Mercurial > cpython-lru_cache-weakref
annotate Python/formatter_unicode.c @ 76564:c5ec907bcad2
Update importlib.h
| author | Brett Cannon <brett@python.org> |
|---|---|
| date | Wed, 25 Apr 2012 20:18:55 -0400 |
| parents | 139c3ae84772 |
| children | 22b56b0b8619 |
| 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 { | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
349 PyObject *decimal_point; |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
350 PyObject *thousands_sep; |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
351 const char *grouping; |
| 72475 | 352 } LocaleInfo; |
| 353 | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
354 #define STATIC_LOCALE_INFO_INIT {0, 0, 0} |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
355 |
| 72475 | 356 /* describes the layout for an integer, see the comment in |
| 357 calc_number_widths() for details */ | |
| 358 typedef struct { | |
| 359 Py_ssize_t n_lpadding; | |
| 360 Py_ssize_t n_prefix; | |
| 361 Py_ssize_t n_spadding; | |
| 362 Py_ssize_t n_rpadding; | |
| 363 char sign; | |
| 364 Py_ssize_t n_sign; /* number of digits needed for sign (0/1) */ | |
| 365 Py_ssize_t n_grouped_digits; /* Space taken up by the digits, including | |
| 366 any grouping chars. */ | |
| 367 Py_ssize_t n_decimal; /* 0 if only an integer */ | |
| 368 Py_ssize_t n_remainder; /* Digits in decimal and/or exponent part, | |
| 369 excluding the decimal itself, if | |
| 370 present. */ | |
| 371 | |
| 372 /* These 2 are not the widths of fields, but are needed by | |
| 373 STRINGLIB_GROUPING. */ | |
| 374 Py_ssize_t n_digits; /* The number of digits before a decimal | |
| 375 or exponent. */ | |
| 376 Py_ssize_t n_min_width; /* The min_width we used when we computed | |
| 377 the n_grouped_digits width. */ | |
| 378 } NumberFieldWidths; | |
|
42817
5355337709e1
Implementation of PEP 3101, Advanced String Formatting.
Eric Smith <eric@trueblade.com>
parents:
diff
changeset
|
379 |
|
47393
0da27fdf0192
Refactor and clean up str.format() code (and helpers) in advance of optimizations.
Eric Smith <eric@trueblade.com>
parents:
42817
diff
changeset
|
380 |
| 72475 | 381 /* Given a number of the form: |
| 382 digits[remainder] | |
| 383 where ptr points to the start and end points to the end, find where | |
| 384 the integer part ends. This could be a decimal, an exponent, both, | |
| 385 or neither. | |
| 386 If a decimal point is present, set *has_decimal and increment | |
| 387 remainder beyond it. | |
| 388 Results are undefined (but shouldn't crash) for improperly | |
| 389 formatted strings. | |
| 390 */ | |
| 391 static void | |
| 392 parse_number(PyObject *s, Py_ssize_t pos, Py_ssize_t end, | |
| 393 Py_ssize_t *n_remainder, int *has_decimal) | |
| 394 { | |
| 395 Py_ssize_t remainder; | |
| 396 | |
| 397 while (pos<end && isdigit(PyUnicode_READ_CHAR(s, pos))) | |
| 398 ++pos; | |
| 399 remainder = pos; | |
| 400 | |
| 401 /* Does remainder start with a decimal point? */ | |
| 402 *has_decimal = pos<end && PyUnicode_READ_CHAR(s, remainder) == '.'; | |
| 403 | |
| 404 /* Skip the decimal point. */ | |
| 405 if (*has_decimal) | |
| 406 remainder++; | |
| 407 | |
| 408 *n_remainder = end - remainder; | |
| 409 } | |
| 410 | |
| 411 /* not all fields of format are used. for example, precision is | |
| 412 unused. should this take discrete params in order to be more clear | |
| 413 about what it does? or is passing a single format parameter easier | |
| 414 and more efficient enough to justify a little obfuscation? */ | |
| 415 static Py_ssize_t | |
| 416 calc_number_widths(NumberFieldWidths *spec, Py_ssize_t n_prefix, | |
| 417 Py_UCS4 sign_char, PyObject *number, Py_ssize_t n_start, | |
| 418 Py_ssize_t n_end, Py_ssize_t n_remainder, | |
| 419 int has_decimal, const LocaleInfo *locale, | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
420 const InternalFormatSpec *format, Py_UCS4 *maxchar) |
| 72475 | 421 { |
| 422 Py_ssize_t n_non_digit_non_padding; | |
| 423 Py_ssize_t n_padding; | |
| 424 | |
| 425 spec->n_digits = n_end - n_start - n_remainder - (has_decimal?1:0); | |
| 426 spec->n_lpadding = 0; | |
| 427 spec->n_prefix = n_prefix; | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
428 spec->n_decimal = has_decimal ? PyUnicode_GET_LENGTH(locale->decimal_point) : 0; |
| 72475 | 429 spec->n_remainder = n_remainder; |
| 430 spec->n_spadding = 0; | |
| 431 spec->n_rpadding = 0; | |
| 432 spec->sign = '\0'; | |
| 433 spec->n_sign = 0; | |
| 434 | |
| 435 /* the output will look like: | |
| 436 | | | |
| 437 | <lpadding> <sign> <prefix> <spadding> <grouped_digits> <decimal> <remainder> <rpadding> | | |
| 438 | | | |
| 439 | |
| 440 sign is computed from format->sign and the actual | |
| 441 sign of the number | |
| 442 | |
| 443 prefix is given (it's for the '0x' prefix) | |
| 444 | |
| 445 digits is already known | |
| 446 | |
| 447 the total width is either given, or computed from the | |
| 448 actual digits | |
| 449 | |
| 450 only one of lpadding, spadding, and rpadding can be non-zero, | |
| 451 and it's calculated from the width and other fields | |
| 452 */ | |
| 453 | |
| 454 /* compute the various parts we're going to write */ | |
| 455 switch (format->sign) { | |
| 456 case '+': | |
| 457 /* always put a + or - */ | |
| 458 spec->n_sign = 1; | |
| 459 spec->sign = (sign_char == '-' ? '-' : '+'); | |
| 460 break; | |
| 461 case ' ': | |
| 462 spec->n_sign = 1; | |
| 463 spec->sign = (sign_char == '-' ? '-' : ' '); | |
| 464 break; | |
| 465 default: | |
| 466 /* Not specified, or the default (-) */ | |
| 467 if (sign_char == '-') { | |
| 468 spec->n_sign = 1; | |
| 469 spec->sign = '-'; | |
| 470 } | |
| 471 } | |
| 472 | |
| 473 /* The number of chars used for non-digits and non-padding. */ | |
| 474 n_non_digit_non_padding = spec->n_sign + spec->n_prefix + spec->n_decimal + | |
| 475 spec->n_remainder; | |
| 476 | |
| 477 /* min_width can go negative, that's okay. format->width == -1 means | |
| 478 we don't care. */ | |
| 479 if (format->fill_char == '0' && format->align == '=') | |
| 480 spec->n_min_width = format->width - n_non_digit_non_padding; | |
| 481 else | |
| 482 spec->n_min_width = 0; | |
| 483 | |
| 484 if (spec->n_digits == 0) | |
| 485 /* This case only occurs when using 'c' formatting, we need | |
| 486 to special case it because the grouping code always wants | |
| 487 to have at least one character. */ | |
| 488 spec->n_grouped_digits = 0; | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
489 else { |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
490 Py_UCS4 grouping_maxchar; |
| 72475 | 491 spec->n_grouped_digits = _PyUnicode_InsertThousandsGrouping( |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
492 NULL, 0, |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
