wikipeg 4.0.2 → 6.0.0

package/lib/utils/objects.js

@@ -2,53 +2,23 @@
 
 /* Object utilities. */
 var objects = {
-  keys: function(object) {
-    var result = [], key;
-
-    for (key in object) {
-      if (object.hasOwnProperty(key)) {
-        result.push(key);
-      }
-    }
-
-    return result;
-  },
-
-  values: function(object) {
-    var result = [], key;
-
-    for (key in object) {
-      if (object.hasOwnProperty(key)) {
-        result.push(object[key]);
-      }
-    }
-
-    return result;
-  },
-
   clone: function(object) {
-    var result = {}, key;
+    var result = Object.create(null);
 
-    for (key in object) {
-      if (object.hasOwnProperty(key)) {
-        result[key] = object[key];
-      }
-    }
+    Object.getOwnPropertyNames(object).forEach((key)=>{
+      result[key] = object[key];
+    });
 
     return result;
   },
 
   defaults: function(object, defaults) {
-    var key;
-
-    for (key in defaults) {
-      if (defaults.hasOwnProperty(key)) {
-        if (!(key in object)) {
-          object[key] = defaults[key];
-        }
+    Object.getOwnPropertyNames(defaults).forEach((key)=>{
+      if (!(key in object)) {
+        object[key] = defaults[key];
       }
-    }
-  }
+    });
+  },
 };
 
 module.exports = objects;

package/lib/utils/unicode.js

@@ -0,0 +1,34 @@
+"use strict";
+
+const fs = require("fs"),
+    casefold = require("./casefold");
+
+// Raw case folding definitions from CaseFolding.txt
+const caseFoldDefs = casefold.parse(
+  fs.readFileSync(__dirname + "/CaseFolding.txt", "utf-8")
+);
+
+// Maps from a character to its canonical "case folded" version
+const simpleCaseFolding = [];
+
+// Maps from a canonical "case folded" character to all characters
+// which map to it.
+const reverseSimpleCaseFolding = [];
+
+// Compute simpleCaseFolding/reverseSimpleCaseFolding
+for (const def of caseFoldDefs) {
+  if (def.status === 'C' || def.status === 'S') {
+    const mapped = def.mapped[0];
+    simpleCaseFolding[def.code] = mapped;
+    if (reverseSimpleCaseFolding[mapped] === undefined) {
+      reverseSimpleCaseFolding[mapped] = [mapped];
+    }
+    reverseSimpleCaseFolding[mapped].push(def.code);
+  }
+}
+
+module.exports = {
+  defs: caseFoldDefs,
+  simpleCaseFolding,
+  reverseSimpleCaseFolding,
+};

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "wikipeg",
-	"version": "4.0.2",
+	"version": "6.0.0",
 	"description": "Parser generator for JavaScript and PHP",
 	"license": "MIT",
 	"homepage": "https://gerrit.wikimedia.org/r/plugins/gitiles/wikipeg/",
@@ -16,7 +16,7 @@
 		}
 	],
 	"files": [
-		"CHANGELOG.md",
+		"HISTORY.md",
 		"LICENSE",
 		"README.md",
 		"VERSION",
@@ -25,6 +25,7 @@
 		"examples/*.pegphp",
 		"lib/**/*.js",
 		"lib/**/*.php",
+		"lib/utils/CaseFolding.txt",
 		"src/**/*.php",
 		"tools/build-browser.js",
 		"package.json"
@@ -33,14 +34,15 @@
 	"bin": "bin/wikipeg",
 	"scripts": {
 		"eslint": "make eslint",
-		"test": "make eslint && make test"
+		"test": "make eslint && make test",
+		"rebuild": "make parser"
 	},
 	"repository": {
 		"type": "git",
 		"url": "https://gerrit.wikimedia.org/r/wikipeg"
 	},
 	"devDependencies": {
-		"eslint": "8.31.0",
+		"eslint": "8.57.0",
 		"jasmine-node": "3.0.0"
 	},
 	"engines": {

package/src/DefaultTracer.php

@@ -5,31 +5,31 @@ namespace Wikimedia\WikiPEG;
 use InvalidArgumentException;
 
 class DefaultTracer implements Tracer {
-	private $indentLevel = 0;
+	protected int $indentLevel = 0;
 
