Chris@0: <?php
Chris@0: 
Chris@0: $grammarFileToName = [
Chris@0:     __DIR__ . '/php5.y' => 'Php5',
Chris@0:     __DIR__ . '/php7.y' => 'Php7',
Chris@0: ];
Chris@0: 
Chris@0: $tokensFile     = __DIR__ . '/tokens.y';
Chris@0: $tokensTemplate = __DIR__ . '/tokens.template';
Chris@0: $skeletonFile   = __DIR__ . '/parser.template';
Chris@0: $tmpGrammarFile = __DIR__ . '/tmp_parser.phpy';
Chris@0: $tmpResultFile  = __DIR__ . '/tmp_parser.php';
Chris@0: $resultDir = __DIR__ . '/../lib/PhpParser/Parser';
Chris@0: $tokensResultsFile = $resultDir . '/Tokens.php';
Chris@0: 
Chris@0: // check for kmyacc.exe binary in this directory, otherwise fall back to global name
Chris@0: $kmyacc = __DIR__ . '/kmyacc.exe';
Chris@0: if (!file_exists($kmyacc)) {
Chris@0:     $kmyacc = 'kmyacc';
Chris@0: }
Chris@0: 
Chris@0: $options = array_flip($argv);
Chris@0: $optionDebug = isset($options['--debug']);
Chris@0: $optionKeepTmpGrammar = isset($options['--keep-tmp-grammar']);
Chris@0: 
Chris@0: ///////////////////////////////
Chris@0: /// Utility regex constants ///
Chris@0: ///////////////////////////////
Chris@0: 
Chris@0: const LIB = '(?(DEFINE)
Chris@0:     (?<singleQuotedString>\'[^\\\\\']*+(?:\\\\.[^\\\\\']*+)*+\')
Chris@0:     (?<doubleQuotedString>"[^\\\\"]*+(?:\\\\.[^\\\\"]*+)*+")
Chris@0:     (?<string>(?&singleQuotedString)|(?&doubleQuotedString))
Chris@0:     (?<comment>/\*[^*]*+(?:\*(?!/)[^*]*+)*+\*/)
Chris@0:     (?<code>\{[^\'"/{}]*+(?:(?:(?&string)|(?&comment)|(?&code)|/)[^\'"/{}]*+)*+})
Chris@0: )';
Chris@0: 
Chris@0: const PARAMS = '\[(?<params>[^[\]]*+(?:\[(?&params)\][^[\]]*+)*+)\]';
Chris@0: const ARGS   = '\((?<args>[^()]*+(?:\((?&args)\)[^()]*+)*+)\)';
Chris@0: 
Chris@0: ///////////////////
Chris@0: /// Main script ///
Chris@0: ///////////////////
Chris@0: 
Chris@0: $tokens = file_get_contents($tokensFile);
Chris@0: 
Chris@0: foreach ($grammarFileToName as $grammarFile => $name) {
Chris@0:     echo "Building temporary $name grammar file.\n";
Chris@0: 
Chris@0:     $grammarCode = file_get_contents($grammarFile);
Chris@0:     $grammarCode = str_replace('%tokens', $tokens, $grammarCode);
Chris@0: 
Chris@0:     $grammarCode = resolveNodes($grammarCode);
Chris@0:     $grammarCode = resolveMacros($grammarCode);
Chris@0:     $grammarCode = resolveStackAccess($grammarCode);
Chris@0: 
Chris@0:     file_put_contents($tmpGrammarFile, $grammarCode);
Chris@0: 
Chris@0:     $additionalArgs = $optionDebug ? '-t -v' : '';
Chris@0: 
Chris@0:     echo "Building $name parser.\n";
Chris@0:     $output = trim(shell_exec("$kmyacc $additionalArgs -l -m $skeletonFile -p $name $tmpGrammarFile 2>&1"));
Chris@0:     echo "Output: \"$output\"\n";
Chris@0: 
Chris@0:     $resultCode = file_get_contents($tmpResultFile);
Chris@0:     $resultCode = removeTrailingWhitespace($resultCode);
Chris@0: 
Chris@0:     ensureDirExists($resultDir);
Chris@0:     file_put_contents("$resultDir/$name.php", $resultCode);
Chris@0:     unlink($tmpResultFile);
Chris@0: 
Chris@0:     echo "Building token definition.\n";
Chris@0:     $output = trim(shell_exec("$kmyacc -l -m $tokensTemplate $tmpGrammarFile 2>&1"));
Chris@0:     assert($output === '');
Chris@0:     rename($tmpResultFile, $tokensResultsFile);
Chris@0: 
Chris@0:     if (!$optionKeepTmpGrammar) {
Chris@0:         unlink($tmpGrammarFile);
Chris@0:     }
Chris@0: }
Chris@0: 