493 0, NULL, |
| 72475 | 494 spec->n_digits, spec->n_min_width, |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
495 locale->grouping, locale->thousands_sep, &grouping_maxchar); |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
496 *maxchar = Py_MAX(*maxchar, grouping_maxchar); |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
497 } |
| 72475 | 498 |
| 499 /* Given the desired width and the total of digit and non-digit | |
| 500 space we consume, see if we need any padding. format->width can | |
| 501 be negative (meaning no padding), but this code still works in | |
| 502 that case. */ | |
| 503 n_padding = format->width - | |
| 504 (n_non_digit_non_padding + spec->n_grouped_digits); | |
| 505 if (n_padding > 0) { | |
| 506 /* Some padding is needed. Determine if it's left, space, or right. */ | |
| 507 switch (format->align) { | |
| 508 case '<': | |
| 509 spec->n_rpadding = n_padding; | |
| 510 break; | |
| 511 case '^': | |
| 512 spec->n_lpadding = n_padding / 2; | |
| 513 spec->n_rpadding = n_padding - spec->n_lpadding; | |
| 514 break; | |
| 515 case '=': | |
| 516 spec->n_spadding = n_padding; | |
| 517 break; | |
| 518 case '>': | |
| 519 spec->n_lpadding = n_padding; | |
| 520 break; | |
| 521 default: | |
| 522 /* Shouldn't get here, but treat it as '>' */ | |
| 523 spec->n_lpadding = n_padding; | |
| 524 assert(0); | |
| 525 break; | |
| 526 } | |
| 527 } | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
528 |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
529 if (spec->n_lpadding || spec->n_spadding || spec->n_rpadding) |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
530 *maxchar = Py_MAX(*maxchar, format->fill_char); |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
531 |
|
75234
a29d20fa85b4
Issue #13706: Fix format(float, "n") for locale with non-ASCII decimal point (e.g. ps_aF)
Victor Stinner <victor.stinner@haypocalc.com>
parents:
75231
diff
changeset
|
532 if (spec->n_decimal) |
|
a29d20fa85b4
Issue #13706: Fix format(float, "n") for locale with non-ASCII decimal point (e.g. ps_aF)
Victor Stinner <victor.stinner@haypocalc.com>
parents:
75231
diff
changeset
|
533 *maxchar = Py_MAX(*maxchar, PyUnicode_MAX_CHAR_VALUE(locale->decimal_point)); |
|
a29d20fa85b4
Issue #13706: Fix format(float, "n") for locale with non-ASCII decimal point (e.g. ps_aF)
Victor Stinner <victor.stinner@haypocalc.com>
parents:
75231
diff
changeset
|
534 |
| 72475 | 535 return spec->n_lpadding + spec->n_sign + spec->n_prefix + |
| 536 spec->n_spadding + spec->n_grouped_digits + spec->n_decimal + | |
| 537 spec->n_remainder + spec->n_rpadding; | |
| 538 } | |
| 539 | |
| 540 /* Fill in the digit parts of a numbers's string representation, | |
| 541 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
|
542 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
|
543 static int |
| 72475 | 544 fill_number(PyObject *out, Py_ssize_t pos, const NumberFieldWidths *spec, |
| 545 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
|
546 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
|
547 Py_UCS4 fill_char, |
| 72475 | 548 LocaleInfo *locale, int toupper) |
| 549 { | |
| 550 /* Used to keep track of digits, decimal, and remainder. */ | |
| 551 Py_ssize_t d_pos = d_start; | |
| 552 unsigned int kind = PyUnicode_KIND(out); | |
| 553 void *data = PyUnicode_DATA(out); | |
| 554 Py_ssize_t r; | |
| 555 | |
| 556 if (spec->n_lpadding) { | |
|
74261
405f65676010
Add a new PyUnicode_Fill() function
Victor Stinner <victor.stinner@haypocalc.com>
parents:
73805
diff
changeset
|
557 PyUnicode_Fill(out, pos, pos + spec->n_lpadding, fill_char); |
| 72475 | 558 pos += spec->n_lpadding; |
| 559 } | |
| 560 if (spec->n_sign == 1) { | |
| 561 PyUnicode_WRITE(kind, data, pos++, spec->sign); | |
| 562 } | |
| 563 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
|
564 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
|
565 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
|
566 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
|
567 return -1; |
| 72475 | 568 if (toupper) { |
| 569 Py_ssize_t t; | |
|
74341
c2153ce1b5dd
remove some usage of Py_UNICODE_TOUPPER/LOWER
Benjamin Peterson <benjamin@python.org>
parents:
74261
diff
changeset
|
570 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
|
571 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
|
572 c = Py_TOUPPER(c); |
|
74557
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
573 assert (c <= 127); |
|
74702
056f5cc8885d
Issue #13706: Add assertions to detect bugs earlier
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74586
diff
changeset
|
574 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
|
575 } |
| 72475 | 576 } |
| 577 pos += spec->n_prefix; | |
| 578 } | |
| 579 if (spec->n_spadding) { | |
|
74261
405f65676010
Add a new PyUnicode_Fill() function
Victor Stinner <victor.stinner@haypocalc.com>
parents:
73805
diff
changeset
|
580 PyUnicode_Fill(out, pos, pos + spec->n_spadding, fill_char); |
| 72475 | 581 pos += spec->n_spadding; |
| 582 } | |
| 583 | |
| 584 /* Only for type 'c' special case, it has no digits. */ | |
| 585 if (spec->n_digits != 0) { | |
| 586 /* 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
|
587 char *pdigits; |
|
ad337be523bf
fill_number() ensures that the 'digits' string is ready
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72495
diff
changeset
|
588 if (PyUnicode_READY(digits)) |
|
ad337be523bf
fill_number() ensures that the 'digits' string is ready
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72495
diff
changeset
|
589 return -1; |
|
ad337be523bf
fill_number() ensures that the 'digits' string is ready
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72495
diff
changeset
|
590 pdigits = PyUnicode_DATA(digits); |
| 72475 | 591 if (PyUnicode_KIND(digits) < kind) { |
| 592 pdigits = _PyUnicode_AsKind(digits, kind); | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
593 if (pdigits == NULL) |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
594 return -1; |
| 72475 | 595 } |
|
75234
a29d20fa85b4
Issue #13706: Fix format(float, "n") for locale with non-ASCII decimal point (e.g. ps_aF)
Victor Stinner <victor.stinner@haypocalc.com>
parents:
75231
diff
changeset
|
596 r = _PyUnicode_InsertThousandsGrouping( |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
597 out, pos, |
| 72475 | 598 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
|
599 pdigits + kind * d_pos, |
| 72475 | 600 spec->n_digits, spec->n_min_width, |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
601 locale->grouping, locale->thousands_sep, NULL); |
|
75234
a29d20fa85b4
Issue #13706: Fix format(float, "n") for locale with non-ASCII decimal point (e.g. ps_aF)
Victor Stinner <victor.stinner@haypocalc.com>
parents:
75231
diff
