TokenMatcherGenerator::buildOnTransition() - Code Metrics - Inspection of "Long conditions moved to private methods in genera..." - remorhaz/php-unilex - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 5848ce...73ad9b )

by Edward

created 2020-04-12 16:28 UTC

TokenMatcherGenerator::buildOnTransition() A

↳ Parent: TokenMatcherGenerator

Complexity

Conditions	1
Paths	1

Size

Total Lines	3
Code Lines	1

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
cc	1
eloc	1
c	1
b	0
f	0
nc	1
nop	0
dl	0
loc	3
rs	10

<?php

declare(strict_types=1);

namespace Remorhaz\UniLex\Lexer;

use ReflectionException;
use Remorhaz\UniLex\AST\Translator;
use Remorhaz\UniLex\AST\Tree;
use Remorhaz\UniLex\Exception;
use Remorhaz\UniLex\RegExp\FSM\Dfa;
use Remorhaz\UniLex\RegExp\FSM\DfaBuilder;
use Remorhaz\UniLex\RegExp\FSM\LanguageBuilder;
use Remorhaz\UniLex\RegExp\FSM\Nfa;
use Remorhaz\UniLex\RegExp\FSM\NfaBuilder;
use Remorhaz\UniLex\RegExp\FSM\Range;
use Remorhaz\UniLex\RegExp\ParserFactory;
use Remorhaz\UniLex\RegExp\PropertyLoader;
use Remorhaz\UniLex\Unicode\CharBufferFactory;
use Throwable;

use function array_diff;
use function array_intersect;
use function array_merge;
use function array_pop;
use function array_unique;
use function count;
use function implode;
use function in_array;

class TokenMatcherGenerator
{

    private $spec;

    private $output;

    private $dfa;

    private $regExpFinishMap = [];

    private $conditionFunctions = [];

    public function __construct(TokenMatcherSpec $spec)
    {
        $this->spec = $spec;
    }

    /**
     * @return string
     * @throws Exception
     * @throws ReflectionException
     */
    private function buildOutput(): string
    {
        $this->conditionFunctions = [];

        return
            "{$this->buildFileComment()}\ndeclare(strict_types=1);\n\n" .
            "{$this->buildHeader()}\n" .
            "class {$this->spec->getTargetShortName()} extends {$this->spec->getTemplateClass()->getShortName()}\n" .
            "{\n" .
            "\n" .
            "    public function match({$this->buildMatchParameters()}): bool\n" .
            "    {\n{$this->buildMatchBody()}" .
            "    }\n" .
            $this->buildConditionFunctions() .
            "}\n";
    }

    /**
     * @return TokenMatcherInterface
     * @throws Exception
     */
    public function load(): TokenMatcherInterface
    {
        $targetClass = $this->spec->getTargetClassName();
        if (!class_exists($targetClass)) {
            try {
                $source = $this->getOutput(false);
                eval($source);

            } catch (Throwable $e) {
                throw new Exception("Invalid PHP code generated", 0, $e);
            }
            if (!class_exists($targetClass)) {
                throw new Exception("Failed to generate target class");
            }
        }

        return new $targetClass();
    }

    /**
     * @param bool $asFile
     * @return string
     * @throws Exception
     * @throws ReflectionException
     */
    public function getOutput(bool $asFile = true): string
    {
        if (!isset($this->output)) {
            $this->output = $this->buildOutput();
        }

        return $asFile ? "<?php\n\n{$this->output}" : $this->output;
    }

    private function buildFileComment(): string
    {
        $content = $this->spec->getFileComment();
        if ('' == $content) {
            return '';
        }
        $comment = "/**\n";
        $commentLineList = explode("\n", $content);
        foreach ($commentLineList as $commentLine) {
            $comment .= rtrim(" * {$commentLine}") . "\n";
        }
        $comment .= " */\n";

