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 $m = 0xfffff000 >> $b;
  13 $b2 = 8;
  14 $x = $m >> $b2;
  15
  16 $f = fopen($i18n, "r");
  17 $c = false;
  18 $a = false;
  19
  20 $ranges = $lables = $gotos = [];
  21
  22 ob_start(null, 0xfffff);
  23
  24 while (!feof($f)) {
  25         $line = fgets($f);
  26         if (!$c && $line !== "LC_CTYPE\n") {
  27                 continue;
  28         }
  29         $c = true;
  30         if ($line === "END LC_CTYPE\n") {
  31                 break;
  32         }
  33         switch($line{0}) {
  34         case "%":
  35                 break;
  36         case "\n":
  37                 if ($a) {
  38                         break 2;
  39                 }
  40                 break;
  41         case " ":
  42                 if ($a) {
  43                         foreach (explode(";", trim($line, "\n/ ;")) as $list) {
  44                                 $range = explode("..", $list);
  45                                 $step = 0;
  46                                 $end = 0;
  47                                 switch (count($range)) {
  48                                 case 3:
  49                                         list($sstart, $sstep, $send) = $range;
  50                                         sscanf($sstart, "<U%X>", $start);
  51                                         sscanf($sstep, "(%d)", $step);
  52                                         sscanf($send, "<U%X>", $end);
  53
  54                                         break;
  55                                 case 2:
  56                                         list($sstart, $send) = $range;
  57                                         $step = 1;
  58                                         sscanf($sstart, "<U%X>", $start);
  59                                         sscanf($send, "<U%X>", $end);
  60                                         break;
  61                                 case 1:
  62                                         list($sstart) = $range;
  63                                         sscanf($sstart, "<U%X>", $start);
  64                                         break;
  65                                 }
  66                                 $sw = $start >> $b;
  67                                 $sw2 = ($start & $m) >> $b2;
  68                                 if (isset($ranges[$sw][$sw2])) {
  69                                         //$ranges[$sw][$sw2] = array_filter($ranges[$sw][$sw2]);
  70                                 }
  71                                 $ranges[$sw][$sw2][]=[$start,$end,$step];
  72                                 if ($end) {
  73                                         $goto = $start;
  74                                 }
  75                                 while (($start += $step) <= $end) {
  76                                         $ssw = $start >> $b;
  77                                         $ssw2 = ($start & $m) >> $b2;
  78                                         if (!isset($ranges[$ssw][$ssw2]) || null !== end($ranges[$ssw][$ssw2])) {
  79                                                 $ranges[$ssw][$ssw2][]=null;
  80                                         }
  81                                         if ($ssw != $sw || $ssw2 != $sw2) {
  82                                                 $gotos[$ssw][$ssw2] = $goto;
  83                                                 $labels[$sw][$sw2] = $goto;
  84                                         }
  85                                 }
  86                         }
  87                 }
  88                 break;
  89         default:
  90                 if ($a) {
  91                         break 2;
  92                 } elseif ($line === "alpha /\n") {
  93                         $a = true;
  94                 }
  95                 break;
  96         }
  97 }
  98
  99 function sp($sp, $ch = " ") { return str_repeat($ch, $sp); }
 100
 101 printf("switch (ch >> %d) {\n", $b);
 102 foreach ($ranges as $sw => $sws) {
 103         printf("case 0x%08X:\n", $sw);
 104         printf(" switch((ch & 0x%08X) >> %d) {\n", $m, $b2);
 105         foreach ($sws as $sw2 => $specs) {
 106                 printf(" case 0x%08X:\n", $sw2);
 107                 $sp = 2;
 108                 $start = null;
 109                 foreach ($specs as $index => $spec) {
 110                         if ($spec) {
 111                                 list($start, $end, $step) = $spec;
 112                                 if (isset($labels[$sw][$sw2])) {
 113                                         $label = $labels[$sw][$sw2];
 114                                         if ((!$end && $label == $start) || ($end && $label >= $start && $label <= $end)) {
 115                                                 printf("%sc_%08X:;\n", sp($sp), $label);
 116                                         }
 117                                 }
 118                                 if ($end) {
 119                                         if ($step > 1) {
 120                                                 die("\nUNEXPECTED: step>1\n");
 121                                                 printf("\tfor (i=0x%08X; i <= 0x%08X; i+= %d) { if (i == ch) return 1; }\n", $start, $end, $step);
 122                                         } else {
 123                                                 printf("%sif (ch >= 0x%08X) {\n", sp($sp), $start);
 124                                                 printf("%sif (ch <= 0x%08X) return 1;\n", sp(++$sp), $end);
 125                                         }
 126                                 } else {
 127                                         printf("%sif (ch == 0x%08X) return 1;\n", sp($sp), $start);
 128                                 }
 129                         } else {
 130                                 if (isset($gotos[$sw][$sw2]) && !$start) {
 131                                         if (isset($specs[$index + 1])) {
 132                                                 list($next) = $specs[$index + 1];
 133                                                 printf("%sif (ch < 0x%08X)\n ", sp($sp), $next);
 134                                         }
 135                                         $goto = $gotos[$sw][$sw2];
 136                                         printf("%sgoto c_%08X;\n", sp($sp), $goto);
 137                                         $start = $goto;
 138                                 }
 139                         }
 140                 }
 141                 if ($sp > 2) {
 142                         printf("  %s\n", sp($sp-2, "}"));
 143                 }
 144                 printf("  break;\n");
 145         }
 146         printf(" }\n break;\n");
 147 }
 148 printf("}\n");
 149
 150 file_put_contents("php_http_utf8.h",
 151         preg_replace('/(\/\* BEGIN::UTF8SWITCH \*\/\n).*(\n\s*\/\* END::UTF8SWITCH \*\/)/s', '$1'. ob_get_contents() .'$2',
 152                 file_get_contents("php_http_utf8.h")));