Mercurial > cpython-lru_cache-weakref
annotate Python/formatter_unicode.c @ 75177:fcd0a67e708e
Issue #14084: Fix a file descriptor leak when importing a module with a bad encoding.
| author | Antoine Pitrou <solipsis@pitrou.net> |
|---|---|
| date | Wed, 22 Feb 2012 18:08:30 +0100 |
| parents | 056f5cc8885d |
| children | f89e2f4cda88 |
| rev | line source |
|---|---|
|
42817
5355337709e1
Implementation of PEP 3101, Advanced String Formatting.
Eric Smith <eric@trueblade.com>
parents:
diff
changeset
|
1 /* implements the unicode (as opposed to string) version of the |
|
5355337709e1
Implementation of PEP 3101, Advanced String Formatting.
Eric Smith <eric@trueblade.com>
parents:
diff
changeset
|
2 built-in formatters for string, int, float. that is, the versions |
|
5355337709e1
Implementation of PEP 3101, Advanced String Formatting.
Eric Smith <eric@trueblade.com>
parents:
diff
changeset
|
3 of int.__float__, etc., that take and return unicode objects */ |
|
5355337709e1
Implementation of PEP 3101, Advanced String Formatting.
Eric Smith <eric@trueblade.com>
parents:
diff
changeset
|
4 |
|
5355337709e1
Implementation of PEP 3101, Advanced String Formatting.
Eric Smith <eric@trueblade.com>
parents:
diff
changeset
|
5 #include "Python.h" |
| 72475 | 6 #include <locale.h> |
| 7 | |
| 8 /* Raises an exception about an unknown presentation type for this | |
| 9 * type. */ | |
| 10 | |
| 11 static void | |
| 12 unknown_presentation_type(Py_UCS4 presentation_type, | |
| 13 const char* type_name) | |
| 14 { | |
| 15 /* %c might be out-of-range, hence the two cases. */ | |
| 16 if (presentation_type > 32 && presentation_type < 128) | |
| 17 PyErr_Format(PyExc_ValueError, | |
| 18 "Unknown format code '%c' " | |
| 19 "for object of type '%.200s'", | |
| 20 (char)presentation_type, | |
| 21 type_name); | |
| 22 else | |
| 23 PyErr_Format(PyExc_ValueError, | |
| 24 "Unknown format code '\\x%x' " | |
| 25 "for object of type '%.200s'", | |
| 26 (unsigned int)presentation_type, | |
| 27 type_name); | |
| 28 } | |
| 29 | |
| 30 static void | |
| 31 invalid_comma_type(Py_UCS4 presentation_type) | |
| 32 { | |
| 33 if (presentation_type > 32 && presentation_type < 128) | |
| 34 PyErr_Format(PyExc_ValueError, | |
| 35 "Cannot specify ',' with '%c'.", | |
| 36 (char)presentation_type); | |
| 37 else | |
| 38 PyErr_Format(PyExc_ValueError, | |
| 39 "Cannot specify ',' with '\\x%x'.", | |
| 40 (unsigned int)presentation_type); | |
| 41 } | |
| 42 | |
| 43 /* | |
| 44 get_integer consumes 0 or more decimal digit characters from an | |
| 45 input string, updates *result with the corresponding positive | |
| 46 integer, and returns the number of digits consumed. | |
| 47 | |
| 48 returns -1 on error. | |
| 49 */ | |
| 50 static int | |
| 51 get_integer(PyObject *str, Py_ssize_t *pos, Py_ssize_t end, | |
| 52 Py_ssize_t *result) | |
| 53 { | |
|
73805
7e37598a25a6
Issue #9530: Fix undefined behaviour due to signed overflow in Python/formatter_unicode.c.
Mark Dickinson <mdickinson@enthought.com>
parents:
72863
diff
changeset
|
54 Py_ssize_t accumulator, digitval; |
| 72475 | 55 int numdigits; |
| 56 accumulator = numdigits = 0; | |
| 57 for (;;(*pos)++, numdigits++) { | |
| 58 if (*pos >= end) | |
| 59 break; | |
| 60 digitval = Py_UNICODE_TODECIMAL(PyUnicode_READ_CHAR(str, *pos)); | |
| 61 if (digitval < 0) | |
| 62 break; | |
| 63 /* | |
|
73805
7e37598a25a6
Issue #9530: Fix undefined behaviour due to signed overflow in Python/formatter_unicode.c.
Mark Dickinson <mdickinson@enthought.com>
parents:
72863
diff
changeset
|
64 Detect possible overflow before it happens: |
|
7e37598a25a6
Issue #9530: Fix undefined behaviour due to signed overflow in Python/formatter_unicode.c.
Mark Dickinson <mdickinson@enthought.com>
parents:
72863
diff
changeset
|
65 |
|
7e37598a25a6
Issue #9530: Fix undefined behaviour due to signed overflow in Python/formatter_unicode.c.
Mark Dickinson <mdickinson@enthought.com>
parents:
72863
diff
changeset
|
66 accumulator * 10 + digitval > PY_SSIZE_T_MAX if and only if |
|
7e37598a25a6
Issue #9530: Fix undefined behaviour due to signed overflow in Python/formatter_unicode.c.
Mark Dickinson <mdickinson@enthought.com>
parents:
72863
diff
changeset
|
67 accumulator > (PY_SSIZE_T_MAX - digitval) / 10. |
| 72475 | 68 */ |
|
73805
7e37598a25a6
Issue #9530: Fix undefined behaviour due to signed overflow in Python/formatter_unicode.c.
Mark Dickinson <mdickinson@enthought.com>
parents:
72863
diff
changeset
|
69 if (accumulator > (PY_SSIZE_T_MAX - digitval) / 10) { |
| 72475 | 70 PyErr_Format(PyExc_ValueError, |
| 71 "Too many decimal digits in format string"); | |
| 72 return -1; | |
| 73 } | |
|
73805
7e37598a25a6
Issue #9530: Fix undefined behaviour due to signed overflow in Python/formatter_unicode.c.
Mark Dickinson <mdickinson@enthought.com>
parents:
72863
diff
changeset
|
74 accumulator = accumulator * 10 + digitval; |
| 72475 | 75 } |
| 76 *result = accumulator; | |
| 77 return numdigits; | |
| 78 } | |
| 79 | |
| 80 /************************************************************************/ | |
| 81 /*********** standard format specifier parsing **************************/ | |
| 82 /************************************************************************/ | |
| 83 | |
| 84 /* returns true if this character is a specifier alignment token */ | |
| 85 Py_LOCAL_INLINE(int) | |
| 86 is_alignment_token(Py_UCS4 c) | |
| 87 { | |
| 88 switch (c) { | |
| 89 case '<': case '>': case '=': case '^': | |
| 90 return 1; | |
| 91 default: | |
| 92 return 0; | |
| 93 } | |
| 94 } | |
| 95 | |
| 96 /* returns true if this character is a sign element */ | |
| 97 Py_LOCAL_INLINE(int) | |
| 98 is_sign_element(Py_UCS4 c) | |
| 99 { | |
| 100 switch (c) { | |
| 101 case ' ': case '+': case '-': | |
| 102 return 1; | |
| 103 default: | |
| 104 return 0; | |
| 105 } | |
| 106 } | |
| 107 | |
| 108 | |
| 109 typedef struct { | |
| 110 Py_UCS4 fill_char; | |
| 111 Py_UCS4 align; | |
| 112 int alternate; | |
| 113 Py_UCS4 sign; | |
| 114 Py_ssize_t width; | |
| 115 int thousands_separators; | |
| 116 Py_ssize_t precision; | |
| 117 Py_UCS4 type; | |
| 118 } InternalFormatSpec; | |
| 119 | |
| 120 #if 0 | |
| 121 /* Occassionally useful for debugging. Should normally be commented out. */ | |
| 122 static void | |
| 123 DEBUG_PRINT_FORMAT_SPEC(InternalFormatSpec *format) | |
| 124 { | |
| 125 printf("internal format spec: fill_char %d\n", format->fill_char); | |
| 126 printf("internal format spec: align %d\n", format->align); | |
| 127 printf("internal format spec: alternate %d\n", format->alternate); | |
| 128 printf("internal format spec: sign %d\n", format->sign); | |
| 129 printf("internal format spec: width %zd\n", format->width); | |
| 130 printf("internal format spec: thousands_separators %d\n", | |
| 131 format->thousands_separators); | |
| 132 printf("internal format spec: precision %zd\n", format->precision); | |
| 133 printf("internal format spec: type %c\n", format->type); | |
| 134 printf("\n"); | |
| 135 } | |
| 136 #endif | |
| 137 | |
| 138 | |
| 139 /* | |
| 140 ptr points to the start of the format_spec, end points just past its end. | |
| 141 fills in format with the parsed information. | |
| 142 returns 1 on success, 0 on failure. | |
| 143 if failure, sets the exception | |
| 144 */ | |
| 145 static int | |
| 146 parse_internal_render_format_spec(PyObject *format_spec, | |
| 147 Py_ssize_t start, Py_ssize_t end, | |
| 148 InternalFormatSpec *format, | |
| 149 char default_type, | |
| 150 char default_align) | |
| 151 { | |
| 152 Py_ssize_t pos = start; | |
| 153 /* end-pos is used throughout this code to specify the length of | |
| 154 the input string */ | |
| 155 #define READ_spec(index) PyUnicode_READ_CHAR(format_spec, index) | |
| 156 | |
| 157 Py_ssize_t consumed; | |
| 158 int align_specified = 0; | |
| 159 | |
| 160 format->fill_char = '\0'; | |
| 161 format->align = default_align; | |
| 162 format->alternate = 0; | |
| 163 format->sign = '\0'; | |
| 164 format->width = -1; | |
| 165 format->thousands_separators = 0; | |
| 166 format->precision = -1; | |