        return $comment;
    }

    /**
     * @return string
     * @throws ReflectionException
     */
    public function buildHeader(): string
    {
        $headerParts = [];
        $namespace = $this->spec->getTargetNamespaceName();
        if ($namespace != '') {
            $headerParts[] = $this->buildMethodPart("namespace {$namespace};", 0);
        }
        $useList = $this->buildUseList();
        if ('' != $useList) {
            $headerParts[] = $useList;
        }
        $header = $this->buildMethodPart($this->spec->getHeader(), 0);
        if ('' != $header) {
            $headerParts[] = $header;
        }

        return implode("\n", $headerParts);
    }

    /**
     * @return string
     * @throws ReflectionException
     */
    private function buildUseList(): string
    {
        $result = '';
        foreach ($this->spec->getUsedClassList() as $alias => $className) {
            $classWithAlias = is_string($alias) ? "{$className} {$alias}" : $className;
            $result .= $this->buildMethodPart("use {$classWithAlias};", 0);
        }

        return $result;
    }

    /**
     * @return string
     * @throws ReflectionException
     */
    private function buildMatchParameters(): string
    {
        $paramList = [];
        foreach ($this->spec->getMatchMethod()->getParameters() as $matchParameter) {
            if ($matchParameter->hasType()) {
                $param = $matchParameter->getType()->isBuiltin()
                    ? $matchParameter->getType()->getName()
                    : $matchParameter->getClass()->getShortName();
                $param .= " ";
            } else {
                $param = "";
            }
            $param .= "\${$matchParameter->getName()}";
            $paramList[] = $param;
        }

        return implode(", ", $paramList);
    }

    /**
     * @return string
     * @throws Exception
     */
    private function buildMatchBody(): string
    {
        $result = $this->buildBeforeMatch();

        foreach ($this->spec->getModeList() as $mode) {
            if (TokenMatcherInterface::DEFAULT_MODE == $mode) {
                continue;
            }
            $result .=
                $this->buildMethodPart("if (\$context->getMode() == '{$mode}') {") .
                $this->buildFsmEntry($mode, 3) .
                $this->buildMethodPart("}");
        }
        foreach ($this->spec->getModeList() as $mode) {
            if (TokenMatcherInterface::DEFAULT_MODE == $mode) {
                $result .= $this->buildFsmEntry(TokenMatcherInterface::DEFAULT_MODE) . "\n";
            }
            $result .= $this->buildFsmMoves($mode);
        }

        $result .= $this->buildErrorState();

        return $result;
    }

    private function buildBeforeMatch(): string
    {
        return
            $this->buildMethodPart("\$context = \$this->createContext(\$buffer, \$tokenFactory);") .
            $this->buildMethodPart($this->spec->getBeforeMatch());
    }

    /**
     * @param string $mode
     * @param int    $indent
     * @return string
     * @throws Exception
     */
    private function buildFsmEntry(string $mode, int $indent = 2): string
    {
        $state = $this->getDfa($mode)->getStateMap()->getStartState();

        return $this->buildMethodPart("goto {$this->buildStateLabel('state', $mode, $state)};", $indent);
    }

    private function buildStateLabel(string $prefix, string $mode, int $state): string
    {
        $contextSuffix = TokenMatcherInterface::DEFAULT_MODE == $mode
            ? ''
            : ucfirst($mode);

        return "{$prefix}{$contextSuffix}{$state}";
    }

    /**
     * @param string $mode
     * @return string
     * @throws Exception
     */
    private function buildFsmMoves(string $mode): string
    {
        $result = '';
        foreach ($this->getDfa($mode)->getStateMap()->getStateList() as $stateIn) {
            if ($this->isFinishStateWithSingleEnteringTransition($mode, $stateIn)) {
                continue;
            }
            $result .=
                $this->buildStateEntry($mode, $stateIn) .
                $this->buildStateTransitionList($mode, $stateIn) .
                $this->buildStateFinish($mode, $stateIn);
        }

