Mercurial > vim

comparison src/mbyte.c @ 21971:0bc43a704f56 v8.2.1535

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
patch 8.2.1535: it is not possible to specify cell widths of characters Commit: https://github.com/vim/vim/commit/08aac3c6192f0103cb87e280270a32b50e653be1 Author: Bram Moolenaar <Bram@vim.org> Date: Fri Aug 28 21:04:24 2020 +0200 patch 8.2.1535: it is not possible to specify cell widths of characters Problem: It is not possible to specify cell widths of characters. Solution: Add setcellwidths().
author Bram Moolenaar <Bram@vim.org>
date Fri, 28 Aug 2020 21:15:03 +0200
parents cea8ae407452
children 85add08e6a2d
comparison
equal deleted inserted replaced
21970:7c76b6dbc14d 21971:0bc43a704f56
130 static int utf_ptr2cells_len(char_u *p, int size); 130 static int utf_ptr2cells_len(char_u *p, int size);
131 static int dbcs_char2cells(int c); 131 static int dbcs_char2cells(int c);
132 static int dbcs_ptr2cells_len(char_u *p, int size); 132 static int dbcs_ptr2cells_len(char_u *p, int size);
133 static int dbcs_ptr2char(char_u *p); 133 static int dbcs_ptr2char(char_u *p);
134 static int dbcs_head_off(char_u *base, char_u *p); 134 static int dbcs_head_off(char_u *base, char_u *p);
135 static int cw_value(int c);
135 136
136 /* 137 /*
137 * Lookup table to quickly get the length in bytes of a UTF-8 character from 138 * Lookup table to quickly get the length in bytes of a UTF-8 character from
138 * the first byte of a UTF-8 string. 139 * the first byte of a UTF-8 string.
139 * Bytes which are illegal when used as the first byte have a 1. 140 * Bytes which are illegal when used as the first byte have a 1.
1485 }; 1486 };
1486 1487
1487 // Sorted list of non-overlapping intervals of Emoji characters that don't 1488 // Sorted list of non-overlapping intervals of Emoji characters that don't
1488 // have ambiguous or double width, 1489 // have ambiguous or double width,
1489 // based on http://unicode.org/emoji/charts/emoji-list.html 1490 // based on http://unicode.org/emoji/charts/emoji-list.html
1490 static struct interval emoji_width[] = 1491 static struct interval emoji_wide[] =
1491 { 1492 {
1492 {0x1f1e6, 0x1f1ff}, 1493 {0x1f1e6, 0x1f1ff},
1493 {0x1f321, 0x1f321}, 1494 {0x1f321, 0x1f321},
1494 {0x1f324, 0x1f32c}, 1495 {0x1f324, 0x1f32c},
1495 {0x1f336, 0x1f336}, 1496 {0x1f336, 0x1f336},
1530 {0x1f6f3, 0x1f6f3} 1531 {0x1f6f3, 0x1f6f3}
1531 }; 1532 };
1532 1533
1533 if (c >= 0x100) 1534 if (c >= 0x100)
1534 { 1535 {
1536 int n;
1537
1538 n = cw_value(c);
1539 if (n != 0)
1540 return n;
1541
1535 #ifdef USE_WCHAR_FUNCTIONS 1542 #ifdef USE_WCHAR_FUNCTIONS
1536 /* 1543 /*
1537 * Assume the library function wcwidth() works better than our own 1544 * Assume the library function wcwidth() works better than our own
1538 * stuff. It should return 1 for ambiguous width chars! 1545 * stuff. It should return 1 for ambiguous width chars!
1539 */ 1546 */
1540 int n = wcwidth(c); 1547 n = wcwidth(c);
1541 1548
1542 if (n < 0) 1549 if (n < 0)
1543 return 6; // unprintable, displays <xxxx> 1550 return 6; // unprintable, displays <xxxx>
1544 if (n > 1) 1551 if (n > 1)
1545 return n; 1552 return n;
1547 if (!utf_printable(c)) 1554 if (!utf_printable(c))
1548 return 6; // unprintable, displays <xxxx> 1555 return 6; // unprintable, displays <xxxx>
1549 if (intable(doublewidth, sizeof(doublewidth), c)) 1556 if (intable(doublewidth, sizeof(doublewidth), c))
1550 return 2; 1557 return 2;
1551 #endif 1558 #endif
1552 if (p_emoji && intable(emoji_width, sizeof(emoji_width), c)) 1559 if (p_emoji && intable(emoji_wide, sizeof(emoji_wide), c))
1553 return 2; 1560 return 2;
1554 } 1561 }
1555 1562
1556 // Characters below 0x100 are influenced by 'isprint' option 1563 // Characters below 0x100 are influenced by 'isprint' option
1557 else if (c >= 0x80 && !vim_isprintc(c)) 1564 else if (c >= 0x80 && !vim_isprintc(c))
2568 #endif 2575 #endif
2569 } 2576 }
2570 2577
2571 // Sorted list of non-overlapping intervals of all Emoji characters, 2578 // Sorted list of non-overlapping intervals of all Emoji characters,
2572 // based on http://unicode.org/emoji/charts/emoji-list.html 2579 // based on http://unicode.org/emoji/charts/emoji-list.html
2580 // Generated by ../runtime/tools/unicode.vim.
2581 // Excludes 0x00a9 and 0x00ae because they are considered latin1.
2573 static struct interval emoji_all[] = 2582 static struct interval emoji_all[] =
2574 { 2583 {
2575 {0x203c, 0x203c}, 2584 {0x203c, 0x203c},
2576 {0x2049, 0x2049}, 2585 {0x2049, 0x2049},
2577 {0x2122, 0x2122}, 2586 {0x2122, 0x2122},
5340 # endif 5349 # endif
5341 } 5350 }
5342 5351
5343 return retval; 5352 return retval;
5344 } 5353 }
5354
5355 /*
5356 * Table set by setcellwidths().
