Chris@0: #!/usr/bin/env php
Chris@0: <?php
Chris@0: 
Chris@0: /*
Chris@0:  * This file is part of Psy Shell.
Chris@0:  *
Chris@0:  * (c) 2012-2017 Justin Hileman
Chris@0:  *
Chris@0:  * For the full copyright and license information, please view the LICENSE
Chris@0:  * file that was distributed with this source code.
Chris@0:  */
Chris@0: 
Chris@0: define('WRAP_WIDTH', 100);
Chris@0: 
Chris@0: $count = 0;
Chris@0: 
Chris@0: if (count($argv) !== 3 || !is_dir($argv[1])) {
Chris@0:     echo "usage: build_manual path/to/manual output_filename.db\n";
Chris@0:     exit(1);
Chris@0: }
Chris@0: 
Chris@0: function htmlwrap($text, $width = null)
Chris@0: {
Chris@0:     if ($width === null) {
Chris@0:         $width = WRAP_WIDTH;
Chris@0:     }
Chris@0: 
Chris@0:     $len = strlen($text);
Chris@0: 
Chris@0:     $return = array();
Chris@0:     $lastSpace = null;
Chris@0:     $inTag = false;
Chris@0:     $i = $tagWidth = 0;
Chris@0:     do {
Chris@0:         switch (substr($text, $i, 1)) {
Chris@0:             case "\n":
Chris@0:                 $return[] = trim(substr($text, 0, $i));
Chris@0:                 $text     = substr($text, $i);
Chris@0:                 $len      = strlen($text);
Chris@0: 
Chris@0:                 $i = $lastSpace = 0;
Chris@0:                 continue;
Chris@0: 
Chris@0:             case ' ':
Chris@0:                 if (!$inTag) {
Chris@0:                     $lastSpace = $i;
Chris@0:                 }
Chris@0:                 break;
Chris@0: 
Chris@0:             case '<':
Chris@0:                 $inTag = true;
Chris@0:                 break;
Chris@0: 
Chris@0:             case '>':
Chris@0:                 $inTag = false;
Chris@0: 
Chris@0:             default:
Chris@0:         }
Chris@0: 
Chris@0:         if ($inTag) {
Chris@0:             $tagWidth++;
Chris@0:         }
Chris@0: 
Chris@0:         $i++;
Chris@0: 
Chris@0:         if (!$inTag && ($i - $tagWidth > $width)) {
Chris@0:             $lastSpace = $lastSpace ?: $width;
Chris@0: 
Chris@0:             $return[] = trim(substr($text, 0, $lastSpace));
Chris@0:             $text     = substr($text, $lastSpace);
Chris@0:             $len      = strlen($text);
Chris@0: 
Chris@0:             $i = $tagWidth = 0;
Chris@0:         }
Chris@0:     } while ($i < $len);
Chris@0: 
Chris@0:     $return[] = trim($text);
Chris@0: 
Chris@0:     return implode("\n", $return);
Chris@0: }
Chris@0: 
Chris@0: function extract_paragraphs($element)
Chris@0: {
Chris@0:     $paragraphs = array();
Chris@0:     foreach ($element->getElementsByTagName('para') as $p) {
Chris@0:         $text = '';
Chris@0:         foreach ($p->childNodes as $child) {
Chris@0:             // @todo figure out if there's something we can do with tables.
Chris@0:             if ($child instanceof DOMElement && $child->tagName === 'table') {
Chris@0:                 continue;
Chris@0:             }
Chris@0: 
Chris@0:             // skip references, because ugh.
