Index: phpdoc/chm/make_chm.php diff -u phpdoc/chm/make_chm.php:1.5 phpdoc/chm/make_chm.php:1.6 --- phpdoc/chm/make_chm.php:1.5 Sun Oct 6 05:35:03 2002 +++ phpdoc/chm/make_chm.php Sun Nov 17 11:11:54 2002 @@ -14,25 +14,79 @@ $FANCY_PATH = getenv("PHP_HELP_COMPILE_FANCYDIR"); $LANGUAGE = getenv("PHP_HELP_COMPILE_LANG"); $INDEX_IN_HTML = "index.html"; +$INTERNAL_CHARSET = "UTF-8"; +$DEFAULT_FONT = "Arial,10,0"; if (empty($FANCY_PATH)) { $FANCY_PATH = $HTML_PATH; } // Array to manual code -> HTML Help Code conversion // Code list: http://www.helpware.net/htmlhelp/hh_info.htm $LANGUAGES = array( - "tw" => "0x404 Traditional Chinese", - "cs" => "0x405 Czech", - "de" => "0x407 German (Germany)", - "en" => "0x809 Enlish (United Kingdom)", - "es" => "0xc0a Spanish (International Sort)", - "fr" => "0x40c French (France)", - "hu" => "0x40e Hungarian", - "it" => "0x410 Italian (Italy)", - "ja" => "0x411 Japanese", - "kr" => "0x412 Korean", - "nl" => "0x413 Dutch (Netherlands)", - "pt_BR" => "0x416 Portuguese (Brazil)", - "zh" => "0x804 Simplified Chinese" + "tw" => array( + "langcode" => "0x404 Traditional Chinese", + "preferred_charset" => "CP950", + "preferred_font" => $DEFAULT_FONT + ), + "cs" => array( + "langcode" => "0x405 Czech", + "preferred_charset" => "Windows-1250", + "preferred_font" => $DEFAULT_FONT + ), + "de" => array( + "langcode" => "0x407 German (Germany)", + "preferred_charset" => "Windows-1252", + "preferred_font" => $DEFAULT_FONT + ), + "en" => array( + "langcode" => "0x809 English (United Kingdom)", + "preferred_charset" => "Windows-1252", + "preferred_font" => $DEFAULT_FONT + ), + "es" => array( + "langcode" => "0xc0a Spanish (International Sort)", + "preferred_charset" => "Windows-1252", + "preferred_font" => $DEFAULT_FONT + ), + "fr" => array( + "langcode" => "0x40c French (France)", + "preferred_charset" => "Windows-1252", + "preferred_font" => $DEFAULT_FONT + ), + "hu" => array( + "langcode" => "0x40e Hungarian", + "preferred_charset" => "Windows-1250", + "preferred_font" => $DEFAULT_FONT + ), + "it" => array( + "langcode" => "0x410 Italian (Italy)", + "preferred_charset" => "Windows-1252", + "preferred_font" => $DEFAULT_FONT + ), + "ja" => array( + "langcode" => "0x411 Japanese", + "preferred_charset" => "CP932", + "preferred_font" => "MS P Gothic,10,0" + ), + "kr" => array( + "langcode" => "0x412 Korean", + "preferred_charset" => "CP949", + "preferred_font" => $DEFAULT_FONT + ), + "nl" => array( + "langcode" => "0x413 Dutch (Netherlands)", + "preferred_charset" => "Windows-1252", + "preferred_font" => $DEFAULT_FONT + ), + "pt_BR" => array( + "langcode" => "0x416 Portuguese (Brazil)", + "preferred_charset" => "Windows-1252", + "preferred_font" => $DEFAULT_FONT + ), + "zh" => array( + "langcode" => "0x804 Simplified Chinese", + "preferred_charset" => "CP936", + "preferred_font" => $DEFAULT_FONT + ) ); // Files on the top level of the TOC @@ -46,6 +100,18 @@ "appendixes.html" ); +// backwards compatibility +if (!function_exists("file_get_contents")) { + function file_get_contents($file) + { + $cnt = file($file); + if ($cnt !== false) { + return join('', $cnt); + } + return false; + } +} + // Header for index and toc $HEADER = ' @@ -72,8 +138,8 @@ $index = fopen("php_manual_$LANGUAGE.hhk", "w"); // Write out file headers - fputs($toc, $HEADER); - fputs($index, $HEADER); + fputs_wrapper($toc, $HEADER); + fputs_wrapper($index, $HEADER); // Read original index file and drop out newlines $indexline = oneLiner("$HTML_PATH/$INDEX_IN_HTML"); @@ -85,7 +151,7 @@ if ($FIRST_PAGE != $INDEX_IN_HTML) { // Find the name of the Table of Contents - preg_match('|CLASS=\"TOC\" >

(.*)
(.*)(.*)|U', $indexline, $match); + preg_match('|([^<]*)|U', $indexline, $match); if (empty($match[1])) { // Fallback $match[1] = "Preface"; } mapAndIndex($match[1], "preface.html", " ", $toc, $index); // Now autofind the main pages + $MAIN_REGEXP = join("|", $MAIN_FILES); - preg_match_all("![IVX]+\. (.+)(.+)(?=[IVX]+\.
