lib-unicode-names.c

   1 /*
   2  * Copyright Neil Brown ©2023 <neil@brown.name>
   3  * May be distributed under terms of GPLv2 - see file:COPYING
   4  *
   5  * Parse the Unicode NamesList.txt file to find names for
   6  * unicode characters.
   7  */
   8
   9 #include <unistd.h>
  10 #include <sys/mman.h>
  11 #include <fcntl.h>
  12 #include "core.h"
  13
  14 struct unicode_data {
  15         struct command c;
  16         char *names;
  17         int len;
  18 };
  19
  20 static void report_names(struct unicode_data *ud safe, const char *name safe,
  21                          int which,
  22                          struct pane *p safe, struct command *c safe)
  23 {
  24         /* name must be start of a word, as either primary or secondary
  25          * name.  Ignore case.
  26          * If "which" is zero, return them all, else only return the
  27          * nth one where which==n
  28          */
  29         char *ptn = strconcat(p, "?i:^([0-9A-F]{4,5}    |       = ).*\\b", name);
  30         int i;
  31
  32         if (!ud->names)
  33                 return;
  34
  35         for (i = 0; i < ud->len; ) {
  36                 int ch, s;
  37                 char *cp, *n, *eol;
  38
  39                 s = call("text-search", p, 0, NULL, ptn,
  40                          ud->len - i, NULL, ud->names + i);
  41                 if (s <= 0)
  42                         break;
  43                 i += s-1;
  44                 /* i is now the start of the match */
  45                 cp = ud->names + i;
  46                 eol = strchr(cp, '\n');
  47                 if (!eol)
  48                         break;
  49                 i = (eol - ud->names) + 1;
  50                 if (eol[-1] == '\r')
  51                         eol -= 1;
  52                 if (*cp == '\t') {
  53                         /* secondary name "\t= "*/
  54                         n = strndup(cp+3, eol-cp-3);
  55                         /* find number */
  56                         while (cp > ud->names &&
  57                                (cp[-1] != '\n' || cp[0] == '\t'))
  58                                 cp -= 1;
  59                 } else {
  60                         /* primary name "XXXXX?\t" */
  61                         if (cp[4] == '\t')
  62                                 n = strndup(cp+5, eol-cp-5);
  63                         else
  64                                 n = strndup(cp+6, eol-cp-6);
  65                 }
  66                 ch = strtoul(cp, &eol, 16);
  67                 if (eol == cp+4 || eol == cp+5) {
  68                         if (which == 0)
  69                                 comm_call(c, "cb", p, ch, NULL, n);
  70                         else {
  71                                 which -= 1;
  72                                 if (which == 0) {
  73                                         comm_call(c, "cb", p, ch, NULL, n);
  74                                         i = ud->len;
  75                                 }
  76                         }
  77                 }
  78                 free(n);
  79         }
  80 }
  81
  82 static void unicode_free(struct command *c safe)
  83 {
  84         struct unicode_data *ud = container_of(c, struct unicode_data, c);
  85
  86         if (ud->names)
  87                 munmap(ud->names, ud->len);
  88 }
  89
  90 DEF_CMD(unicode_names)
  91 {
  92         struct unicode_data *ud;
  93         if (ci->comm == &unicode_names) {
  94                 /* This is the first call - need to allocate storage,
  95                  * load the NamesList file, and register a new command.
  96                  */
  97                 char *p;
  98                 int fd;
  99
 100                 alloc(ud, pane);
 101                 ud->c = unicode_names;
 102                 ud->c.free = unicode_free;
 103                 call_comm("global-set-command", ci->home, &ud->c, 0, NULL,
 104                           "Unicode-names");
 105                 p = call_ret(str, "xdg-find-edlib-file", ci->focus, 0, NULL,
 106                              "NamesList.txt", 0, NULL, "data");
 107                 if (!p)
 108                         return Efail;
 109                 fd = open(p, O_RDONLY);
 110                 free(p);
 111                 if (fd < 0)
 112                         return Efail;
 113                 ud->len = lseek(fd, 0, 2);
 114                 ud->names = mmap(NULL, ud->len, PROT_READ, MAP_SHARED, fd, 0);
 115                 close(fd);
 116         } else {
 117                 ud = container_of(ci->comm, struct unicode_data, c);
 118         }
 119         if (!ud->names)
 120                 return Efail;
 121         if (ci->str && ci->comm2)
 122                 report_names(ud, ci->str, ci->num, ci->focus, ci->comm2);
 123         return 1;
 124 }
 125
 126 void edlib_init(struct pane *ed safe)
 127 {
 128         call_comm("global-set-command", ed, &unicode_names,
 129                   0, NULL, "Unicode-names");
 130 }