changeset
|
602 if (r == -1) |
|
a29d20fa85b4
Issue #13706: Fix format(float, "n") for locale with non-ASCII decimal point (e.g. ps_aF)
Victor Stinner <victor.stinner@haypocalc.com>
parents:
75231
diff
changeset
|
603 return -1; |
| 72475 | 604 assert(r == spec->n_grouped_digits); |
| 605 if (PyUnicode_KIND(digits) < kind) | |
| 606 PyMem_Free(pdigits); | |
| 607 d_pos += spec->n_digits; | |
| 608 } | |
| 609 if (toupper) { | |
| 610 Py_ssize_t t; | |
|
74341
c2153ce1b5dd
remove some usage of Py_UNICODE_TOUPPER/LOWER
Benjamin Peterson <benjamin@python.org>
parents:
74261
diff
changeset
|
611 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
|
612 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
|
613 c = Py_TOUPPER(c); |
|
74341
c2153ce1b5dd
remove some usage of Py_UNICODE_TOUPPER/LOWER
Benjamin Peterson <benjamin@python.org>
parents:
74261
diff
changeset
|
614 if (c > 127) { |
|
c2153ce1b5dd
remove some usage of Py_UNICODE_TOUPPER/LOWER
Benjamin Peterson <benjamin@python.org>
parents:
74261
diff
changeset
|
615 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
|
616 return -1; |
|
c2153ce1b5dd
remove some usage of Py_UNICODE_TOUPPER/LOWER
Benjamin Peterson <benjamin@python.org>
parents:
74261
diff
changeset
|
617 } |
|
74702
056f5cc8885d
Issue #13706: Add assertions to detect bugs earlier
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74586
diff
changeset
|
618 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
|
619 } |
| 72475 | 620 } |
| 621 pos += spec->n_grouped_digits; | |
| 622 | |
| 623 if (spec->n_decimal) { | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
624 if (PyUnicode_CopyCharacters(out, pos, locale->decimal_point, 0, spec->n_decimal) < 0) |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
625 return -1; |
| 72475 | 626 pos += spec->n_decimal; |
| 627 d_pos += 1; | |
| 628 } | |
| 629 | |
| 630 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
|
631 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
|
632 return -1; |
| 72475 | 633 pos += spec->n_remainder; |
| 634 d_pos += spec->n_remainder; | |
| 635 } | |
| 636 | |
| 637 if (spec->n_rpadding) { | |
|
74261
405f65676010
Add a new PyUnicode_Fill() function
Victor Stinner <victor.stinner@haypocalc.com>
parents:
73805
diff
changeset
|
638 PyUnicode_Fill(out, pos, pos + spec->n_rpadding, fill_char); |
| 72475 | 639 pos += spec->n_rpadding; |
| 640 } | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
641 return 0; |
| 72475 | 642 } |
| 643 | |
| 644 static char no_grouping[1] = {CHAR_MAX}; | |
| 645 | |
| 646 /* Find the decimal point character(s?), thousands_separator(s?), and | |
| 647 grouping description, either for the current locale if type is | |
| 648 LT_CURRENT_LOCALE, a hard-coded locale if LT_DEFAULT_LOCALE, or | |
| 649 none if LT_NO_LOCALE. */ | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
650 static int |
| 72475 | 651 get_locale_info(int type, LocaleInfo *locale_info) |
| 652 { | |
| 653 switch (type) { | |
| 654 case LT_CURRENT_LOCALE: { | |
| 655 struct lconv *locale_data = localeconv(); | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
656 locale_info->decimal_point = PyUnicode_DecodeLocale( |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
657 locale_data->decimal_point, |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
658 NULL); |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
659 if (locale_info->decimal_point == NULL) |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
660 return -1; |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
661 locale_info->thousands_sep = PyUnicode_DecodeLocale( |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
662 locale_data->thousands_sep, |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
663 NULL); |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
664 if (locale_info->thousands_sep == NULL) { |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
665 Py_DECREF(locale_info->decimal_point); |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
666 return -1; |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
667 } |
| 72475 | 668 locale_info->grouping = locale_data->grouping; |
| 669 break; | |
| 670 } | |
| 671 case LT_DEFAULT_LOCALE: | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
672 locale_info->decimal_point = PyUnicode_FromOrdinal('.'); |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
673 locale_info->thousands_sep = PyUnicode_FromOrdinal(','); |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
674 if (!locale_info->decimal_point || !locale_info->thousands_sep) { |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
675 Py_XDECREF(locale_info->decimal_point); |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
676 Py_XDECREF(locale_info->thousands_sep); |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
677 return -1; |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
678 } |
| 72475 | 679 locale_info->grouping = "\3"; /* Group every 3 characters. The |
| 680 (implicit) trailing 0 means repeat | |
| 681 infinitely. */ | |
| 682 break; | |
| 683 case LT_NO_LOCALE: | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
684 locale_info->decimal_point = PyUnicode_FromOrdinal('.'); |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
685 locale_info->thousands_sep = PyUnicode_New(0, 0); |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
686 if (!locale_info->decimal_point || !locale_info->thousands_sep) { |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
687 Py_XDECREF(locale_info->decimal_point); |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
688 Py_XDECREF(locale_info->thousands_sep); |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
689 return -1; |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
690 } |
| 72475 | 691 locale_info->grouping = no_grouping; |
| 692 break; | |
| 693 default: | |
| 694 assert(0); | |
| 695 } | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
696 return 0; |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
697 } |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
698 |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
699 static void |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
700 free_locale_info(LocaleInfo *locale_info) |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
701 { |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
702 Py_XDECREF(locale_info->decimal_point); |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
703 Py_XDECREF(locale_info->thousands_sep); |
| 72475 | 704 } |
| 705 | |
| 706 /************************************************************************/ | |
| 707 /*********** string formatting ******************************************/ | |
| 708 /************************************************************************/ | |