([^<]+)(.+)!Ui", $indexline, $matches, PREG_SET_ORDER); // Go through the main files, and link in subpages foreach ($matches as $matchinfo) { mapAndIndex($matchinfo[2], $matchinfo[1], " ", $toc, $index); - fputs($toc, "\n
    \n"); - preg_match_all("!(.+)!U", $matchinfo[3], $subpages, PREG_SET_ORDER); + fputs_wrapper($toc, "\n
      \n"); + preg_match_all("!([^<]*)!iU", $matchinfo[3], $subpages, PREG_SET_ORDER); + foreach ($subpages as $spinfo) { mapAndIndex($spinfo[2], $spinfo[1], " ", $toc, $index); findDeeperLinks($spinfo[1], $toc, $index); } - fputs($toc, "\n
    \n"); + fputs_wrapper($toc, "\n
\n"); } // Link in directly the copyright page $copyline = oneLiner("$HTML_PATH/copyright.html"); - preg_match('|

(.*)([^<]*)\n\n"); - fputs($toc, " \n\n"); + fputs_wrapper($index, " \n\n"); + fputs_wrapper($toc, " \n\n"); fclose($index); fclose($toc); } // makeContentfiles() function end @@ -148,41 +217,41 @@ // Start writing the project file $project = fopen("php_manual_$LANGUAGE.hhp", "w"); - fputs($project, "[OPTIONS]\n"); - fputs($project, "Compatibility=1.1 or later\n"); - fputs($project, "Compiled file=php_manual_$LANGUAGE.chm\n"); - fputs($project, "Contents file=php_manual_$LANGUAGE.hhc\n"); - fputs($project, "Index file=php_manual_$LANGUAGE.hhk\n"); - fputs($project, "Default Font=Arial,10,0\n"); - fputs($project, "Default Window=phpdoc\n"); - fputs($project, "Default topic=$FANCY_PATH\\$FIRST_PAGE\n"); - fputs($project, "Display compile progress=Yes\n"); - fputs($project, "Full-text search=Yes\n"); + fputs_wrapper($project, "[OPTIONS]\n"); + fputs_wrapper($project, "Compatibility=1.1 or later\n"); + fputs_wrapper($project, "Compiled file=php_manual_$LANGUAGE.chm\n"); + fputs_wrapper($project, "Contents file=php_manual_$LANGUAGE.hhc\n"); + fputs_wrapper($project, "Index file=php_manual_$LANGUAGE.hhk\n"); + fputs_wrapper($project, "Default Font={$LANGUAGES[$LANGUAGE]['preferred_font']}\n"); + fputs_wrapper($project, "Default Window=phpdoc\n"); + fputs_wrapper($project, "Default topic=$FANCY_PATH\\$FIRST_PAGE\n"); + fputs_wrapper($project, "Display compile progress=Yes\n"); + fputs_wrapper($project, "Full-text search=Yes\n"); // Get the proper language code from the array - fputs($project, "Language=" . $LANGUAGES[$LANGUAGE] . "\n"); + fputs_wrapper($project, "Language={$LANGUAGES[$LANGUAGE]["langcode"]}\n"); // Now try to find out how the manual named in the actual language // this must be in the index.html file as the title (DSSSL generated) $content = oneLiner("$HTML_PATH/$INDEX_IN_HTML"); - if (preg_match("|(.*)|U", $content, $found)) { + if (preg_match("|([^<]*)|U", $content, $found)) { $MANUAL_TITLE = $found[1]; } else { // Fallback $MANUAL_TITLE = "PHP Manual"; } - fputs($project, "Title=$MANUAL_TITLE\n"); + fputs_wrapper($project, "Title=$MANUAL_TITLE\n"); // Define the phpdoc window style (adds more functionality) - fputs($project, "\n[WINDOWS]\nphpdoc=\"$MANUAL_TITLE\",\"php_manual_$LANGUAGE.hhc\",\"php_manual_$LANGUAGE.hhk\"," . + fputs_wrapper($project, "\n[WINDOWS]\nphpdoc=\"$MANUAL_TITLE\",\"php_manual_$LANGUAGE.hhc\",\"php_manual_$LANGUAGE.hhk\"," . "\"$FANCY_PATH\\$FIRST_PAGE\",\"$FANCY_PATH\\$FIRST_PAGE\",,,,,0x23520,,0x386e,,,,,,,,0\n"); // Write out all the filenames as in FANCY_PATH - fputs($project, "\n[FILES]\n"); + fputs_wrapper($project, "\n[FILES]\n"); $handle = opendir($FANCY_PATH); while (false !