git.m6w6.name Git - m6w6/ext-http/blob - ctype.php

   1 <?php
   2
   3 error_reporting(E_ALL);
   4 set_error_handler(function($c, $e, $f, $l) {
   5         throw new Exception("$e in $f on line $l");
   6 });
   7
   8 $i18n = $argc >= 2 ? $argv[1] : "/usr/share/i18n/locales/i18n";
   9
  10 $f = fopen($i18n, "r");
  11 $c = false;
  12 $a = false;
  13 $r = array();
  14
  15 print <<<C
  16 typedef struct utf8_range {
  17         unsigned int start;
  18         unsigned int end;
  19         unsigned char step;
  20 } utf8_range_t;
  21
  22 static const utf8_range_t utf8_ranges[] = {
  23
  24 C;
  25 while (!feof($f)) {
  26         $line = fgets($f);
  27         if (!$c && $line !== "LC_CTYPE\n") {
  28                 continue;
  29         }
  30         $c = true;
  31         if ($line === "END LC_CTYPE\n") {
  32                 break;
  33         }
  34         switch($line{0}) {
  35         case "%":
  36                 if ($a) {
  37                         printf("/* %s */\n", trim($line, "%\n/ "));
  38                 }
  39                 break;
  40         case "\n":
  41                 if ($a) {
  42                         break 2;
  43                 }
  44                 break;
  45         case " ":
  46                 if ($a) {
  47                         foreach (explode(";", trim($line, "\n/ ;")) as $ranges) {
  48                                 $range = explode("..", $ranges);
  49                                 $step = 0;
  50                                 $end = 0;
  51                                 switch (count($range)) {
  52                                 case 3:
  53                                         list($sstart, $sstep, $send) = $range;
  54                                         sscanf($sstart, "<U%X>", $start);
  55                                         sscanf($sstep, "(%d)", $step);
  56                                         sscanf($send, "<U%X>", $end);
  57
  58                                         break;
  59                                 case 2:
  60                                         list($sstart, $send) = $range;
  61                                         $step = 1;
  62                                         sscanf($sstart, "<U%X>", $start);
  63                                         sscanf($send, "<U%X>", $end);
  64                                         break;
  65                                 case 1:
  66                                         list($sstart) = $range;
  67                                         sscanf($sstart, "<U%X>", $start);
  68                                         break;
  69                                 }
  70                                 print "\t{";
  71                                 if ($start >= 0xffff) {
  72                                         printf("0x%08X, ", $start);
  73                                         if ($end) {
  74                                                 printf("0x%08X, ", $end);
  75                                         } else {
  76                                                 print("         0, ");
  77                                         }
  78                                 } else {
  79                                         printf("    0x%04X, ", $start);
  80                                         if ($end) {
  81                                                 printf("    0x%04X, ", $end);
  82                                         } else {
  83                                                 print("         0, ");
  84                                         }
  85                                 }
  86                                 printf("%d},\n", $step);
  87                         }
  88                 }
  89                 break;
  90         default:
  91                 if ($a) {
  92                         break 2;
  93                 } elseif ($line === "alpha /\n") {
  94                         $a = true;
  95                 }
  96                 break;
  97         }
  98 }
  99
 100 print <<<C
 101         {0, 0, 0}
 102 };
 103
 104 C;