        return $result;
    }

    /**
     * @param string $mode
     * @param int    $stateIn
     * @return string
     * @throws Exception
     */
    private function buildStateEntry(string $mode, int $stateIn): string
    {
        $result = '';
        $result .= $this->buildMethodPart("{$this->buildStateLabel('state', $mode, $stateIn)}:");
        $moves = $this->getDfa($mode)->getTransitionMap()->getExitList($stateIn);
        if (empty($moves)) {
            return $result;
        }
        $result .= $this->buildMethodPart("if (\$context->getBuffer()->isEnd()) {");
        $result .= $this->getDfa($mode)->getStateMap()->isFinishState($stateIn)
            ? $this->buildMethodPart("goto {$this->buildStateLabel('finish', $mode, $stateIn)};", 3)
            : $this->buildMethodPart("goto error;", 3);
        $result .=
            $this->buildMethodPart("}") .
            $this->buildMethodPart("\$char = \$context->getBuffer()->getSymbol();");

        return $result;
    }

    /**
     * @param string $mode
     * @param int    $stateIn
     * @return string
     * @throws Exception
     */
    private function buildStateTransitionList(string $mode, int $stateIn): string
    {
        $result = '';
        foreach ($this->getDfa($mode)->getTransitionMap()->getExitList($stateIn) as $stateOut => $symbolList) {
            foreach ($symbolList as $symbol) {
                $result .=
                    $this->buildMethodPart("if ({$this->buildRangeSetCondition($mode, $symbol)}) {") .
                    $this->buildOnTransition() .
                    $this->buildMethodPart("\$context->getBuffer()->nextSymbol();", 3);
                $result .= $this->isFinishStateWithSingleEnteringTransition($mode, $stateOut)
                    ? $this->buildToken($mode, $stateOut, 3)
                    : $this->buildStateTransition($mode, $stateOut, 3);
                $result .= $this->buildMethodPart("}");
            }
        }

        return $result;
    }

    /**
     * @param string $mode
     * @param int    $stateOut
     * @param int    $indent
     * @return string
     */
    private function buildStateTransition(string $mode, int $stateOut, int $indent = 3): string
    {
        return $this->buildMethodPart("goto {$this->buildStateLabel('state', $mode, $stateOut)};", $indent);
    }

    /**
     * @param string $mode
     * @param int    $stateOut
     * @return bool
     * @throws Exception
     */
    private function isFinishStateWithSingleEnteringTransition(string $mode, int $stateOut): bool
    {
        if (!$this->getDfa($mode)->getStateMap()->isFinishState($stateOut)) {
            return false;
        }
        $enters = $this->getDfa($mode)->getTransitionMap()->getEnterList($stateOut);
        $exits = $this->getDfa($mode)->getTransitionMap()->getExitList($stateOut);
        if (!(count($enters) == 1 && count($exits) == 0)) {
            return false;
        }
        $symbolList = array_pop($enters);

        return count($symbolList) == 1;
    }

    private function buildHex(int $char): string
    {
        $hexChar = strtoupper(dechex($char));
        if (strlen($hexChar) % 2 != 0) {
            $hexChar = "0{$hexChar}";
        }

        return "0x{$hexChar}";
    }

    private function buildRangeCondition(Range $range): array
    {
        $startChar = $this->buildHex($range->getStart());
        if ($range->getStart() == $range->getFinish()) {
            return ["{$startChar} == \$char"];
        }
        $finishChar = $this->buildHex($range->getFinish());
        if ($range->getStart() + 1 == $range->getFinish()) {
            return [
                "{$startChar} == \$char",
                "{$finishChar} == \$char",
            ];
        }

        return ["{$startChar} <= \$char && \$char <= {$finishChar}"];
    }

    /**
     * @param string $mode
     * @param int    $symbol
     * @return string
     * @throws Exception
     */
    private function buildRangeSetCondition(string $mode, int $symbol): string
    {
        $rangeSet = $this->getDfa($mode)->getSymbolTable()->getRangeSet($symbol);

