// Collator.startswith {{{
static PyObject *
icu_Collator_startswith(icu_Collator *self, PyObject *args, PyObject *kwargs) {
PyObject *a_ = NULL, *b_ = NULL;
int32_t asz = 0, bsz = 0;
UChar *a = NULL, *b = NULL;
uint8_t ans = 0;
if (!PyArg_ParseTuple(args, "OO", &a_, &b_)) return NULL;
a = python_to_icu(a_, &asz, 1);
if (a == NULL) goto end;
b = python_to_icu(b_, &bsz, 1);
if (b == NULL) goto end;
if (asz < bsz) goto end;
if (bsz == 0) { ans = 1; goto end; }
ans = ucol_equal(self->collator, a, bsz, b, bsz);
end:
if (a != NULL) free(a);
if (b != NULL) free(b);
if (PyErr_Occurred()) return NULL;
if (ans) { Py_RETURN_TRUE; }
Py_RETURN_FALSE;
} // }}}
// Collator.contains {{{
static PyObject *
icu_Collator_contains(icu_Collator *self, PyObject *args, PyObject *kwargs) {
PyObject *a_ = NULL, *b_ = NULL;
UChar *a = NULL, *b = NULL;
int32_t asz = 0, bsz = 0, pos = -1;
uint8_t found = 0;
UErrorCode status = U_ZERO_ERROR;
UStringSearch *search = NULL;
if (!PyArg_ParseTuple(args, "OO", &a_, &b_)) return NULL;
a = python_to_icu(a_, &asz, 1);
if (a == NULL) goto end;
if (asz == 0) { found = TRUE; goto end; }
b = python_to_icu(b_, &bsz, 1);
if (b == NULL) goto end;
search = usearch_openFromCollator(a, asz, b, bsz, self->collator, NULL, &status);
if (U_SUCCESS(status)) {
pos = usearch_first(search, &status);
if (pos != USEARCH_DONE) found = TRUE;
}
end:
if (search != NULL) usearch_close(search);
if (a != NULL) free(a);
if (b != NULL) free(b);
if (PyErr_Occurred()) return NULL;
if (found) Py_RETURN_TRUE;
Py_RETURN_FALSE;
} // }}}
static int
Matcher_init(Matcher *self, PyObject *args, PyObject *kwds)
{
PyObject *items = NULL, *p = NULL, *py_items = NULL, *level1 = NULL, *level2 = NULL, *level3 = NULL;
int32_t i = 0;
if (!PyArg_ParseTuple(args, "OOOO", &items, &level1, &level2, &level3)) return -1;
py_items = PySequence_Fast(items, "Must pass in two sequence objects");
if (py_items == NULL) goto end;
self->item_count = (uint32_t)PySequence_Size(items);
self->items = (UChar**)calloc(self->item_count, sizeof(UChar*));
self->item_lengths = (int32_t*)calloc(self->item_count, sizeof(uint32_t));
self->level1 = python_to_icu(level1, NULL, 1);
self->level2 = python_to_icu(level2, NULL, 1);
self->level3 = python_to_icu(level3, NULL, 1);
if (self->items == NULL || self->item_lengths == NULL ) { PyErr_NoMemory(); goto end; }
if (self->level1 == NULL || self->level2 == NULL || self->level3 == NULL) goto end;
for (i = 0; i < (int32_t)self->item_count; i++) {
p = PySequence_Fast_GET_ITEM(py_items, i);
self->items[i] = python_to_icu(p, self->item_lengths + i, 1);
if (self->items[i] == NULL) { PyErr_NoMemory(); goto end; }
}
end:
Py_XDECREF(py_items);
if (PyErr_Occurred()) { free_matcher(self); }
return (PyErr_Occurred()) ? -1 : 0;
}
static PyObject* icu_swap_case(PyObject *self, PyObject *input) {
PyObject *result = NULL;
