aboutsummaryrefslogtreecommitdiffstatshomepage
path: root/Objects/stringlib
diff options
context:
space:
mode:
Diffstat (limited to 'Objects/stringlib')
-rw-r--r--Objects/stringlib/codecs.h201
-rw-r--r--Objects/stringlib/fastsearch.h150
-rw-r--r--Objects/stringlib/find.h13
-rw-r--r--Objects/stringlib/unicode_format.h2
4 files changed, 199 insertions, 167 deletions
diff --git a/Objects/stringlib/codecs.h b/Objects/stringlib/codecs.h
index 0fc6b582d28..2beb604f116 100644
--- a/Objects/stringlib/codecs.h
+++ b/Objects/stringlib/codecs.h
@@ -263,50 +263,34 @@ STRINGLIB(utf8_encoder)(PyObject *unicode,
#define MAX_SHORT_UNICHARS 300 /* largest size we'll do on the stack */
Py_ssize_t i; /* index into s of next input byte */
- PyObject *result; /* result string object */
char *p; /* next free byte in output buffer */
- Py_ssize_t nallocated; /* number of result bytes allocated */
- Py_ssize_t nneeded; /* number of result bytes needed */
#if STRINGLIB_SIZEOF_CHAR > 1
- PyObject *errorHandler = NULL;
+ PyObject *error_handler_obj = NULL;
PyObject *exc = NULL;
PyObject *rep = NULL;
+ _Py_error_handler error_handler = _Py_ERROR_UNKNOWN;
#endif
#if STRINGLIB_SIZEOF_CHAR == 1
const Py_ssize_t max_char_size = 2;
- char stackbuf[MAX_SHORT_UNICHARS * 2];
#elif STRINGLIB_SIZEOF_CHAR == 2
const Py_ssize_t max_char_size = 3;
- char stackbuf[MAX_SHORT_UNICHARS * 3];
#else /* STRINGLIB_SIZEOF_CHAR == 4 */
const Py_ssize_t max_char_size = 4;
- char stackbuf[MAX_SHORT_UNICHARS * 4];
#endif
+ _PyBytesWriter writer;
assert(size >= 0);
+ _PyBytesWriter_Init(&writer);
- if (size <= MAX_SHORT_UNICHARS) {
- /* Write into the stack buffer; nallocated can't overflow.
- * At the end, we'll allocate exactly as much heap space as it
- * turns out we need.
- */
- nallocated = Py_SAFE_DOWNCAST(sizeof(stackbuf), size_t, int);
- result = NULL; /* will allocate after we're done */
- p = stackbuf;
- }
- else {
- if (size > PY_SSIZE_T_MAX / max_char_size) {
- /* integer overflow */
- return PyErr_NoMemory();
- }
- /* Overallocate on the heap, and give the excess back at the end. */
- nallocated = size * max_char_size;
- result = PyBytes_FromStringAndSize(NULL, nallocated);
- if (result == NULL)
- return NULL;
- p = PyBytes_AS_STRING(result);
+ if (size > PY_SSIZE_T_MAX / max_char_size) {
+ /* integer overflow */
+ return PyErr_NoMemory();
}
+ p = _PyBytesWriter_Alloc(&writer, size * max_char_size);
+ if (p == NULL)
+ return NULL;
+
for (i = 0; i < size;) {
Py_UCS4 ch = data[i++];
@@ -326,72 +310,118 @@ STRINGLIB(utf8_encoder)(PyObject *unicode,
}
#if STRINGLIB_SIZEOF_CHAR > 1
else if (Py_UNICODE_IS_SURROGATE(ch)) {
- Py_ssize_t newpos;
- Py_ssize_t repsize, k, startpos;
+ Py_ssize_t startpos, endpos, newpos;
+ Py_ssize_t k;
+ if (error_handler == _Py_ERROR_UNKNOWN)
+ error_handler = get_error_handler(errors);
+