        $conditionList = [];
        foreach ($rangeSet->getRanges() as $range) {
            $conditionList = array_merge($conditionList, $this->buildRangeCondition($range));
        }
        $result = implode(" || ", $conditionList);
        if (strlen($result) + 15 <= 120 || count($conditionList) == 1) {
            return ltrim($result);
        }
        $result = $this->buildMethodPart(implode(" ||\n", $conditionList), 1);
        if (count($conditionList) > 10) {
            $method = "isMode" . ucfirst($mode) . "Symbol{$symbol}";
            $this->conditionFunctions[$method] = $result;

            return "\$this->{$method}(\$char)";
        }

        return "\n    " . ltrim($result);
    }

    private function buildConditionFunctions(): string
    {
        $result = '';

        foreach ($this->conditionFunctions as $method => $conditionList) {
            $result .=
                "\n    private function {$method}(int \$char): bool\n    {\n" .
                $this->buildMethodPart("return") .
                $this->buildMethodPart(rtrim($conditionList) . ';') .
                "    }\n";
        }

        return $result;
    }

    /**
     * @param string $mode
     * @param int    $stateIn
     * @return string
     * @throws Exception
     */
    private function buildStateFinish(string $mode, int $stateIn): string
    {
        if (!$this->getDfa($mode)->getStateMap()->isFinishState($stateIn)) {
            return $this->buildMethodPart("goto error;\n");
        }
        $result = '';
        if (!empty($this->getDfa($mode)->getTransitionMap()->getExitList($stateIn))) {
            $result .= $this->buildMethodPart("{$this->buildStateLabel('finish', $mode, $stateIn)}:");
        }
        $result .= "{$this->buildToken($mode, $stateIn)}\n";

        return $result;
    }

    /**
     * @param string $mode
     * @param int    $stateIn
     * @param int    $indent
     * @return string
     * @throws Exception
     */
    private function buildToken(string $mode, int $stateIn, int $indent = 2): string
    {
        if (!isset($this->regExpFinishMap[$mode][$stateIn])) {
            throw new Exception("No regular expressions found for state {$mode}:{$stateIn}");
        }
        $tokenSpec = $this->spec->getTokenSpec($mode, $this->regExpFinishMap[$mode][$stateIn]);

        return
            $this->buildMethodPart("// {$tokenSpec->getRegExp()}", $indent) .
            $this->buildSingleToken($tokenSpec, $indent);
    }

    private function buildSingleToken(TokenSpec $tokenSpec, int $indent): string
    {
        return
            $this->buildMethodPart($tokenSpec->getCode(), $indent) .
            $this->buildOnToken($indent) . "\n" .
            $this->buildMethodPart("return true;", $indent);
    }

    private function buildErrorState(): string
    {
        $code = $this->spec->getOnError();

        return
            $this->buildMethodPart("error:") .
            $this->buildMethodPart('' == $code ? "return false;" : $code);
    }

    private function buildMethodPart(string $code, int $indent = 2): string
    {
        if ('' == $code) {
            return '';
        }
        $result = '';
        $codeLineList = explode("\n", $code);
        foreach ($codeLineList as $codeLine) {
            $line = '';
            for ($i = 0; $i < $indent; $i++) {
                $line .= "    ";
            }
            $result .= rtrim($line . $codeLine) . "\n";
        }

        return $result;
    }

    private function buildOnTransition(): string
    {
        return $this->buildMethodPart($this->spec->getOnTransition(), 3);
    }

    private function buildOnToken(int $indent = 2): string
    {
        return $this->buildMethodPart($this->spec->getOnToken(), $indent);
    }