| 167 format->type = default_type; | |
| 168 | |
| 169 /* If the second char is an alignment token, | |
| 170 then parse the fill char */ | |
| 171 if (end-pos >= 2 && is_alignment_token(READ_spec(pos+1))) { | |
| 172 format->align = READ_spec(pos+1); | |
| 173 format->fill_char = READ_spec(pos); | |
| 174 align_specified = 1; | |
| 175 pos += 2; | |
| 176 } | |
| 177 else if (end-pos >= 1 && is_alignment_token(READ_spec(pos))) { | |
| 178 format->align = READ_spec(pos); | |
| 179 align_specified = 1; | |
| 180 ++pos; | |
| 181 } | |
| 182 | |
| 183 /* Parse the various sign options */ | |
| 184 if (end-pos >= 1 && is_sign_element(READ_spec(pos))) { | |
| 185 format->sign = READ_spec(pos); | |
| 186 ++pos; | |
| 187 } | |
| 188 | |
| 189 /* If the next character is #, we're in alternate mode. This only | |
| 190 applies to integers. */ | |
| 191 if (end-pos >= 1 && READ_spec(pos) == '#') { | |
| 192 format->alternate = 1; | |
| 193 ++pos; | |
| 194 } | |
| 195 | |
| 196 /* The special case for 0-padding (backwards compat) */ | |
| 197 if (format->fill_char == '\0' && end-pos >= 1 && READ_spec(pos) == '0') { | |
| 198 format->fill_char = '0'; | |
| 199 if (!align_specified) { | |
| 200 format->align = '='; | |
| 201 } | |
| 202 ++pos; | |
| 203 } | |
| 204 | |
| 205 consumed = get_integer(format_spec, &pos, end, &format->width); | |
| 206 if (consumed == -1) | |
| 207 /* Overflow error. Exception already set. */ | |
| 208 return 0; | |
| 209 | |
| 210 /* If consumed is 0, we didn't consume any characters for the | |
| 211 width. In that case, reset the width to -1, because | |
| 212 get_integer() will have set it to zero. -1 is how we record | |
| 213 that the width wasn't specified. */ | |
| 214 if (consumed == 0) | |
| 215 format->width = -1; | |
| 216 | |
| 217 /* Comma signifies add thousands separators */ | |
| 218 if (end-pos && READ_spec(pos) == ',') { | |
| 219 format->thousands_separators = 1; | |
| 220 ++pos; | |
| 221 } | |
| 222 | |
| 223 /* Parse field precision */ | |
| 224 if (end-pos && READ_spec(pos) == '.') { | |
| 225 ++pos; | |
| 226 | |
| 227 consumed = get_integer(format_spec, &pos, end, &format->precision); | |
| 228 if (consumed == -1) | |
| 229 /* Overflow error. Exception already set. */ | |
| 230 return 0; | |
| 231 | |
| 232 /* Not having a precision after a dot is an error. */ | |
| 233 if (consumed == 0) { | |
| 234 PyErr_Format(PyExc_ValueError, | |
| 235 "Format specifier missing precision"); | |
| 236 return 0; | |
| 237 } | |
| 238 | |
| 239 } | |
| 240 | |
| 241 /* Finally, parse the type field. */ | |
| 242 | |
| 243 if (end-pos > 1) { | |
|
74533
5c33ebb50702
Improve exception text. Closes issue 13811.
Eric V. Smith <eric@trueblade.com>
parents:
74341
diff
changeset
|
244 /* More than one char remain, invalid format specifier. */ |
|
5c33ebb50702
Improve exception text. Closes issue 13811.
Eric V. Smith <eric@trueblade.com>
parents:
74341
diff
changeset
|
245 PyErr_Format(PyExc_ValueError, "Invalid format specifier"); |
| 72475 | 246 return 0; |
| 247 } | |
| 248 | |
| 249 if (end-pos == 1) { | |
| 250 format->type = READ_spec(pos); | |
| 251 ++pos; | |
| 252 } | |
| 253 | |
| 254 /* Do as much validating as we can, just by looking at the format | |
| 255 specifier. Do not take into account what type of formatting | |
| 256 we're doing (int, float, string). */ | |
| 257 | |
| 258 if (format->thousands_separators) { | |
| 259 switch (format->type) { | |
| 260 case 'd': | |
| 261 case 'e': | |
| 262 case 'f': | |
| 263 case 'g': | |
| 264 case 'E': | |
| 265 case 'G': | |
| 266 case '%': | |
| 267 case 'F': | |
| 268 case '\0': | |
| 269 /* These are allowed. See PEP 378.*/ | |
| 270 break; | |
| 271 default: | |
| 272 invalid_comma_type(format->type); | |
| 273 return 0; | |
| 274 } | |
| 275 } | |
| 276 | |
|
74557
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
277 assert (format->align <= 127); |
|
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
278 assert (format->sign <= 127); |
| 72475 | 279 return 1; |
| 280 } | |
| 281 | |
| 282 /* Calculate the padding needed. */ | |
| 283 static void | |
| 284 calc_padding(Py_ssize_t nchars, Py_ssize_t width, Py_UCS4 align, | |
| 285 Py_ssize_t *n_lpadding, Py_ssize_t *n_rpadding, | |
| 286 Py_ssize_t *n_total) | |
| 287 { | |
| 288 if (width >= 0) { | |
| 289 if (nchars > width) | |
| 290 *n_total = nchars; | |
| 291 else | |
| 292 *n_total = width; | |
| 293 } | |
| 294 else { | |
| 295 /* not specified, use all of the chars and no more */ | |
| 296 *n_total = nchars; | |
| 297 } | |
| 298 | |
| 299 /* Figure out how much leading space we need, based on the | |
| 300 aligning */ | |
| 301 if (align == '>') | |
| 302 *n_lpadding = *n_total - nchars; | |
| 303 else if (align == '^') | |
| 304 *n_lpadding = (*n_total - nchars) / 2; | |
| 305 else if (align == '<' || align == '=') | |
| 306 *n_lpadding = 0; | |
| 307 else { | |
| 308 /* We should never have an unspecified alignment. */ | |
| 309 *n_lpadding = 0; | |
| 310 assert(0); | |
| 311 } | |
| 312 | |
| 313 *n_rpadding = *n_total - nchars - *n_lpadding; | |
| 314 } | |
| 315 | |
| 316 /* Do the padding, and return a pointer to where the caller-supplied | |
| 317 content goes. */ | |
| 318 static Py_ssize_t | |
| 319 fill_padding(PyObject *s, Py_ssize_t start, Py_ssize_t nchars, | |
| 320 Py_UCS4 fill_char, Py_ssize_t n_lpadding, | |
| 321 Py_ssize_t n_rpadding) | |
| 322 { | |
| 323 /* Pad on left. */ | |
| 324 if (n_lpadding) | |
|
74261
405f65676010
Add a new PyUnicode_Fill() function
Victor Stinner <victor.stinner@haypocalc.com>
parents:
73805
diff
changeset
|
325 PyUnicode_Fill(s, start, start + n_lpadding, fill_char); |
| 72475 | 326 |
| 327 /* Pad on right. */ | |
| 328 if (n_rpadding) | |
|
74261
405f65676010
Add a new PyUnicode_Fill() function
Victor Stinner <victor.stinner@haypocalc.com>
parents:
73805
diff
changeset
|
329 PyUnicode_Fill(s, start + nchars + n_lpadding, |
|
405f65676010
Add a new PyUnicode_Fill() function
Victor Stinner <victor.stinner@haypocalc.com>
parents:
73805
diff
changeset
|
330 start + nchars + n_lpadding + n_rpadding, fill_char); |
| 72475 | 331 |
| 332 /* Pointer to the user content. */ | |
| 333 return start + n_lpadding; | |
| 334 } | |
| 335 | |
| 336 /************************************************************************/ | |
| 337 /*********** common routines for numeric formatting *********************/ | |
| 338 /************************************************************************/ | |
| 339 | |
| 340 /* Locale type codes. */ | |
| 341 #define LT_CURRENT_LOCALE 0 | |
| 342 #define LT_DEFAULT_LOCALE 1 | |
| 343 #define LT_NO_LOCALE 2 | |
| 344 | |
| 345 /* Locale info needed for formatting integers and the part of floats | |
| 346 before and including the decimal. Note that locales only support | |
| 347 8-bit chars, not unicode. */ | |
| 348 typedef struct { | |
| 349 char *decimal_point; | |
| 350 char *thousands_sep; | |
| 351 char *grouping; | |
| 352 } LocaleInfo; | |
| 353 | |
| 354 /* describes the layout for an integer, see the comment in | |
| 355 calc_number_widths() for details */ | |
| 356 typedef struct { | |
| 357 Py_ssize_t n_lpadding; | |
| 358 Py_ssize_t n_prefix; | |
| 359 Py_ssize_t n_spadding; | |
| 360 Py_ssize_t n_rpadding; | |
| 361 char sign; | |
| 362 Py_ssize_t n_sign; /* number of digits needed for sign (0/1) */ | |
| 363 Py_ssize_t n_grouped_digits; /* Space taken up by the digits, including | |
| 364 any grouping chars. */ | |
| 365 Py_ssize_t n_decimal; /* 0 if only an integer */ | |
| 366 Py_ssize_t n_remainder; /* Digits in decimal and/or exponent part, | |
| 367 excluding the decimal itself, if | |
| 368 present. */ | |
| 369 | |
| 370 /* These 2 are not the widths of fields, but are needed by | |
| 371 STRINGLIB_GROUPING. */ | |
| 372 Py_ssize_t n_digits; /* The number of digits before a decimal | |
| 373 or exponent. */ | |
| 374 Py_ssize_t n_min_width; /* The min_width we used when we computed | |
| 375 the n_grouped_digits width. */ | |
| 376 } NumberFieldWidths; | |
|
42817
5355337709e1
Implementation of PEP 3101, Advanced String Formatting.
