libgrapheme

grapheme cluster utility library
git clone git://git.suckless.org/libgrapheme
Log | Files | Refs | LICENSE

grapheme_boundary.c (2737B)

      1 /* See LICENSE file for copyright and license details. */
      2 #include <stddef.h>
      3 #include <stdio.h>
      4 #include <string.h>
      5 
      6 #include "datautil.h"
      7 
      8 static struct {
      9 	char         *enumname;
     10 	char         *identifier;
     11 	struct range *table;
     12 	size_t        tablelen;
     13 } properties[] = {
     14 	{
     15 		/* carriage return */
     16 		.enumname   = "GB_PROP_CR",
     17 		.identifier = "CR",
     18 	},
     19 	{
     20 		/* line feed */
     21 		.enumname   = "GB_PROP_LF",
     22 		.identifier = "LF",
     23 	},
     24 	{
     25 		/* control character */
     26 		.enumname   = "GB_PROP_CONTROL",
     27 		.identifier = "Control",
     28 	},
     29 	{
     30 		/* grapheme extender */
     31 		.enumname   = "GB_PROP_EXTEND",
     32 		.identifier = "Extend",
     33 	},
     34 	{
     35 		/* zero width joiner */
     36 		.enumname   = "GB_PROP_ZWJ",
     37 		.identifier = "ZWJ",
     38 	},
     39 	{
     40 		/* regional indicator */
     41 		.enumname   = "GB_PROP_REGIONAL_INDICATOR",
     42 		.identifier = "Regional_Indicator",
     43 	},
     44 	{
     45 		/* prepend character */
     46 		.enumname   = "GB_PROP_PREPEND",
     47 		.identifier = "Prepend",
     48 	},
     49 	{
     50 		/* spacing mark */
     51 		.enumname   = "GB_PROP_SPACINGMARK",
     52 		.identifier = "SpacingMark",
     53 	},
     54 	{
     55 		/* hangul syllable type L */
     56 		.enumname   = "GB_PROP_L",
     57 		.identifier = "L",
     58 	},
     59 	{
     60 		/* hangul syllable type V */
     61 		.enumname   = "GB_PROP_V",
     62 		.identifier = "V",
     63 	},
     64 	{
     65 		/* hangul syllable type T */
     66 		.enumname   = "GB_PROP_T",
     67 		.identifier = "T",
     68 	},
     69 	{
     70 		/* hangul syllable type LV */
     71 		.enumname   = "GB_PROP_LV",
     72 		.identifier = "LV",
     73 	},
     74 	{
     75 		/* hangul syllable type LVT */
     76 		.enumname   = "GB_PROP_LVT",
     77 		.identifier = "LVT",
     78 	},
     79 };
     80 
     81 int
     82 process_line(char **field, size_t nfields, char *comment)
     83 {
     84 	size_t i;
     85 	struct range r;
     86 
     87 	(void)comment;
     88 
     89 	if (nfields < 2) {
     90 		return 1;
     91 	}
     92 
     93 	for (i = 0; i < LEN(properties); i++) {
     94 		if (!strcmp(field[1], properties[i].identifier)) {
     95 			if (range_parse(field[0], &r)) {
     96 				return 1;
     97 			}
     98 			range_list_append(&(properties[i].table),
     99 			                  &(properties[i].tablelen), &r);
    100 			break;
    101 		}
    102 	}
    103 
    104 	return 0;
    105 }
    106 
    107 int
    108 main(void)
    109 {
    110 	size_t i, j;
    111 
    112 	printf("/* Automatically generated by data/gbp */\n"
    113 	       "#include <stdint.h>\n\n#include \"../src/util.h\"\n\n");
    114 
    115 	parse_input(process_line);
    116 
    117 	/* output enum */
    118 	printf("enum gb_prop {\n");
    119 	for (i = 0; i < LEN(properties); i++) {
    120 		printf("\t%s,\n", properties[i].enumname);
    121 	}
    122 	printf("};\n\n");
    123 
    124 	/* output table */
    125 	printf("static const struct range_list gb_prop[] = {\n");
    126 	for (i = 0; i < LEN(properties); i++) {
    127 		printf("\t[%s] = {\n\t\t.data = (struct range[]){\n", properties[i].enumname);
    128 		for (j = 0; j < properties[i].tablelen; j++) {
    129 			printf("\t\t\t{ UINT32_C(0x%06X), UINT32_C(0x%06X) },\n",
    130 			       properties[i].table[j].lower,
    131 			       properties[i].table[j].upper);
    132 		}
    133 		printf("\t\t},\n\t\t.len = %zu,\n\t},\n", properties[i].tablelen);
    134 	}
    135 	printf("};\n");
    136 
    137 	return 0;
    138 }