    /**
     * @param string $context
     * @return Dfa
     * @throws Exception
     */
    private function getDfa(string $context): Dfa
    {
        if (!isset($this->dfa[$context])) {
            $this->dfa[$context] = $this->buildDfa($context);
        }

        return $this->dfa[$context];
    }

    /**
     * @param string $mode
     * @return Dfa
     * @throws Exception
     */
    private function buildDfa(string $mode): Dfa
    {
        $nfa = new Nfa();
        $startState = $nfa->getStateMap()->createState();
        $nfa->getStateMap()->addStartState($startState);
        $nfaRegExpMap = [];
        /** @var Dfa[] $dfaList */
        $dfaList = [];
        foreach ($this->spec->getTokenSpecList($mode) as $tokenSpec) {
            $existingStates = $nfa->getStateMap()->getStateList();
            $regExpEntryState = $nfa->getStateMap()->createState();
            $nfa
                ->getEpsilonTransitionMap()
                ->addTransition($startState, $regExpEntryState, true);
            $this->buildRegExp($nfa, $regExpEntryState, $tokenSpec->getRegExp());
            $regExpStates = array_diff($nfa->getStateMap()->getStateList(), $existingStates);
            $nfaRegExpMap[$tokenSpec->getRegExp()] = $regExpStates;
            $dfaList[$tokenSpec->getRegExp()] = $this->buildIndependentRegExp($tokenSpec->getRegExp());
        }

        $joinedNfa = new Nfa();
        $startState = $joinedNfa->getStateMap()->createState();
        $joinedNfa->getStateMap()->addStartState($startState);
        $languageBuilder = LanguageBuilder::forNfa($joinedNfa);
        $joinedNfaStates = [];
        $nfaRegExpMap = [];
        $regExpFinishMap = [];
        foreach ($dfaList as $regExp => $dfa) {
            $nfaRegExpMap[$regExp] = [];
            foreach ($dfa->getStateMap()->getStateList() as $dfaState) {
                $nfaState = $joinedNfa->getStateMap()->createState();
                $nfaRegExpMap[$regExp][] = $nfaState;
                $joinedNfaStates[$dfaState] = $nfaState;
                if ($dfa->getStateMap()->isStartState($dfaState)) {
                    $joinedNfa->getEpsilonTransitionMap()->addTransition($startState, $nfaState, true);
                }
                if ($dfa->getStateMap()->isFinishState($dfaState)) {
                    $regExpFinishMap[$regExp][] = $nfaState;
                    $joinedNfa->getStateMap()->addFinishState($nfaState);
                }
            }
            foreach ($dfa->getTransitionMap()->getTransitionList() as $dfaStateIn => $transitions) {
                foreach ($transitions as $dfaStateOut => $symbols) {
                    foreach ($symbols as $symbol) {
                        $rangeSet = $dfa->getSymbolTable()->getRangeSet($symbol);
                        $newSymbols = $languageBuilder->getSymbolList(...$rangeSet->getRanges());
                        $oldSymbols = $joinedNfa
                            ->getSymbolTransitionMap()
                            ->transitionExists($joinedNfaStates[$dfaStateIn], $joinedNfaStates[$dfaStateOut])
                            ? $joinedNfa
                                ->getSymbolTransitionMap()
                                ->getTransition($joinedNfaStates[$dfaStateIn], $joinedNfaStates[$dfaStateOut])
                            : [];
                        $joinedNfa->getSymbolTransitionMap()->replaceTransition(
                            $joinedNfaStates[$dfaStateIn],
                            $joinedNfaStates[$dfaStateOut],
                            array_unique(array_merge($oldSymbols, $newSymbols))
                        );
                    }
                }
            }
        }

        $dfa = new Dfa();
        (new DfaBuilder($dfa, $joinedNfa))->run();

        $dfaRegExpFinishMap = [];
        foreach ($dfa->getStateMap()->getFinishStateList() as $dfaFinishState) {
            $nfaFinishStates = array_intersect(
                $dfa->getStateMap()->getStateValue($dfaFinishState),
                $joinedNfa->getStateMap()->getFinishStateList()
            );
            foreach ($regExpFinishMap as $regExp => $regExpFinishStates) {
                foreach ($nfaFinishStates as $nfaFinishState) {
                    if (in_array($nfaFinishState, $regExpFinishStates)) {
                        $dfaRegExpFinishMap[$dfaFinishState] = (string) $regExp;
                        break 2;
                    }
                }
            }
        }
        foreach ($this->spec->getTokenSpecList($mode) as $tokenSpec) {
            if (!in_array($tokenSpec->getRegExp(), $dfaRegExpFinishMap)) {
                throw new Exception("Token not reachable for regular expression: {$tokenSpec->getRegExp()} ");
            }
        }
        $this->regExpFinishMap[$mode] = $dfaRegExpFinishMap;