UErrorCode status = U_ZERO_ERROR;
UChar *input_buf = NULL, *output_buf = NULL;
UChar32 *buf = NULL;
int32_t sz = 0, sz32 = 0, i = 0;
input_buf = python_to_icu(input, &sz);
if (input_buf == NULL) goto end;
output_buf = (UChar*) calloc(3 * sz, sizeof(UChar));
buf = (UChar32*) calloc(2 * sz, sizeof(UChar32));
if (output_buf == NULL || buf == NULL) { PyErr_NoMemory(); goto end; }
u_strToUTF32(buf, 2 * sz, &sz32, input_buf, sz, &status);
for (i = 0; i < sz32; i++) {
if (u_islower(buf[i])) buf[i] = u_toupper(buf[i]);
else if (u_isupper(buf[i])) buf[i] = u_tolower(buf[i]);
}
u_strFromUTF32(output_buf, 3*sz, &sz, buf, sz32, &status);
if (U_FAILURE(status)) { PyErr_SetString(PyExc_ValueError, u_errorName(status)); goto end; }
result = icu_to_python(output_buf, sz);
end:
if (input_buf != NULL) free(input_buf);
if (output_buf != NULL) free(output_buf);
if (buf != NULL) free(buf);
return result;
} // }}}
// character_name {{{
static PyObject *
icu_character_name(PyObject *self, PyObject *args) {
char name[512] = {0};
int32_t sz = 0, alias = 0;
UChar *buf;
UErrorCode status = U_ZERO_ERROR;
PyObject *palias = NULL, *result = NULL, *input = NULL;
UChar32 code = 0;
if (!PyArg_ParseTuple(args, "O|O", &input, &palias)) return NULL;
if (palias != NULL && PyObject_IsTrue(palias)) alias = 1;
buf = python_to_icu(input, &sz, 1);
if (buf == NULL) goto end;
U16_GET(buf, 0, 0, sz, code);
if (alias) {
sz = u_charName(code, U_CHAR_NAME_ALIAS, name, 511, &status);
} else {
sz = u_charName(code, U_UNICODE_CHAR_NAME, name, 511, &status);
}
if (U_FAILURE(status)) { PyErr_SetString(PyExc_ValueError, "Failed to get name for code"); goto end; }
result = PyUnicode_DecodeUTF8(name, sz, "strict");
end:
if (buf != NULL) free(buf);
return result;
} // }}}
// Collator.sort_key {{{
static PyObject *
icu_Collator_sort_key(icu_Collator *self, PyObject *args, PyObject *kwargs) {
int32_t sz = 0, key_size = 0, bsz = 0;
UChar *buf = NULL;
uint8_t *buf2 = NULL;
PyObject *ans = NULL, *input = NULL;
if (!PyArg_ParseTuple(args, "O", &input)) return NULL;
buf = python_to_icu(input, &sz, 1);
if (buf == NULL) return NULL;
bsz = 7 * sz + 1;
buf2 = (uint8_t*)calloc(bsz, sizeof(uint8_t));
if (buf2 == NULL) { PyErr_NoMemory(); goto end; }
key_size = ucol_getSortKey(self->collator, buf, sz, buf2, bsz);
if (key_size > bsz) {
buf2 = realloc(buf2, (key_size + 1) * sizeof(uint8_t));
if (buf2 == NULL) { PyErr_NoMemory(); goto end; }
key_size = ucol_getSortKey(self->collator, buf, sz, buf2, key_size + 1);
}
ans = PyBytes_FromStringAndSize((char*)buf2, key_size);
end:
if (buf != NULL) free(buf);
if (buf2 != NULL) free(buf2);
return ans;
} // }}}
static PyObject *
icu_normalize(PyObject *self, PyObject *args) {
UErrorCode status = U_ZERO_ERROR;
int32_t sz = 0, cap = 0, rsz = 0;
NORM_MODES mode;
UChar *dest = NULL, *source = NULL;
PyObject *ret = NULL, *src = NULL;