startpos = i-1;
- rep = unicode_encode_call_errorhandler(
- errors, &errorHandler, "utf-8", "surrogates not allowed",
- unicode, &exc, startpos, startpos+1, &newpos);
- if (!rep)
- goto error;
-
- if (PyBytes_Check(rep))
- repsize = PyBytes_GET_SIZE(rep);
- else
- repsize = PyUnicode_GET_LENGTH(rep);
+ endpos = startpos+1;
+
+ while ((endpos < size) && Py_UNICODE_IS_SURROGATE(data[endpos]))
+ endpos++;
+
+ /* Only overallocate the buffer if it's not the last write */
+ writer.overallocate = (endpos < size);
+
+ switch (error_handler)
+ {
+ case _Py_ERROR_REPLACE:
+ memset(p, '?', endpos - startpos);
+ p += (endpos - startpos);
+ /* fall through the ignore handler */
+ case _Py_ERROR_IGNORE:
+ i += (endpos - startpos - 1);
+ break;
- if (repsize > max_char_size) {
- Py_ssize_t offset;
+ case _Py_ERROR_SURROGATEPASS:
+ for (k=startpos; k<endpos; k++) {
+ ch = data[k];
+ *p++ = (char)(0xe0 | (ch >> 12));
+ *p++ = (char)(0x80 | ((ch >> 6) & 0x3f));
+ *p++ = (char)(0x80 | (ch & 0x3f));
+ }
+ i += (endpos - startpos - 1);
+ break;
- if (result == NULL)
- offset = p - stackbuf;
- else
- offset = p - PyBytes_AS_STRING(result);
+ case _Py_ERROR_BACKSLASHREPLACE:
+ /* substract preallocated bytes */
+ writer.min_size -= max_char_size * (endpos - startpos);
+ p = backslashreplace(&writer, p,
+ unicode, startpos, endpos);
+ if (p == NULL)
+ goto error;
+ i += (endpos - startpos - 1);
+ break;
- if (nallocated > PY_SSIZE_T_MAX - repsize + max_char_size) {
- /* integer overflow */
- PyErr_NoMemory();
+ case _Py_ERROR_XMLCHARREFREPLACE:
+ /* substract preallocated bytes */
+ writer.min_size -= max_char_size * (endpos - startpos);
+ p = xmlcharrefreplace(&writer, p,
+ unicode, startpos, endpos);
+ if (p == NULL)
goto error;
+ i += (endpos - startpos - 1);
+ break;
+
+ case _Py_ERROR_SURROGATEESCAPE:
+ for (k=startpos; k<endpos; k++) {
+ ch = data[k];
+ if (!(0xDC80 <= ch && ch <= 0xDCFF))
+ break;
+ *p++ = (char)(ch & 0xff);
}
- nallocated += repsize - max_char_size;
- if (result != NULL) {
- if (_PyBytes_Resize(&result, nallocated) < 0)
- goto error;
- } else {
- result = PyBytes_FromStringAndSize(NULL, nallocated);
- if (result == NULL)
- goto error;
- Py_MEMCPY(PyBytes_AS_STRING(result), stackbuf, offset);
+ if (k >= endpos) {
+ i += (endpos - startpos - 1);
+ break;
}
- p = PyBytes_AS_STRING(result) + offset;
- }
+ startpos = k;
+ assert(startpos < endpos);
+ /* fall through the default handler */
+ default:
+ rep = unicode_encode_call_errorhandler(
+ errors, &error_handler_obj, "utf-8", "surrogates not allowed",
+ unicode, &exc, startpos, endpos, &newpos);
+ if (!rep)
+ goto error;
- if (PyBytes_Check(rep)) {
- char *prep = PyBytes_AS_STRING(rep);
- for(k = repsize; k > 0; k--)
- *p++ = *prep++;
- } else /* rep is unicode */ {
- enum PyUnicode_Kind repkind;
- void *repdata;