Eric Smith <eric@trueblade.com>
parents:
diff
changeset
|
377 |
|
47393
0da27fdf0192
Refactor and clean up str.format() code (and helpers) in advance of optimizations.
Eric Smith <eric@trueblade.com>
parents:
42817
diff
changeset
|
378 |
| 72475 | 379 /* Given a number of the form: |
| 380 digits[remainder] | |
| 381 where ptr points to the start and end points to the end, find where | |
| 382 the integer part ends. This could be a decimal, an exponent, both, | |
| 383 or neither. | |
| 384 If a decimal point is present, set *has_decimal and increment | |
| 385 remainder beyond it. | |
| 386 Results are undefined (but shouldn't crash) for improperly | |
| 387 formatted strings. | |
| 388 */ | |
| 389 static void | |
| 390 parse_number(PyObject *s, Py_ssize_t pos, Py_ssize_t end, | |
| 391 Py_ssize_t *n_remainder, int *has_decimal) | |
| 392 { | |
| 393 Py_ssize_t remainder; | |
| 394 | |
| 395 while (pos<end && isdigit(PyUnicode_READ_CHAR(s, pos))) | |
| 396 ++pos; | |
| 397 remainder = pos; | |
| 398 | |
| 399 /* Does remainder start with a decimal point? */ | |
| 400 *has_decimal = pos<end && PyUnicode_READ_CHAR(s, remainder) == '.'; | |
| 401 | |
| 402 /* Skip the decimal point. */ | |
| 403 if (*has_decimal) | |
| 404 remainder++; | |
| 405 | |
| 406 *n_remainder = end - remainder; | |
| 407 } | |
| 408 | |
| 409 /* not all fields of format are used. for example, precision is | |
| 410 unused. should this take discrete params in order to be more clear | |
| 411 about what it does? or is passing a single format parameter easier | |
| 412 and more efficient enough to justify a little obfuscation? */ | |
| 413 static Py_ssize_t | |
| 414 calc_number_widths(NumberFieldWidths *spec, Py_ssize_t n_prefix, | |
| 415 Py_UCS4 sign_char, PyObject *number, Py_ssize_t n_start, | |
| 416 Py_ssize_t n_end, Py_ssize_t n_remainder, | |
| 417 int has_decimal, const LocaleInfo *locale, | |
| 418 const InternalFormatSpec *format) | |
| 419 { | |
| 420 Py_ssize_t n_non_digit_non_padding; | |
| 421 Py_ssize_t n_padding; | |
| 422 | |
| 423 spec->n_digits = n_end - n_start - n_remainder - (has_decimal?1:0); | |
| 424 spec->n_lpadding = 0; | |
| 425 spec->n_prefix = n_prefix; | |
| 426 spec->n_decimal = has_decimal ? strlen(locale->decimal_point) : 0; | |
| 427 spec->n_remainder = n_remainder; | |
| 428 spec->n_spadding = 0; | |
| 429 spec->n_rpadding = 0; | |
| 430 spec->sign = '\0'; | |
| 431 spec->n_sign = 0; | |
| 432 | |
| 433 /* the output will look like: | |
| 434 | | | |
| 435 | <lpadding> <sign> <prefix> <spadding> <grouped_digits> <decimal> <remainder> <rpadding> | | |
| 436 | | | |
| 437 | |
| 438 sign is computed from format->sign and the actual | |
| 439 sign of the number | |
| 440 | |
| 441 prefix is given (it's for the '0x' prefix) | |
| 442 | |
| 443 digits is already known | |
| 444 | |
| 445 the total width is either given, or computed from the | |
| 446 actual digits | |
| 447 | |
| 448 only one of lpadding, spadding, and rpadding can be non-zero, | |
| 449 and it's calculated from the width and other fields | |
| 450 */ | |
| 451 | |
| 452 /* compute the various parts we're going to write */ | |
| 453 switch (format->sign) { | |
| 454 case '+': | |
| 455 /* always put a + or - */ | |
| 456 spec->n_sign = 1; | |
| 457 spec->sign = (sign_char == '-' ? '-' : '+'); | |
| 458 break; | |
| 459 case ' ': | |
| 460 spec->n_sign = 1; | |
| 461 spec->sign = (sign_char == '-' ? '-' : ' '); | |
| 462 break; | |
| 463 default: | |
| 464 /* Not specified, or the default (-) */ | |
| 465 if (sign_char == '-') { | |
| 466 spec->n_sign = 1; | |
| 467 spec->sign = '-'; | |
| 468 } | |
| 469 } | |
| 470 | |
| 471 /* The number of chars used for non-digits and non-padding. */ | |
| 472 n_non_digit_non_padding = spec->n_sign + spec->n_prefix + spec->n_decimal + | |
| 473 spec->n_remainder; | |
| 474 | |
| 475 /* min_width can go negative, that's okay. format->width == -1 means | |
| 476 we don't care. */ | |
| 477 if (format->fill_char == '0' && format->align == '=') | |
| 478 spec->n_min_width = format->width - n_non_digit_non_padding; | |
| 479 else | |
| 480 spec->n_min_width = 0; | |
| 481 | |
| 482 if (spec->n_digits == 0) | |
| 483 /* This case only occurs when using 'c' formatting, we need | |
| 484 to special case it because the grouping code always wants | |
| 485 to have at least one character. */ | |
| 486 spec->n_grouped_digits = 0; | |
| 487 else | |
| 488 spec->n_grouped_digits = _PyUnicode_InsertThousandsGrouping( | |
|
72705
05ed6e5f2cf4
Add asciilib: similar to ucs1, ucs2 and ucs4 library, but specialized to ASCII
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72497
diff
changeset
|
489 NULL, PyUnicode_1BYTE_KIND, NULL, 0, NULL, |
| 72475 | 490 spec->n_digits, spec->n_min_width, |
| 491 locale->grouping, locale->thousands_sep); | |
| 492 | |
| 493 /* Given the desired width and the total of digit and non-digit | |
| 494 space we consume, see if we need any padding. format->width can | |
| 495 be negative (meaning no padding), but this code still works in | |
| 496 that case. */ | |
| 497 n_padding = format->width - | |
| 498 (n_non_digit_non_padding + spec->n_grouped_digits); | |
| 499 if (n_padding > 0) { | |
| 500 /* Some padding is needed. Determine if it's left, space, or right. */ | |
| 501 switch (format->align) { | |
| 502 case '<': | |
| 503 spec->n_rpadding = n_padding; | |
| 504 break; | |
| 505 case '^': | |
| 506 spec->n_lpadding = n_padding / 2; | |
| 507 spec->n_rpadding = n_padding - spec->n_lpadding; | |
| 508 break; | |
| 509 case '=': | |
| 510 spec->n_spadding = n_padding; | |
| 511 break; | |
| 512 case '>': | |
| 513 spec->n_lpadding = n_padding; | |
| 514 break; | |
| 515 default: | |
| 516 /* Shouldn't get here, but treat it as '>' */ | |
| 517 spec->n_lpadding = n_padding; | |
| 518 assert(0); | |
| 519 break; | |
| 520 } | |
| 521 } | |
| 522 return spec->n_lpadding + spec->n_sign + spec->n_prefix + | |
| 523 spec->n_spadding + spec->n_grouped_digits + spec->n_decimal + | |
| 524 spec->n_remainder + spec->n_rpadding; | |
| 525 } | |
| 526 | |
| 527 /* Fill in the digit parts of a numbers's string representation, | |
| 528 as determined in calc_number_widths(). | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
529 Return -1 on error, or 0 on success. */ |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
530 static int |
| 72475 | 531 fill_number(PyObject *out, Py_ssize_t pos, const NumberFieldWidths *spec, |
| 532 PyObject *digits, Py_ssize_t d_start, Py_ssize_t d_end, | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
533 PyObject *prefix, Py_ssize_t p_start, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
534 Py_UCS4 fill_char, |
| 72475 | 535 LocaleInfo *locale, int toupper) |
| 536 { | |
| 537 /* Used to keep track of digits, decimal, and remainder. */ | |
| 538 Py_ssize_t d_pos = d_start; | |
| 539 unsigned int kind = PyUnicode_KIND(out); | |
| 540 void *data = PyUnicode_DATA(out); | |
| 541 | |
| 542 #ifndef NDEBUG | |
| 543 Py_ssize_t r; | |
| 544 #endif | |
| 545 | |
| 546 if (spec->n_lpadding) { | |
|
74261
405f65676010
Add a new PyUnicode_Fill() function
Victor Stinner <victor.stinner@haypocalc.com>
parents:
73805
diff
changeset
|
547 PyUnicode_Fill(out, pos, pos + spec->n_lpadding, fill_char); |
| 72475 | 548 pos += spec->n_lpadding; |
| 549 } | |
| 550 if (spec->n_sign == 1) { | |
| 551 PyUnicode_WRITE(kind, data, pos++, spec->sign); | |
| 552 } | |
| 553 if (spec->n_prefix) { | |
|
72497
f05c2df39ea8
fill_number() and format_string_internal() check for PyUnicode_CopyCharacters() failure
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72496
diff
changeset
|
554 if (PyUnicode_CopyCharacters(out, pos, |
|
f05c2df39ea8
fill_number() and format_string_internal() check for PyUnicode_CopyCharacters() failure
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72496
diff
changeset
|
555 prefix, p_start, |
|
f05c2df39ea8
fill_number() and format_string_internal() check for PyUnicode_CopyCharacters() failure
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72496
diff
changeset
|
556 spec->n_prefix) < 0) |
|
f05c2df39ea8
fill_number() and format_string_internal() check for PyUnicode_CopyCharacters() failure
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72496
diff
changeset
|
557 return -1; |
| 72475 | 558 if (toupper) { |
| 559 Py_ssize_t t; | |
|
74341
c2153ce1b5dd
remove some usage of Py_UNICODE_TOUPPER/LOWER
Benjamin Peterson <benjamin@python.org>
parents:
74261
diff
changeset
|
560 for (t = 0; t < spec->n_prefix; t++) { |
|
c2153ce1b5dd
remove some usage of Py_UNICODE_TOUPPER/LOWER
Benjamin Peterson <benjamin@python.org>
parents:
74261
diff
changeset
|
561 Py_UCS4 c = PyUnicode_READ(kind, data, pos + t); |
|
74702
056f5cc8885d
Issue #13706: Add assertions to detect bugs earlier
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74586
diff
changeset
|
562 c = Py_TOUPPER(c); |
|
74557
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
563 assert (c <= 127); |
|
74702
056f5cc8885d
Issue #13706: Add assertions to detect bugs earlier
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74586
diff
changeset
|
564 PyUnicode_WRITE(kind, data, pos + t, c); |
|
74341
c2153ce1b5dd
remove some usage of Py_UNICODE_TOUPPER/LOWER
Benjamin Peterson <benjamin@python.org>
parents:
74261
diff
changeset
|
565 } |
| 72475 | 566 } |
| 567 pos += spec->n_prefix; | |
| 568 } | |
| 569 if (spec->n_spadding) { | |
|
74261
405f65676010
Add a new PyUnicode_Fill() function
Victor Stinner <victor.stinner@haypocalc.com>
parents:
73805
diff
changeset
|
570 PyUnicode_Fill(out, pos, pos + spec->n_spadding, fill_char); |
| 72475 | 571 pos += spec->n_spadding; |
| 572 } | |
| 573 | |
| 574 /* Only for type 'c' special case, it has no digits. */ | |
| 575 if (spec->n_digits != 0) { | |
| 576 /* Fill the digits with InsertThousandsGrouping. */ | |
|
72496
ad337be523bf
fill_number() ensures that the 'digits' string is ready
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72495
diff
changeset
|
577 char *pdigits; |
|
ad337be523bf
fill_number() ensures that the 'digits' string is ready
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72495
diff
changeset
|
578 if (PyUnicode_READY(digits)) |
|
ad337be523bf
fill_number() ensures that the 'digits' string is ready
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72495
diff
changeset
|
579 return -1; |
|
ad337be523bf
fill_number() ensures that the 'digits' string is ready
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72495
diff
changeset
|
580 pdigits = PyUnicode_DATA(digits); |
| 72475 | 581 if (PyUnicode_KIND(digits) < kind) { |
| 582 pdigits = _PyUnicode_AsKind(digits, kind); | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
583 if (pdigits == NULL) |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
584 return -1; |
| 72475 | 585 } |
| 586 #ifndef NDEBUG | |
| 587 r = | |
| 588 #endif | |
| 589 _PyUnicode_InsertThousandsGrouping( | |
|
72705
05ed6e5f2cf4
Add asciilib: similar to ucs1, ucs2 and ucs4 library, but specialized to ASCII
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72497
diff
changeset
|
590 out, kind, |
|
72796
c25262e97304
Change PyUnicode_KIND to 1,2,4. Drop _KIND_SIZE and _CHARACTER_SIZE.
Martin v. Löwis <martin@v.loewis.de>
parents:
72714
diff
changeset
|
591 (char*)data + kind * pos, |
| 72475 | 592 spec->n_grouped_digits, |
|
72796
c25262e97304
Change PyUnicode_KIND to 1,2,4. Drop _KIND_SIZE and _CHARACTER_SIZE.
Martin v. Löwis <martin@v.loewis.de>
parents:
72714
diff
changeset
|
593 pdigits + kind * d_pos, |
| 72475 | 594 spec->n_digits, spec->n_min_width, |
| 595 locale->grouping, locale->thousands_sep); | |
| 596 #ifndef NDEBUG | |
| 597 assert(r == spec->n_grouped_digits); | |
| 598 #endif | |
| 599 if (PyUnicode_KIND(digits) < kind) | |
| 600 PyMem_Free(pdigits); | |
| 601 d_pos += spec->n_digits; | |
| 602 } | |
| 603 if (toupper) { | |
| 604 Py_ssize_t t; | |
|
74341
c2153ce1b5dd
remove some usage of Py_UNICODE_TOUPPER/LOWER
Benjamin Peterson <benjamin@python.org>
parents:
74261
diff
changeset
|
605 for (t = 0; t < spec->n_grouped_digits; t++) { |
|
c2153ce1b5dd
remove some usage of Py_UNICODE_TOUPPER/LOWER
Benjamin Peterson <benjamin@python.org>
parents:
74261
diff
changeset
|
606 Py_UCS4 c = PyUnicode_READ(kind, data, pos + t); |
|
74702
056f5cc8885d
Issue #13706: Add assertions to detect bugs earlier
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74586
diff
changeset
|
607 c = Py_TOUPPER(c); |
|
74341
c2153ce1b5dd
remove some usage of Py_UNICODE_TOUPPER/LOWER
Benjamin Peterson <benjamin@python.org>
parents:
74261
diff
changeset
|
608 if (c > 127) { |
|
c2153ce1b5dd
remove some usage of Py_UNICODE_TOUPPER/LOWER
Benjamin Peterson <benjamin@python.org>
parents:
74261
diff
changeset
|
609 PyErr_SetString(PyExc_SystemError, "non-ascii grouped digit"); |
|
c2153ce1b5dd
remove some usage of Py_UNICODE_TOUPPER/LOWER
Benjamin Peterson <benjamin@python.org>
parents:
74261
diff
changeset
|
610 return -1; |
|
c2153ce1b5dd
remove some usage of Py_UNICODE_TOUPPER/LOWER
Benjamin Peterson <benjamin@python.org>
parents:
74261
diff
changeset
|
611 } |
|
74702
056f5cc8885d
Issue #13706: Add assertions to detect bugs earlier
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74586
diff
changeset
|
612 PyUnicode_WRITE(kind, data, pos + t, c); |
|
74341
c2153ce1b5dd
remove some usage of Py_UNICODE_TOUPPER/LOWER
Benjamin Peterson <benjamin@python.org>
parents:
74261
diff
changeset
|
613 } |
| 72475 | 614 } |
| 615 pos += spec->n_grouped_digits; | |
| 616 | |
| 617 if (spec->n_decimal) { | |
| 618 Py_ssize_t t; | |
| 619 for (t = 0; t < spec->n_decimal; ++t) | |
| 620 PyUnicode_WRITE(kind, data, pos + t, | |
| 621 locale->decimal_point[t]); | |
| 622 pos += spec->n_decimal; | |
| 623 d_pos += 1; | |
| 624 } | |
| 625 | |
| 626 if (spec->n_remainder) { | |
|
72497
f05c2df39ea8
fill_number() and format_string_internal() check for PyUnicode_CopyCharacters() failure
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72496
diff
changeset
|
627 if (PyUnicode_CopyCharacters(out, pos, digits, d_pos, spec->n_remainder) < 0) |
|
f05c2df39ea8
fill_number() and format_string_internal() check for PyUnicode_CopyCharacters() failure
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72496
diff
changeset
|
628 return -1; |
| 72475 | 629 pos += spec->n_remainder; |
| 630 d_pos += spec->n_remainder; | |
| 631 } | |
| 632 | |
| 633 if (spec->n_rpadding) { | |
|
74261
405f65676010
Add a new PyUnicode_Fill() function
Victor Stinner <victor.stinner@haypocalc.com>
parents:
73805
diff
changeset
|
634 PyUnicode_Fill(out, pos, pos + spec->n_rpadding, fill_char); |
| 72475 | 635 pos += spec->n_rpadding; |
| 636 } | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
637 return 0; |
| 72475 | 638 } |
| 639 | |
| 640 static char no_grouping[1] = {CHAR_MAX}; | |
| 641 | |
| 642 /* Find the decimal point character(s?), thousands_separator(s?), and | |
| 643 grouping description, either for the current locale if type is | |
| 644 LT_CURRENT_LOCALE, a hard-coded locale if LT_DEFAULT_LOCALE, or | |
| 645 none if LT_NO_LOCALE. */ | |
| 646 static void | |
| 647 get_locale_info(int type, LocaleInfo *locale_info) | |
| 648 { | |
| 649 switch (type) { | |
| 650 case LT_CURRENT_LOCALE: { | |
| 651 struct lconv *locale_data = localeconv(); | |
| 652 locale_info->decimal_point = locale_data->decimal_point; | |
| 653 locale_info->thousands_sep = locale_data->thousands_sep; | |
| 654 locale_info->grouping = locale_data->grouping; | |
| 655 break; | |
| 656 } | |
| 657 case LT_DEFAULT_LOCALE: | |
| 658 locale_info->decimal_point = "."; | |
| 659 locale_info->thousands_sep = ","; | |
| 660 locale_info->grouping = "\3"; /* Group every 3 characters. The | |
| 661 (implicit) trailing 0 means repeat | |
| 662 infinitely. */ | |
| 663 break; | |
| 664 case LT_NO_LOCALE: | |
| 665 locale_info->decimal_point = "."; | |
| 666 locale_info->thousands_sep = ""; | |
| 667 locale_info->grouping = no_grouping; | |
| 668 break; | |
| 669 default: | |
| 670 assert(0); | |
| 671 } | |
| 672 } | |
| 673 | |
| 674 /************************************************************************/ | |
| 675 /*********** string formatting ******************************************/ | |
| 676 /************************************************************************/ | |
| 677 | |
| 678 static PyObject * | |
| 679 format_string_internal(PyObject *value, const InternalFormatSpec *format) | |
| 680 { | |
| 681 Py_ssize_t lpad; | |
| 682 Py_ssize_t rpad; | |
| 683 Py_ssize_t total; | |
| 684 Py_ssize_t pos; | |
|
72863
afa42c04f0a3
Fix misuse of PyUnicode_GET_SIZE, use PyUnicode_GET_LENGTH instead
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72796
diff
changeset
|
685 Py_ssize_t len = PyUnicode_GET_LENGTH(value); |
| 72475 | 686 PyObject *result = NULL; |
|
74585
04b565d912c2
Fix compilation warnings (seen on win32 buildbot)
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents:
74557
diff
changeset
|
687 Py_UCS4 maxchar = 127; |
| 72475 | 688 |
| 689 /* sign is not allowed on strings */ | |
| 690 if (format->sign != '\0') { | |
| 691 PyErr_SetString(PyExc_ValueError, | |
| 692 "Sign not allowed in string format specifier"); | |
| 693 goto done; | |
| 694 } | |
| 695 | |
| 696 /* alternate is not allowed on strings */ | |
| 697 if (format->alternate) { | |
| 698 PyErr_SetString(PyExc_ValueError, | |
| 699 "Alternate form (#) not allowed in string format " | |
| 700 "specifier"); | |
| 701 goto done; | |
| 702 } | |
| 703 | |
| 704 /* '=' alignment not allowed on strings */ | |
| 705 if (format->align == '=') { | |
| 706 PyErr_SetString(PyExc_ValueError, | |
| 707 "'=' alignment not allowed " | |
| 708 "in string format specifier"); | |
| 709 goto done; | |
| 710 } | |
| 711 | |
| 712 /* if precision is specified, output no more that format.precision | |
| 713 characters */ | |
| 714 if (format->precision >= 0 && len >= format->precision) { | |
| 715 len = format->precision; | |
| 716 } | |
| 717 | |
| 718 calc_padding(len, format->width, format->align, &lpad, &rpad, &total); | |
| 719 | |
|
74557
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
720 if (lpad != 0 || rpad != 0) |
|
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
721 maxchar = Py_MAX(maxchar, format->fill_char); |
|
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
722 |
| 72475 | 723 /* allocate the resulting string */ |
| 724 result = PyUnicode_New(total, maxchar); | |
| 725 if (result == NULL) | |
| 726 goto done; | |
| 727 | |
| 728 /* Write into that space. First the padding. */ | |
| 729 pos = fill_padding(result, 0, len, | |
| 730 format->fill_char=='\0'?' ':format->fill_char, | |
| 731 lpad, rpad); | |
| 732 | |
| 733 /* Then the source string. */ | |
|
72497
f05c2df39ea8
fill_number() and format_string_internal() check for PyUnicode_CopyCharacters() failure
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72496
diff
changeset
|
734 if (PyUnicode_CopyCharacters(result, pos, value, 0, len) < 0) |
|
f05c2df39ea8
fill_number() and format_string_internal() check for PyUnicode_CopyCharacters() failure
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72496
diff
changeset
|
735 Py_CLEAR(result); |
| 72475 | 736 |
| 737 done: | |
|
74702
056f5cc8885d
Issue #13706: Add assertions to detect bugs earlier
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74586
diff
changeset
|
738 assert(!result || _PyUnicode_CheckConsistency(result, 1)); |
| 72475 | 739 return result; |
| 740 } | |
| 741 | |
| 742 | |
| 743 /************************************************************************/ | |
| 744 /*********** long formatting ********************************************/ | |
| 745 /************************************************************************/ | |
| 746 | |
| 747 typedef PyObject* | |
| 748 (*IntOrLongToString)(PyObject *value, int base); | |
| 749 | |
| 750 static PyObject * | |
| 751 format_int_or_long_internal(PyObject *value, const InternalFormatSpec *format, | |
| 752 IntOrLongToString tostring) | |
| 753 { | |
| 754 PyObject *result = NULL; | |
|
74585
04b565d912c2
Fix compilation warnings (seen on win32 buildbot)
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents:
74557
diff
changeset
|
755 Py_UCS4 maxchar = 127; |
| 72475 | 756 PyObject *tmp = NULL; |
| 757 Py_ssize_t inumeric_chars; | |
| 758 Py_UCS4 sign_char = '\0'; | |
| 759 Py_ssize_t n_digits; /* count of digits need from the computed | |
| 760 string */ | |
| 761 Py_ssize_t n_remainder = 0; /* Used only for 'c' formatting, which | |
| 762 produces non-digits */ | |
| 763 Py_ssize_t n_prefix = 0; /* Count of prefix chars, (e.g., '0x') */ | |
| 764 Py_ssize_t n_total; | |
|
74702
056f5cc8885d
Issue #13706: Add assertions to detect bugs earlier
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74586
diff
changeset
|
765 Py_ssize_t prefix = 0; |
| 72475 | 766 NumberFieldWidths spec; |
| 767 long x; | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
768 int err; |
| 72475 | 769 |
| 770 /* Locale settings, either from the actual locale or | |
| 771 from a hard-code pseudo-locale */ | |
| 772 LocaleInfo locale; | |
| 773 | |
| 774 /* no precision allowed on integers */ | |
| 775 if (format->precision != -1) { | |
| 776 PyErr_SetString(PyExc_ValueError, | |
| 777 "Precision not allowed in integer format specifier"); | |
| 778 goto done; | |
| 779 } | |
| 780 | |
| 781 /* special case for character formatting */ | |
| 782 if (format->type == 'c') { | |
| 783 /* error to specify a sign */ | |
| 784 if (format->sign != '\0') { | |
| 785 PyErr_SetString(PyExc_ValueError, | |
| 786 "Sign not allowed with integer" | |
| 787 " format specifier 'c'"); | |
| 788 goto done; | |
| 789 } | |
| 790 | |
| 791 /* taken from unicodeobject.c formatchar() */ | |
| 792 /* Integer input truncated to a character */ | |
| 793 x = PyLong_AsLong(value); | |
| 794 if (x == -1 && PyErr_Occurred()) | |
| 795 goto done; | |
| 796 if (x < 0 || x > 0x10ffff) { | |
| 797 PyErr_SetString(PyExc_OverflowError, | |
|
74557
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
798 "%c arg not in range(0x110000)"); |
| 72475 | 799 goto done; |
| 800 } | |
| 801 tmp = PyUnicode_FromOrdinal(x); | |
| 802 inumeric_chars = 0; | |
| 803 n_digits = 1; | |
|
74586
728cfc671d15
Silence last compilation warning.
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents:
74585
diff
changeset
|
804 maxchar = Py_MAX(maxchar, (Py_UCS4)x); |
| 72475 | 805 |
| 806 /* As a sort-of hack, we tell calc_number_widths that we only | |
| 807 have "remainder" characters. calc_number_widths thinks | |
| 808 these are characters that don't get formatted, only copied | |
| 809 into the output string. We do this for 'c' formatting, | |
| 810 because the characters are likely to be non-digits. */ | |
| 811 n_remainder = 1; | |
| 812 } | |
| 813 else { | |
| 814 int base; | |
| 815 int leading_chars_to_skip = 0; /* Number of characters added by | |
| 816 PyNumber_ToBase that we want to | |
| 817 skip over. */ | |
| 818 | |
| 819 /* Compute the base and how many characters will be added by | |
| 820 PyNumber_ToBase */ | |
| 821 switch (format->type) { | |
| 822 case 'b': | |
| 823 base = 2; | |
| 824 leading_chars_to_skip = 2; /* 0b */ | |
| 825 break; | |
| 826 case 'o': | |
| 827 base = 8; | |
| 828 leading_chars_to_skip = 2; /* 0o */ | |
| 829 break; | |
| 830 case 'x': | |
| 831 case 'X': | |
| 832 base = 16; | |
| 833 leading_chars_to_skip = 2; /* 0x */ | |
| 834 break; | |
| 835 default: /* shouldn't be needed, but stops a compiler warning */ | |
| 836 case 'd': | |
| 837 case 'n': | |
| 838 base = 10; | |
| 839 break; | |
| 840 } | |
| 841 | |
| 842 /* The number of prefix chars is the same as the leading | |
| 843 chars to skip */ | |
| 844 if (format->alternate) | |
| 845 n_prefix = leading_chars_to_skip; | |
| 846 | |
| 847 /* Do the hard part, converting to a string in a given base */ | |
| 848 tmp = tostring(value, base); | |
| 849 if (tmp == NULL || PyUnicode_READY(tmp) == -1) | |
| 850 goto done; | |
| 851 | |
| 852 inumeric_chars = 0; | |
| 853 n_digits = PyUnicode_GET_LENGTH(tmp); | |
| 854 | |
| 855 prefix = inumeric_chars; | |
| 856 | |
| 857 /* Is a sign character present in the output? If so, remember it | |
| 858 and skip it */ | |
| 859 if (PyUnicode_READ_CHAR(tmp, inumeric_chars) == '-') { | |
| 860 sign_char = '-'; | |
| 861 ++prefix; | |
| 862 ++leading_chars_to_skip; | |
| 863 } | |
| 864 | |
| 865 /* Skip over the leading chars (0x, 0b, etc.) */ | |
| 866 n_digits -= leading_chars_to_skip; | |
| 867 inumeric_chars += leading_chars_to_skip; | |
| 868 } | |
| 869 | |
| 870 /* Determine the grouping, separator, and decimal point, if any. */ | |
| 871 get_locale_info(format->type == 'n' ? LT_CURRENT_LOCALE : | |
| 872 (format->thousands_separators ? | |
| 873 LT_DEFAULT_LOCALE : | |
| 874 LT_NO_LOCALE), | |
| 875 &locale); | |
| 876 | |
| 877 /* Calculate how much memory we'll need. */ | |
| 878 n_total = calc_number_widths(&spec, n_prefix, sign_char, tmp, inumeric_chars, | |
| 879 inumeric_chars + n_digits, n_remainder, 0, &locale, format); | |
| 880 | |
|
74557
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
881 if (spec.n_lpadding || spec.n_spadding || spec.n_rpadding) |
|
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
882 maxchar = Py_MAX(maxchar, format->fill_char); |
|
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
883 |
| 72475 | 884 /* Allocate the memory. */ |
| 885 result = PyUnicode_New(n_total, maxchar); | |
| 886 if (!result) | |
| 887 goto done; | |
| 888 | |
| 889 /* Populate the memory. */ | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
890 err = fill_number(result, 0, &spec, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
891 tmp, inumeric_chars, inumeric_chars + n_digits, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
892 tmp, prefix, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
893 format->fill_char == '\0' ? ' ' : format->fill_char, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
894 &locale, format->type == 'X'); |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
895 if (err) |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
896 Py_CLEAR(result); |
| 72475 | 897 |
| 898 done: | |
| 899 Py_XDECREF(tmp); | |
|
74702
056f5cc8885d
Issue #13706: Add assertions to detect bugs earlier
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74586
diff
changeset
|
900 assert(!result || _PyUnicode_CheckConsistency(result, 1)); |
| 72475 | 901 return result; |
| 902 } | |
| 903 | |
| 904 /************************************************************************/ | |
| 905 /*********** float formatting *******************************************/ | |
| 906 /************************************************************************/ | |
| 907 | |
| 908 static PyObject* | |
| 909 strtounicode(char *charbuffer, Py_ssize_t len) | |
| 910 { | |
| 911 return PyUnicode_FromKindAndData(PyUnicode_1BYTE_KIND, charbuffer, len); | |
| 912 } | |
|
47393
0da27fdf0192
Refactor and clean up str.format() code (and helpers) in advance of optimizations.
