git.m6w6.name Git - m6w6/ext-http/blob - scripts/gen_switch_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 $b = 10;
  12
  13 $f = fopen($i18n, "r");
  14 $c = false;
  15 $a = false;
  16
  17 ob_start(null, 0xfffff);
  18 while (!feof($f)) {
  19         $line = fgets($f);
  20         if (!$c && $line !== "LC_CTYPE\n") {
  21                 continue;
  22         }
  23         $c = true;
  24         if ($line === "END LC_CTYPE\n") {
  25                 break;
  26         }
  27         switch($line{0}) {
  28         case "%":
  29                 break;
  30         case "\n":
  31                 if ($a) {
  32                         break 2;
  33                 }
  34                 break;
  35         case " ":
  36                 if ($a) {
  37                         foreach (explode(";", trim($line, "\n/ ;")) as $ranges) {
  38                                 $range = explode("..", $ranges);
  39                                 $step = 0;
  40                                 $end = 0;
  41                                 switch (count($range)) {
  42                                 case 3:
  43                                         list($sstart, $sstep, $send) = $range;
  44                                         sscanf($sstart, "<U%X>", $start);
  45                                         sscanf($sstep, "(%d)", $step);
  46                                         sscanf($send, "<U%X>", $end);
  47
  48                                         break;
  49                                 case 2:
  50                                         list($sstart, $send) = $range;
  51                                         $step = 1;
  52                                         sscanf($sstart, "<U%X>", $start);
  53                                         sscanf($send, "<U%X>", $end);
  54                                         break;
  55                                 case 1:
  56                                         list($sstart) = $range;
  57                                         sscanf($sstart, "<U%X>", $start);
  58                                         break;
  59                                 }
  60                                 $r[$start >> $b][]=[$start,$end,$step];
  61                         }
  62                 }
  63                 break;
  64         default:
  65                 if ($a) {
  66                         break 2;
  67                 } elseif ($line === "alpha /\n") {
  68                         $a = true;
  69                 }
  70                 break;
  71         }
  72 }
  73
  74 function sp($sp, $ch = " ") { return str_repeat($ch, $sp); }
  75 printf("switch (ch >> %d) {\n", $b);
  76 foreach ($r as $sw => $specs) {
  77         printf("case 0x%08X:\n", $sw);
  78         $sp = 1;
  79         foreach ($specs as list($start, $end, $step)) {
  80                 if ($end) {
  81                         if ($step > 1) {
  82                                 die("\nUNEXPECTED: step>1\n");
  83                                 printf("\tfor (i=0x%08X; i <= 0x%08X; i+= %d) { if (i == ch) return 1; }\n", $start, $end, $step);
  84                         } else {
  85                                 //printf(" if (ch >= 0x%08X && ch <= 0x%08X) return 1;\n", $start, $end);
  86                                 printf("%sif (ch >= 0x%08X) {\n", sp($sp), $start);
  87                                 printf("%sif (ch <= 0x%08X) return 1;\n", sp(++$sp), $end);
  88                         }
  89                 } else {
  90                         printf("%sif (ch == 0x%08X) return 1;\n", sp($sp), $start);
  91                 }
  92         }
  93         printf(" %s\n break;\n", sp(--$sp, "}"));
  94 }
  95 printf("}\n");
  96
  97 file_put_contents("php_http_utf8.h",
  98         preg_replace('/(\/\* BEGIN::UTF8SWITCH \*\/\n).*(\n\s*\/\* END::UTF8SWITCH \*\/)/s', '$1'. ob_get_contents() .'$2',
  99                 file_get_contents("php_http_utf8.h")));