+ /* substract preallocated bytes */
+ writer.min_size -= max_char_size;
- if (PyUnicode_READY(rep) < 0)
- goto error;
- repkind = PyUnicode_KIND(rep);
- repdata = PyUnicode_DATA(rep);
+ if (PyBytes_Check(rep)) {
+ p = _PyBytesWriter_WriteBytes(&writer, p,
+ PyBytes_AS_STRING(rep),
+ PyBytes_GET_SIZE(rep));
+ }
+ else {
+ /* rep is unicode */
+ if (PyUnicode_READY(rep) < 0)
+ goto error;
- for(k=0; k<repsize; k++) {
- Py_UCS4 c = PyUnicode_READ(repkind, repdata, k);
- if (0x80 <= c) {
+ if (!PyUnicode_IS_ASCII(rep)) {
raise_encode_exception(&exc, "utf-8",
unicode,
i-1, i,
"surrogates not allowed");
goto error;
}
- *p++ = (char)c;
+
+ assert(PyUnicode_KIND(rep) == PyUnicode_1BYTE_KIND);
+ p = _PyBytesWriter_WriteBytes(&writer, p,
+ PyUnicode_DATA(rep),
+ PyUnicode_GET_LENGTH(rep));
}
+
+ if (p == NULL)
+ goto error;
+ Py_CLEAR(rep);
+
+ i = newpos;
}
- Py_CLEAR(rep);
+
+ /* If overallocation was disabled, ensure that it was the last
+ write. Otherwise, we missed an optimization */
+ assert(writer.overallocate || i == size);
}
else
#if STRINGLIB_SIZEOF_CHAR > 2
@@ -416,31 +446,18 @@ STRINGLIB(utf8_encoder)(PyObject *unicode,
#endif /* STRINGLIB_SIZEOF_CHAR > 1 */
}
- if (result == NULL) {
- /* This was stack allocated. */
- nneeded = p - stackbuf;
- assert(nneeded <= nallocated);
- result = PyBytes_FromStringAndSize(stackbuf, nneeded);
- }
- else {
- /* Cut back to size actually needed. */
- nneeded = p - PyBytes_AS_STRING(result);
- assert(nneeded <= nallocated);
- _PyBytes_Resize(&result, nneeded);
- }
-
#if STRINGLIB_SIZEOF_CHAR > 1
- Py_XDECREF(errorHandler);
+ Py_XDECREF(error_handler_obj);
Py_XDECREF(exc);
#endif
- return result;
+ return _PyBytesWriter_Finish(&writer, p);
#if STRINGLIB_SIZEOF_CHAR > 1
error:
Py_XDECREF(rep);
- Py_XDECREF(errorHandler);
+ Py_XDECREF(error_handler_obj);
Py_XDECREF(exc);
- Py_XDECREF(result);
+ _PyBytesWriter_Dealloc(&writer);
return NULL;
#endif
diff --git a/Objects/stringlib/fastsearch.h b/Objects/stringlib/fastsearch.h
index cda68e77c8d..98165ad1146 100644
--- a/Objects/stringlib/fastsearch.h
+++ b/Objects/stringlib/fastsearch.h
@@ -32,52 +32,98 @@
#define STRINGLIB_BLOOM(mask, ch) \
((mask & (1UL << ((ch) & (STRINGLIB_BLOOM_WIDTH -1)))))
-
Py_LOCAL_INLINE(Py_ssize_t)
-STRINGLIB(fastsearch_memchr_1char)(const STRINGLIB_CHAR* s, Py_ssize_t n,
- STRINGLIB_CHAR ch, unsigned char needle,
- int mode)
+STRINGLIB(find_char)(const STRINGLIB_CHAR* s, Py_ssize_t n, STRINGLIB_CHAR ch)
{
- if (mode == FAST_SEARCH) {
- const STRINGLIB_CHAR *ptr = s;
- const STRINGLIB_CHAR *e = s + n;
- while (ptr < e) {
- void *candidate = memchr((const void *) ptr, needle, (e - ptr) * sizeof(STRINGLIB_CHAR));
- if (candidate == NULL)
- return -1;
- ptr = (const STRINGLIB_CHAR *) _Py_ALIGN_DOWN(candidate, sizeof(STRINGLIB_CHAR));
- if (sizeof(STRINGLIB_CHAR) == 1 || *ptr == ch)
- return (ptr - s);
- /* False positive */
- ptr++;
- }
+ const STRINGLIB_CHAR *p, *e;
+
+ p = s;
+ e = s + n;
+ if (n > 10) {
+#if STRINGLIB_SIZEOF_CHAR == 1
+ p = memchr(s, ch, n);
+ if (p != NULL)
+ return (p - s);
return -1;
+#else
+ /* use memchr if we can choose a needle without two many likely
+ false positives */
+ unsigned char needle = ch & 0xff;
+ /* If looking for a multiple of 256, we'd have too
+ many false positives looking for the '\0' byte in UCS2
+ and UCS4 representations. */
+ if (needle != 0) {
+ while (p < e) {
+ void *candidate = memchr(p, needle,
+ (e - p) * sizeof(STRINGLIB_CHAR));
+ if (candidate == NULL)
+ return -1;
+ p = (const STRINGLIB_CHAR *)
+ _Py_ALIGN_DOWN(candidate, sizeof(STRINGLIB_CHAR));
+ if (*p == ch)
+ return (p - s);
+ /* False positive */
+ p++;
+ }
+ return -1;
+ }
+#endif
}
+ while (p < e) {
+ if (*p == ch)
+ return (p - s);
+ p++;
+ }
+ return -1;
+}
+
+Py_LOCAL_INLINE(Py_ssize_t)
+STRINGLIB(rfind_char)(const STRINGLIB_CHAR* s, Py_ssize_t n, STRINGLIB_CHAR ch)
+{
+ const STRINGLIB_CHAR *p;
#ifdef HAVE_MEMRCHR
/* memrchr() is a GNU extension, available since glibc 2.1.91.
it doesn't seem as optimized as memchr(), but is still quite
- faster than our hand-written loop in FASTSEARCH below */
- else if (mode == FAST_RSEARCH) {
- while (n > 0) {
- const STRINGLIB_CHAR *found;
- void *candidate = memrchr((const void *) s, needle, n * sizeof(STRINGLIB_CHAR));
- if (candidate == NULL)
- return -1;
- found = (const STRINGLIB_CHAR *) _Py_ALIGN_DOWN(candidate, sizeof(STRINGLIB_CHAR));
- n = found - s;
- if (sizeof(STRINGLIB_CHAR) == 1 || *found == ch)
- return n;
- /* False positive */
- }
+ faster than our hand-written loop below */
+
+ if (n > 10) {
+#if STRINGLIB_SIZEOF_CHAR == 1
+ p = memrchr(s, ch, n);
+ if (p != NULL)
+ return (p - s);
return -1;
- }
+#else
+ /* use memrchr if we can choose a needle without two many likely
+ false positives */
+ unsigned char needle = ch & 0xff;
+ /* If looking for a multiple of 256, we'd have too
+ many false positives looking for the '\0' byte in UCS2
+ and UCS4 representations. */
+ if (needle != 0) {
+ while (n > 0) {
+ void *candidate = memrchr(s, needle,
+ n * sizeof(STRINGLIB_CHAR));
+ if (candidate == NULL)
+ return -1;
+ p = (const STRINGLIB_CHAR *)
+ _Py_ALIGN_DOWN(candidate, sizeof(STRINGLIB_CHAR));
+ n = p - s;
+ if (*p == ch)
+ return n;
+ /* False positive */
+ }
+ return -1;
+ }
#endif
- else {
- assert(0); /* Should never get here */
- return 0;
}
-
-#undef DO_MEMCHR
+#endif /* HAVE_MEMRCHR */
+ p = s + n;
+ while (p > s) {
+ p--;
+ if (*p == ch)
+ return (p - s);
+ }
+ return -1;
}
Py_LOCAL_INLINE(Py_ssize_t)
@@ -99,25 +145,11 @@ FASTSEARCH(const STRINGLIB_CHAR* s, Py_ssize_t n,
if (m <= 0)
return -1;
/* use special case for 1-character strings */
- if (n > 10 && (mode == FAST_SEARCH
-#ifdef HAVE_MEMRCHR
- || mode == FAST_RSEARCH
-#endif
- )) {
- /* use memchr if we can choose a needle without two many likely
- false positives */
- unsigned char needle;
- needle = p[0] & 0xff;
-#if STRINGLIB_SIZEOF_CHAR > 1
- /* If looking for a multiple of 256, we'd have too
- many false positives looking for the '\0' byte in UCS2
- and UCS4 representations. */
- if (needle != 0)
-#endif
- return STRINGLIB(fastsearch_memchr_1char)
- (s, n, p[0], needle, mode);
- }
- if (mode == FAST_COUNT) {
+ if (mode == FAST_SEARCH)
+ return STRINGLIB(find_char)(s, n, p[0]);
+ else if (mode == FAST_RSEARCH)
+ return STRINGLIB(rfind_char)(s, n, p[0]);
+ else { /* FAST_COUNT */
for (i = 0; i < n; i++)
if (s[i] == p[0]) {
count++;
@@ -125,14 +157,6 @@ FASTSEARCH(const STRINGLIB_CHAR* s, Py_ssize_t n,
return maxcount;
}
return count;
- } else if (mode == FAST_SEARCH) {
- for (i = 0; i < n; i++)
- if (s[i] == p[0])
- return i;
- } else { /* FAST_RSEARCH */
- for (i = n - 1; i > -1; i--)
- if (s[i] == p[0])
- return i;
}
return -1;
}
diff --git a/Objects/stringlib/find.h b/Objects/stringlib/find.h
index 14815f6e62b..a7065fcbc06 100644
--- a/Objects/stringlib/find.h
+++ b/Objects/stringlib/find.h
@@ -123,11 +123,6 @@ STRINGLIB(parse_args_finds)(const char * function_name, PyObject *args,
/*
Wraps stringlib_parse_args_finds() and additionally ensures that the
first argument is a unicode object.
-
-Note that we receive a pointer to the pointer of the substring object,
-so when we create that object in this function we don't DECREF it,
-because it continues living in the caller functions (those functions,
-after finishing using the substring, must DECREF it).
*/
Py_LOCAL_INLINE(int)
@@ -135,14 +130,10 @@ STRINGLIB(parse_args_finds_unicode)(const char * function_name, PyObject *args,
PyObject **substring,
Py_ssize_t *start, Py_ssize_t *end)
{
- PyObject *tmp_substring;
-
- if(STRINGLIB(parse_args_finds)(function_name, args, &tmp_substring,
+ if(STRINGLIB(parse_args_finds)(function_name, args, substring,
start, end)) {
- tmp_substring = PyUnicode_FromObject(tmp_substring);
- if (!tmp_substring)
+ if (ensure_unicode(*substring) < 0)
return 0;
- *substring = tmp_substring;
return 1;
}
return 0;
diff --git a/Objects/stringlib/unicode_format.h b/Objects/stringlib/unicode_format.h
index be09b5f6fa7..14fa28ea54a 100644
--- a/Objects/stringlib/unicode_format.h
+++ b/Objects/stringlib/unicode_format.h
@@ -67,7 +67,7 @@ SubString_new_object(SubString *str)
return PyUnicode_Substring(str->str, str->start, str->end);
}
-/* return a new string. if str->str is NULL, return None */
+/* return a new string. if str->str is NULL, return a new empty string */
Py_LOCAL_INLINE(PyObject *)
SubString_new_object_or_empty(SubString *str)
{
generated by cgit v1.2.3 (git 2.39.1) at 2025-06-02 20:09:13 +0000