if (!PyArg_ParseTuple(args, "iO", &mode, &src)) return NULL;
const UNormalizer2 *n = NULL;
switch (mode) {
case NFC:
n = unorm2_getNFCInstance(&status);
break;
case NFKC:
n = unorm2_getNFKCInstance(&status);
break;
case NFD:
n = unorm2_getNFDInstance(&status);
break;
case NFKD:
n = unorm2_getNFKDInstance(&status);
break;
}
if (U_FAILURE(status)) {
PyErr_SetString(PyExc_ValueError, u_errorName(status));
goto end;
}
source = python_to_icu(src, &sz);
if (source == NULL) goto end;
cap = 2 * sz;
dest = (UChar*) calloc(cap, sizeof(UChar));
if (dest == NULL) { PyErr_NoMemory(); goto end; }
while (1) {
rsz = unorm2_normalize(n, source, sz, dest, cap, &status);
if (status == U_BUFFER_OVERFLOW_ERROR) {
cap *= 2;
dest = (UChar*) realloc(dest, cap*sizeof(UChar));
if (dest == NULL) { PyErr_NoMemory(); goto end; }
continue;
}
break;
}
if (U_FAILURE(status)) {
PyErr_SetString(PyExc_ValueError, u_errorName(status));
goto end;
}
ret = icu_to_python(dest, rsz);
end:
if (source != NULL) free(source);
if (dest != NULL) free(dest);
return ret;
} // }}}
// BreakIterator.index {{{
static PyObject *
icu_BreakIterator_index(icu_BreakIterator *self, PyObject *token) {
#if PY_VERSION_HEX >= 0x03030000
#error Not implemented for python >= 3.3
#endif
UChar *buf = NULL, *needle = NULL;
int32_t word_start = 0, p = 0, sz = 0, ans = -1, leading_hyphen = 0, trailing_hyphen = 0;
buf = python_to_icu(token, &sz);
if (buf == NULL) return NULL;
if (sz < 1) goto end;
needle = buf;
if (sz > 1 && IS_HYPHEN_CHAR(buf[0])) { needle = buf + 1; leading_hyphen = 1; sz -= 1; }
if (sz > 1 && IS_HYPHEN_CHAR(buf[sz-1])) trailing_hyphen = 1;
Py_BEGIN_ALLOW_THREADS;
p = ubrk_first(self->break_iterator);
while (p != UBRK_DONE) {
word_start = p; p = ubrk_next(self->break_iterator);
if (self->type == UBRK_WORD && ubrk_getRuleStatus(self->break_iterator) == UBRK_WORD_NONE)
continue; // We are not at the start of a word
if (self->text_len >= word_start + sz && memcmp(self->text + word_start, needle, sz * sizeof(UChar)) == 0) {
if (word_start > 0 && (
(leading_hyphen && !IS_HYPHEN_CHAR(self->text[word_start-1])) ||
(!leading_hyphen && IS_HYPHEN_CHAR(self->text[word_start-1]))
)) continue;
if (!trailing_hyphen && IS_HYPHEN_CHAR(self->text[word_start + sz])) continue;
if (p == UBRK_DONE || self->text_len <= word_start + sz) { ans = word_start; break; }
if (
// Check that the found word is followed by a word boundary
ubrk_isBoundary(self->break_iterator, word_start + sz) &&
// If there is a leading hyphen check that the leading
// hyphen is preceded by a word boundary
(!leading_hyphen || (word_start > 1 && ubrk_isBoundary(self->break_iterator, word_start - 2))) &&
// Check that there is a word boundary *after* the trailing
// hyphen. We cannot rely on ubrk_isBoundary() as that
// always returns true because of the trailing hyphen.