| 709 | |
| 710 static PyObject * | |
| 711 format_string_internal(PyObject *value, const InternalFormatSpec *format) | |
| 712 { | |
| 713 Py_ssize_t lpad; | |
| 714 Py_ssize_t rpad; | |
| 715 Py_ssize_t total; | |
| 716 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
|
717 Py_ssize_t len = PyUnicode_GET_LENGTH(value); |
| 72475 | 718 PyObject *result = NULL; |
|
76509
139c3ae84772
Close #14648: Compute correctly maxchar in str.format() for substrin
Victor Stinner <victor.stinner@gmail.com>
parents:
76484
diff
changeset
|
719 Py_UCS4 maxchar; |
| 72475 | 720 |
| 721 /* sign is not allowed on strings */ | |
| 722 if (format->sign != '\0') { | |
| 723 PyErr_SetString(PyExc_ValueError, | |
| 724 "Sign not allowed in string format specifier"); | |
| 725 goto done; | |
| 726 } | |
| 727 | |
| 728 /* alternate is not allowed on strings */ | |
| 729 if (format->alternate) { | |
| 730 PyErr_SetString(PyExc_ValueError, | |
| 731 "Alternate form (#) not allowed in string format " | |
| 732 "specifier"); | |
| 733 goto done; | |
| 734 } | |
| 735 | |
| 736 /* '=' alignment not allowed on strings */ | |
| 737 if (format->align == '=') { | |
| 738 PyErr_SetString(PyExc_ValueError, | |
| 739 "'=' alignment not allowed " | |
| 740 "in string format specifier"); | |
| 741 goto done; | |
| 742 } | |
| 743 | |
| 744 /* if precision is specified, output no more that format.precision | |
| 745 characters */ | |
| 746 if (format->precision >= 0 && len >= format->precision) { | |
| 747 len = format->precision; | |
| 748 } | |
| 749 | |
| 750 calc_padding(len, format->width, format->align, &lpad, &rpad, &total); | |
| 751 | |
|
76509
139c3ae84772
Close #14648: Compute correctly maxchar in str.format() for substrin
Victor Stinner <victor.stinner@gmail.com>
parents:
76484
diff
changeset
|
752 maxchar = _PyUnicode_FindMaxChar(value, 0, len); |
|
74557
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
753 if (lpad != 0 || rpad != 0) |
|
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
754 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
|
755 |
| 72475 | 756 /* allocate the resulting string */ |
| 757 result = PyUnicode_New(total, maxchar); | |
| 758 if (result == NULL) | |
| 759 goto done; | |
| 760 | |
| 761 /* Write into that space. First the padding. */ | |
| 762 pos = fill_padding(result, 0, len, | |
| 763 format->fill_char=='\0'?' ':format->fill_char, | |
| 764 lpad, rpad); | |
| 765 | |
| 766 /* 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
|
767 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
|
768 Py_CLEAR(result); |
| 72475 | 769 |
| 770 done: | |
|
74702
056f5cc8885d
Issue #13706: Add assertions to detect bugs earlier
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74586
diff
changeset
|
771 assert(!result || _PyUnicode_CheckConsistency(result, 1)); |
| 72475 | 772 return result; |
| 773 } | |
| 774 | |
| 775 | |
| 776 /************************************************************************/ | |
| 777 /*********** long formatting ********************************************/ | |
| 778 /************************************************************************/ | |
| 779 | |
| 780 typedef PyObject* | |
| 781 (*IntOrLongToString)(PyObject *value, int base); | |
| 782 | |
| 783 static PyObject * | |
| 784 format_int_or_long_internal(PyObject *value, const InternalFormatSpec *format, | |
| 785 IntOrLongToString tostring) | |
| 786 { | |
| 787 PyObject *result = NULL; | |
|
74585
04b565d912c2
Fix compilation warnings (seen on win32 buildbot)
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents:
74557
diff
changeset
|
788 Py_UCS4 maxchar = 127; |
| 72475 | 789 PyObject *tmp = NULL; |
| 790 Py_ssize_t inumeric_chars; | |
| 791 Py_UCS4 sign_char = '\0'; | |
| 792 Py_ssize_t n_digits; /* count of digits need from the computed | |
| 793 string */ | |
| 794 Py_ssize_t n_remainder = 0; /* Used only for 'c' formatting, which | |
| 795 produces non-digits */ | |
| 796 Py_ssize_t n_prefix = 0; /* Count of prefix chars, (e.g., '0x') */ | |
| 797 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
|
798 Py_ssize_t prefix = 0; |
| 72475 | 799 NumberFieldWidths spec; |
| 800 long x; | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
801 int err; |
| 72475 | 802 |
| 803 /* Locale settings, either from the actual locale or | |
| 804 from a hard-code pseudo-locale */ | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
805 LocaleInfo locale = STATIC_LOCALE_INFO_INIT; |
| 72475 | 806 |
| 807 /* no precision allowed on integers */ | |
| 808 if (format->precision != -1) { | |
| 809 PyErr_SetString(PyExc_ValueError, | |
| 810 "Precision not allowed in integer format specifier"); | |
| 811 goto done; | |
| 812 } | |
| 813 | |
| 814 /* special case for character formatting */ | |
| 815 if (format->type == 'c') { | |
| 816 /* error to specify a sign */ | |
| 817 if (format->sign != '\0') { | |
| 818 PyErr_SetString(PyExc_ValueError, | |
| 819 "Sign not allowed with integer" | |
| 820 " format specifier 'c'"); | |
| 821 goto done; | |
| 822 } | |
| 823 | |
| 824 /* taken from unicodeobject.c formatchar() */ | |
| 825 /* Integer input truncated to a character */ | |
| 826 x = PyLong_AsLong(value); | |
| 827 if (x == -1 && PyErr_Occurred()) | |
| 828 goto done; | |
| 829 if (x < 0 || x > 0x10ffff) { | |
| 830 PyErr_SetString(PyExc_OverflowError, | |
|
74557
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
831 "%c arg not in range(0x110000)"); |
| 72475 | 832 goto done; |
| 833 } | |
| 834 tmp = PyUnicode_FromOrdinal(x); | |
| 835 inumeric_chars = 0; | |
| 836 n_digits = 1; | |
|
74586
728cfc671d15
Silence last compilation warning.
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents:
74585
diff
changeset
|
837 maxchar = Py_MAX(maxchar, (Py_UCS4)x); |
| 72475 | 838 |
| 839 /* As a sort-of hack, we tell calc_number_widths that we only | |
| 840 have "remainder" characters. calc_number_widths thinks | |
| 841 these are characters that don't get formatted, only copied | |
| 842 into the output string. We do this for 'c' formatting, | |
| 843 because the characters are likely to be non-digits. */ | |
| 844 n_remainder = 1; | |
| 845 } | |
| 846 else { | |
| 847 int base; | |
| 848 int leading_chars_to_skip = 0; /* Number of characters added by | |
| 849 PyNumber_ToBase that we want to | |
| 850 skip over. */ | |
| 851 | |
| 852 /* Compute the base and how many characters will be added by | |
| 853 PyNumber_ToBase */ | |
| 854 switch (format->type) { | |
| 855 case 'b': | |
| 856 base = 2; | |
| 857 leading_chars_to_skip = 2; /* 0b */ | |
| 858 break; | |
| 859 case 'o': | |
| 860 base = 8; | |
| 861 leading_chars_to_skip = 2; /* 0o */ | |
| 862 break; | |
| 863 case 'x': | |
| 864 case 'X': | |
| 865 base = 16; | |