5357 */
5358 typedef struct
5359 {
5360 long first;
5361 long last;
5362 char width;
5363 } cw_interval_T;
5364
5365 static cw_interval_T *cw_table = NULL;
5366 static size_t cw_table_size = 0;
5367
5368 /*
5369 * Return 1 or 2 when "c" is in the cellwidth table.
5370 * Return 0 if not.
5371 */
5372 static int
5373 cw_value(int c)
5374 {
5375 int mid, bot, top;
5376
5377 if (cw_table == NULL)
5378 return 0;
5379
5380 // first quick check for Latin1 etc. characters
5381 if (c < cw_table[0].first)
5382 return 0;
5383
5384 // binary search in table
5385 bot = 0;
5386 top = (int)cw_table_size - 1;
5387 while (top >= bot)
5388 {
5389 mid = (bot + top) / 2;
5390 if (cw_table[mid].last < c)
5391 bot = mid + 1;
5392 else if (cw_table[mid].first > c)
5393 top = mid - 1;
5394 else
5395 return cw_table[mid].width;
5396 }
5397 return 0;
5398 }
5399
5400 static int
5401 tv_nr_compare(const void *a1, const void *a2)
5402 {
5403 listitem_T *li1 = (listitem_T *)a1;
5404 listitem_T *li2 = (listitem_T *)a2;
5405
5406 return li1->li_tv.vval.v_number - li2->li_tv.vval.v_number;
5407 }
5408
5409 void
5410 f_setcellwidths(typval_T *argvars, typval_T *rettv UNUSED)
5411 {
5412 list_T *l;
5413 listitem_T *li;
5414 int item;
5415 int i;
5416 listitem_T **ptrs;
5417 cw_interval_T *table;
5418
5419 if (argvars[0].v_type != VAR_LIST || argvars[0].vval.v_list == NULL)
5420 {
5421 emsg(_(e_listreq));
5422 return;
5423 }
5424 l = argvars[0].vval.v_list;
5425 if (l->lv_len == 0)
5426 {
5427 // Clearing the table.
5428 vim_free(cw_table);
5429 cw_table = NULL;
5430 cw_table_size = 0;
5431 return;
5432 }
5433
5434 ptrs = ALLOC_MULT(listitem_T *, l->lv_len);
5435 if (ptrs == NULL)
5436 return;
5437
5438 // Check that all entries are a list with three numbers, the range is
5439 // valid and the cell width is valid.
5440 item = 0;
5441 for (li = l->lv_first; li != NULL; li = li->li_next)
5442 {
5443 listitem_T *lili;
5444 varnumber_T n1;
5445
5446 if (li->li_tv.v_type != VAR_LIST || li->li_tv.vval.v_list == NULL)
5447 {
5448 semsg(_(e_list_item_nr_is_not_list), item);
5449 vim_free(ptrs);
5450 return;
5451 }
5452 for (lili = li->li_tv.vval.v_list->lv_first, i = 0; lili != NULL;
5453 lili = lili->li_next, ++i)
5454 {
5455 if (lili->li_tv.v_type != VAR_NUMBER)
5456 break;
5457 if (i == 0)
5458 {
5459 n1 = lili->li_tv.vval.v_number;
5460 if (n1 < 0x100)
5461 {
5462 emsg(_(e_only_values_of_0x100_and_higher_supported));
5463 vim_free(ptrs);
5464 return;
5465 }
5466 }
5467 else if (i == 1 && lili->li_tv.vval.v_number < n1)
5468 {
5469 semsg(_(e_list_item_nr_range_invalid), item);
5470 vim_free(ptrs);
5471 return;
5472 }
5473 else if (i == 2 && (lili->li_tv.vval.v_number < 1
5474 || lili->li_tv.vval.v_number > 2))
5475 {
5476 semsg(_(e_list_item_nr_cell_width_invalid), item);
5477 vim_free(ptrs);
5478 return;
5479 }
5480 }
5481 if (i != 3)
5482 {
5483 semsg(_(e_list_item_nr_does_not_contain_3_numbers), item);
5484 vim_free(ptrs);
5485 return;
5486 }
5487 ptrs[item++] = lili;
5488 }
5489
5490 // Sort the list on the first number.
5491 qsort((void *)ptrs, (size_t)l->lv_len, sizeof(listitem_T *), tv_nr_compare);
5492
5493 table = ALLOC_MULT(cw_interval_T, l->lv_len);
5494 if (table == NULL)
5495 {
5496 vim_free(ptrs);
5497 return;
5498 }
5499
5500 // Store the items in the new table.
5501 item = 0;
5502 for (li = l->lv_first; li != NULL; li = li->li_next)
5503 {
5504 listitem_T *lili = li->li_tv.vval.v_list->lv_first;
5505 varnumber_T n1;
5506
5507 n1 = lili->li_tv.vval.v_number;
5508 if (item > 0 && n1 <= table[item - 1].last)
5509 {
5510 semsg(_(e_overlapping_ranges_for_nr), (long)n1);
5511 vim_free(ptrs);
5512 vim_free(table);
5513 return;
5514 }
5515 table[item].first = n1;
5516 lili = lili->li_next;
5517 table[item].last = lili->li_tv.vval.v_number;
5518 lili = lili->li_next;
5519 table[item].width = lili->li_tv.vval.v_number;
5520 ++item;
5521 }
5522
5523 vim_free(ptrs);
5524 vim_free(cw_table);
5525 cw_table = table;
5526 cw_table_size = l->lv_len;
5527 }