(!trailing_hyphen || ubrk_following(self->break_iterator, word_start + sz) == UBRK_DONE || ubrk_getRuleStatus(self->break_iterator) == UBRK_WORD_NONE)
) { ans = word_start; break; }
if (p != UBRK_DONE) ubrk_isBoundary(self->break_iterator, p); // Reset the iterator to its position before the call to ubrk_isBoundary()
}
}
if (leading_hyphen && ans > -1) ans -= 1;
#ifdef Py_UNICODE_WIDE
if (ans > 0) ans = u_countChar32(self->text, ans);
#endif
Py_END_ALLOW_THREADS;
end:
free(buf);
return Py_BuildValue("l", (long)ans);
} // }}}
// Collator.strcmp {{{
static PyObject *
icu_Collator_strcmp(icu_Collator *self, PyObject *args, PyObject *kwargs) {
PyObject *a_ = NULL, *b_ = NULL;
int32_t asz = 0, bsz = 0;
UChar *a = NULL, *b = NULL;
UCollationResult res = UCOL_EQUAL;
if (!PyArg_ParseTuple(args, "OO", &a_, &b_)) return NULL;
a = python_to_icu(a_, &asz, 1);
if (a == NULL) goto end;
b = python_to_icu(b_, &bsz, 1);
if (b == NULL) goto end;
res = ucol_strcoll(self->collator, a, asz, b, bsz);
end:
if (a != NULL) free(a); if (b != NULL) free(b);
return (PyErr_Occurred()) ? NULL : Py_BuildValue("i", res);
} // }}}
static int
Matcher_init(Matcher *self, PyObject *args, PyObject *kwds)
{
PyObject *items = NULL, *p = NULL, *py_items = NULL, *level1 = NULL, *level2 = NULL, *level3 = NULL, *collator = NULL;
int32_t i = 0;
UErrorCode status = U_ZERO_ERROR;
UCollator *col = NULL;
if (!PyArg_ParseTuple(args, "OOOOO", &items, &collator, &level1, &level2, &level3)) return -1;
// Clone the passed in collator (cloning is needed as collators are not thread safe)
if (!PyCapsule_CheckExact(collator)) { PyErr_SetString(PyExc_TypeError, "Collator must be a capsule"); return -1; }
col = (UCollator*)PyCapsule_GetPointer(collator, NULL);
if (col == NULL) return -1;
self->collator = ucol_safeClone(col, NULL, NULL, &status);
col = NULL;
if (U_FAILURE(status)) { self->collator = NULL; PyErr_SetString(PyExc_ValueError, u_errorName(status)); return -1; }
py_items = PySequence_Fast(items, "Must pass in two sequence objects");
if (py_items == NULL) goto end;
self->item_count = (uint32_t)PySequence_Size(items);
self->items = (UChar**)calloc(self->item_count, sizeof(UChar*));
self->item_lengths = (int32_t*)calloc(self->item_count, sizeof(uint32_t));
self->level1 = python_to_icu(level1, NULL);
self->level2 = python_to_icu(level2, NULL);
self->level3 = python_to_icu(level3, NULL);
if (self->items == NULL || self->item_lengths == NULL ) { PyErr_NoMemory(); goto end; }
if (self->level1 == NULL || self->level2 == NULL || self->level3 == NULL) goto end;
for (i = 0; i < (int32_t)self->item_count; i++) {
p = PySequence_Fast_GET_ITEM(py_items, i);
self->items[i] = python_to_icu(p, self->item_lengths + i);
if (self->items[i] == NULL) { PyErr_NoMemory(); goto end; }
}
end:
Py_XDECREF(py_items);
if (PyErr_Occurred()) { free_matcher(self); }
return (PyErr_Occurred()) ? -1 : 0;
}
// string_length {{{
static PyObject *
icu_string_length(PyObject *self, PyObject *src) {
int32_t sz = 0;
UChar *icu = NULL;
icu = python_to_icu(src, &sz);