Chris@0:             if (preg_match('{^\s*&[a-z][a-z\.]+;\s*$}', $child->textContent)) {
Chris@0:                 continue;
Chris@0:             }
Chris@0: 
Chris@0:             $text .= $child->ownerDocument->saveXML($child);
Chris@0:         }
Chris@0: 
Chris@0:         if ($text = trim(preg_replace('{\n[ \t]+}', ' ', $text))) {
Chris@0:             $paragraphs[] = $text;
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     return implode("\n\n", $paragraphs);
Chris@0: }
Chris@0: 
Chris@0: function format_doc($doc)
Chris@0: {
Chris@0:     $chunks   = array();
Chris@0: 
Chris@0:     if (!empty($doc['description'])) {
Chris@0:         $chunks[] = '<comment>Description:</comment>';
Chris@0:         $chunks[] = indent_text(htmlwrap(thunk_tags($doc['description']), WRAP_WIDTH - 2));
Chris@0:         $chunks[] = '';
Chris@0:     }
Chris@0: 
Chris@0:     if (!empty($doc['params'])) {
Chris@0:         $chunks[] = '<comment>Param:</comment>';
Chris@0: 
Chris@0:         $typeMax = max(array_map(function ($param) {
Chris@0:             return strlen($param['type']);
Chris@0:         }, $doc['params']));
Chris@0: 
Chris@0:         $max = max(array_map(function ($param) {
Chris@0:             return strlen($param['name']);
Chris@0:         }, $doc['params']));
Chris@0: 
Chris@0:         $template  = '  <info>%-' . $typeMax . 's</info>  <strong>%-' . $max . 's</strong>  %s';
Chris@0:         $indent    = str_repeat(' ', $typeMax + $max + 6);
Chris@0:         $wrapWidth = WRAP_WIDTH - strlen($indent);
Chris@0: 
Chris@0:         foreach ($doc['params'] as $param) {
Chris@0:             $desc = indent_text(htmlwrap(thunk_tags($param['description']), $wrapWidth), $indent, false);
Chris@0:             $chunks[] = sprintf($template, $param['type'], $param['name'], $desc);
Chris@0:         }
Chris@0:         $chunks[] = '';
Chris@0:     }
Chris@0: 
Chris@0:     if (isset($doc['return']) || isset($doc['return_type'])) {
Chris@0:         $chunks[] = '<comment>Return:</comment>';
Chris@0: 
Chris@0:         $type   = isset($doc['return_type']) ? $doc['return_type'] : 'unknown';
Chris@0:         $desc   = isset($doc['return']) ? $doc['return'] : '';
Chris@0: 
Chris@0:         $indent    = str_repeat(' ', strlen($type) + 4);
Chris@0:         $wrapWidth = WRAP_WIDTH - strlen($indent);
Chris@0: 
Chris@0:         if (!empty($desc)) {
Chris@0:             $desc = indent_text(htmlwrap(thunk_tags($doc['return']), $wrapWidth), $indent, false);
Chris@0:         }
Chris@0: 
Chris@0:         $chunks[] = sprintf('  <info>%s</info>  %s', $type, $desc);
Chris@0:         $chunks[] = '';
Chris@0:     }
Chris@0: 
Chris@0:     array_pop($chunks); // get rid of the trailing newline
Chris@0: 
Chris@0:     return implode("\n", $chunks);
Chris@0: }
Chris@0: 
Chris@0: function thunk_tags($text)
Chris@0: {
Chris@0:     $tagMap = array(
Chris@0:         'parameter>' => 'strong>',
Chris@0:         'function>'  => 'strong>',
Chris@0:         'literal>'   => 'return>',
Chris@0:         'type>'      => 'info>',
Chris@0:         'constant>'  => 'info>',
Chris@0:     );
Chris@0: 
Chris@0:     $andBack = array(
Chris@0:         '&amp;'       => '&',
Chris@0:         '&amp;true;'  => '<return>true</return>',
Chris@0:         '&amp;false;' => '<return>false</return>',
Chris@0:         '&amp;null;'  => '<return>null</return>',
Chris@0:     );
Chris@0: 
Chris@0:     return strtr(strip_tags(strtr($text, $tagMap), '<strong><return><info>'), $andBack);
Chris@0: }
Chris@0: 
Chris@0: function indent_text($text, $indent = '  ', $leading = true)
Chris@0: {
Chris@0:     return ($leading ? $indent : '') . str_replace("\n", "\n" . $indent, $text);
Chris@0: }
Chris@0: 
Chris@0: function find_type($xml, $paramName)
Chris@0: {
Chris@0:     foreach ($xml->getElementsByTagName('methodparam') as $param) {
Chris@0:         if ($type = $param->getElementsByTagName('type')->item(0)) {
Chris@0:             if ($parameter = $param->getElementsByTagName('parameter')->item(0)) {
Chris@0:                 if ($paramName === $parameter->textContent) {
Chris@0:                     return $type->textContent;
Chris@0:                 }
Chris@0:             }
Chris@0:         }
Chris@0:     }
Chris@0: }
Chris@0: 
Chris@0: function format_function_doc($xml)
Chris@0: {
Chris@0:     $doc = array();
Chris@0:     $refsect1s = $xml->getElementsByTagName('refsect1');
Chris@0:     foreach ($refsect1s as $refsect1) {
Chris@0:         $role = $refsect1->getAttribute('role');
Chris@0:         switch ($role) {
Chris@0:             case 'description':
Chris@0:                 $doc['description'] = extract_paragraphs($refsect1);
Chris@0: 
Chris@0:                 if ($synopsis = $refsect1->getElementsByTagName('methodsynopsis')->item(0)) {
Chris@0:                     foreach ($synopsis->childNodes as $node) {
Chris@0:                         if ($node instanceof DOMElement && $node->tagName === 'type') {
Chris@0:                             $doc['return_type'] = $node->textContent;
Chris@0:                             break;
Chris@0:                         }
Chris@0:                     }
Chris@0:                 }
Chris@0:                 break;
Chris@0: 
Chris@0:             case 'returnvalues':
Chris@0:                 // do nothing.