| 866 leading_chars_to_skip = 2; /* 0x */ | |
| 867 break; | |
| 868 default: /* shouldn't be needed, but stops a compiler warning */ | |
| 869 case 'd': | |
| 870 case 'n': | |
| 871 base = 10; | |
| 872 break; | |
| 873 } | |
| 874 | |
| 875 /* The number of prefix chars is the same as the leading | |
| 876 chars to skip */ | |
| 877 if (format->alternate) | |
| 878 n_prefix = leading_chars_to_skip; | |
| 879 | |
| 880 /* Do the hard part, converting to a string in a given base */ | |
| 881 tmp = tostring(value, base); | |
| 882 if (tmp == NULL || PyUnicode_READY(tmp) == -1) | |
| 883 goto done; | |
| 884 | |
| 885 inumeric_chars = 0; | |
| 886 n_digits = PyUnicode_GET_LENGTH(tmp); | |
| 887 | |
| 888 prefix = inumeric_chars; | |
| 889 | |
| 890 /* Is a sign character present in the output? If so, remember it | |
| 891 and skip it */ | |
| 892 if (PyUnicode_READ_CHAR(tmp, inumeric_chars) == '-') { | |
| 893 sign_char = '-'; | |
| 894 ++prefix; | |
| 895 ++leading_chars_to_skip; | |
| 896 } | |
| 897 | |
| 898 /* Skip over the leading chars (0x, 0b, etc.) */ | |
| 899 n_digits -= leading_chars_to_skip; | |
| 900 inumeric_chars += leading_chars_to_skip; | |
| 901 } | |
| 902 | |
| 903 /* Determine the grouping, separator, and decimal point, if any. */ | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
904 if (get_locale_info(format->type == 'n' ? LT_CURRENT_LOCALE : |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
905 (format->thousands_separators ? |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
906 LT_DEFAULT_LOCALE : |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
907 LT_NO_LOCALE), |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
908 &locale) == -1) |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
909 goto done; |
| 72475 | 910 |
| 911 /* Calculate how much memory we'll need. */ | |
| 912 n_total = calc_number_widths(&spec, n_prefix, sign_char, tmp, inumeric_chars, | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
913 inumeric_chars + n_digits, n_remainder, 0, |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
914 &locale, format, &maxchar); |
|
74557
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
915 |
| 72475 | 916 /* Allocate the memory. */ |
| 917 result = PyUnicode_New(n_total, maxchar); | |
| 918 if (!result) | |
| 919 goto done; | |
| 920 | |
| 921 /* Populate the memory. */ | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
922 err = fill_number(result, 0, &spec, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
923 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
|
924 tmp, prefix, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
925 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
|
926 &locale, format->type == 'X'); |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
927 if (err) |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
928 Py_CLEAR(result); |
| 72475 | 929 |
| 930 done: | |
| 931 Py_XDECREF(tmp); | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
932 free_locale_info(&locale); |
|
74702
056f5cc8885d
Issue #13706: Add assertions to detect bugs earlier
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74586
diff
changeset
|
933 assert(!result || _PyUnicode_CheckConsistency(result, 1)); |
| 72475 | 934 return result; |
| 935 } | |
| 936 | |
| 937 /************************************************************************/ | |
| 938 /*********** float formatting *******************************************/ | |
| 939 /************************************************************************/ | |
| 940 | |
| 941 static PyObject* | |
| 942 strtounicode(char *charbuffer, Py_ssize_t len) | |
| 943 { | |
| 944 return PyUnicode_FromKindAndData(PyUnicode_1BYTE_KIND, charbuffer, len); | |
| 945 } | |
|
47393
0da27fdf0192
Refactor and clean up str.format() code (and helpers) in advance of optimizations.
Eric Smith <eric@trueblade.com>
parents:
42817
diff
changeset
|
946 |
| 72475 | 947 /* much of this is taken from unicodeobject.c */ |
| 948 static PyObject * | |
| 949 format_float_internal(PyObject *value, | |
| 950 const InternalFormatSpec *format) | |
| 951 { | |
| 952 char *buf = NULL; /* buffer returned from PyOS_double_to_string */ | |
| 953 Py_ssize_t n_digits; | |
| 954 Py_ssize_t n_remainder; | |
| 955 Py_ssize_t n_total; | |
| 956 int has_decimal; | |
| 957 double val; | |
| 958 Py_ssize_t precision = format->precision; | |
| 959 Py_ssize_t default_precision = 6; | |
| 960 Py_UCS4 type = format->type; | |
| 961 int add_pct = 0; | |
| 962 Py_ssize_t index; | |
| 963 NumberFieldWidths spec; | |
| 964 int flags = 0; | |
| 965 PyObject *result = NULL; | |
|
74585
04b565d912c2
Fix compilation warnings (seen on win32 buildbot)
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents:
74557
diff
changeset
|
966 Py_UCS4 maxchar = 127; |
| 72475 | 967 Py_UCS4 sign_char = '\0'; |
| 968 int float_type; /* Used to see if we have a nan, inf, or regular float. */ | |
| 969 PyObject *unicode_tmp = NULL; | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
970 int err; |
| 72475 | 971 |
| 972 /* Locale settings, either from the actual locale or | |
| 973 from a hard-code pseudo-locale */ | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
974 LocaleInfo locale = STATIC_LOCALE_INFO_INIT; |
| 72475 | 975 |
| 976 if (format->alternate) | |
| 977 flags |= Py_DTSF_ALT; | |
| 978 | |
| 979 if (type == '\0') { | |
| 980 /* Omitted type specifier. Behaves in the same way as repr(x) | |
| 981 and str(x) if no precision is given, else like 'g', but with | |
| 982 at least one digit after the decimal point. */ | |
| 983 flags |= Py_DTSF_ADD_DOT_0; | |
| 984 type = 'r'; | |
| 985 default_precision = 0; | |
| 986 } | |
| 987 | |
| 988 if (type == 'n') | |
| 989 /* 'n' is the same as 'g', except for the locale used to | |
| 990 format the result. We take care of that later. */ | |
| 991 type = 'g'; | |
| 992 | |
| 993 val = PyFloat_AsDouble(value); | |
| 994 if (val == -1.0 && PyErr_Occurred()) | |
| 995 goto done; | |
| 996 | |
| 997 if (type == '%') { | |
| 998 type = 'f'; | |
| 999 val *= 100; | |
| 1000 add_pct = 1; | |
| 1001 } | |
| 1002 | |
| 1003 if (precision < 0) | |
| 1004 precision = default_precision; | |
| 1005 else if (type == 'r') | |
| 1006 type = 'g'; | |
| 1007 | |
| 1008 /* Cast "type", because if we're in unicode we need to pass a | |
| 1009 8-bit char. This is safe, because we've restricted what "type" | |
| 1010 can be. */ | |
| 1011 buf = PyOS_double_to_string(val, (char)type, precision, flags, | |
| 1012 &float_type); | |
| 1013 if (buf == NULL) | |
| 1014 goto done; | |
| 1015 n_digits = strlen(buf); | |
| 1016 | |
| 1017 if (add_pct) { | |
| 1018 /* We know that buf has a trailing zero (since we just called | |
| 1019 strlen() on it), and we don't use that fact any more. So we | |