-	public function trace( $event ) {
+	public function trace( array $event ): void {
 		switch ( $event['type'] ) {
-		case 'rule.enter':
-			$this->log( $event );
-			$this->indentLevel++;
-			break;
+			case 'rule.enter':
+				$this->log( $event );
+				$this->indentLevel++;
+				break;
 
-		case 'rule.match':
-			$this->indentLevel--;
-			$this->log( $event );
-			break;
+			case 'rule.match':
+				$this->indentLevel--;
+				$this->log( $event );
+				break;
 
-		case 'rule.fail':
-			$this->indentLevel--;
-			$this->log( $event );
-			break;
+			case 'rule.fail':
+				$this->indentLevel--;
+				$this->log( $event );
+				break;
 
-		default:
-			throw new InvalidArgumentException( "Invalid event type {$event['type']}" );
+			default:
+				throw new InvalidArgumentException( "Invalid event type {$event['type']}" );
 		}
 	}
 
-	private function log( $event ) {
+	protected function log( array $event ) {
 		print str_pad(
 			'' . $event['location'],
 			20
@@ -40,7 +40,7 @@ class DefaultTracer implements Tracer {
 			. "\n";
 	}
 
-	private function formatArgs( $argMap ) {
+	protected function formatArgs( ?array $argMap ): string {
 		if ( !$argMap ) {
 			return '';
 		}

package/src/PEGParserBase.php

@@ -2,29 +2,31 @@
 
 namespace Wikimedia\WikiPEG;
 
+use stdClass;
+
 abstract class PEGParserBase {
-	protected static $FAILED;
-	protected static $UNDEFINED;
-	protected $currPos;
-	protected $savedPos;
-	protected $input;
-	protected $inputLength;
-	protected $options;
+	protected static ?stdClass $FAILED = null;
+	protected static ?stdClass $UNDEFINED = null;
+	protected int $currPos;
+	protected int $savedPos;
+	protected string $input;
+	protected int $inputLength;
+	protected array $options;
+	/** @var array */
 	protected $cache;
 
 	/** @var array<int,array{line:int,column:int,seenCR:bool}> */
-	protected $posDetailsCache;
-	protected $maxFailPos;
-	protected $maxFailExpected;
+	protected array $posDetailsCache;
+	protected int $maxFailPos;
+	protected array $maxFailExpected;
 
 	/** @var array Associative arrays of expectation info */
 	protected $expectations;
 
 	/** @var Expectation[] */
-	private $expectationCache;
+	private array $expectationCache;
 
-	/** @var Tracer */
-	protected $tracer;
+	protected Tracer $tracer;
 
 	public function __construct() {
 		if ( !self::$FAILED ) {
@@ -35,7 +37,8 @@ abstract class PEGParserBase {
 		}
 	}
 
-	protected function traceCall( $parseFunc, $name, $argNames, $args ) {
+	/** @return mixed */
+	protected function traceCall( callable $parseFunc, string $name, array $argNames, array $args ) {
 		$argMap = [];
 		foreach ( $args as $i => $argValue ) {
 			$argMap[$argNames[$i]] = $argValue;
@@ -47,7 +50,7 @@ abstract class PEGParserBase {
 			'location' => $this->computeLocation( $startPos, $startPos ),
 			'args' => $argMap
 		] );
-		$result = call_user_func_array( $parseFunc, $args );
+		$result = $parseFunc( ...$args );
 		if ( $result !== self::$FAILED ) {
 			$this->tracer->trace( [
 				'type' => 'rule.match',
@@ -65,17 +68,18 @@ abstract class PEGParserBase {
 		return $result;
 	}
 
-	protected function text() {
+	protected function text(): string {
 		return substr( $this->input, $this->savedPos, $this->currPos - $this->savedPos );
 	}
 
-	protected function location() {
+	protected function location(): LocationRange {
 		return $this->computeLocation( $this->savedPos, $this->currPos );
 	}
 