Chris@0: ///////////////////////////////
Chris@0: /// Preprocessing functions ///
Chris@0: ///////////////////////////////
Chris@0: 
Chris@0: function resolveNodes($code) {
Chris@0:     return preg_replace_callback(
Chris@0:         '~\b(?<name>[A-Z][a-zA-Z_\\\\]++)\s*' . PARAMS . '~',
Chris@0:         function($matches) {
Chris@0:             // recurse
Chris@0:             $matches['params'] = resolveNodes($matches['params']);
Chris@0: 
Chris@0:             $params = magicSplit(
Chris@0:                 '(?:' . PARAMS . '|' . ARGS . ')(*SKIP)(*FAIL)|,',
Chris@0:                 $matches['params']
Chris@0:             );
Chris@0: 
Chris@0:             $paramCode = '';
Chris@0:             foreach ($params as $param) {
Chris@0:                 $paramCode .= $param . ', ';
Chris@0:             }
Chris@0: 
Chris@0:             return 'new ' . $matches['name'] . '(' . $paramCode . 'attributes())';
Chris@0:         },
Chris@0:         $code
Chris@0:     );
Chris@0: }
Chris@0: 
Chris@0: function resolveMacros($code) {
Chris@0:     return preg_replace_callback(
Chris@0:         '~\b(?<!::|->)(?!array\()(?<name>[a-z][A-Za-z]++)' . ARGS . '~',
Chris@0:         function($matches) {
Chris@0:             // recurse
Chris@0:             $matches['args'] = resolveMacros($matches['args']);
Chris@0: 
Chris@0:             $name = $matches['name'];
Chris@0:             $args = magicSplit(
Chris@0:                 '(?:' . PARAMS . '|' . ARGS . ')(*SKIP)(*FAIL)|,',
Chris@0:                 $matches['args']
Chris@0:             );
Chris@0: 
Chris@0:             if ('attributes' == $name) {
Chris@0:                 assertArgs(0, $args, $name);
Chris@0:                 return '$this->startAttributeStack[#1] + $this->endAttributes';
Chris@0:             }
Chris@0: 
Chris@0:             if ('stackAttributes' == $name) {
Chris@0:                 assertArgs(1, $args, $name);
Chris@0:                 return '$this->startAttributeStack[' . $args[0] . ']'
Chris@0:                      . ' + $this->endAttributeStack[' . $args[0] . ']';
Chris@0:             }
Chris@0: 
Chris@0:             if ('init' == $name) {
Chris@0:                 return '$$ = array(' . implode(', ', $args) . ')';
Chris@0:             }
Chris@0: 
Chris@0:             if ('push' == $name) {
Chris@0:                 assertArgs(2, $args, $name);
Chris@0: 
Chris@0:                 return $args[0] . '[] = ' . $args[1] . '; $$ = ' . $args[0];
Chris@0:             }
Chris@0: 
Chris@0:             if ('pushNormalizing' == $name) {
Chris@0:                 assertArgs(2, $args, $name);
Chris@0: 
Chris@0:                 return 'if (is_array(' . $args[1] . ')) { $$ = array_merge(' . $args[0] . ', ' . $args[1] . '); }'
Chris@0:                      . ' else { ' . $args[0] . '[] = ' . $args[1] . '; $$ = ' . $args[0] . '; }';
Chris@0:             }
Chris@0: 
Chris@0:             if ('toArray' == $name) {
Chris@0:                 assertArgs(1, $args, $name);
Chris@0: 
Chris@0:                 return 'is_array(' . $args[0] . ') ? ' . $args[0] . ' : array(' . $args[0] . ')';
Chris@0:             }
Chris@0: 
Chris@0:             if ('parseVar' == $name) {
Chris@0:                 assertArgs(1, $args, $name);
Chris@0: 
Chris@0:                 return 'substr(' . $args[0] . ', 1)';
Chris@0:             }
Chris@0: 
Chris@0:             if ('parseEncapsed' == $name) {