| 1020 can just write over the trailing zero. */ | |
| 1021 buf[n_digits] = '%'; | |
| 1022 n_digits += 1; | |
| 1023 } | |
| 1024 | |
| 1025 /* Since there is no unicode version of PyOS_double_to_string, | |
| 1026 just use the 8 bit version and then convert to unicode. */ | |
| 1027 unicode_tmp = strtounicode(buf, n_digits); | |
| 1028 if (unicode_tmp == NULL) | |
| 1029 goto done; | |
| 1030 index = 0; | |
| 1031 | |
| 1032 /* Is a sign character present in the output? If so, remember it | |
| 1033 and skip it */ | |
| 1034 if (PyUnicode_READ_CHAR(unicode_tmp, index) == '-') { | |
| 1035 sign_char = '-'; | |
| 1036 ++index; | |
| 1037 --n_digits; | |
| 1038 } | |
| 1039 | |
| 1040 /* Determine if we have any "remainder" (after the digits, might include | |
| 1041 decimal or exponent or both (or neither)) */ | |
| 1042 parse_number(unicode_tmp, index, index + n_digits, &n_remainder, &has_decimal); | |
| 1043 | |
| 1044 /* Determine the grouping, separator, and decimal point, if any. */ | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1045 if (get_locale_info(format->type == 'n' ? LT_CURRENT_LOCALE : |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1046 (format->thousands_separators ? |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1047 LT_DEFAULT_LOCALE : |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1048 LT_NO_LOCALE), |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1049 &locale) == -1) |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1050 goto done; |
| 72475 | 1051 |
| 1052 /* 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
|
1053 n_total = calc_number_widths(&spec, 0, sign_char, unicode_tmp, index, |
| 72475 | 1054 index + n_digits, n_remainder, has_decimal, |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1055 &locale, format, &maxchar); |
|
74557
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
1056 |
| 72475 | 1057 /* Allocate the memory. */ |
| 1058 result = PyUnicode_New(n_total, maxchar); | |
| 1059 if (result == NULL) | |
| 1060 goto done; | |
| 1061 | |
| 1062 /* Populate the memory. */ | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1063 err = fill_number(result, 0, &spec, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1064 unicode_tmp, index, index + n_digits, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1065 NULL, 0, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1066 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
|
1067 &locale, 0); |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1068 if (err) |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1069 Py_CLEAR(result); |
| 72475 | 1070 |
| 1071 done: | |
| 1072 PyMem_Free(buf); | |
| 1073 Py_DECREF(unicode_tmp); | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1074 free_locale_info(&locale); |
|
74702
056f5cc8885d
Issue #13706: Add assertions to detect bugs earlier
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74586
diff
changeset
|
1075 assert(!result || _PyUnicode_CheckConsistency(result, 1)); |
| 72475 | 1076 return result; |
| 1077 } | |
| 1078 | |
| 1079 /************************************************************************/ | |
| 1080 /*********** complex formatting *****************************************/ | |
| 1081 /************************************************************************/ | |
| 1082 | |
| 1083 static PyObject * | |
| 1084 format_complex_internal(PyObject *value, | |
| 1085 const InternalFormatSpec *format) | |
| 1086 { | |
| 1087 double re; | |
| 1088 double im; | |
| 1089 char *re_buf = NULL; /* buffer returned from PyOS_double_to_string */ | |
| 1090 char *im_buf = NULL; /* buffer returned from PyOS_double_to_string */ | |
| 1091 | |
| 1092 InternalFormatSpec tmp_format = *format; | |
| 1093 Py_ssize_t n_re_digits; | |
| 1094 Py_ssize_t n_im_digits; | |
| 1095 Py_ssize_t n_re_remainder; | |
| 1096 Py_ssize_t n_im_remainder; | |
| 1097 Py_ssize_t n_re_total; | |
| 1098 Py_ssize_t n_im_total; | |
| 1099 int re_has_decimal; | |
| 1100 int im_has_decimal; | |
| 1101 Py_ssize_t precision = format->precision; | |
| 1102 Py_ssize_t default_precision = 6; | |
| 1103 Py_UCS4 type = format->type; | |
| 1104 Py_ssize_t i_re; | |
| 1105 Py_ssize_t i_im; | |
| 1106 NumberFieldWidths re_spec; | |
| 1107 NumberFieldWidths im_spec; | |
| 1108 int flags = 0; | |
| 1109 PyObject *result = NULL; | |
|
74585
04b565d912c2
Fix compilation warnings (seen on win32 buildbot)
Amaury Forgeot d'Arc <amauryfa@gmail.com>
parents:
74557
diff
changeset
|
1110 Py_UCS4 maxchar = 127; |
| 72475 | 1111 int rkind; |
| 1112 void *rdata; | |
| 1113 Py_ssize_t index; | |
| 1114 Py_UCS4 re_sign_char = '\0'; | |
| 1115 Py_UCS4 im_sign_char = '\0'; | |
| 1116 int re_float_type; /* Used to see if we have a nan, inf, or regular float. */ | |
| 1117 int im_float_type; | |
| 1118 int add_parens = 0; | |
| 1119 int skip_re = 0; | |
| 1120 Py_ssize_t lpad; | |
| 1121 Py_ssize_t rpad; | |
| 1122 Py_ssize_t total; | |
| 1123 PyObject *re_unicode_tmp = NULL; | |
| 1124 PyObject *im_unicode_tmp = NULL; | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1125 int err; |
| 72475 | 1126 |
| 1127 /* Locale settings, either from the actual locale or | |
| 1128 from a hard-code pseudo-locale */ | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1129 LocaleInfo locale = STATIC_LOCALE_INFO_INIT; |
| 72475 | 1130 |
| 1131 /* Zero padding is not allowed. */ | |
| 1132 if (format->fill_char == '0') { | |
| 1133 PyErr_SetString(PyExc_ValueError, | |
| 1134 "Zero padding is not allowed in complex format " | |
| 1135 "specifier"); | |
| 1136 goto done; | |
| 1137 } | |
| 1138 | |
| 1139 /* Neither is '=' alignment . */ | |
| 1140 if (format->align == '=') { | |
| 1141 PyErr_SetString(PyExc_ValueError, | |
| 1142 "'=' alignment flag is not allowed in complex format " | |
| 1143 "specifier"); | |
| 1144 goto done; | |
| 1145 } | |
| 1146 | |
| 1147 re = PyComplex_RealAsDouble(value); | |
| 1148 if (re == -1.0 && PyErr_Occurred()) | |
| 1149 goto done; | |
| 1150 im = PyComplex_ImagAsDouble(value); | |
| 1151 if (im == -1.0 && PyErr_Occurred()) | |
| 1152 goto done; | |
| 1153 | |
| 1154 if (format->alternate) | |
| 1155 flags |= Py_DTSF_ALT; | |
| 1156 | |
| 1157 if (type == '\0') { | |
| 1158 /* Omitted type specifier. Should be like str(self). */ | |
| 1159 type = 'r'; | |
| 1160 default_precision = 0; | |
| 1161 if (re == 0.0 && copysign(1.0, re) == 1.0) | |
| 1162 skip_re = 1; | |
| 1163 else | |
| 1164 add_parens = 1; | |
| 1165 } | |
| 1166 | |
| 1167 if (type == 'n') | |
| 1168 /* 'n' is the same as 'g', except for the locale used to | |
| 1169 format the result. We take care of that later. */ | |
| 1170 type = 'g'; | |
| 1171 | |
| 1172 if (precision < 0) | |
| 1173 precision = default_precision; | |
