MensBeam
/
Lit
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
Browse Source

Started playing around with a VSCode-style parser

main
Dustin Wilson 3 years ago
parent
86b96565f5
commit
38504d13ce
2 changed files with 103 additions and 342 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 405
      lib/Scope/Parser.php
  2. 40
      lib/Scope/Tokenizer.php

405
lib/Scope/Parser.php
View File

@ -7,392 +7,113 @@ declare(strict_types=1);
namespace dW\Highlighter\Scope;
class Parser {
public static $debug = false;
protected Data $data;
protected int $debugCount = 1;
protected array $lastExceptionData = [];
protected string $token;
protected Tokenizer $tokenizer;
protected static Parser $instance;
protected function __construct(string $selector) {
$this->data = new Data($selector);
$this->tokenizer = new Tokenizer($selector);
}
public static function parse(string $selector): Matcher|false {
public static function parse(string $selector): array {
self::$instance = new self($selector);
$result = false;
self::parseSpace();
$s2 = self::parseSelector();
if ($s2 !== false) {
self::parseSpace();
$result = $s2;
}
if ($result === false && self::$instance->lastExceptionData !== []) {
throw new Exception(self::$instance->lastExceptionData['expected'], self::$instance->lastExceptionData['found']);
}
return $result;
}
protected static function parseComposite(): Matcher|false {
if (self::$debug === true) {
self::debug();
}
$result = false;
$position = self::$instance->data->position;
$s1 = self::parseExpression();
if ($s1 !== false) {
self::parseSpace();
$s3 = self::$instance->data->consumeIf('|&-');
if (in_array($s3, [ '|', '&', '-' ])) {
self::parseSpace();
$s5 = self::parseComposite();
if ($s5 !== false) {
$result = new CompositeMatcher($s1, $s3, $s5);
}
} else {
self::fail('|&-');
}
}
if ($result === false) {
self::$instance->data->unconsumeTo($position);
$result = self::parseExpression();
}
if (self::$debug === true) {
self::debugResult($result);
}
return $result;
}
protected static function parseExpression(): Matcher|false {
if (self::$debug === true) {
self::debug();
}
$result = false;
$s1 = self::$instance->data->consumeIf('-');
if ($s1 === '-') {
self::parseSpace();
$s3 = self::parseGroup();
if ($s3 !== false) {
self::parseSpace();
$result = new NegateMatcher($s3);
}
} else {
self::fail('-');
}
if ($result === false) {
$s1 = self::$instance->data->consumeIf('-');
if ($s1 === '' || $s1 === false) {
$s1 = false;
self::fail('-');
}
if ($s1 !== false) {
self::parseSpace();
$s3 = self::parsePath();
if ($s3 !== false) {
self::parseSpace();
$result = new NegateMatcher($s3);
$result = [];
while (self::$instance->token = self::$instance->tokenizer->next()) {
$priority = 0;
if (strlen(self::$instance->token) === 2 && self::$instance->token[1] === ':') {
switch (self::$instance->token[0]) {
case 'R': $priority = 1;
break;
case 'L': $priority = -1;
break;
default: die('OOK!');
}
}
if ($result === false) {
$result = self::parseGroup();
if ($result === false) {
$result = self::parsePath();
self::$instance->token = self::$instance->tokenizer->next();
if (self::$instance->token === false) {
break;
}
}
}
if (self::$debug === true) {
self::debugResult($result);
}
return $result;
}
protected static function parseGroup(): Matcher|false {
if (self::$debug === true) {
self::debug();
}
$result = false;
$prefix = null;
$position = self::$instance->data->position;
$s2 = self::$instance->data->consumeIf('LRB');
if ($s2 === '' || $s2 === false) {
$s2 = false;
self::fail('LRB');
}
if ($s2 !== false) {
$s3 = self::$instance->data->consumeIf(':');