Chris@0:                 assertArgs(3, $args, $name);
Chris@0: 
Chris@0:                 return 'foreach (' . $args[0] . ' as $s) { if ($s instanceof Node\Scalar\EncapsedStringPart) {'
Chris@0:                      . ' $s->value = Node\Scalar\String_::parseEscapeSequences($s->value, ' . $args[1] . ', ' . $args[2] . '); } }';
Chris@0:             }
Chris@0: 
Chris@0:             if ('makeNop' == $name) {
Chris@13:                 assertArgs(3, $args, $name);
Chris@0: 
Chris@0:                 return '$startAttributes = ' . $args[1] . ';'
Chris@0:                 . ' if (isset($startAttributes[\'comments\']))'
Chris@13:                 . ' { ' . $args[0] . ' = new Stmt\Nop($startAttributes + ' . $args[2] . '); }'
Chris@0:                 . ' else { ' . $args[0] . ' = null; }';
Chris@0:             }
Chris@0: 
Chris@0:             if ('strKind' == $name) {
Chris@0:                 assertArgs(1, $args, $name);
Chris@0: 
Chris@0:                 return '(' . $args[0] . '[0] === "\'" || (' . $args[0] . '[1] === "\'" && '
Chris@0:                      . '(' . $args[0] . '[0] === \'b\' || ' . $args[0] . '[0] === \'B\')) '
Chris@0:                      . '? Scalar\String_::KIND_SINGLE_QUOTED : Scalar\String_::KIND_DOUBLE_QUOTED)';
Chris@0:             }
Chris@0: 
Chris@0:             if ('prependLeadingComments' == $name) {
Chris@0:                 assertArgs(1, $args, $name);
Chris@0: 
Chris@0:                 return '$attrs = $this->startAttributeStack[#1]; $stmts = ' . $args[0] . '; '
Chris@0:                 . 'if (!empty($attrs[\'comments\'])) {'
Chris@0:                 . '$stmts[0]->setAttribute(\'comments\', '
Chris@0:                 . 'array_merge($attrs[\'comments\'], $stmts[0]->getAttribute(\'comments\', []))); }';
Chris@0:             }
Chris@0: 
Chris@0:             return $matches[0];
Chris@0:         },
Chris@0:         $code
Chris@0:     );
Chris@0: }
Chris@0: 
Chris@0: function assertArgs($num, $args, $name) {
Chris@0:     if ($num != count($args)) {
Chris@0:         die('Wrong argument count for ' . $name . '().');
Chris@0:     }
Chris@0: }
Chris@0: 
Chris@0: function resolveStackAccess($code) {
Chris@0:     $code = preg_replace('/\$\d+/', '$this->semStack[$0]', $code);
Chris@0:     $code = preg_replace('/#(\d+)/', '$$1', $code);
Chris@0:     return $code;
Chris@0: }
Chris@0: 
Chris@0: function removeTrailingWhitespace($code) {
Chris@0:     $lines = explode("\n", $code);
Chris@0:     $lines = array_map('rtrim', $lines);
Chris@0:     return implode("\n", $lines);
Chris@0: }
Chris@0: 
Chris@0: function ensureDirExists($dir) {
Chris@0:     if (!is_dir($dir)) {
Chris@0:         mkdir($dir, 0777, true);
Chris@0:     }
Chris@0: }
Chris@0: 
Chris@0: //////////////////////////////
Chris@0: /// Regex helper functions ///
Chris@0: //////////////////////////////
Chris@0: 
Chris@0: function regex($regex) {
Chris@0:     return '~' . LIB . '(?:' . str_replace('~', '\~', $regex) . ')~';
Chris@0: }
Chris@0: 
Chris@0: function magicSplit($regex, $string) {
Chris@0:     $pieces = preg_split(regex('(?:(?&string)|(?&comment)|(?&code))(*SKIP)(*FAIL)|' . $regex), $string);
Chris@0: 
Chris@0:     foreach ($pieces as &$piece) {
Chris@0:         $piece = trim($piece);
Chris@0:     }
Chris@0: 
Chris@0:     if ($pieces === ['']) {
Chris@0:         return [];
Chris@0:     }
Chris@0: 
Chris@0:     return $pieces;
Chris@0: }