diff --git a/scripts/symbols/generate_ctags_symbols.php b/scripts/symbols/generate_ctags_symbols.php index f9a65d00ea..a665d6082a 100755 --- a/scripts/symbols/generate_ctags_symbols.php +++ b/scripts/symbols/generate_ctags_symbols.php @@ -1,130 +1,135 @@ #!/usr/bin/env php limit(8) as $file => $future) { $tags = $future->resolve(); $tags = explode("\n", $tags[1]); foreach ($tags as $tag) { $parts = explode(";", $tag); // skip lines that we can not parse if (count($parts) < 2) { continue; } // split ctags information $tag_info = explode("\t", $parts[0]); // split exuberant ctags "extension fields" (additional information) $parts[1] = trim($parts[1], "\t \""); $extension_fields = explode("\t", $parts[1]); // skip lines that we can not parse if (count($tag_info) < 3 || count($extension_fields) < 2) { continue; } + // default $context to empty + $extension_fields[] = ''; list($token, $file_path, $line_num) = $tag_info; - list($type, $language) = $extension_fields; + list($type, $language, $context) = $extension_fields; // strip "language:" $language = substr($language, 9); // To keep consistent with "Separate with commas, for example: php, py" // in Arcanist Project edit form. $language = str_ireplace("python", "py", $language); // also, "normalize" c++ and c# $language = str_ireplace("c++", "cpp", $language); $language = str_ireplace("c#", "csharp", $language); - switch ($type) { - case 'class': - print_symbol($file_path, $line_num, 'class', $token, $language); - break; - case 'function': - print_symbol($file_path, $line_num, 'function', $token, $language); - break; - default: + // Ruby has "singleton method", for example + $type = substr(str_replace(' ', '_', $type), 0, 12); + // class:foo, struct:foo, union:foo, enum:foo, ... + $context = last(explode(':', $context, 2)); + + $ignore = array( + 'variable' => true, + ); + if (empty($ignore[$type])) { + print_symbol($file_path, $line_num, $type, $token, $context, $language); } } } function ctags_get_parser_future($file_path) { - $future = new ExecFuture('ctags -n --fields=Kl -o - %s', + $future = new ExecFuture('ctags -n --fields=Kls -o - %s', $file_path); return $future; } function ctags_check_executable() { $future = new ExecFuture('ctags --version'); $result = $future->resolve(); if (empty($result[1])) { return false; } return true; } -function print_symbol($file, $line_num, $type, $token, $language) { +function print_symbol($file, $line_num, $type, $token, $context, $language) { // get rid of relative path $file = explode('/', $file); if ($file[0] == '.' || $file[0] == "..") { array_shift($file); } $file = '/' . implode('/', $file); $parts = array( + $context, $token, $type, strtolower($language), $line_num, $file, ); echo implode(' ', $parts)."\n"; } diff --git a/scripts/symbols/generate_php_symbols.php b/scripts/symbols/generate_php_symbols.php index 39125bea71..7be7a7c553 100755 --- a/scripts/symbols/generate_php_symbols.php +++ b/scripts/symbols/generate_php_symbols.php @@ -1,77 +1,124 @@ #!/usr/bin/env php limit(8) as $file => $future) { $tree = XHPASTTree::newFromDataAndResolvedExecFuture( $data[$file], $future->resolve()); $root = $tree->getRootNode(); + $scopes = array(); $functions = $root->selectDescendantsOfType('n_FUNCTION_DECLARATION'); foreach ($functions as $function) { $name = $function->getChildByIndex(2); print_symbol($file, 'function', $name); } $classes = $root->selectDescendantsOfType('n_CLASS_DECLARATION'); foreach ($classes as $class) { $class_name = $class->getChildByIndex(1); print_symbol($file, 'class', $class_name); + $scopes[] = array($class, $class_name); } $interfaces = $root->selectDescendantsOfType('n_INTERFACE_DECLARATION'); foreach ($interfaces as $interface) { $interface_name = $interface->getChildByIndex(1); print_symbol($file, 'interface', $interface_name); + $scopes[] = array($interface, $interface_name); + } + + $constants = $root->selectDescendantsOfType('n_CONSTANT_DECLARATION_LIST'); + foreach ($constants as $constant_list) { + foreach ($constant_list->getChildren() as $constant) { + $constant_name = $constant->getChildByIndex(0); + print_symbol($file, 'constant', $constant_name); + } + } + + foreach ($scopes as $scope) { + // this prints duplicate symbols in the case of nested classes + // luckily, PHP doesn't allow those + list($class, $class_name) = $scope; + + $consts = $class->selectDescendantsOfType( + 'n_CLASS_CONSTANT_DECLARATION_LIST'); + foreach ($consts as $const_list) { + foreach ($const_list->getChildren() as $const) { + $const_name = $const->getChildByIndex(0); + print_symbol($file, 'class_const', $const_name, $class_name); + } + } + + $members = $class->selectDescendantsOfType( + 'n_CLASS_MEMBER_DECLARATION_LIST'); + foreach ($members as $member_list) { + foreach ($member_list->getChildren() as $member) { + if ($member->getTypeName() == 'n_CLASS_MEMBER_MODIFIER_LIST') { + continue; + } + $member_name = $member->getChildByIndex(0); + print_symbol($file, 'member', $member_name, $class_name); + } + } + + $methods = $class->selectDescendantsOfType('n_METHOD_DECLARATION'); + foreach ($methods as $method) { + $method_name = $method->getChildByIndex(2); + print_symbol($file, 'method', $method_name, $class_name); + } } } -function print_symbol($file, $type, $token) { +function print_symbol($file, $type, $token, $context=null) { $parts = array( - $token->getConcreteString(), + $context ? $context->getConcreteString() : '', + // variable tokens are `$name`, not just `name`, so strip the $ off of + // class field names + ltrim($token->getConcreteString(), '$'), $type, 'php', $token->getLineNumber(), '/'.ltrim($file, './'), ); echo implode(' ', $parts)."\n"; }