_states = defaultdict(set)
     while i < len(text):
         orig_i = i
@@ -24,7 +29,12 @@ def run_loop(user_data, state_map, formats, text):
         fmt = state_map[state.parse](state, text, i, formats, user_data)
         for num, f in fmt:
             if num > 0:
-                yield i, num, f
+                if fix_offsets:
+                    # We need to map offsets/lengths from UCS-4 to UTF-16 in
+                    # which non-BMP characters are two code points wide
+                    yield utf16_length(text[:i]), utf16_length(text[i:i+num]), f
+                else:
+                    yield i, num, f
                 i += num
         if orig_i == i and state.parse in seen_states[i]:
             # Something went wrong in the syntax highlighter
diff --git a/src/calibre/gui2/tweak_book/editor/syntax/html.py b/src/calibre/gui2/tweak_book/editor/syntax/html.py
index e506d4229b..9d92394e70 100644
--- a/src/calibre/gui2/tweak_book/editor/syntax/html.py
+++ b/src/calibre/gui2/tweak_book/editor/syntax/html.py
@@ -486,6 +486,7 @@ if __name__ == '__main__':
         <input disabled><input disabled /><span attr=<></span>
         <!-- Non-breaking spaces are rendered differently from normal spaces, so that they stand out -->
         <p>Some\xa0words\xa0separated\xa0by\xa0non\u2011breaking\xa0spaces and non\u2011breaking hyphens.</p>
+        <p>Some non-BMP unicode text:\U0001f431\U0001f431\U0001f431</p>
     </body>
 </html>
 ''', path_is_raw=True)