if (icu == NULL) return NULL;
sz = u_countChar32(icu, sz);
free(icu);
return Py_BuildValue("l", (long)sz);
} // }}}
// roundtrip {{{
static PyObject *
icu_roundtrip(PyObject *self, PyObject *src) {
int32_t sz = 0;
UChar *icu = NULL;
PyObject *ret = NULL;
icu = python_to_icu(src, &sz);
if (icu != NULL) {
ret = icu_to_python(icu, sz);
free(icu);
}
return ret;
} // }}}
// string_length {{{
static PyObject *
icu_string_length(PyObject *self, PyObject *args) {
int32_t sz = 0;
UChar *icu = NULL;
PyObject *src = NULL;
if (!PyArg_ParseTuple(args, "O", &src)) return NULL;
icu = python_to_icu(src, &sz, 1);
if (icu == NULL) return NULL;
sz = u_countChar32(icu, sz);
free(icu);
return Py_BuildValue("i", sz);
} // }}}
// Collator.find {{{
static PyObject *
icu_Collator_find(icu_Collator *self, PyObject *args, PyObject *kwargs) {
#if PY_VERSION_HEX >= 0x03030000
#error Not implemented for python >= 3.3
#endif
PyObject *a_ = NULL, *b_ = NULL;
UChar *a = NULL, *b = NULL;
int32_t asz = 0, bsz = 0, pos = -1, length = -1;
UErrorCode status = U_ZERO_ERROR;
UStringSearch *search = NULL;
if (!PyArg_ParseTuple(args, "OO", &a_, &b_)) return NULL;
a = python_to_icu(a_, &asz, 1);
if (a == NULL) goto end;
b = python_to_icu(b_, &bsz, 1);
if (b == NULL) goto end;
search = usearch_openFromCollator(a, asz, b, bsz, self->collator, NULL, &status);
if (U_SUCCESS(status)) {
pos = usearch_first(search, &status);
if (pos != USEARCH_DONE) {
length = usearch_getMatchedLength(search);
#ifdef Py_UNICODE_WIDE
// We have to return number of unicode characters since the string
// could contain surrogate pairs which are represented as a single
// character in python wide builds
length = u_countChar32(b + pos, length);
pos = u_countChar32(b, pos);
#endif
} else pos = -1;
}
end:
if (search != NULL) usearch_close(search);
if (a != NULL) free(a);
if (b != NULL) free(b);
return (PyErr_Occurred()) ? NULL : Py_BuildValue("ii", pos, length);
} // }}}
// Matcher.calculate_scores {{{
static PyObject *
Matcher_calculate_scores(Matcher *self, PyObject *args) {
int32_t *final_positions = NULL, *p;
Match *matches = NULL;
bool ok = FALSE;
uint32_t i = 0, needle_char_len = 0, j = 0;
PyObject *items = NULL, *score = NULL, *positions = NULL, *pneedle = NULL;
UChar *needle = NULL;
if (!PyArg_ParseTuple(args, "O", &pneedle)) return NULL;
needle = python_to_icu(pneedle, NULL);
if (needle == NULL) return NULL;
needle_char_len = u_countChar32(needle, -1);
items = PyTuple_New(self->item_count);
positions = PyTuple_New(self->item_count);
matches = (Match*)calloc(self->item_count, sizeof(Match));
final_positions = (int32_t*) calloc(needle_char_len * self->item_count, sizeof(int32_t));
if (items == NULL || matches == NULL || final_positions == NULL || positions == NULL) {PyErr_NoMemory(); goto end;}
for (i = 0; i < self->item_count; i++) {
score = PyTuple_New(needle_char_len);
if (score == NULL) { PyErr_NoMemory(); goto end; }
PyTuple_SET_ITEM(positions, (Py_ssize_t)i, score);
}
Py_BEGIN_ALLOW_THREADS;