| 1174 else if (type == 'r') | |
| 1175 type = 'g'; | |
| 1176 | |
| 1177 /* Cast "type", because if we're in unicode we need to pass a | |
| 1178 8-bit char. This is safe, because we've restricted what "type" | |
| 1179 can be. */ | |
| 1180 re_buf = PyOS_double_to_string(re, (char)type, precision, flags, | |
| 1181 &re_float_type); | |
| 1182 if (re_buf == NULL) | |
| 1183 goto done; | |
| 1184 im_buf = PyOS_double_to_string(im, (char)type, precision, flags, | |
| 1185 &im_float_type); | |
| 1186 if (im_buf == NULL) | |
| 1187 goto done; | |
| 1188 | |
| 1189 n_re_digits = strlen(re_buf); | |
| 1190 n_im_digits = strlen(im_buf); | |
| 1191 | |
| 1192 /* Since there is no unicode version of PyOS_double_to_string, | |
| 1193 just use the 8 bit version and then convert to unicode. */ | |
| 1194 re_unicode_tmp = strtounicode(re_buf, n_re_digits); | |
| 1195 if (re_unicode_tmp == NULL) | |
| 1196 goto done; | |
| 1197 i_re = 0; | |
| 1198 | |
| 1199 im_unicode_tmp = strtounicode(im_buf, n_im_digits); | |
| 1200 if (im_unicode_tmp == NULL) | |
| 1201 goto done; | |
| 1202 i_im = 0; | |
| 1203 | |
| 1204 /* Is a sign character present in the output? If so, remember it | |
| 1205 and skip it */ | |
| 1206 if (PyUnicode_READ_CHAR(re_unicode_tmp, i_re) == '-') { | |
| 1207 re_sign_char = '-'; | |
| 1208 ++i_re; | |
| 1209 --n_re_digits; | |
| 1210 } | |
| 1211 if (PyUnicode_READ_CHAR(im_unicode_tmp, i_im) == '-') { | |
| 1212 im_sign_char = '-'; | |
| 1213 ++i_im; | |
| 1214 --n_im_digits; | |
| 1215 } | |
| 1216 | |
| 1217 /* Determine if we have any "remainder" (after the digits, might include | |
| 1218 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
|
1219 parse_number(re_unicode_tmp, i_re, i_re + n_re_digits, |
| 72475 | 1220 &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
|
1221 parse_number(im_unicode_tmp, i_im, i_im + n_im_digits, |
| 72475 | 1222 &n_im_remainder, &im_has_decimal); |
| 1223 | |
| 1224 /* Determine the grouping, separator, and decimal point, if any. */ | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1225 if (get_locale_info(format->type == 'n' ? LT_CURRENT_LOCALE : |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1226 (format->thousands_separators ? |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1227 LT_DEFAULT_LOCALE : |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1228 LT_NO_LOCALE), |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1229 &locale) == -1) |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1230 goto done; |
| 72475 | 1231 |
| 1232 /* Turn off any padding. We'll do it later after we've composed | |
| 1233 the numbers without padding. */ | |
| 1234 tmp_format.fill_char = '\0'; | |
| 1235 tmp_format.align = '<'; | |
| 1236 tmp_format.width = -1; | |
| 1237 | |
| 1238 /* Calculate how much memory we'll need. */ | |
| 1239 n_re_total = calc_number_widths(&re_spec, 0, re_sign_char, re_unicode_tmp, | |
| 1240 i_re, i_re + n_re_digits, n_re_remainder, | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1241 re_has_decimal, &locale, &tmp_format, |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1242 &maxchar); |
| 72475 | 1243 |
| 1244 /* Same formatting, but always include a sign, unless the real part is | |
| 1245 * going to be omitted, in which case we use whatever sign convention was | |
| 1246 * requested by the original format. */ | |
| 1247 if (!skip_re) | |
| 1248 tmp_format.sign = '+'; | |
| 1249 n_im_total = calc_number_widths(&im_spec, 0, im_sign_char, im_unicode_tmp, | |
| 1250 i_im, i_im + n_im_digits, n_im_remainder, | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1251 im_has_decimal, &locale, &tmp_format, |
|
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1252 &maxchar); |
| 72475 | 1253 |
| 1254 if (skip_re) | |
| 1255 n_re_total = 0; | |
| 1256 | |
| 1257 /* Add 1 for the 'j', and optionally 2 for parens. */ | |
| 1258 calc_padding(n_re_total + n_im_total + 1 + add_parens * 2, | |
| 1259 format->width, format->align, &lpad, &rpad, &total); | |
| 1260 | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1261 if (lpad || rpad) |
|
74557
231c6042c40c
Issue #13706: Support non-ASCII fill characters
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74533
diff
changeset
|
1262 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
|
1263 |
| 72475 | 1264 result = PyUnicode_New(total, maxchar); |
| 1265 if (result == NULL) | |
| 1266 goto done; | |
| 1267 rkind = PyUnicode_KIND(result); | |
| 1268 rdata = PyUnicode_DATA(result); | |
| 1269 | |
| 1270 /* Populate the memory. First, the padding. */ | |
| 1271 index = fill_padding(result, 0, | |
| 1272 n_re_total + n_im_total + 1 + add_parens * 2, | |
| 1273 format->fill_char=='\0' ? ' ' : format->fill_char, | |
| 1274 lpad, rpad); | |
| 1275 | |
| 1276 if (add_parens) | |
| 1277 PyUnicode_WRITE(rkind, rdata, index++, '('); | |
| 1278 | |
| 1279 if (!skip_re) { | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1280 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
|
1281 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
|
1282 NULL, 0, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1283 0, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1284 &locale, 0); |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1285 if (err) { |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1286 Py_CLEAR(result); |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1287 goto done; |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1288 } |
| 72475 | 1289 index += n_re_total; |
| 1290 } | |
|
72495
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1291 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
|
1292 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
|
1293 NULL, 0, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1294 0, |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1295 &locale, 0); |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1296 if (err) { |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1297 Py_CLEAR(result); |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1298 goto done; |
|
cf3557b65ae1
fill_char() can now propagate an error
Victor Stinner <victor.stinner@haypocalc.com>
parents:
72475
diff
changeset
|
1299 } |
| 72475 | 1300 index += n_im_total; |
| 1301 PyUnicode_WRITE(rkind, rdata, index++, 'j'); | |
| 1302 | |
| 1303 if (add_parens) | |
| 1304 PyUnicode_WRITE(rkind, rdata, index++, ')'); | |
| 1305 | |
| 1306 done: | |
| 1307 PyMem_Free(re_buf); | |
| 1308 PyMem_Free(im_buf); | |
| 1309 Py_XDECREF(re_unicode_tmp); | |
| 1310 Py_XDECREF(im_unicode_tmp); | |
|
75231
f89e2f4cda88
Issue #13706: Fix format(int, "n") for locale with non-ASCII thousands separator
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74702
diff
changeset
|
1311 free_locale_info(&locale); |