Chris@0:                 $doc['return'] = extract_paragraphs($refsect1);
Chris@0:                 break;
Chris@0: 
Chris@0:             case 'parameters':
Chris@0:                 $params = array();
Chris@0:                 $vars = $refsect1->getElementsByTagName('varlistentry');
Chris@0:                 foreach ($vars as $var) {
Chris@0:                     if ($name = $var->getElementsByTagName('parameter')->item(0)) {
Chris@0:                         $params[] = array(
Chris@0:                             'name'        => '$' . $name->textContent,
Chris@0:                             'type'        => find_type($xml, $name->textContent),
Chris@0:                             'description' => extract_paragraphs($var),
Chris@0:                         );
Chris@0:                     }
Chris@0:                 }
Chris@0: 
Chris@0:                 $doc['params'] = $params;
Chris@0:                 break;
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     // and the purpose
Chris@0:     if ($purpose = $xml->getElementsByTagName('refpurpose')->item(0)) {
Chris@0:         $desc = htmlwrap($purpose->textContent);
Chris@0:         if (isset($doc['description'])) {
Chris@0:             $desc .= "\n\n" . $doc['description'];
Chris@0:         }
Chris@0: 
Chris@0:         $doc['description'] = trim($desc);
Chris@0:     }
Chris@0: 
Chris@0:     $ids = array();
Chris@0:     foreach ($xml->getElementsByTagName('refname') as $ref) {
Chris@0:         $ids[] = $ref->textContent;
Chris@0:     }
Chris@0: 
Chris@0:     return array($ids, format_doc($doc));
Chris@0: }
Chris@0: 
Chris@0: function format_class_doc($xml)
Chris@0: {
Chris@0:     // @todo implement this
Chris@0:     return array(array(), null);
Chris@0: }
Chris@0: 
Chris@0: $dir = new RecursiveDirectoryIterator($argv[1]);
Chris@0: $filter = new RecursiveCallbackFilterIterator($dir, function ($current, $key, $iterator) {
Chris@0:     return $current->getFilename()[0] !== '.' &&
Chris@0:         ($current->isDir() || $current->getExtension() === 'xml') &&
Chris@0:         strpos($current->getFilename(), 'entities.') !== 0 &&
Chris@0:         $current->getFilename() !== 'pdo_4d'; // Temporarily blacklist this one, the docs are weird.
Chris@0: });
Chris@0: $iterator = new RecursiveIteratorIterator($filter);
Chris@0: 
Chris@0: $docs = array();
Chris@0: foreach ($iterator as $file) {
Chris@0:     $xmlstr = str_replace('&', '&amp;', file_get_contents($file));
Chris@0: 
Chris@0:     $xml = new DOMDocument();
Chris@0:     $xml->preserveWhiteSpace = false;
Chris@0: 
Chris@0:     if (!@$xml->loadXml($xmlstr)) {
Chris@0:         echo "XML Parse Error: $file\n";
Chris@0:         continue;
Chris@0:     }
Chris@0: 
Chris@0:     if ($xml->getElementsByTagName('refentry')->length !== 0) {
Chris@0:         list($ids, $doc) = format_function_doc($xml);
Chris@0:     } elseif ($xml->getElementsByTagName('classref')->length !== 0) {
Chris@0:         list($ids, $doc) = format_class_doc($xml);
Chris@0:     } else {
Chris@0:         $ids = array();
Chris@0:         $doc = null;
Chris@0:     }
Chris@0: 
Chris@0:     foreach ($ids as $id) {
Chris@0:         $docs[$id] = $doc;
Chris@0:     }
Chris@0: }
Chris@0: 
Chris@0: if (is_file($argv[2])) {
Chris@0:     unlink($argv[2]);
Chris@0: }
Chris@0: 
Chris@0: $db = new PDO('sqlite:' . $argv[2]);
Chris@0: 
Chris@0: $db->query('CREATE TABLE php_manual (id char(256) PRIMARY KEY, doc TEXT)');
Chris@0: $cmd = $db->prepare('INSERT INTO php_manual (id, doc) VALUES (?, ?)');
Chris@0: foreach ($docs as $id => $doc) {
Chris@0:     $cmd->execute(array($id, $doc));
Chris@0: }