ok = match(self->items, self->item_lengths, self->item_count, needle, matches, final_positions, needle_char_len, self->collator, self->level1, self->level2, self->level3);
Py_END_ALLOW_THREADS;
if (ok) {
for (i = 0; i < self->item_count; i++) {
score = PyFloat_FromDouble(matches[i].score);
if (score == NULL) { PyErr_NoMemory(); goto end; }
PyTuple_SET_ITEM(items, (Py_ssize_t)i, score);
p = final_positions + (i * needle_char_len);
for (j = 0; j < needle_char_len; j++) {
score = PyLong_FromLong((long)p[j]);
if (score == NULL) { PyErr_NoMemory(); goto end; }
PyTuple_SET_ITEM(PyTuple_GET_ITEM(positions, (Py_ssize_t)i), (Py_ssize_t)j, score);
}
}
} else { PyErr_NoMemory(); goto end; }
end:
nullfree(needle);
nullfree(matches);
nullfree(final_positions);
if (PyErr_Occurred()) { Py_XDECREF(items); items = NULL; Py_XDECREF(positions); positions = NULL; return NULL; }
return Py_BuildValue("NN", items, positions);
} // }}}
// roundtrip {{{
static PyObject *
icu_roundtrip(PyObject *self, PyObject *args) {
int32_t sz = 0;
UChar *icu = NULL;
PyObject *ret = NULL, *src = NULL;
if (!PyArg_ParseTuple(args, "O", &src)) return NULL;
icu = python_to_icu(src, &sz, 1);
if (icu != NULL) {
ret = icu_to_python(icu, sz);
free(icu);
}
return ret;
} // }}}
// BreakIterator.set_text {{{
static PyObject *
icu_BreakIterator_set_text(icu_BreakIterator *self, PyObject *input) {
int32_t sz = 0;
UChar *buf = NULL;
UErrorCode status = U_ZERO_ERROR;
buf = python_to_icu(input, &sz);
if (buf == NULL) return NULL;
ubrk_setText(self->break_iterator, buf, sz, &status);
if (U_FAILURE(status)) {
free(buf); buf = NULL;
PyErr_SetString(PyExc_ValueError, u_errorName(status));
} else { self->text = buf; self->text_len = sz; }
Py_RETURN_NONE;
} // }}}
// string_length {{{
static PyObject *
icu_string_length(PyObject *self, PyObject *args) {
#if PY_VERSION_HEX >= 0x03030000
#error Not implemented for python >= 3.3
#endif
int32_t sz = 0;
UChar *icu = NULL;
PyObject *src = NULL;
if (!PyArg_ParseTuple(args, "O", &src)) return NULL;
icu = python_to_icu(src, &sz, 1);
if (icu == NULL) return NULL;
sz = u_countChar32(icu, sz);
free(icu);
return Py_BuildValue("i", sz);
} // }}}
// Collator.collation_order {{{
static PyObject *
icu_Collator_collation_order(icu_Collator *self, PyObject *a_) {
int32_t asz = 0;
UChar *a = NULL;
UErrorCode status = U_ZERO_ERROR;
UCollationElements *iter = NULL;
int order = 0, len = -1;
a = python_to_icu(a_, &asz);
if (a == NULL) goto end;
iter = ucol_openElements(self->collator, a, asz, &status);
if (U_FAILURE(status)) { PyErr_SetString(PyExc_ValueError, u_errorName(status)); goto end; }
order = ucol_next(iter, &status);
len = ucol_getOffset(iter);
end:
if (iter != NULL) ucol_closeElements(iter); iter = NULL;
if (a != NULL) free(a);
if (PyErr_Occurred()) return NULL;
return Py_BuildValue("ii", order, len);
} // }}}
static PyObject* icu_change_case(PyObject *self, PyObject *args) {
char *locale = NULL;
PyObject *input = NULL, *result = NULL;
int which = UPPER_CASE;
UErrorCode status = U_ZERO_ERROR;
UChar *input_buf = NULL, *output_buf = NULL;
