text-util.c

   1 #include "text-util.h"
   2 #include "util.h"
   3 #include <wchar.h>
   4 #include <errno.h>
   5 #include <stdlib.h>
   6
   7 bool text_range_valid(const Filerange *r) {
   8         return r->start != EPOS && r->end != EPOS && r->start <= r->end;
   9 }
  10
  11 size_t text_range_size(const Filerange *r) {
  12         return text_range_valid(r) ? r->end - r->start : 0;
  13 }
  14
  15 Filerange text_range_empty(void) {
  16         return (Filerange){ .start = EPOS, .end = EPOS };
  17 }
  18
  19 Filerange text_range_union(const Filerange *r1, const Filerange *r2) {
  20         if (!text_range_valid(r1))
  21                 return *r2;
  22         if (!text_range_valid(r2))
  23                 return *r1;
  24         return (Filerange) {
  25                 .start = MIN(r1->start, r2->start),
  26                 .end = MAX(r1->end, r2->end),
  27         };
  28 }
  29
  30 Filerange text_range_new(size_t a, size_t b) {
  31         return (Filerange) {
  32                 .start = MIN(a, b),
  33                 .end = MAX(a, b),
  34         };
  35 }
  36
  37 bool text_range_equal(const Filerange *r1, const Filerange *r2) {
  38         if (!text_range_valid(r1) && !text_range_valid(r2))
  39                 return true;
  40         return r1->start == r2->start && r1->end == r2->end;
  41 }
  42
  43 bool text_range_overlap(const Filerange *r1, const Filerange *r2) {
  44         if (!text_range_valid(r1) || !text_range_valid(r2))
  45                 return false;
  46         return r1->start <= r2->end && r2->start <= r1->end;
  47 }
  48
  49 bool text_range_contains(const Filerange *r, size_t pos) {
  50         return text_range_valid(r) && r->start <= pos && pos <= r->end;
  51 }
  52
  53 int text_char_count(const char *data, size_t len) {
  54         int count = 0;
  55         mbstate_t ps = { 0 };
  56         while (len > 0) {
  57                 wchar_t wc;
  58                 size_t wclen = mbrtowc(&wc, data, len, &ps);
  59                 if (wclen == (size_t)-1 && errno == EILSEQ) {
  60                         count++;
  61                         while (!ISUTF8(*data))
  62                                 data++, len--;
  63                 } else if (wclen == (size_t)-2) {
  64                         break;
  65                 } else if (wclen == 0) {
  66                         count++;
  67                         data++;
  68                         len--;
  69                 } else {
  70                         int width = wcwidth(wc);
  71                         if (width != 0)
  72                                 count++;
  73                         data += wclen;
  74                         len -= wclen;
  75                 }
  76         }
  77         return count;
  78 }
  79
  80 int text_string_width(const char *data, size_t len) {
  81
  82         int width = 0;
  83         mbstate_t ps = { 0 };
  84         const char *s = data;
  85
  86         while (len > 0) {
  87                 wchar_t wc;
  88                 size_t wclen = mbrtowc(&wc, s, len, &ps);
  89                 if (wclen == (size_t)-1 && errno == EILSEQ) {
  90                         /* assume a replacement symbol will be displayed */
  91                         width++;
  92                         wclen = 1;
  93                 } else if (wclen == (size_t)-2) {
  94                         /* do nothing, advance to next character */
  95                         wclen = 1;
  96                 } else if (wclen == 0) {
  97                         /* assume NUL byte will be displayed as ^@ */
  98                         width += 2;
  99                         wclen = 1;
 100                 } else if (wc == L'\t') {
 101                         width++;
 102                         wclen = 1;
 103                 } else {
 104                         int w = wcwidth(wc);
 105                         if (w == -1)
 106                                 w = 2; /* assume non-printable will be displayed as ^{char} */
 107                         width += w;
 108                 }
 109                 len -= wclen;
 110                 s += wclen;
 111         }
 112
 113         return width;
 114 }