Ensure that the default encoding used by python is never ASCII (needed when running a non frozen version of calibre on linux)

src/calibre/utils/icu.c

@@ -329,7 +329,7 @@ icu_lower(PyObject *self, PyObject *args) {
     PyMem_Free(input);
 
     return ret;
-}
+} // }}}
 
 // title {{{
 static PyObject *
@@ -374,9 +374,22 @@ icu_title(PyObject *self, PyObject *args) {
     PyMem_Free(input);
 
     return ret;
+} // }}}
+
+
+// set_default_encoding {{{
+static PyObject *
+icu_set_default_encoding(PyObject *self, PyObject *args) {
+    char *encoding;
+    if (!PyArg_ParseTuple(args, "s:setdefaultencoding", &encoding))
+        return NULL;
+    if (PyUnicode_SetDefaultEncoding(encoding))
+        return NULL;
+    Py_INCREF(Py_None);
+    return Py_None;
+
 }
-
-
+// }}}
 
 static PyMethodDef icu_methods[] = {
     {"upper", icu_upper, METH_VARARGS,
@@ -391,6 +404,10 @@ static PyMethodDef icu_methods[] = {
         "title(locale, unicode object) -> Title cased unicode object using locale rules."
     },
 
+    {"set_default_encoding", icu_set_default_encoding, METH_VARARGS,
+        "set_default_encoding(encoding) -> Set the default encoding for the python unicode implementation."
+    },
+
     {NULL}  /* Sentinel */
 };
 

src/calibre/utils/icu.py

@@ -6,6 +6,7 @@ __copyright__ = '2010, Kovid Goyal <kovid@kovidgoyal.net>'
 __docformat__ = 'restructuredtext en'
 
 # Setup code {{{
+import sys
 from functools import partial
 
 from calibre.constants import plugins
@@ -85,6 +86,12 @@ load_icu()
 load_collator()
 _icu_not_ok = _icu is None or _collator is None
 
+try:
+    if sys.getdefaultencoding().lower() == 'ascii':
+        _icu.set_default_encoding('utf-8')
+except:
+    pass
+
 # }}}
 
 ################# The string functions ########################################