        return $dfa;
    }

    /**
     * @param Nfa    $nfa
     * @param int    $entryState
     * @param string $regExp
     * @throws Exception
     */
    private function buildRegExp(Nfa $nfa, int $entryState, string $regExp): void
    {
        $buffer = CharBufferFactory::createFromString($regExp);
        $tree = new Tree();
        ParserFactory::createFromBuffer($tree, $buffer)->run();
        $nfaBuilder = new NfaBuilder($nfa, PropertyLoader::create());
        $nfaBuilder->setStartState($entryState);
        (new Translator($tree, $nfaBuilder))->run();
    }

    /**
     * @param string $regExp
     * @return Dfa
     * @throws Exception
     */
    private function buildIndependentRegExp(string $regExp): Dfa
    {
        $buffer = CharBufferFactory::createFromString($regExp);
        $tree = new Tree();
        ParserFactory::createFromBuffer($tree, $buffer)->run();
        $nfa = new Nfa();
        $nfaBuilder = new NfaBuilder($nfa, PropertyLoader::create());
        (new Translator($tree, $nfaBuilder))->run();

        return DfaBuilder::fromNfa($nfa);
    }
}


1			<?php
2
3			declare(strict_types=1);
4
5			namespace Remorhaz\UniLex\Lexer;
6
7			use ReflectionException;
8			use Remorhaz\UniLex\AST\Translator;
9			use Remorhaz\UniLex\AST\Tree;
10			use Remorhaz\UniLex\Exception;
11			use Remorhaz\UniLex\RegExp\FSM\Dfa;
12			use Remorhaz\UniLex\RegExp\FSM\DfaBuilder;
13			use Remorhaz\UniLex\RegExp\FSM\LanguageBuilder;
14			use Remorhaz\UniLex\RegExp\FSM\Nfa;
15			use Remorhaz\UniLex\RegExp\FSM\NfaBuilder;
16			use Remorhaz\UniLex\RegExp\FSM\Range;
17			use Remorhaz\UniLex\RegExp\ParserFactory;
18			use Remorhaz\UniLex\RegExp\PropertyLoader;
19			use Remorhaz\UniLex\Unicode\CharBufferFactory;
20			use Throwable;
21
22			use function array_diff;
23			use function array_intersect;
24			use function array_merge;
25			use function array_pop;
26			use function array_unique;
27			use function count;
28			use function implode;
29			use function in_array;
30
31			class TokenMatcherGenerator
32			{
33
34			private $spec;
35
36			private $output;
37
38			private $dfa;
39
40			private $regExpFinishMap = [];
41
42			private $conditionFunctions = [];
43
44			public function __construct(TokenMatcherSpec $spec)
45			{
46			$this->spec = $spec;
47			}
48
49			/**
50			* @return string
51			* @throws Exception
52			* @throws ReflectionException
53			*/
54			private function buildOutput(): string
55			{
56			$this->conditionFunctions = [];
57
58			return
59			"{$this->buildFileComment()}\ndeclare(strict_types=1);\n\n" .
60			"{$this->buildHeader()}\n" .
61			"class {$this->spec->getTargetShortName()} extends {$this->spec->getTemplateClass()->getShortName()}\n" .
62			"{\n" .
63			"\n" .
64			" public function match({$this->buildMatchParameters()}): bool\n" .
65			" {\n{$this->buildMatchBody()}" .
66			" }\n" .
67			$this->buildConditionFunctions() .
68			"}\n";
69			}
70
71			/**
72			* @return TokenMatcherInterface
73			* @throws Exception
74			*/
75			public function load(): TokenMatcherInterface
76			{
77			$targetClass = $this->spec->getTargetClassName();
78			if (!class_exists($targetClass)) {
79			try {
80			$source = $this->getOutput(false);
81			eval($source);
			0 ignored issues – show introduced 2020-02-17 16:36 UTC by Report Bug Copy Issue Report The use of eval() is discouraged. Loading history...
82			} catch (Throwable $e) {