if ($s3 === ':') {
$prefix = "$s2$s3";
} else {
self::fail(':');
$matcher = self::parseConjunction();
if ($matcher === false) {
$matcher = self::parseOperand();
}
}
$s2 = self::$instance->data->consumeIf('(');
if ($s2 === '(') {
self::parseSpace();
$s4 = self::parseSelector();
if ($s4 !== false) {
self::parseSpace();
$s6 = self::$instance->data->consumeIf(')');
if ($s6 === '' || $s6 === false) {
$s6 = false;
self::fail(')');
}
if ($s6 !== false) {
$result = new GroupMatcher($prefix, $s4);
}
}
} else {
self::fail('(');
}
if ($result === false) {
self::$instance->data->unconsumeTo($position);
}
$result[] = [
'matcher' => $matcher,
'priority' => $priority
];
if (self::$debug === true) {
self::debugResult($result);
if (self::$instance->token !== ',') {
break;
}
}
return $result;
}
protected static function parsePath(): Matcher|false {
if (self::$debug === true) {
self::debug();
}
$result = false;
$prefix = null;
$s2 = self::$instance->data->consumeIf('LRB');
if (in_array($s2, [ 'L', 'R', 'B' ])) {
$s3 = self::$instance->data->consumeIf(':');
if ($s3 === '' || $s3 === false) {
$s3 = false;
self::fail(':');
}
if ($s3 !== false) {
$prefix = "$s2$s3";
}
} else {
self::fail('LRB');
}
$s2 = self::parseScope();
if ($s2 !== false) {
$s3 = [$s2];
do {
$s6 = false;
self::parseSpace();
$s6 = self::parseScope();
if ($s6 !== false) {
$s3[] = $s6;
}
} while ($s6 !== false);
$result = new PathMatcher($prefix, ...$s3);
}
if (self::$debug === true) {
self::debugResult($result);
protected static function parseConjunction(): AndMatcher|false {
$matchers = [];
while ($matcher = self::parseOperand()) {
$matchers[] = $matcher;
}
return $result;
return (count($matchers) > 1) ? new AndMatcher($matchers[0], $matchers[1]) : false;
}
protected static function parseSegment(): SegmentMatcher|TrueMatcher|false {
if (self::$debug === true) {
self::debug();
}
$result = false;
self::parseSpace();
$s2 = self::$instance->data->consumeWhile('ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+_');
if ($s2 === '' || $s2 === false) {
$s2 = false;
self::fail('ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+_');
}
if ($s2 !== false) {
$s3 = self::$instance->data->consumeWhile('ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-+_');
if ($s3 === '' || $s2 === false) {
$s3 = false;
self::fail('ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-+_');
protected static function parseInnerExpression(): Matcher|false {
$matchers = [];
while ($matcher = self::parseConjunction()) {
$matchers[] = $matcher;
if (self::$instance->token === '|' || self::$instance->token === ',') {
do {
self::$instance->token = self::$instance->tokenizer->next();
} while (self::$instance->token === '|' || self::$instance->token === ',');
} else {
$s2 .= $s3;
break;
}
}
if ($s2 !== false) {
self::parseSpace();
$result = new SegmentMatcher($s2);
}
if ($result === false) {
self::parseSpace();
$s2 = self::$instance->data->consumeIf('*');
if ($s2 === '*') {
self::parseSpace();
$result = new TrueMatcher($s2);
} else {
self::fail('*');
}
}
if (self::$debug === true) {
self::debugResult($result);
}
return $result;
return (count($matchers) > 1) ? new OrMatcher($matchers[0], $matchers[1]) : false;
}
protected static function parseSelector(): Matcher|false {
if (self::$debug === true) {
self::debug();
}
$result = false;
$position = self::$instance->data->position;
protected static function parseOperand(): Matcher|false {
if (self::$instance->token === '-') {
self::$instance->token = self::$instance->tokenizer->next();
$s1 = self::parseComposite();