 	/**
 	 * @param string $description
 	 * @return never
+	 * @throws SyntaxError
 	 */
 	protected function expected( $description ) {
 		throw $this->buildException(
@@ -89,6 +93,7 @@ abstract class PEGParserBase {
 	/**
 	 * @param string $message
 	 * @return never
+	 * @throws SyntaxError
 	 */
 	protected function error( $message ) {
 		throw $this->buildException(
@@ -99,7 +104,7 @@ abstract class PEGParserBase {
 		);
 	}
 
-	public static function charAt( $s, $byteOffset ) {
+	public static function charAt( string $s, int $byteOffset ): string {
 		if ( !isset( $s[$byteOffset] ) ) {
 			return '';
 		}
@@ -117,7 +122,7 @@ abstract class PEGParserBase {
 		return $char;
 	}
 
-	public static function charsAt( $s, $byteOffset, $numChars ) {
+	public static function charsAt( string $s, int $byteOffset, int $numChars ): string {
 		$ret = '';
 		for ( $i = 0; $i < $numChars; $i++ ) {
 			$ret .= self::consumeChar( $s, $byteOffset );
@@ -125,7 +130,7 @@ abstract class PEGParserBase {
 		return $ret;
 	}
 
-	public static function consumeChar( $s, &$byteOffset ) {
+	public static function consumeChar( string $s, int &$byteOffset ): string {
 		if ( !isset( $s[$byteOffset] ) ) {
 			return '';
 		}
@@ -143,6 +148,22 @@ abstract class PEGParserBase {
 		return $char;
 	}
 
+	public static function advanceChar( string $s, int &$byteOffset ): void {
+		if ( !isset( $s[$byteOffset] ) ) {
+			return;
+		}
+		$byteOffset += match ( ord( $s[$byteOffset] ) & 0xf0 ) {
+			default => 1,
+			0xc0, 0xd0 => 2,
+			0xe0 => 3,
+			0xf0 => 4,
+		};
+	}
+
+	/**
+	 * @param mixed $value
+	 * @return mixed
+	 */
 	public static function &newRef( $value ) {
 		return $value;
 	}
@@ -165,7 +186,8 @@ abstract class PEGParserBase {
 		while ( $p < $pos ) {
 			$ch = self::charAt( $this->input, $p );
 			if ( $ch === "\n" ) {
-				if ( !$details['seenCR'] ) { $details['line']++;
+				if ( !$details['seenCR'] ) {
+					$details['line']++;
 				}
 				$details['column'] = 1;
 				$details['seenCR'] = false;
@@ -185,7 +207,7 @@ abstract class PEGParserBase {
 		return $details;
 	}
 
-	protected function computeLocation( $startPos, $endPos ) {
+	protected function computeLocation( int $startPos, int $endPos ): LocationRange {
 		if ( $endPos > $this->inputLength ) {
 			$endPos--;
 		}
@@ -202,7 +224,7 @@ abstract class PEGParserBase {
 		);
 	}
 
-	protected function fail( $expected ) {
+	protected function fail( int $expected ) {
 		if ( $this->currPos < $this->maxFailPos ) {
 			return;
 		}
@@ -234,7 +256,7 @@ abstract class PEGParserBase {
 		return $expanded;
 	}
 
-	private function buildMessage( $expected, $found ) {
+	private function buildMessage( array $expected, ?string $found ): string {
 		$expectedDescs = [];
 
 		foreach ( $expected as $info ) {
@@ -251,7 +273,9 @@ abstract class PEGParserBase {
 		return "Expected " . $expectedDesc . " but " . $foundDesc . " found.";
 	}
 
-	protected function buildException( $message, $expected, $found, $location ) {
+	protected function buildException(
+		?string $message, ?array $expected, ?string $found, LocationRange $location
+	): SyntaxError {
 		if ( $expected !== null ) {
 			sort( $expected );
 			$expected = array_unique( $expected );
@@ -271,7 +295,7 @@ abstract class PEGParserBase {
 		);
 	}
 
