From 76c0aeb57c5b863de091a0bc0b55cfd78895fd70 Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Sat, 12 May 2007 00:22:53 +0000
Subject: [PATCH] Fix handling of filenames that have non ascii characters on
 systems with encodings other than utf8

---
 src/libprs500/__init__.py              | 5 +++++
 src/libprs500/lrf/html/convert_from.py | 4 ++--
 src/libprs500/lrf/txt/convert_from.py  | 4 ++--
 3 files changed, 9 insertions(+), 4 deletions(-)

diff --git a/src/libprs500/__init__.py b/src/libprs500/__init__.py
index 50154a71a4..a06dc9626a 100644
--- a/src/libprs500/__init__.py
+++ b/src/libprs500/__init__.py
@@ -41,6 +41,11 @@ import sys
 iswindows = 'win32' in sys.platform.lower()
 isosx     = 'darwin' in sys.platform.lower()
 
+def filename_to_utf8(name):
+    '''Return C{name} encoded in utf8. Unhandled characters are replaced. '''
+    codec = 'cp1252' if iswindows else 'utf8'
+    return name.decode(codec, 'replace').encode('utf8')
+
 def extract(path, dir):
     import os
     ext = os.path.splitext(path)[1][1:].lower()
diff --git a/src/libprs500/lrf/html/convert_from.py b/src/libprs500/lrf/html/convert_from.py
index 24f0e30314..0ab43a2bd1 100644
--- a/src/libprs500/lrf/html/convert_from.py
+++ b/src/libprs500/lrf/html/convert_from.py
@@ -40,7 +40,7 @@ from libprs500.lrf.pylrs.pylrs import Paragraph, CR, Italic, ImageStream, TextBl
                                       RuledLine, BookSetting
 from libprs500.lrf.pylrs.pylrs import Span as _Span
 from libprs500.lrf import ConversionError, option_parser, Book
-from libprs500 import extract
+from libprs500 import extract, filename_to_utf8
 from libprs500.ptempfile import PersistentTemporaryFile
 
 class Span(_Span):
@@ -1063,7 +1063,7 @@ def main():
         sys.exit(1)
     src = args[0]
     if options.title == None:
-        options.title = os.path.splitext(os.path.basename(src))[0]
+        options.title = filename_to_utf8(os.path.splitext(os.path.basename(src))[0])
     process_file(src, options)
 
 def console_query(dirpath, candidate, docs):
diff --git a/src/libprs500/lrf/txt/convert_from.py b/src/libprs500/lrf/txt/convert_from.py
index 82899c8f10..7c3c6aace7 100644
--- a/src/libprs500/lrf/txt/convert_from.py
+++ b/src/libprs500/lrf/txt/convert_from.py
@@ -20,7 +20,7 @@ import os, sys
 from libprs500.lrf import ConversionError, option_parser
 from libprs500.lrf import Book
 from libprs500.lrf.pylrs.pylrs import Paragraph, Italic, Bold, BookSetting
-
+from libprs500 import filename_to_utf8
 
 def main():
     """ CLI for txt -> lrf conversions """
@@ -42,7 +42,7 @@ def main():
         sys.exit(1)
     src = os.path.abspath(os.path.expanduser(args[0]))
     if options.title == None:
-        options.title = os.path.splitext(os.path.basename(src))[0]
+        options.title = filename_to_utf8(os.path.splitext(os.path.basename(src))[0])
     try:
         convert_txt(src, options)
     except ConversionError, err: