maxchar_str2 = PyUnicode_MAX_CHAR_VALUE(str2);
/* Replacing str1 with str2 may cause a maxchar reduction in the
result string. */
- mayshrink = (maxchar_str2 < maxchar);
+ mayshrink = (maxchar_str2 < maxchar_str1) && (maxchar == maxchar_str1);
- maxchar = MAX_MAXCHAR(maxchar, maxchar_str2);
+ maxchar = Py_MAX(maxchar, maxchar_str2);
if (len1 == len2) {
/* same length */
}
newlen = writer->pos + length;
- maxchar = MAX_MAXCHAR(maxchar, writer->min_char);
++ maxchar = Py_MAX(maxchar, writer->min_char);
+
if (writer->buffer == NULL) {
- if (writer->overallocate) {
+ assert(!writer->readonly);
+ if (writer->overallocate && newlen <= (PY_SSIZE_T_MAX - newlen / 4)) {
/* overallocate 25% to limit the number of resize */
- if (newlen <= (PY_SSIZE_T_MAX - newlen / 4))
- newlen += newlen / 4;
- if (newlen < writer->min_length)
- newlen = writer->min_length;
+ newlen += newlen / 4;
}
+ if (newlen < writer->min_length)
+ newlen = writer->min_length;
+
writer->buffer = PyUnicode_New(newlen, maxchar);
if (writer->buffer == NULL)
return -1;
return (Py_UCS4) -1;
}
-PyObject *
-PyUnicode_Format(PyObject *format, PyObject *args)
-{
- Py_ssize_t fmtcnt, fmtpos, arglen, argidx;
- int args_owned = 0;
- PyObject *dict = NULL;
- PyObject *temp = NULL;
- PyObject *second = NULL;
- PyObject *uformat;
- void *fmt;
- enum PyUnicode_Kind kind, fmtkind;
- _PyUnicodeWriter writer;
- Py_ssize_t sublen;
- Py_UCS4 maxchar;
+/* Parse options of an argument: flags, width, precision.
+ Handle also "%(name)" syntax.
- if (format == NULL || args == NULL) {
- PyErr_BadInternalCall();
- return NULL;
- }
- uformat = PyUnicode_FromObject(format);
- if (uformat == NULL)
- return NULL;
- if (PyUnicode_READY(uformat) == -1) {
- Py_DECREF(uformat);
- return NULL;
- }
+ Return 0 if the argument has been formatted into arg->str.
+ Return 1 if the argument has been written into ctx->writer,
+ Raise an exception and return -1 on error. */
+static int
+unicode_format_arg_parse(struct unicode_formatter_t *ctx,
+ struct unicode_format_arg_t *arg)
+{
+#define FORMAT_READ(ctx) \
+ PyUnicode_READ((ctx)->fmtkind, (ctx)->fmtdata, (ctx)->fmtpos)
- fmt = PyUnicode_DATA(uformat);
- fmtkind = PyUnicode_KIND(uformat);
- fmtcnt = PyUnicode_GET_LENGTH(uformat);
- fmtpos = 0;
+ PyObject *v;
- _PyUnicodeWriter_Init(&writer, fmtcnt + 100);
+ if (arg->ch == '(') {
+ /* Get argument value from a dictionary. Example: "%(name)s". */
+ Py_ssize_t keystart;
+ Py_ssize_t keylen;
+ PyObject *key;
+ int pcount = 1;
- if (PyTuple_Check(args)) {
- arglen = PyTuple_Size(args);
- argidx = 0;
- }
- else {
- arglen = -1;
- argidx = -2;
+ if (ctx->dict == NULL) {
+ PyErr_SetString(PyExc_TypeError,
+ "format requires a mapping");
+ return -1;
+ }
+ ++ctx->fmtpos;
+ --ctx->fmtcnt;
+ keystart = ctx->fmtpos;
+ /* Skip over balanced parentheses */
+ while (pcount > 0 && --ctx->fmtcnt >= 0) {
+ arg->ch = FORMAT_READ(ctx);
+ if (arg->ch == ')')
+ --pcount;
+ else if (arg->ch == '(')
+ ++pcount;
+ ctx->fmtpos++;
+ }
+ keylen = ctx->fmtpos - keystart - 1;
+ if (ctx->fmtcnt < 0 || pcount > 0) {
+ PyErr_SetString(PyExc_ValueError,
+ "incomplete format key");
+ return -1;
+ }
+ key = PyUnicode_Substring(ctx->fmtstr,
+ keystart, keystart + keylen);
+ if (key == NULL)
+ return -1;
+ if (ctx->args_owned) {
+ Py_DECREF(ctx->args);
+ ctx->args_owned = 0;
+ }
+ ctx->args = PyObject_GetItem(ctx->dict, key);
+ Py_DECREF(key);
+ if (ctx->args == NULL)
+ return -1;
+ ctx->args_owned = 1;
+ ctx->arglen = -1;
+ ctx->argidx = -2;
+ }
+
+ /* Parse flags. Example: "%+i" => flags=F_SIGN. */
+ while (--ctx->fmtcnt >= 0) {
+ arg->ch = FORMAT_READ(ctx);
+ ctx->fmtpos++;
+ switch (arg->ch) {
+ case '-': arg->flags |= F_LJUST; continue;
+ case '+': arg->flags |= F_SIGN; continue;
+ case ' ': arg->flags |= F_BLANK; continue;
+ case '#': arg->flags |= F_ALT; continue;
+ case '0': arg->flags |= F_ZERO; continue;
+ }
+ break;
}
- if (PyMapping_Check(args) && !PyTuple_Check(args) && !PyUnicode_Check(args))
- dict = args;
- while (--fmtcnt >= 0) {
- if (PyUnicode_READ(fmtkind, fmt, fmtpos) != '%') {
- Py_ssize_t nonfmtpos;
- nonfmtpos = fmtpos++;
- while (fmtcnt >= 0 &&
- PyUnicode_READ(fmtkind, fmt, fmtpos) != '%') {
- fmtpos++;
- fmtcnt--;
+ /* Parse width. Example: "%10s" => width=10 */
+ if (arg->ch == '*') {
+ v = unicode_format_getnextarg(ctx);
+ if (v == NULL)
+ return -1;
+ if (!PyLong_Check(v)) {
+ PyErr_SetString(PyExc_TypeError,
+ "* wants int");
+ return -1;
+ }
+ arg->width = PyLong_AsSsize_t(v);
+ if (arg->width == -1 && PyErr_Occurred())
+ return -1;
+ if (arg->width < 0) {
+ arg->flags |= F_LJUST;
+ arg->width = -arg->width;
+ }
+ if (--ctx->fmtcnt >= 0) {
+ arg->ch = FORMAT_READ(ctx);
+ ctx->fmtpos++;
+ }
+ }
+ else if (arg->ch >= '0' && arg->ch <= '9') {
+ arg->width = arg->ch - '0';
+ while (--ctx->fmtcnt >= 0) {
+ arg->ch = FORMAT_READ(ctx);
+ ctx->fmtpos++;
+ if (arg->ch < '0' || arg->ch > '9')
+ break;
+ /* Since arg->ch is unsigned, the RHS would end up as unsigned,
+ mixing signed and unsigned comparison. Since arg->ch is between
+ '0' and '9', casting to int is safe. */
+ if (arg->width > (PY_SSIZE_T_MAX - ((int)arg->ch - '0')) / 10) {
+ PyErr_SetString(PyExc_ValueError,
+ "width too big");
+ return -1;
}
- if (fmtcnt < 0)
- fmtpos--;
- sublen = fmtpos - nonfmtpos;
- maxchar = _PyUnicode_FindMaxChar(uformat,
- nonfmtpos, nonfmtpos + sublen);
- if (_PyUnicodeWriter_Prepare(&writer, sublen, maxchar) == -1)
- goto onError;
+ arg->width = arg->width*10 + (arg->ch - '0');
+ }
+ }
- _PyUnicode_FastCopyCharacters(writer.buffer, writer.pos,
- uformat, nonfmtpos, sublen);
- writer.pos += sublen;
+ /* Parse precision. Example: "%.3f" => prec=3 */
+ if (arg->ch == '.') {
+ arg->prec = 0;
+ if (--ctx->fmtcnt >= 0) {
+ arg->ch = FORMAT_READ(ctx);
+ ctx->fmtpos++;
}
- else {
- /* Got a format specifier */
- int flags = 0;
- Py_ssize_t width = -1;
- int prec = -1;
- Py_UCS4 c = '\0';
- Py_UCS4 fill;
- int sign;
- Py_UCS4 signchar;
- int isnumok;
- PyObject *v = NULL;
- void *pbuf = NULL;
- Py_ssize_t pindex, len;
- Py_UCS4 bufmaxchar;
- Py_ssize_t buflen;
-
- fmtpos++;
- c = PyUnicode_READ(fmtkind, fmt, fmtpos);
- if (c == '(') {
- Py_ssize_t keystart;
- Py_ssize_t keylen;
- PyObject *key;
- int pcount = 1;
-
- if (dict == NULL) {
- PyErr_SetString(PyExc_TypeError,
- "format requires a mapping");
- goto onError;
- }
- ++fmtpos;
- --fmtcnt;
- keystart = fmtpos;
- /* Skip over balanced parentheses */
- while (pcount > 0 && --fmtcnt >= 0) {
- c = PyUnicode_READ(fmtkind, fmt, fmtpos);
- if (c == ')')
- --pcount;
- else if (c == '(')
- ++pcount;
- fmtpos++;
- }
- keylen = fmtpos - keystart - 1;
- if (fmtcnt < 0 || pcount > 0) {
- PyErr_SetString(PyExc_ValueError,
- "incomplete format key");
- goto onError;
- }
- key = PyUnicode_Substring(uformat,
- keystart, keystart + keylen);
- if (key == NULL)
- goto onError;
- if (args_owned) {
- Py_DECREF(args);
- args_owned = 0;
- }
- args = PyObject_GetItem(dict, key);
- Py_DECREF(key);
- if (args == NULL) {
- goto onError;
- }
- args_owned = 1;
- arglen = -1;
- argidx = -2;
- }
- while (--fmtcnt >= 0) {
- c = PyUnicode_READ(fmtkind, fmt, fmtpos++);
- switch (c) {
- case '-': flags |= F_LJUST; continue;
- case '+': flags |= F_SIGN; continue;
- case ' ': flags |= F_BLANK; continue;
- case '#': flags |= F_ALT; continue;
- case '0': flags |= F_ZERO; continue;
- }
- break;
- }
- if (c == '*') {
- v = getnextarg(args, arglen, &argidx);
- if (v == NULL)
- goto onError;
- if (!PyLong_Check(v)) {
- PyErr_SetString(PyExc_TypeError,
- "* wants int");
- goto onError;
- }
- width = PyLong_AsSsize_t(v);
- if (width == -1 && PyErr_Occurred())
- goto onError;
- if (width < 0) {
- flags |= F_LJUST;
- width = -width;
- }
- if (--fmtcnt >= 0)
- c = PyUnicode_READ(fmtkind, fmt, fmtpos++);
- }
- else if (c >= '0' && c <= '9') {
- width = c - '0';
- while (--fmtcnt >= 0) {
- c = PyUnicode_READ(fmtkind, fmt, fmtpos++);
- if (c < '0' || c > '9')
- break;
- /* Since c is unsigned, the RHS would end up as unsigned,
- mixing signed and unsigned comparison. Since c is between
- '0' and '9', casting to int is safe. */
- if (width > (PY_SSIZE_T_MAX - ((int)c - '0')) / 10) {
- PyErr_SetString(PyExc_ValueError,
- "width too big");
- goto onError;
- }
- width = width*10 + (c - '0');
- }
+ if (arg->ch == '*') {
+ v = unicode_format_getnextarg(ctx);
+ if (v == NULL)
+ return -1;
+ if (!PyLong_Check(v)) {
+ PyErr_SetString(PyExc_TypeError,
+ "* wants int");
+ return -1;
}
- if (c == '.') {
- prec = 0;
- if (--fmtcnt >= 0)
- c = PyUnicode_READ(fmtkind, fmt, fmtpos++);
- if (c == '*') {
- v = getnextarg(args, arglen, &argidx);
- if (v == NULL)
- goto onError;
- if (!PyLong_Check(v)) {
- PyErr_SetString(PyExc_TypeError,
- "* wants int");
- goto onError;
- }
- prec = _PyLong_AsInt(v);
- if (prec == -1 && PyErr_Occurred())
- goto onError;
- if (prec < 0)
- prec = 0;
- if (--fmtcnt >= 0)
- c = PyUnicode_READ(fmtkind, fmt, fmtpos++);
- }
- else if (c >= '0' && c <= '9') {
- prec = c - '0';
- while (--fmtcnt >= 0) {
- c = PyUnicode_READ(fmtkind, fmt, fmtpos++);
- if (c < '0' || c > '9')
- break;
- if (prec > (INT_MAX - ((int)c - '0')) / 10) {
- PyErr_SetString(PyExc_ValueError,
- "prec too big");
- goto onError;
- }
- prec = prec*10 + (c - '0');
- }
- }
- } /* prec */
- if (fmtcnt >= 0) {
- if (c == 'h' || c == 'l' || c == 'L') {
- if (--fmtcnt >= 0)
- c = PyUnicode_READ(fmtkind, fmt, fmtpos++);
+ arg->prec = _PyLong_AsInt(v);
+ if (arg->prec == -1 && PyErr_Occurred())
+ return -1;
+ if (arg->prec < 0)
+ arg->prec = 0;
+ if (--ctx->fmtcnt >= 0) {
+ arg->ch = FORMAT_READ(ctx);
+ ctx->fmtpos++;
+ }
+ }
+ else if (arg->ch >= '0' && arg->ch <= '9') {
+ arg->prec = arg->ch - '0';
+ while (--ctx->fmtcnt >= 0) {
+ arg->ch = FORMAT_READ(ctx);
+ ctx->fmtpos++;
+ if (arg->ch < '0' || arg->ch > '9')
+ break;
+ if (arg->prec > (INT_MAX - ((int)arg->ch - '0')) / 10) {
+ PyErr_SetString(PyExc_ValueError,
+ "precision too big");
+ return -1;
}
+ arg->prec = arg->prec*10 + (arg->ch - '0');
}
- if (fmtcnt < 0) {
- PyErr_SetString(PyExc_ValueError,
- "incomplete format");
- goto onError;
- }
- if (fmtcnt == 0)
- writer.overallocate = 0;
+ }
+ }
- if (c == '%') {
- if (_PyUnicodeWriter_Prepare(&writer, 1, '%') == -1)
- goto onError;
- PyUnicode_WRITE(writer.kind, writer.data, writer.pos, '%');
- writer.pos += 1;
- continue;
+ /* Ignore "h", "l" and "L" format prefix (ex: "%hi" or "%ls") */
+ if (ctx->fmtcnt >= 0) {
+ if (arg->ch == 'h' || arg->ch == 'l' || arg->ch == 'L') {
+ if (--ctx->fmtcnt >= 0) {
+ arg->ch = FORMAT_READ(ctx);
+ ctx->fmtpos++;
}
+ }
+ }
+ if (ctx->fmtcnt < 0) {
+ PyErr_SetString(PyExc_ValueError,
+ "incomplete format");
+ return -1;
+ }
+ return 0;
- v = getnextarg(args, arglen, &argidx);
- if (v == NULL)
- goto onError;
+#undef FORMAT_READ
+}
- sign = 0;
- signchar = '\0';
- fill = ' ';
- switch (c) {
-
- case 's':
- case 'r':
- case 'a':
- if (PyLong_CheckExact(v) && width == -1 && prec == -1) {
- /* Fast path */
- if (_PyLong_FormatWriter(&writer, v, 10, flags & F_ALT) == -1)
- goto onError;
- goto nextarg;
- }
+/* Format one argument. Supported conversion specifiers:
- if (PyUnicode_CheckExact(v) && c == 's') {
- temp = v;
- Py_INCREF(temp);
- }
- else {
- if (c == 's')
- temp = PyObject_Str(v);
- else if (c == 'r')
- temp = PyObject_Repr(v);
- else
- temp = PyObject_ASCII(v);
- }
- break;
+ - "s", "r", "a": any type
+ - "i", "d", "u", "o", "x", "X": int
+ - "e", "E", "f", "F", "g", "G": float
+ - "c": int or str (1 character)
- case 'i':
- case 'd':
- case 'u':
- case 'o':
- case 'x':
- case 'X':
- if (PyLong_CheckExact(v)
- && width == -1 && prec == -1
- && !(flags & (F_SIGN | F_BLANK)))
- {
- /* Fast path */
- switch(c)
- {
- case 'd':
- case 'i':
- case 'u':
- if (_PyLong_FormatWriter(&writer, v, 10, flags & F_ALT) == -1)
- goto onError;
- goto nextarg;
- case 'x':
- if (_PyLong_FormatWriter(&writer, v, 16, flags & F_ALT) == -1)
- goto onError;
- goto nextarg;
- case 'o':
- if (_PyLong_FormatWriter(&writer, v, 8, flags & F_ALT) == -1)
- goto onError;
- goto nextarg;
- default:
- break;
- }
- }
+ When possible, the output is written directly into the Unicode writer
+ (ctx->writer). A string is created when padding is required.
- isnumok = 0;
- if (PyNumber_Check(v)) {
- PyObject *iobj=NULL;
+ Return 0 if the argument has been formatted into *p_str,
+ 1 if the argument has been written into ctx->writer,
+ -1 on error. */
+static int
+unicode_format_arg_format(struct unicode_formatter_t *ctx,
+ struct unicode_format_arg_t *arg,
+ PyObject **p_str)
+{
+ PyObject *v;
+ _PyUnicodeWriter *writer = &ctx->writer;
- if (PyLong_Check(v)) {
- iobj = v;
- Py_INCREF(iobj);
- }
- else {
- iobj = PyNumber_Long(v);
- }
- if (iobj!=NULL) {
- if (PyLong_Check(iobj)) {
- isnumok = 1;
- sign = 1;
- temp = formatlong(iobj, flags, prec, (c == 'i'? 'd': c));
- Py_DECREF(iobj);
- }
- else {
- Py_DECREF(iobj);
- }
- }
- }
- if (!isnumok) {
- PyErr_Format(PyExc_TypeError,
- "%%%c format: a number is required, "
- "not %.200s", (char)c, Py_TYPE(v)->tp_name);
- goto onError;
- }
- if (flags & F_ZERO)
- fill = '0';
- break;
+ if (ctx->fmtcnt == 0)
+ ctx->writer.overallocate = 0;
- case 'e':
- case 'E':
- case 'f':
- case 'F':
- case 'g':
- case 'G':
- if (width == -1 && prec == -1
- && !(flags & (F_SIGN | F_BLANK)))
- {
- /* Fast path */
- if (formatfloat(v, flags, prec, c, NULL, &writer) == -1)
- goto onError;
- goto nextarg;
- }
+ if (arg->ch == '%') {
+ if (_PyUnicodeWriter_WriteCharInline(writer, '%') < 0)
+ return -1;
+ return 1;
+ }
- sign = 1;
- if (flags & F_ZERO)
- fill = '0';
- if (formatfloat(v, flags, prec, c, &temp, NULL) == -1)
- temp = NULL;
- break;
+ v = unicode_format_getnextarg(ctx);
+ if (v == NULL)
+ return -1;
- case 'c':
- {
- Py_UCS4 ch = formatchar(v);
- if (ch == (Py_UCS4) -1)
- goto onError;
- if (width == -1 && prec == -1) {
- /* Fast path */
- if (_PyUnicodeWriter_Prepare(&writer, 1, ch) == -1)
- goto onError;
- PyUnicode_WRITE(writer.kind, writer.data, writer.pos, ch);
- writer.pos += 1;
- goto nextarg;
- }
- temp = PyUnicode_FromOrdinal(ch);
- break;
- }
- default:
- PyErr_Format(PyExc_ValueError,
- "unsupported format character '%c' (0x%x) "
- "at index %zd",
- (31<=c && c<=126) ? (char)c : '?',
- (int)c,
- fmtpos - 1);
- goto onError;
- }
- if (temp == NULL)
- goto onError;
- assert (PyUnicode_Check(temp));
+ switch (arg->ch) {
+ case 's':
+ case 'r':
+ case 'a':
+ if (PyLong_CheckExact(v) && arg->width == -1 && arg->prec == -1) {
+ /* Fast path */
+ if (_PyLong_FormatWriter(writer, v, 10, arg->flags & F_ALT) == -1)
+ return -1;
+ return 1;
+ }
- if (width == -1 && prec == -1
- && !(flags & (F_SIGN | F_BLANK)))
- {
- /* Fast path */
- if (_PyUnicodeWriter_WriteStr(&writer, temp) == -1)
- goto onError;
- goto nextarg;
- }
+ if (PyUnicode_CheckExact(v) && arg->ch == 's') {
+ *p_str = v;
+ Py_INCREF(*p_str);
+ }
+ else {
+ if (arg->ch == 's')
+ *p_str = PyObject_Str(v);
+ else if (arg->ch == 'r')
+ *p_str = PyObject_Repr(v);
+ else
+ *p_str = PyObject_ASCII(v);
+ }
+ break;
- if (PyUnicode_READY(temp) == -1) {
- Py_CLEAR(temp);
- goto onError;
- }
- kind = PyUnicode_KIND(temp);
- pbuf = PyUnicode_DATA(temp);
- len = PyUnicode_GET_LENGTH(temp);
+ case 'i':
+ case 'd':
+ case 'u':
+ case 'o':
+ case 'x':
+ case 'X':
+ {
+ int ret = mainformatlong(v, arg, p_str, writer);
+ if (ret != 0)
+ return ret;
+ arg->sign = 1;
+ break;
+ }
- if (c == 's' || c == 'r' || c == 'a') {
- if (prec >= 0 && len > prec)
- len = prec;
- }
+ case 'e':
+ case 'E':
+ case 'f':
+ case 'F':
+ case 'g':
+ case 'G':
+ if (arg->width == -1 && arg->prec == -1
+ && !(arg->flags & (F_SIGN | F_BLANK)))
+ {
+ /* Fast path */
+ if (formatfloat(v, arg, NULL, writer) == -1)
+ return -1;