int32_t sz = 0;
if (!PyArg_ParseTuple(args, "Oiz", &input, &which, &locale)) return NULL;
if (locale == NULL) {
PyErr_SetString(PyExc_NotImplementedError, "You must specify a locale"); // We deliberately use NotImplementedError so that this error can be unambiguously identified
return NULL;
}
input_buf = python_to_icu(input, &sz, 1);
if (input_buf == NULL) goto end;
output_buf = (UChar*) calloc(3 * sz, sizeof(UChar));
if (output_buf == NULL) { PyErr_NoMemory(); goto end; }
switch (which) {
case TITLE_CASE:
sz = u_strToTitle(output_buf, 3 * sz, input_buf, sz, NULL, locale, &status);
break;
case UPPER_CASE:
sz = u_strToUpper(output_buf, 3 * sz, input_buf, sz, locale, &status);
break;
default:
sz = u_strToLower(output_buf, 3 * sz, input_buf, sz, locale, &status);
}
if (U_FAILURE(status)) { PyErr_SetString(PyExc_ValueError, u_errorName(status)); goto end; }
result = icu_to_python(output_buf, sz);
end:
if (input_buf != NULL) free(input_buf);
if (output_buf != NULL) free(output_buf);
return result;
} // }}}
// BreakIterator.index {{{
static PyObject *
icu_BreakIterator_index(icu_BreakIterator *self, PyObject *args, PyObject *kwargs) {
#if PY_VERSION_HEX >= 0x03030000
#error Not implemented for python >= 3.3
#endif
UChar *buf = NULL;
int32_t prev = 0, p = 0, sz = 0, tsz = 0, ans = -1;
PyObject *token = NULL;
if (!PyArg_ParseTuple(args, "O", &token)) return NULL;
buf = python_to_icu(token, &sz, 1);
if (buf == NULL) return NULL;
if (sz < 1) goto end;
Py_BEGIN_ALLOW_THREADS;
p = ubrk_first(self->break_iterator);
while (p != UBRK_DONE) {
prev = p; p = ubrk_next(self->break_iterator);
if (self->type == UBRK_WORD && ubrk_getRuleStatus(self->break_iterator) == UBRK_WORD_NONE)
continue; // We are not at the start of a word
tsz = (p == UBRK_DONE) ? self->text_len - prev : p - prev;
if (sz == tsz && memcmp(self->text + prev, buf, sz * sizeof(UChar)) == 0) {
#ifdef PY_UNICODE_WIDE
ans = u_countChar32(self->text, prev);
#else
ans = prev;
#endif
break;
}
}
Py_END_ALLOW_THREADS;
end:
free(buf);
return Py_BuildValue("i", ans);
} // }}}
// normalize {{{
static PyObject *
icu_normalize(PyObject *self, PyObject *args) {
UErrorCode status = U_ZERO_ERROR;
int32_t sz = 0, mode = UNORM_DEFAULT, cap = 0, rsz = 0;
UChar *dest = NULL, *source = NULL;
PyObject *ret = NULL, *src = NULL;
if (!PyArg_ParseTuple(args, "iO", &mode, &src)) return NULL;
source = python_to_icu(src, &sz, 1);
if (source == NULL) goto end;
cap = 2 * sz;
dest = (UChar*) calloc(cap, sizeof(UChar));
if (dest == NULL) { PyErr_NoMemory(); goto end; }
while (1) {
rsz = unorm_normalize(source, sz, (UNormalizationMode)mode, 0, dest, cap, &status);
if (status == U_BUFFER_OVERFLOW_ERROR) {
cap *= 2;
dest = (UChar*) realloc(dest, cap*sizeof(UChar));
if (dest == NULL) { PyErr_NoMemory(); goto end; }
continue;
}
break;
}
if (U_FAILURE(status)) {
PyErr_SetString(PyExc_ValueError, u_errorName(status));
goto end;
}
ret = icu_to_python(dest, rsz);
end:
if (source != NULL) free(source);
if (dest != NULL) free(dest);
return ret;
} // }}}
