git.m6w6.name Git - m6w6/ext-http/blob - scripts/gen_if_utf8.php

   1 #!/usr/bin/env php
   2 <?php
   3
   4 error_reporting(E_ALL);
   5 set_error_handler(function($c, $e, $f, $l) {
   6         throw new Exception("$e in $f on line $l");
   7 });
   8
   9 $i18n = $argc >= 2 ? $argv[1] : "/usr/share/i18n/locales/i18n";
  10
  11 $f = fopen($i18n, "r");
  12 $c = false;
  13 $a = false;
  14
  15 ob_start(null, 0xfffff);
  16
  17 while (!feof($f)) {
  18         $line = fgets($f);
  19         if (!$c && $line !== "LC_CTYPE\n") {
  20                 continue;
  21         }
  22         $c = true;
  23         if ($line === "END LC_CTYPE\n") {
  24                 break;
  25         }
  26         switch($line{0}) {
  27         case "%":
  28                 break;
  29         case "\n":
  30                 if ($a) {
  31                         break 2;
  32                 }
  33                 break;
  34         case " ":
  35                 if ($a) {
  36                         foreach (explode(";", trim($line, "\n/ ;")) as $ranges) {
  37                                 $range = explode("..", $ranges);
  38                                 $step = 0;
  39                                 $end = 0;
  40                                 switch (count($range)) {
  41                                 case 3:
  42                                         list($sstart, $sstep, $send) = $range;
  43                                         sscanf($sstart, "<U%X>", $start);
  44                                         sscanf($sstep, "(%d)", $step);
  45                                         sscanf($send, "<U%X>", $end);
  46
  47                                         break;
  48                                 case 2:
  49                                         list($sstart, $send) = $range;
  50                                         $step = 1;
  51                                         sscanf($sstart, "<U%X>", $start);
  52                                         sscanf($send, "<U%X>", $end);
  53                                         break;
  54                                 case 1:
  55                                         list($sstart) = $range;
  56                                         sscanf($sstart, "<U%X>", $start);
  57                                         break;
  58                                 }
  59                                 $r[]=[$start,$end,$step];
  60                         }
  61                 }
  62                 break;
  63         default:
  64                 if ($a) {
  65                         break 2;
  66                 } elseif ($line === "alpha /\n") {
  67                         $a = true;
  68                 }
  69                 break;
  70         }
  71 }
  72 function sp($sp, $ch = " ") { return str_repeat($ch, $sp); }
  73 $sp = 1;
  74 foreach ($r as list($start, $end, $step)) {
  75         if ($end) {
  76                 if ($step > 1) {
  77                         die("\nUNEXPECTED: step>1\n");
  78                         printf("\tfor (i=0x%08X; i <= 0x%08X; i+= %d) { if (i == ch) return 1; }\n", $start, $end, $step);
  79                 } else {
  80                         //printf(" if (ch >= 0x%08X && ch <= 0x%08X) return 1;\n", $start, $end);
  81                         printf("%sif (ch >= 0x%08X) {\n", sp($sp), $start);
  82                         printf("%sif (ch <= 0x%08X) return 1;\n", sp(++$sp), $end);
  83                 }
  84         } else {
  85                 printf("%sif (ch == 0x%08X) return 1;\n", sp($sp), $start);
  86         }
  87 }
  88 printf("  %s\n", sp($sp-1, "}"));
  89
  90 file_put_contents("php_http_utf8.h",
  91         preg_replace('/(\/\* BEGIN::UTF8SWITCH \*\/\n).*(\n\s*\/\* END::UTF8SWITCH \*\/)/s', '$1'. ob_get_contents() .'$2',
  92                 file_get_contents("php_http_utf8.h")));