Eric Smith <eric@trueblade.com>
parents:
42817
diff
changeset
|
913 |
| 72475 | 914 /* much of this is taken from unicodeobject.c */ |
| 915 static PyObject * | |
| 916 format_float_internal(PyObject *value, | |
| 917 const InternalFormatSpec *format) | |
| 918 { | |
| 919 char *buf = NULL; /* buffer returned from PyOS_double_to_string */ | |
| 920 Py_ssize_t n_digits; | |
| 921 Py_ssize_t n_remainder; | |
| 922 Py_ssize_t n_total; | |
| 923 int has_decimal; | |
| 924 double val; | |
| 925 Py_ssize_t precision = format->precision; | |
| 926 Py_ssize_t default_precision = 6; | |
| 927 Py_UCS4 type = format->type; | |
| 928 int add_pct = 0; | |
| 929 Py_ssize_t index; | |
| 930 NumberFieldWidths spec; | |
| 931 int flags = 0; | |
| 932 PyObject *result = NULL; | |
|
74585
04b565d912c2
Fix compilation warnings (seen on win32 buildbot)
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents:
74557
diff
changeset
|
933 Py_UCS4 maxchar = 127; |
| 72475 | 934 Py_UCS4 sign_char = '\0'; |
| 935 int float_type; /* Used to see if we have a nan, inf, or regular float. */ | |
| 936 PyObject *unicode_tmp = NULL; | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
937 int err; |
| 72475 | 938 |
| 939 /* Locale settings, either from the actual locale or | |
| 940 from a hard-code pseudo-locale */ | |
| 941 LocaleInfo locale; | |
| 942 | |
| 943 if (format->alternate) | |
| 944 flags |= Py_DTSF_ALT; | |
| 945 | |
| 946 if (type == '\0') { | |
| 947 /* Omitted type specifier. Behaves in the same way as repr(x) | |
| 948 and str(x) if no precision is given, else like 'g', but with | |
| 949 at least one digit after the decimal point. */ | |
| 950 flags |= Py_DTSF_ADD_DOT_0; | |
| 951 type = 'r'; | |
| 952 default_precision = 0; | |
| 953 } | |
| 954 | |
| 955 if (type == 'n') | |
| 956 /* 'n' is the same as 'g', except for the locale used to | |
| 957 format the result. We take care of that later. */ | |
| 958 type = 'g'; | |
| 959 | |
| 960 val = PyFloat_AsDouble(value); | |
| 961 if (val == -1.0 && PyErr_Occurred()) | |
| 962 goto done; | |
| 963 | |
| 964 if (type == '%') { | |
| 965 type = 'f'; | |
| 966 val *= 100; | |
| 967 add_pct = 1; | |
| 968 } | |
| 969 | |
| 970 if (precision < 0) | |
| 971 precision = default_precision; | |
| 972 else if (type == 'r') | |
| 973 type = 'g'; | |
| 974 | |
| 975 /* Cast "type", because if we're in unicode we need to pass a | |
| 976 8-bit char. This is safe, because we've restricted what "type" | |
| 977 can be. */ | |
| 978 buf = PyOS_double_to_string(val, (char)type, precision, flags, | |
| 979 &float_type); | |
| 980 if (buf == NULL) | |
| 981 goto done; | |
| 982 n_digits = strlen(buf); | |
| 983 | |
| 984 if (add_pct) { | |
| 985 /* We know that buf has a trailing zero (since we just called | |
| 986 strlen() on it), and we don't use that fact any more. So we | |
| 987 can just write over the trailing zero. */ | |
| 988 buf[n_digits] = '%'; | |
| 989 n_digits += 1; | |
| 990 } | |
| 991 | |
| 992 /* Since there is no unicode version of PyOS_double_to_string, | |
| 993 just use the 8 bit version and then convert to unicode. */ | |
| 994 unicode_tmp = strtounicode(buf, n_digits); | |
| 995 if (unicode_tmp == NULL) | |
| 996 goto done; | |
| 997 index = 0; | |
| 998 | |
| 999 /* Is a sign character present in the output? If so, remember it | |
| 1000 and skip it */ | |
| 1001 if (PyUnicode_READ_CHAR(unicode_tmp, index) == '-') { | |
| 1002 sign_char = '-'; | |
| 1003 ++index; | |
| 1004 --n_digits; | |
| 1005 } | |
| 1006 | |
| 1007 /* Determine if we have any "remainder" (after the digits, might include | |
| 1008 decimal or exponent or both (or neither)) */ | |
| 1009 parse_number(unicode_tmp, index, index + n_digits, &n_remainder, &has_decimal); | |
| 1010 | |
| 1011 /* Determine the grouping, separator, and decimal point, if any. */ | |
| 1012 get_locale_info(format->type == 'n' ? LT_CURRENT_LOCALE : | |
| 1013 (format->thousands_separators ? | |
| 1014 LT_DEFAULT_LOCALE : | |
| 1015 LT_NO_LOCALE), | |
| 1016 &locale); | |
| 1017 | |
| 1018 /* Calculate how much memory we'll need. */ | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1019 n_total = calc_number_widths(&spec, 0, sign_char, unicode_tmp, index, |
| 72475 | 1020 index + n_digits, n_remainder, has_decimal, |
| 1021 &locale, format); | |
| 1022 | |
|
74557
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
1023 if (spec.n_lpadding || spec.n_spadding || spec.n_rpadding) |
|
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
1024 maxchar = Py_MAX(maxchar, format->fill_char); |
|
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
1025 |
| 72475 | 1026 /* Allocate the memory. */ |
| 1027 result = PyUnicode_New(n_total, maxchar); | |
| 1028 if (result == NULL) | |
| 1029 goto done; | |
| 1030 | |
| 1031 /* Populate the memory. */ | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1032 err = fill_number(result, 0, &spec, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1033 unicode_tmp, index, index + n_digits, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1034 NULL, 0, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1035 format->fill_char == '\0' ? ' ' : format->fill_char, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1036 &locale, 0); |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1037 if (err) |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1038 Py_CLEAR(result); |
| 72475 | 1039 |
| 1040 done: | |
| 1041 PyMem_Free(buf); | |
| 1042 Py_DECREF(unicode_tmp); | |
|
74702
056f5cc8885d
Issue #13706: Add assertions to detect bugs earlier
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74586
diff
changeset
|
1043 assert(!result || _PyUnicode_CheckConsistency(result, 1)); |
| 72475 | 1044 return result; |
| 1045 } | |
| 1046 | |
| 1047 /************************************************************************/ | |
| 1048 /*********** complex formatting *****************************************/ | |
| 1049 /************************************************************************/ | |
| 1050 | |
| 1051 static PyObject * | |
| 1052 format_complex_internal(PyObject *value, | |
| 1053 const InternalFormatSpec *format) | |
| 1054 { | |
| 1055 double re; | |
| 1056 double im; | |
| 1057 char *re_buf = NULL; /* buffer returned from PyOS_double_to_string */ | |
| 1058 char *im_buf = NULL; /* buffer returned from PyOS_double_to_string */ | |
| 1059 | |
| 1060 InternalFormatSpec tmp_format = *format; | |
| 1061 Py_ssize_t n_re_digits; | |
| 1062 Py_ssize_t n_im_digits; | |
| 1063 Py_ssize_t n_re_remainder; | |
| 1064 Py_ssize_t n_im_remainder; | |
| 1065 Py_ssize_t n_re_total; | |
| 1066 Py_ssize_t n_im_total; | |
| 1067 int re_has_decimal; | |
| 1068 int im_has_decimal; | |
| 1069 Py_ssize_t precision = format->precision; | |
| 1070 Py_ssize_t default_precision = 6; | |
| 1071 Py_UCS4 type = format->type; | |
| 1072 Py_ssize_t i_re; | |
| 1073 Py_ssize_t i_im; | |
| 1074 NumberFieldWidths re_spec; | |
| 1075 NumberFieldWidths im_spec; | |
| 1076 int flags = 0; | |
| 1077 PyObject *result = NULL; | |
|
74585
04b565d912c2
Fix compilation warnings (seen on win32 buildbot)
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents:
74557
diff
changeset
|
1078 Py_UCS4 maxchar = 127; |
| 72475 | 1079 int rkind; |
| 1080 void *rdata; | |
| 1081 Py_ssize_t index; | |
| 1082 Py_UCS4 re_sign_char = '\0'; | |
| 1083 Py_UCS4 im_sign_char = '\0'; | |
| 1084 int re_float_type; /* Used to see if we have a nan, inf, or regular float. */ | |
| 1085 int im_float_type; | |
| 1086 int add_parens = 0; | |
| 1087 int skip_re = 0; | |
| 1088 Py_ssize_t lpad; | |
| 1089 Py_ssize_t rpad; | |
| 1090 Py_ssize_t total; | |
| 1091 PyObject *re_unicode_tmp = NULL; | |
| 1092 PyObject *im_unicode_tmp = NULL; | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1093 int err; |
| 72475 | 1094 |
| 1095 /* Locale settings, either from the actual locale or | |
| 1096 from a hard-code pseudo-locale */ | |
| 1097 LocaleInfo locale; | |
| 1098 | |
| 1099 /* Zero padding is not allowed. */ | |
| 1100 if (format->fill_char == '0') { | |
| 1101 PyErr_SetString(PyExc_ValueError, | |
| 1102 "Zero padding is not allowed in complex format " | |
| 1103 "specifier"); | |
| 1104 goto done; | |
| 1105 } | |
| 1106 | |
| 1107 /* Neither is '=' alignment . */ | |
| 1108 if (format->align == '=') { | |
| 1109 PyErr_SetString(PyExc_ValueError, | |
| 1110 "'=' alignment flag is not allowed in complex format " | |
| 1111 "specifier"); | |
| 1112 goto done; | |
| 1113 } | |
| 1114 | |
| 1115 re = PyComplex_RealAsDouble(value); | |
| 1116 if (re == -1.0 && PyErr_Occurred()) | |
| 1117 goto done; | |
| 1118 im = PyComplex_ImagAsDouble(value); | |
| 1119 if (im == -1.0 && PyErr_Occurred()) | |
| 1120 goto done; | |
| 1121 | |
| 1122 if (format->alternate) | |
| 1123 flags |= Py_DTSF_ALT; | |
| 1124 | |
| 1125 if (type == '\0') { | |
| 1126 /* Omitted type specifier. Should be like str(self). */ | |
| 1127 type = 'r'; | |
| 1128 default_precision = 0; | |
| 1129 if (re == 0.0 && copysign(1.0, re) == 1.0) | |
| 1130 skip_re = 1; | |
| 1131 else | |
| 1132 add_parens = 1; | |
| 1133 } | |
| 1134 | |
| 1135 if (type == 'n') | |
| 1136 /* 'n' is the same as 'g', except for the locale used to | |
| 1137 format the result. We take care of that later. */ | |
| 1138 type = 'g'; | |
| 1139 | |
| 1140 if (precision < 0) | |
| 1141 precision = default_precision; | |
| 1142 else if (type == 'r') | |
| 1143 type = 'g'; | |
| 1144 | |
| 1145 /* Cast "type", because if we're in unicode we need to pass a | |
| 1146 8-bit char. This is safe, because we've restricted what "type" | |
| 1147 can be. */ | |
| 1148 re_buf = PyOS_double_to_string(re, (char)type, precision, flags, | |
| 1149 &re_float_type); | |
| 1150 if (re_buf == NULL) | |
| 1151 goto done; | |
| 1152 im_buf = PyOS_double_to_string(im, (char)type, precision, flags, | |
| 1153 &im_float_type); | |
| 1154 if (im_buf == NULL) | |
| 1155 goto done; | |
| 1156 | |
| 1157 n_re_digits = strlen(re_buf); | |
| 1158 n_im_digits = strlen(im_buf); | |
| 1159 | |
| 1160 /* Since there is no unicode version of PyOS_double_to_string, | |
| 1161 just use the 8 bit version and then convert to unicode. */ | |
| 1162 re_unicode_tmp = strtounicode(re_buf, n_re_digits); | |
| 1163 if (re_unicode_tmp == NULL) | |
| 1164 goto done; | |
| 1165 i_re = 0; | |
| 1166 | |
| 1167 im_unicode_tmp = strtounicode(im_buf, n_im_digits); | |
| 1168 if (im_unicode_tmp == NULL) | |
| 1169 goto done; | |
| 1170 i_im = 0; | |
| 1171 | |
| 1172 /* Is a sign character present in the output? If so, remember it | |
| 1173 and skip it */ | |
| 1174 if (PyUnicode_READ_CHAR(re_unicode_tmp, i_re) == '-') { | |
| 1175 re_sign_char = '-'; | |
| 1176 ++i_re; | |
| 1177 --n_re_digits; | |
| 1178 } | |
| 1179 if (PyUnicode_READ_CHAR(im_unicode_tmp, i_im) == '-') { | |
| 1180 im_sign_char = '-'; | |
| 1181 ++i_im; | |
| 1182 --n_im_digits; | |
| 1183 } | |
| 1184 | |
| 1185 /* Determine if we have any "remainder" (after the digits, might include | |
| 1186 decimal or exponent or both (or neither)) */ | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1187 parse_number(re_unicode_tmp, i_re, i_re + n_re_digits, |
| 72475 | 1188 &n_re_remainder, &re_has_decimal); |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1189 parse_number(im_unicode_tmp, i_im, i_im + n_im_digits, |
| 72475 | 1190 &n_im_remainder, &im_has_decimal); |
| 1191 | |
| 1192 /* Determine the grouping, separator, and decimal point, if any. */ | |
| 1193 get_locale_info(format->type == 'n' ? LT_CURRENT_LOCALE : | |
| 1194 (format->thousands_separators ? | |
| 1195 LT_DEFAULT_LOCALE : | |
| 1196 LT_NO_LOCALE), | |
| 1197 &locale); | |
| 1198 | |
| 1199 /* Turn off any padding. We'll do it later after we've composed | |
| 1200 the numbers without padding. */ | |
| 1201 tmp_format.fill_char = '\0'; | |
| 1202 tmp_format.align = '<'; | |
| 1203 tmp_format.width = -1; | |
| 1204 | |
| 1205 /* Calculate how much memory we'll need. */ | |
| 1206 n_re_total = calc_number_widths(&re_spec, 0, re_sign_char, re_unicode_tmp, | |
| 1207 i_re, i_re + n_re_digits, n_re_remainder, | |
| 1208 re_has_decimal, &locale, &tmp_format); | |
| 1209 | |
| 1210 /* Same formatting, but always include a sign, unless the real part is | |
| 1211 * going to be omitted, in which case we use whatever sign convention was | |
| 1212 * requested by the original format. */ | |
| 1213 if (!skip_re) | |
| 1214 tmp_format.sign = '+'; | |
| 1215 n_im_total = calc_number_widths(&im_spec, 0, im_sign_char, im_unicode_tmp, | |
| 1216 i_im, i_im + n_im_digits, n_im_remainder, | |
| 1217 im_has_decimal, &locale, &tmp_format); | |
| 1218 | |
| 1219 if (skip_re) | |
| 1220 n_re_total = 0; | |
| 1221 | |
| 1222 /* Add 1 for the 'j', and optionally 2 for parens. */ | |
| 1223 calc_padding(n_re_total + n_im_total + 1 + add_parens * 2, | |
| 1224 format->width, format->align, &lpad, &rpad, &total); | |
| 1225 | |
|
74557
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
1226 if (re_spec.n_lpadding || re_spec.n_spadding || re_spec.n_rpadding |
|
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
1227 || im_spec.n_lpadding || im_spec.n_spadding || im_spec.n_rpadding |
|
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
1228 || lpad || rpad) |
|
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
1229 maxchar = Py_MAX(maxchar, format->fill_char); |
|
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
1230 |
| 72475 | 1231 result = PyUnicode_New(total, maxchar); |
| 1232 if (result == NULL) | |
| 1233 goto done; | |
| 1234 rkind = PyUnicode_KIND(result); | |
| 1235 rdata = PyUnicode_DATA(result); | |
| 1236 | |
| 1237 /* Populate the memory. First, the padding. */ | |
| 1238 index = fill_padding(result, 0, | |
| 1239 n_re_total + n_im_total + 1 + add_parens * 2, | |
| 1240 format->fill_char=='\0' ? ' ' : format->fill_char, | |
| 1241 lpad, rpad); | |
| 1242 | |
| 1243 if (add_parens) | |
| 1244 PyUnicode_WRITE(rkind, rdata, index++, '('); | |
| 1245 | |
| 1246 if (!skip_re) { | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1247 err = fill_number(result, index, &re_spec, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1248 re_unicode_tmp, i_re, i_re + n_re_digits, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1249 NULL, 0, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1250 0, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1251 &locale, 0); |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1252 if (err) { |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1253 Py_CLEAR(result); |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1254 goto done; |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1255 } |
| 72475 | 1256 index += n_re_total; |
| 1257 } | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1258 err = fill_number(result, index, &im_spec, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1259 im_unicode_tmp, i_im, i_im + n_im_digits, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1260 NULL, 0, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1261 0, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1262 &locale, 0); |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1263 if (err) { |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1264 Py_CLEAR(result); |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1265 goto done; |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1266 } |
| 72475 | 1267 index += n_im_total; |
| 1268 PyUnicode_WRITE(rkind, rdata, index++, 'j'); | |
| 1269 | |
| 1270 if (add_parens) | |
| 1271 PyUnicode_WRITE(rkind, rdata, index++, ')'); | |
| 1272 | |
| 1273 done: | |
| 1274 PyMem_Free(re_buf); | |
| 1275 PyMem_Free(im_buf); | |