+ return 1;
+ }
- /* pbuf is initialized here. */
- pindex = 0;
- if (sign) {
- Py_UCS4 ch = PyUnicode_READ(kind, pbuf, pindex);
- if (ch == '-' || ch == '+') {
- signchar = ch;
- len--;
- pindex++;
- }
- else if (flags & F_SIGN)
- signchar = '+';
- else if (flags & F_BLANK)
- signchar = ' ';
- else
- sign = 0;
- }
- if (width < len)
- width = len;
-
- /* Compute the length and maximum character of the
- written characters */
- bufmaxchar = 127;
- if (!(flags & F_LJUST)) {
- if (sign) {
- if ((width-1) > len)
- bufmaxchar = Py_MAX(bufmaxchar, fill);
- }
- else {
- if (width > len)
- bufmaxchar = Py_MAX(bufmaxchar, fill);
- }
- }
- maxchar = _PyUnicode_FindMaxChar(temp, 0, pindex+len);
- bufmaxchar = Py_MAX(bufmaxchar, maxchar);
+ arg->sign = 1;
+ if (formatfloat(v, arg, p_str, NULL) == -1)
+ return -1;
+ break;
- buflen = width;
- if (sign && len == width)
- buflen++;
+ case 'c':
+ {
+ Py_UCS4 ch = formatchar(v);
+ if (ch == (Py_UCS4) -1)
+ return -1;
+ if (arg->width == -1 && arg->prec == -1) {
+ /* Fast path */
+ if (_PyUnicodeWriter_WriteCharInline(writer, ch) < 0)
+ return -1;
+ return 1;
+ }
+ *p_str = PyUnicode_FromOrdinal(ch);
+ break;
+ }
- if (_PyUnicodeWriter_Prepare(&writer, buflen, bufmaxchar) == -1)
- goto onError;
+ default:
+ PyErr_Format(PyExc_ValueError,
+ "unsupported format character '%c' (0x%x) "
+ "at index %zd",
+ (31<=arg->ch && arg->ch<=126) ? (char)arg->ch : '?',
+ (int)arg->ch,
+ ctx->fmtpos - 1);
+ return -1;
+ }
+ if (*p_str == NULL)
+ return -1;
+ assert (PyUnicode_Check(*p_str));
+ return 0;
+}
- /* Write characters */
- if (sign) {
- if (fill != ' ') {
- PyUnicode_WRITE(writer.kind, writer.data, writer.pos, signchar);
- writer.pos += 1;
- }
- if (width > len)
- width--;
- }
- if ((flags & F_ALT) && (c == 'x' || c == 'X' || c == 'o')) {
- assert(PyUnicode_READ(kind, pbuf, pindex) == '0');
- assert(PyUnicode_READ(kind, pbuf, pindex + 1) == c);
- if (fill != ' ') {
- PyUnicode_WRITE(writer.kind, writer.data, writer.pos, '0');
- PyUnicode_WRITE(writer.kind, writer.data, writer.pos+1, c);
- writer.pos += 2;
- pindex += 2;
- }
- width -= 2;
- if (width < 0)
- width = 0;
- len -= 2;
- }
- if (width > len && !(flags & F_LJUST)) {
- sublen = width - len;
- FILL(writer.kind, writer.data, fill, writer.pos, sublen);
- writer.pos += sublen;
- width = len;
- }
- if (fill == ' ') {
- if (sign) {
- PyUnicode_WRITE(writer.kind, writer.data, writer.pos, signchar);
- writer.pos += 1;
- }
- if ((flags & F_ALT) && (c == 'x' || c == 'X' || c == 'o')) {
- assert(PyUnicode_READ(kind, pbuf, pindex) == '0');
- assert(PyUnicode_READ(kind, pbuf, pindex+1) == c);
- PyUnicode_WRITE(writer.kind, writer.data, writer.pos, '0');
- PyUnicode_WRITE(writer.kind, writer.data, writer.pos+1, c);
- writer.pos += 2;
- pindex += 2;
- }
- }
+static int
+unicode_format_arg_output(struct unicode_formatter_t *ctx,
+ struct unicode_format_arg_t *arg,
+ PyObject *str)
+{
+ Py_ssize_t len;
+ enum PyUnicode_Kind kind;
+ void *pbuf;
+ Py_ssize_t pindex;
+ Py_UCS4 signchar;
+ Py_ssize_t buflen;
+ Py_UCS4 maxchar;
+ Py_ssize_t sublen;
+ _PyUnicodeWriter *writer = &ctx->writer;
+ Py_UCS4 fill;
+
+ fill = ' ';
+ if (arg->sign && arg->flags & F_ZERO)
+ fill = '0';
+
+ if (PyUnicode_READY(str) == -1)
+ return -1;
+
+ len = PyUnicode_GET_LENGTH(str);
+ if ((arg->width == -1 || arg->width <= len)
+ && (arg->prec == -1 || arg->prec >= len)
+ && !(arg->flags & (F_SIGN | F_BLANK)))
+ {
+ /* Fast path */
+ if (_PyUnicodeWriter_WriteStr(writer, str) == -1)
+ return -1;
+ return 0;
+ }
+
+ /* Truncate the string for "s", "r" and "a" formats
+ if the precision is set */
+ if (arg->ch == 's' || arg->ch == 'r' || arg->ch == 'a') {
+ if (arg->prec >= 0 && len > arg->prec)
+ len = arg->prec;
+ }
+
+ /* Adjust sign and width */
+ kind = PyUnicode_KIND(str);
+ pbuf = PyUnicode_DATA(str);
+ pindex = 0;
+ signchar = '\0';
+ if (arg->sign) {
+ Py_UCS4 ch = PyUnicode_READ(kind, pbuf, pindex);
+ if (ch == '-' || ch == '+') {
+ signchar = ch;
+ len--;
+ pindex++;
+ }
+ else if (arg->flags & F_SIGN)
+ signchar = '+';
+ else if (arg->flags & F_BLANK)
+ signchar = ' ';
+ else
+ arg->sign = 0;
+ }
+ if (arg->width < len)
+ arg->width = len;
+
+ /* Prepare the writer */
+ maxchar = writer->maxchar;
+ if (!(arg->flags & F_LJUST)) {
+ if (arg->sign) {
+ if ((arg->width-1) > len)
- maxchar = MAX_MAXCHAR(maxchar, fill);
++ maxchar = Py_MAX(maxchar, fill);
+ }
+ else {
+ if (arg->width > len)
- maxchar = MAX_MAXCHAR(maxchar, fill);
++ maxchar = Py_MAX(maxchar, fill);
+ }
+ }
+ if (PyUnicode_MAX_CHAR_VALUE(str) > maxchar) {
+ Py_UCS4 strmaxchar = _PyUnicode_FindMaxChar(str, 0, pindex+len);
- maxchar = MAX_MAXCHAR(maxchar, strmaxchar);
++ maxchar = Py_MAX(maxchar, strmaxchar);
+ }
+
+ buflen = arg->width;
+ if (arg->sign && len == arg->width)
+ buflen++;
+ if (_PyUnicodeWriter_Prepare(writer, buflen, maxchar) == -1)
+ return -1;
+
+ /* Write the sign if needed */
+ if (arg->sign) {
+ if (fill != ' ') {
+ PyUnicode_WRITE(writer->kind, writer->data, writer->pos, signchar);
+ writer->pos += 1;
+ }
+ if (arg->width > len)
+ arg->width--;
+ }
+
+ /* Write the numeric prefix for "x", "X" and "o" formats
+ if the alternate form is used.
+ For example, write "0x" for the "%#x" format. */
+ if ((arg->flags & F_ALT) && (arg->ch == 'x' || arg->ch == 'X' || arg->ch == 'o')) {
+ assert(PyUnicode_READ(kind, pbuf, pindex) == '0');
+ assert(PyUnicode_READ(kind, pbuf, pindex + 1) == arg->ch);
+ if (fill != ' ') {
+ PyUnicode_WRITE(writer->kind, writer->data, writer->pos, '0');
+ PyUnicode_WRITE(writer->kind, writer->data, writer->pos+1, arg->ch);
+ writer->pos += 2;
+ pindex += 2;
+ }
+ arg->width -= 2;
+ if (arg->width < 0)
+ arg->width = 0;
+ len -= 2;
+ }
+
+ /* Pad left with the fill character if needed */
+ if (arg->width > len && !(arg->flags & F_LJUST)) {
+ sublen = arg->width - len;
+ FILL(writer->kind, writer->data, fill, writer->pos, sublen);
+ writer->pos += sublen;
+ arg->width = len;
+ }
+
+ /* If padding with spaces: write sign if needed and/or numeric prefix if
+ the alternate form is used */
+ if (fill == ' ') {
+ if (arg->sign) {
+ PyUnicode_WRITE(writer->kind, writer->data, writer->pos, signchar);
+ writer->pos += 1;
+ }
+ if ((arg->flags & F_ALT) && (arg->ch == 'x' || arg->ch == 'X' || arg->ch == 'o')) {
+ assert(PyUnicode_READ(kind, pbuf, pindex) == '0');
+ assert(PyUnicode_READ(kind, pbuf, pindex+1) == arg->ch);
+ PyUnicode_WRITE(writer->kind, writer->data, writer->pos, '0');
+ PyUnicode_WRITE(writer->kind, writer->data, writer->pos+1, arg->ch);
+ writer->pos += 2;
+ pindex += 2;
+ }
+ }
+
+ /* Write characters */
+ if (len) {
+ _PyUnicode_FastCopyCharacters(writer->buffer, writer->pos,
+ str, pindex, len);
+ writer->pos += len;
+ }
+
+ /* Pad right with the fill character if needed */
+ if (arg->width > len) {
+ sublen = arg->width - len;
+ FILL(writer->kind, writer->data, ' ', writer->pos, sublen);
+ writer->pos += sublen;
+ }
+ return 0;
+}
+
+/* Helper of PyUnicode_Format(): format one arg.
+ Return 0 on success, raise an exception and return -1 on error. */
+static int
+unicode_format_arg(struct unicode_formatter_t *ctx)
+{
+ struct unicode_format_arg_t arg;
+ PyObject *str;
+ int ret;
+
+ arg.ch = PyUnicode_READ(ctx->fmtkind, ctx->fmtdata, ctx->fmtpos);
+ arg.flags = 0;
+ arg.width = -1;
+ arg.prec = -1;
+ arg.sign = 0;
+ str = NULL;
+
+ ret = unicode_format_arg_parse(ctx, &arg);
+ if (ret == -1)
+ return -1;
+
+ ret = unicode_format_arg_format(ctx, &arg, &str);
+ if (ret == -1)
+ return -1;
+
+ if (ret != 1) {
+ ret = unicode_format_arg_output(ctx, &arg, str);
+ Py_DECREF(str);
+ if (ret == -1)
+ return -1;
+ }
+
+ if (ctx->dict && (ctx->argidx < ctx->arglen) && arg.ch != '%') {
+ PyErr_SetString(PyExc_TypeError,
+ "not all arguments converted during string formatting");
+ return -1;
+ }
+ return 0;
+}
+
+PyObject *
+PyUnicode_Format(PyObject *format, PyObject *args)
+{
+ struct unicode_formatter_t ctx;
- if (len) {
- _PyUnicode_FastCopyCharacters(writer.buffer, writer.pos,
- temp, pindex, len);
- writer.pos += len;
+ if (format == NULL || args == NULL) {
+ PyErr_BadInternalCall();
+ return NULL;
+ }
+
+ ctx.fmtstr = PyUnicode_FromObject(format);
+ if (ctx.fmtstr == NULL)
+ return NULL;
+ if (PyUnicode_READY(ctx.fmtstr) == -1) {
+ Py_DECREF(ctx.fmtstr);
+ return NULL;
+ }
+ ctx.fmtdata = PyUnicode_DATA(ctx.fmtstr);
+ ctx.fmtkind = PyUnicode_KIND(ctx.fmtstr);
+ ctx.fmtcnt = PyUnicode_GET_LENGTH(ctx.fmtstr);
+ ctx.fmtpos = 0;
+
+ _PyUnicodeWriter_Init(&ctx.writer);
+ ctx.writer.min_length = ctx.fmtcnt + 100;
+ ctx.writer.overallocate = 1;
+
+ if (PyTuple_Check(args)) {
+ ctx.arglen = PyTuple_Size(args);
+ ctx.argidx = 0;
+ }
+ else {
+ ctx.arglen = -1;
+ ctx.argidx = -2;
+ }
+ ctx.args_owned = 0;
+ if (PyMapping_Check(args) && !PyTuple_Check(args) && !PyUnicode_Check(args))
+ ctx.dict = args;
+ else
+ ctx.dict = NULL;
+ ctx.args = args;
+
+ while (--ctx.fmtcnt >= 0) {
+ if (PyUnicode_READ(ctx.fmtkind, ctx.fmtdata, ctx.fmtpos) != '%') {
+ Py_ssize_t nonfmtpos;
+
+ nonfmtpos = ctx.fmtpos++;
+ while (ctx.fmtcnt >= 0 &&
+ PyUnicode_READ(ctx.fmtkind, ctx.fmtdata, ctx.fmtpos) != '%') {
+ ctx.fmtpos++;
+ ctx.fmtcnt--;
}
- if (width > len) {
- sublen = width - len;
- FILL(writer.kind, writer.data, ' ', writer.pos, sublen);
- writer.pos += sublen;
+ if (ctx.fmtcnt < 0) {
+ ctx.fmtpos--;
+ ctx.writer.overallocate = 0;
}
-nextarg:
- if (dict && (argidx < arglen) && c != '%') {
- PyErr_SetString(PyExc_TypeError,
- "not all arguments converted during string formatting");
+ if (_PyUnicodeWriter_WriteSubstring(&ctx.writer, ctx.fmtstr,
+ nonfmtpos, ctx.fmtpos) < 0)
goto onError;
- }
- Py_CLEAR(temp);
- } /* '%' */
- } /* until end */
- if (argidx < arglen && !dict) {
+ }
+ else {
+ ctx.fmtpos++;
+ if (unicode_format_arg(&ctx) == -1)
+ goto onError;
+ }
+ }
+
+ if (ctx.argidx < ctx.arglen && !ctx.dict) {
PyErr_SetString(PyExc_TypeError,
"not all arguments converted during string formatting");
goto onError;
projects / python / commitdiff