== ($file = readdir($handle))) { if ($file != "." && $file != "..") { - fputs($project, "$FANCY_PATH\\$file\n"); + fputs_wrapper($project, "$FANCY_PATH\\$file\n"); } } closedir($handle); @@ -195,18 +264,18 @@ global $FANCY_PATH; $name = str_replace('"', '"', $name); - fputs($toc, " + fputs_wrapper($toc, " $tabs

  • $tabs $tabs "); if ($imgnum != "auto") { - fputs($toc, "$tabs \n"); + fputs_wrapper($toc, "$tabs \n"); } - fputs($toc, "$tabs \n"); + fputs_wrapper($toc, "$tabs \n"); - fputs($index, " + fputs_wrapper($index, "
  • @@ -223,10 +292,10 @@ $contents = oneLiner("$HTML_PATH/$filename"); // Find all sublinks - if (preg_match_all("!(.*)!U", $contents, $matches, PREG_SET_ORDER)) { + if (preg_match_all("!([^<]*)!U", $contents, $matches, PREG_SET_ORDER)) { // Print out the file informations for all the links - fputs($toc, "\n
      "); + fputs_wrapper($toc, "\n
        "); foreach ($matches as $onematch) { $param["html"] = $onematch[1]; if (!empty($onematch[3])) { @@ -235,7 +304,7 @@ $param["title"] = strip_tags($onematch[4]); mapAndIndex($param["title"], $param["html"], " ", $toc, $index); } - fputs($toc, "
      \n"); + fputs_wrapper($toc, "
    \n"); } else { @@ -250,11 +319,61 @@ } // findDeeperLinks() function end +function fputs_wrapper($fp, $str) +{ + fputs($fp, convertCharset($str)); +} // Return a file joined on one line function oneLiner($filename) { - return preg_replace("/[\r|\n]{1,2}/", " ", join("", file($filename))); + global $INTERNAL_CHARSET; + + $buf = preg_replace("/[\r|\n]{1,2}/U", " ", file_get_contents($filename)); + $charset = detectDocumentCharset($buf); + + if ($charset === false) $charset = "UTF-8"; + + if ($charset != $INTERNAL_CHARSET) { + if (function_exists("iconv")) { + $buf = iconv($charset, $INTERNAL_CHARSET, $buf); + } elseif (function_exists("mb_convert_encoding")) { + $buf = mb_convert_encoding($buf, $INTERNAL_CHARSET, $charset); + } elseif (preg_match("/^UTF-?8$/i", $INTERNAL_CHARSET) && preg_match("/^(ISO-8859-1|WINDOWS-1252)$/i", $charset)) { + $buf = utf8_encode($buf); + } else { + die("charset conversion function is not available."); + } + } + return $buf; +} + +function convertCharset($buf) +{ + global $LANGUAGE, $LANGUAGES, $INTERNAL_CHARSET; + + $charset = $LANGUAGES[$LANGUAGE]['preferred_charset']; + + if ($charset != $INTERNAL_CHARSET) { + if (function_exists("iconv")) { + $buf = iconv($INTERNAL_CHARSET, $charset, $buf); + } elseif (function_exists("mb_convert_encoding")) { + $buf = mb_convert_encoding($buf, $charset, $INTERNAL_CHARSET); + } elseif (preg_match("/^UTF-?8$/i", $INTERNAL_CHARSET) && preg_match("/^(ISO-8859-1|WINDOWS-1252)$/i", $charset)) { + $buf = utf8_decode($buf); + } else { + die("$LANGUAGE locale is not supported."); + } + } + return $buf; } // oneLiner() function end +// Returns the name of character set in the given document +function detectDocumentCharset($doc) +{ + if (preg_match("//iU", $doc, $reg)) { + return $reg[1]; + } + return false; +} ?>