| 1276 Py_XDECREF(re_unicode_tmp); | |
| 1277 Py_XDECREF(im_unicode_tmp); | |
|
74702
056f5cc8885d
Issue #13706: Add assertions to detect bugs earlier
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74586
diff
changeset
|
1278 assert(!result || _PyUnicode_CheckConsistency(result, 1)); |
| 72475 | 1279 return result; |
| 1280 } | |
| 1281 | |
| 1282 /************************************************************************/ | |
| 1283 /*********** built in formatters ****************************************/ | |
| 1284 /************************************************************************/ | |
| 1285 PyObject * | |
| 1286 _PyUnicode_FormatAdvanced(PyObject *obj, | |
| 1287 PyObject *format_spec, | |
| 1288 Py_ssize_t start, Py_ssize_t end) | |
| 1289 { | |
| 1290 InternalFormatSpec format; | |
|
72714
6f03716079a9
Don't check for the maximum character when copying from unicodeobject.c
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72705
diff
changeset
|
1291 PyObject *result; |
| 72475 | 1292 |
| 1293 /* check for the special case of zero length format spec, make | |
| 1294 it equivalent to str(obj) */ | |
|
72714
6f03716079a9
Don't check for the maximum character when copying from unicodeobject.c
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72705
diff
changeset
|
1295 if (start == end) |
|
6f03716079a9
Don't check for the maximum character when copying from unicodeobject.c
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72705
diff
changeset
|
1296 return PyObject_Str(obj); |
| 72475 | 1297 |
| 1298 /* parse the format_spec */ | |
| 1299 if (!parse_internal_render_format_spec(format_spec, start, end, | |
| 1300 &format, 's', '<')) | |
|
72714
6f03716079a9
Don't check for the maximum character when copying from unicodeobject.c
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72705
diff
changeset
|
1301 return NULL; |
| 72475 | 1302 |
| 1303 /* type conversion? */ | |
| 1304 switch (format.type) { | |
| 1305 case 's': | |
| 1306 /* no type conversion needed, already a string. do the formatting */ | |
| 1307 result = format_string_internal(obj, &format); | |
|
72714
6f03716079a9
Don't check for the maximum character when copying from unicodeobject.c
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72705
diff
changeset
|
1308 if (result != NULL) |
|
6f03716079a9
Don't check for the maximum character when copying from unicodeobject.c
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72705
diff
changeset
|
1309 assert(_PyUnicode_CheckConsistency(result, 1)); |
| 72475 | 1310 break; |
| 1311 default: | |
| 1312 /* unknown */ | |
| 1313 unknown_presentation_type(format.type, obj->ob_type->tp_name); | |
|
72714
6f03716079a9
Don't check for the maximum character when copying from unicodeobject.c
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72705
diff
changeset
|
1314 result = NULL; |
| 72475 | 1315 } |
| 1316 return result; | |
| 1317 } | |
| 1318 | |
| 1319 static PyObject* | |
| 1320 format_int_or_long(PyObject* obj, PyObject* format_spec, | |
| 1321 Py_ssize_t start, Py_ssize_t end, | |
| 1322 IntOrLongToString tostring) | |
| 1323 { | |
| 1324 PyObject *result = NULL; | |
| 1325 PyObject *tmp = NULL; | |
| 1326 InternalFormatSpec format; | |
| 1327 | |
| 1328 /* check for the special case of zero length format spec, make | |
| 1329 it equivalent to str(obj) */ | |
| 1330 if (start == end) { | |
| 1331 result = PyObject_Str(obj); | |
| 1332 goto done; | |
| 1333 } | |
| 1334 | |
| 1335 /* parse the format_spec */ | |
| 1336 if (!parse_internal_render_format_spec(format_spec, start, end, | |
| 1337 &format, 'd', '>')) | |
| 1338 goto done; | |
| 1339 | |
| 1340 /* type conversion? */ | |
| 1341 switch (format.type) { | |
| 1342 case 'b': | |
| 1343 case 'c': | |
| 1344 case 'd': | |
| 1345 case 'o': | |
| 1346 case 'x': | |
| 1347 case 'X': | |
| 1348 case 'n': | |
| 1349 /* no type conversion needed, already an int (or long). do | |
| 1350 the formatting */ | |
| 1351 result = format_int_or_long_internal(obj, &format, tostring); | |
| 1352 break; | |
| 1353 | |
| 1354 case 'e': | |
| 1355 case 'E': | |
| 1356 case 'f': | |
| 1357 case 'F': | |
| 1358 case 'g': | |
| 1359 case 'G': | |
| 1360 case '%': | |
| 1361 /* convert to float */ | |
| 1362 tmp = PyNumber_Float(obj); | |
| 1363 if (tmp == NULL) | |
| 1364 goto done; | |
| 1365 result = format_float_internal(tmp, &format); | |
| 1366 break; | |
| 1367 | |
| 1368 default: | |
| 1369 /* unknown */ | |
| 1370 unknown_presentation_type(format.type, obj->ob_type->tp_name); | |
| 1371 goto done; | |
| 1372 } | |
| 1373 | |
| 1374 done: | |
| 1375 Py_XDECREF(tmp); | |
| 1376 return result; | |
| 1377 } | |
| 1378 | |
| 1379 /* Need to define long_format as a function that will convert a long | |
| 1380 to a string. In 3.0, _PyLong_Format has the correct signature. */ | |
| 1381 #define long_format _PyLong_Format | |
| 1382 | |
| 1383 PyObject * | |
| 1384 _PyLong_FormatAdvanced(PyObject *obj, | |
| 1385 PyObject *format_spec, | |
| 1386 Py_ssize_t start, Py_ssize_t end) | |
| 1387 { | |
| 1388 return format_int_or_long(obj, format_spec, start, end, | |
| 1389 long_format); | |
| 1390 } | |
| 1391 | |
| 1392 PyObject * | |
| 1393 _PyFloat_FormatAdvanced(PyObject *obj, | |
| 1394 PyObject *format_spec, | |
| 1395 Py_ssize_t start, Py_ssize_t end) | |
| 1396 { | |
| 1397 PyObject *result = NULL; | |
| 1398 InternalFormatSpec format; | |
| 1399 | |
| 1400 /* check for the special case of zero length format spec, make | |
| 1401 it equivalent to str(obj) */ | |
| 1402 if (start == end) { | |
| 1403 result = PyObject_Str(obj); | |
| 1404 goto done; | |
| 1405 } | |
| 1406 | |
| 1407 /* parse the format_spec */ | |
| 1408 if (!parse_internal_render_format_spec(format_spec, start, end, | |
| 1409 &format, '\0', '>')) | |
| 1410 goto done; | |
| 1411 | |
| 1412 /* type conversion? */ | |
| 1413 switch (format.type) { | |
| 1414 case '\0': /* No format code: like 'g', but with at least one decimal. */ | |
| 1415 case 'e': | |
| 1416 case 'E': | |
| 1417 case 'f': | |
| 1418 case 'F': | |
| 1419 case 'g': | |
| 1420 case 'G': | |
| 1421 case 'n': | |
| 1422 case '%': | |
| 1423 /* no conversion, already a float. do the formatting */ | |
| 1424 result = format_float_internal(obj, &format); | |
| 1425 break; | |
| 1426 | |
| 1427 default: | |
| 1428 /* unknown */ | |
| 1429 unknown_presentation_type(format.type, obj->ob_type->tp_name); | |
| 1430 goto done; | |
| 1431 } | |
| 1432 | |
| 1433 done: | |
| 1434 return result; | |
| 1435 } | |
| 1436 | |
| 1437 PyObject * | |
| 1438 _PyComplex_FormatAdvanced(PyObject *obj, | |
| 1439 PyObject *format_spec, | |
| 1440 Py_ssize_t start, Py_ssize_t end) | |
| 1441 { | |
| 1442 PyObject *result = NULL; | |
| 1443 InternalFormatSpec format; | |
| 1444 | |
| 1445 /* check for the special case of zero length format spec, make | |
| 1446 it equivalent to str(obj) */ | |
| 1447 if (start == end) { | |
| 1448 result = PyObject_Str(obj); | |
| 1449 goto done; | |
| 1450 } | |
| 1451 | |
| 1452 /* parse the format_spec */ | |
| 1453 if (!parse_internal_render_format_spec(format_spec, start, end, | |
| 1454 &format, '\0', '>')) | |
| 1455 goto done; | |
| 1456 | |
| 1457 /* type conversion? */ | |
| 1458 switch (format.type) { | |
| 1459 case '\0': /* No format code: like 'g', but with at least one decimal. */ | |
| 1460 case 'e': | |
| 1461 case 'E': | |
| 1462 case 'f': | |
| 1463 case 'F': | |
| 1464 case 'g': | |
| 1465 case 'G': | |
| 1466 case 'n': | |
| 1467 /* no conversion, already a complex. do the formatting */ | |
| 1468 result = format_complex_internal(obj, &format); | |
| 1469 break; | |
| 1470 | |
| 1471 default: | |
| 1472 /* unknown */ | |
| 1473 unknown_presentation_type(format.type, obj->ob_type->tp_name); | |
| 1474 goto done; | |
| 1475 } | |
| 1476 | |
| 1477 done: | |
| 1478 return result; | |
| 1479 } |