-	protected function buildParseException() {
+	protected function buildParseException(): SyntaxError {
 		$char = self::charAt( $this->input, $this->maxFailPos );
 		return $this->buildException(
 			null,
@@ -284,7 +308,7 @@ abstract class PEGParserBase {
 	protected function initialize() {
 	}
 
-	protected function initInternal( $input, $options ) {
+	protected function initInternal( string $input, array $options ) {
 		$this->currPos = 0;
 		$this->savedPos = 0;
 		$this->input = $input;
@@ -299,5 +323,6 @@ abstract class PEGParserBase {
 		$this->initialize();
 	}
 
-	abstract public function parse( $input, $options = [] );
+	/** @return mixed */
+	abstract public function parse( string $input, array $options = [] );
 }

package/src/SyntaxError.php

@@ -3,9 +3,9 @@
 namespace Wikimedia\WikiPEG;
 
 class SyntaxError extends \Exception implements \JsonSerializable {
-	public $expected;
-	public $found;
-	public $location;
+	public array $expected;
+	public ?string $found;
+	public LocationRange $location;
 
 	/**
 	 * @param string $message
@@ -13,7 +13,7 @@ class SyntaxError extends \Exception implements \JsonSerializable {
 	 * @param string|null $found
 	 * @param LocationRange $location
 	 */
-	public function __construct( string $message, array $expected, $found, LocationRange $location ) {
+	public function __construct( string $message, array $expected, ?string $found, LocationRange $location ) {
 		parent::__construct( $message );
 		$this->expected = $expected;
 		$this->found    = $found;

package/src/Tracer.php

@@ -3,5 +3,5 @@
 namespace Wikimedia\WikiPEG;
 
 interface Tracer {
-	public function trace( $event );
+	public function trace( array $event ): void;
 }

package/lib/compiler/opcodes.js

@@ -1,54 +0,0 @@
-"use strict";
-
-/* Bytecode instruction opcodes. */
-var opcodes = {
-  /* Stack Manipulation */
-
-  PUSH:             0,    // PUSH c
-  PUSH_UNDEFINED:   26,   // PUSH_UNDEFINED
-  PUSH_NULL:        27,   // PUSH_NULL
-  PUSH_FAILED:      28,   // PUSH_FAILED
-  PUSH_EMPTY_ARRAY: 29,   // PUSH_EMPTY_ARRAY
-  PUSH_CURR_POS:    1,    // PUSH_CURR_POS
-  POP:              2,    // POP
-  POP_CURR_POS:     3,    // POP_CURR_POS
-  POP_N:            4,    // POP_N n
-  NIP:              5,    // NIP
-  APPEND:           6,    // APPEND
-  WRAP:             7,    // WRAP n
-  TEXT:             8,    // TEXT
-
-  /* Conditions and Loops */
-
-  IF:               9,    // IF t, f
-  IF_ERROR:         10,   // IF_ERROR t, f
-  IF_NOT_ERROR:     11,   // IF_NOT_ERROR t, f
-  WHILE_NOT_ERROR:  12,   // WHILE_NOT_ERROR b
-
-  /* Matching */
-
-  MATCH_ANY:        13,   // MATCH_ANY a, f, ...
-  MATCH_STRING:     14,   // MATCH_STRING s, a, f, ...
-  MATCH_STRING_IC:  15,   // MATCH_STRING_IC s, a, f, ...
-  MATCH_REGEXP:     16,   // MATCH_REGEXP r, a, f, ...
-  ACCEPT_N:         17,   // ACCEPT_N n
-  ACCEPT_STRING:    18,   // ACCEPT_STRING s
-  FAIL:             19,   // FAIL e
-
-  /* Calls */
-
-  LOAD_SAVED_POS:   20,   // LOAD_SAVED_POS p
-  UPDATE_SAVED_POS: 21,   // UPDATE_SAVED_POS
-  CALL:             22,   // CALL f, n, pc, p1, p2, ..., pN
-
-  /* Rules */
-
-  RULE:             23,   // RULE r
-
-  /* Failure Reporting */
-
-  SILENT_FAILS_ON:  24,   // SILENT_FAILS_ON
-  SILENT_FAILS_OFF: 25    // SILENT_FAILS_FF
-};
-
-module.exports = opcodes;