if ($s1 !== false) {
self::parseSpace();
$s3 = self::$instance->data->consumeIf(',');
if ($s3 === ',') {
self::parseSpace();
$s5 = self::parseSelector();
$result = ($s5 === false) ? $s1 : new OrMatcher($s1, $s5);
} else {
self::fail(',');
$matcher = self::parseOperand();
if ($matcher === false) {
die('OH SHIT');
}
}
if ($result === false) {
self::$instance->data->unconsumeTo($position);
$result = self::parseComposite();
return new NegateMatcher($matcher);
}
if (self::$debug === true) {
self::debugResult($result);
}
return $result;
}
protected static function parseScope(): ScopeMatcher|false {
if (self::$debug === true) {
self::debug();
if (self::$instance->token === '(') {
self::$instance->token = self::$instance->tokenizer->next();
$expressionInParents = self::parseInnerExpression();
if (self::$instance->token === ')') {
self::$instance->token = self::$instance->tokenizer->next();
}
return $expressionInParents;
}
$result = false;
$s1 = self::parseSegment();
if ($s1 !== false) {
$s2 = [$s1];
if (self::$instance->tokenizer->tokenIsIdentifier()) {
$identifiers = [];
do {
$s3 = false;
$s4 = self::$instance->data->consumeIf('.');
if ($s4 === '.') {
$s3 = self::parseSegment();
if ($s3 !== false) {
$s2[] = $s3;
}
} else {
self::fail('.');
}
} while ($s3 !== false);
$result = new ScopeMatcher(...$s2);
}
if (self::$debug === true) {
self::debugResult($result);
}
return $result;
}
protected static function parseSpace(): string|false {
if (self::$debug === true) {
self::debug();
}
$identifiers[] = self::$instance->token;
self::$instance->token = self::$instance->tokenizer->next();
} while (self::$instance->tokenizer->tokenIsIdentifier());
$result = self::$instance->data->consumeWhile(" \t");
if ($result === false) {
self::fail(" \t");
return new ScopeMatcher(...$identifiers);
}
if (self::$debug === true) {
self::debugResult($result);
}
return $result;
}
protected static function debug() {
$message = <<<DEBUG
------------------------------
%s
Method: %s
Position: %s
Char: %s
DEBUG;
$methodTree = '';
$backtrace = debug_backtrace();
array_shift($backtrace);
array_pop($backtrace);
foreach ($backtrace as $b) {
$methodTree = "->{$b['function']}$methodTree";
}
printf($message,
self::$instance->debugCount++,
ltrim($methodTree, '->'),
self::$instance->data->position,
var_export(self::$instance->data->peek(), true)
);
}
protected static function debugResult($result) {
printf("%s Result: %s\n",
debug_backtrace()[1]['function'],
var_export($result, true));
}
protected static function fail(string $expected) {
self::$instance->lastExceptionData = [
'expected' => $expected,
'found' => self::$instance->data->peek()
];
return false;
}
}

40
lib/Scope/Tokenizer.php
View File

@ -0,0 +1,40 @@
<?php
/** @license MIT
* Copyright 2021 Dustin Wilson et al.
* See LICENSE and AUTHORS files for details */
declare(strict_types=1);
namespace dW\Highlighter\Scope;
class Tokenizer {
protected array $matches = [];
protected int $position = 0;
public function __construct(string $scope) {
preg_match_all('/([LR]:|[\w\.:][\w\.:\-]*|[\,\|\-\(\)])/', $scope, $matches);
$this->matches = $matches[1];
}
public function next(): string|false {
if (count($this->matches) === 0) {
return false;
}
$result = $this->matches[$this->position] ?? false;
if ($result !== false) {
$this->position++;
}
return $result;
}
public function tokenIsIdentifier(): bool {
if (!isset($this->matches[$this->position])) {
return false;
}
return (!!$this->matches[$this->position] && !!preg_match('/[\w\.:]+/', $this->matches[$this->position]));
}
}
Write Preview
Loading…
Cancel
Save