|
74702
056f5cc8885d
Issue #13706: Add assertions to detect bugs earlier
Victor Stinner <victor.stinner@haypocalc.com>
parents:
74586
diff
changeset
|
1312 assert(!result || _PyUnicode_CheckConsistency(result, 1)); |
| 72475 | 1313 return result; |
| 1314 } | |
| 1315 | |
| 1316 /************************************************************************/ | |
| 1317 /*********** built in formatters ****************************************/ | |
| 1318 /************************************************************************/ | |
| 1319 PyObject * | |
| 1320 _PyUnicode_FormatAdvanced(PyObject *obj, | |
| 1321 PyObject *format_spec, | |
| 1322 Py_ssize_t start, Py_ssize_t end) | |
| 1323 { | |
| 1324 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
|
1325 PyObject *result; |
| 72475 | 1326 |
| 1327 /* check for the special case of zero length format spec, make | |
| 1328 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
|
1329 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
|
1330 return PyObject_Str(obj); |
| 72475 | 1331 |
| 1332 /* parse the format_spec */ | |
| 1333 if (!parse_internal_render_format_spec(format_spec, start, end, | |
| 1334 &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
|
1335 return NULL; |
| 72475 | 1336 |
| 1337 /* type conversion? */ | |
| 1338 switch (format.type) { | |
| 1339 case 's': | |
| 1340 /* no type conversion needed, already a string. do the formatting */ | |
| 1341 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
|
1342 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
|
1343 assert(_PyUnicode_CheckConsistency(result, 1)); |
| 72475 | 1344 break; |
| 1345 default: | |
| 1346 /* unknown */ | |
| 1347 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
|
1348 result = NULL; |
| 72475 | 1349 } |
| 1350 return result; | |
| 1351 } | |
| 1352 | |
| 1353 static PyObject* | |
| 1354 format_int_or_long(PyObject* obj, PyObject* format_spec, | |
| 1355 Py_ssize_t start, Py_ssize_t end, | |
| 1356 IntOrLongToString tostring) | |
| 1357 { | |
| 1358 PyObject *result = NULL; | |
| 1359 PyObject *tmp = NULL; | |
| 1360 InternalFormatSpec format; | |
| 1361 | |
| 1362 /* check for the special case of zero length format spec, make | |
| 1363 it equivalent to str(obj) */ | |
| 1364 if (start == end) { | |
| 1365 result = PyObject_Str(obj); | |
| 1366 goto done; | |
| 1367 } | |
| 1368 | |
| 1369 /* parse the format_spec */ | |
| 1370 if (!parse_internal_render_format_spec(format_spec, start, end, | |
| 1371 &format, 'd', '>')) | |
| 1372 goto done; | |
| 1373 | |
| 1374 /* type conversion? */ | |
| 1375 switch (format.type) { | |
| 1376 case 'b': | |
| 1377 case 'c': | |
| 1378 case 'd': | |
| 1379 case 'o': | |
| 1380 case 'x': | |
| 1381 case 'X': | |
| 1382 case 'n': | |
| 1383 /* no type conversion needed, already an int (or long). do | |
| 1384 the formatting */ | |
| 1385 result = format_int_or_long_internal(obj, &format, tostring); | |
| 1386 break; | |
| 1387 | |
| 1388 case 'e': | |
| 1389 case 'E': | |
| 1390 case 'f': | |
| 1391 case 'F': | |
| 1392 case 'g': | |
| 1393 case 'G': | |
| 1394 case '%': | |
| 1395 /* convert to float */ | |
| 1396 tmp = PyNumber_Float(obj); | |
| 1397 if (tmp == NULL) | |
| 1398 goto done; | |
| 1399 result = format_float_internal(tmp, &format); | |
| 1400 break; | |
| 1401 | |
| 1402 default: | |
| 1403 /* unknown */ | |
| 1404 unknown_presentation_type(format.type, obj->ob_type->tp_name); | |
| 1405 goto done; | |
| 1406 } | |
| 1407 | |
| 1408 done: | |
| 1409 Py_XDECREF(tmp); | |
| 1410 return result; | |
| 1411 } | |
| 1412 | |
| 1413 /* Need to define long_format as a function that will convert a long | |
| 1414 to a string. In 3.0, _PyLong_Format has the correct signature. */ | |
| 1415 #define long_format _PyLong_Format | |
| 1416 | |
| 1417 PyObject * | |
| 1418 _PyLong_FormatAdvanced(PyObject *obj, | |
| 1419 PyObject *format_spec, | |
| 1420 Py_ssize_t start, Py_ssize_t end) | |
| 1421 { | |
| 1422 return format_int_or_long(obj, format_spec, start, end, | |
| 1423 long_format); | |
| 1424 } | |
| 1425 | |
| 1426 PyObject * | |
| 1427 _PyFloat_FormatAdvanced(PyObject *obj, | |
| 1428 PyObject *format_spec, | |
| 1429 Py_ssize_t start, Py_ssize_t end) | |
| 1430 { | |
| 1431 PyObject *result = NULL; | |
| 1432 InternalFormatSpec format; | |
| 1433 | |
| 1434 /* check for the special case of zero length format spec, make | |
| 1435 it equivalent to str(obj) */ | |
| 1436 if (start == end) { | |
| 1437 result = PyObject_Str(obj); | |
| 1438 goto done; | |
| 1439 } | |
| 1440 | |
| 1441 /* parse the format_spec */ | |
| 1442 if (!parse_internal_render_format_spec(format_spec, start, end, | |
| 1443 &format, '\0', '>')) | |
| 1444 goto done; | |
| 1445 | |
| 1446 /* type conversion? */ | |
| 1447 switch (format.type) { | |
| 1448 case '\0': /* No format code: like 'g', but with at least one decimal. */ | |
| 1449 case 'e': | |
| 1450 case 'E': | |
| 1451 case 'f': | |
| 1452 case 'F': | |
| 1453 case 'g': | |
| 1454 case 'G': | |
| 1455 case 'n': | |
| 1456 case '%': | |
| 1457 /* no conversion, already a float. do the formatting */ | |
| 1458 result = format_float_internal(obj, &format); | |
| 1459 break; | |
| 1460 | |
| 1461 default: | |
| 1462 /* unknown */ | |
| 1463 unknown_presentation_type(format.type, obj->ob_type->tp_name); | |
| 1464 goto done; | |
| 1465 } | |
| 1466 | |
| 1467 done: | |
| 1468 return result; | |
| 1469 } | |
| 1470 | |
| 1471 PyObject * | |
| 1472 _PyComplex_FormatAdvanced(PyObject *obj, | |
| 1473 PyObject *format_spec, | |
| 1474 Py_ssize_t start, Py_ssize_t end) | |
| 1475 { | |
| 1476 PyObject *result = NULL; | |
| 1477 InternalFormatSpec format; | |
| 1478 | |
| 1479 /* check for the special case of zero length format spec, make | |
| 1480 it equivalent to str(obj) */ | |
| 1481 if (start == end) { | |
| 1482 result = PyObject_Str(obj); | |
| 1483 goto done; | |
| 1484 } | |
| 1485 | |
| 1486 /* parse the format_spec */ | |
| 1487 if (!parse_internal_render_format_spec(format_spec, start, end, | |
| 1488 &format, '\0', '>')) | |
| 1489 goto done; | |
| 1490 | |
| 1491 /* type conversion? */ | |
| 1492 switch (format.type) { | |
| 1493 case '\0': /* No format code: like 'g', but with at least one decimal. */ | |
| 1494 case 'e': | |
| 1495 case 'E': | |
| 1496 case 'f': | |
| 1497 case 'F': | |
| 1498 case 'g': | |
| 1499 case 'G': | |
| 1500 case 'n': | |
| 1501 /* no conversion, already a complex. do the formatting */ | |
| 1502 result = format_complex_internal(obj, &format); | |
| 1503 break; | |
| 1504 | |
| 1505 default: | |
| 1506 /* unknown */ | |
| 1507 unknown_presentation_type(format.type, obj->ob_type->tp_name); | |
| 1508 goto done; | |
| 1509 } | |
| 1510 | |
| 1511 done: | |
| 1512 return result; | |
| 1513 } |