83			throw new Exception("Invalid PHP code generated", 0, $e);
84			}
85			if (!class_exists($targetClass)) {
86			throw new Exception("Failed to generate target class");
87			}
88			}
89
90			return new $targetClass();
91			}
92
93			/**
94			* @param bool $asFile
95			* @return string
96			* @throws Exception
97			* @throws ReflectionException
98			*/
99			public function getOutput(bool $asFile = true): string
100			{
101			if (!isset($this->output)) {
102			$this->output = $this->buildOutput();
103			}
104
105			return $asFile ? "<?php\n\n{$this->output}" : $this->output;
106			}
107
108			private function buildFileComment(): string
109			{
110			$content = $this->spec->getFileComment();
111			if ('' == $content) {
112			return '';
113			}
114			$comment = "/**\n";
115			$commentLineList = explode("\n", $content);
116			foreach ($commentLineList as $commentLine) {
117			$comment .= rtrim(" * {$commentLine}") . "\n";
118			}
119			$comment .= " */\n";
120
121			return $comment;
122			}
123
124			/**
125			* @return string
126			* @throws ReflectionException
127			*/
128			public function buildHeader(): string
129			{
130			$headerParts = [];
131			$namespace = $this->spec->getTargetNamespaceName();
132			if ($namespace != '') {
133			$headerParts[] = $this->buildMethodPart("namespace {$namespace};", 0);
134			}
135			$useList = $this->buildUseList();
136			if ('' != $useList) {
137			$headerParts[] = $useList;
138			}
139			$header = $this->buildMethodPart($this->spec->getHeader(), 0);
140			if ('' != $header) {
141			$headerParts[] = $header;
142			}
143
144			return implode("\n", $headerParts);
145			}
146
147			/**
148			* @return string
149			* @throws ReflectionException
150			*/
151			private function buildUseList(): string
152			{
153			$result = '';
154			foreach ($this->spec->getUsedClassList() as $alias => $className) {
155			$classWithAlias = is_string($alias) ? "{$className} {$alias}" : $className;
156			$result .= $this->buildMethodPart("use {$classWithAlias};", 0);
157			}
158
159			return $result;
160			}
161
162			/**
163			* @return string
164			* @throws ReflectionException
165			*/
166			private function buildMatchParameters(): string
167			{
168			$paramList = [];
169			foreach ($this->spec->getMatchMethod()->getParameters() as $matchParameter) {
170			if ($matchParameter->hasType()) {
171			$param = $matchParameter->getType()->isBuiltin()
172			? $matchParameter->getType()->getName()
173			: $matchParameter->getClass()->getShortName();
174			$param .= " ";
175			} else {
176			$param = "";
177			}
178			$param .= "\${$matchParameter->getName()}";
179			$paramList[] = $param;
180			}
181
182			return implode(", ", $paramList);
183			}
184
185			/**
186			* @return string
187			* @throws Exception
188			*/
189			private function buildMatchBody(): string
190			{
191			$result = $this->buildBeforeMatch();
192
193			foreach ($this->spec->getModeList() as $mode) {
194			if (TokenMatcherInterface::DEFAULT_MODE == $mode) {
195			continue;
196			}
197			$result .=
198			$this->buildMethodPart("if (\$context->getMode() == '{$mode}') {") .
199			$this->buildFsmEntry($mode, 3) .
200			$this->buildMethodPart("}");
201			}
202			foreach ($this->spec->getModeList() as $mode) {
203			if (TokenMatcherInterface::DEFAULT_MODE == $mode) {
204			$result .= $this->buildFsmEntry(TokenMatcherInterface::DEFAULT_MODE) . "\n";
205			}
206			$result .= $this->buildFsmMoves($mode);
207			}
208
209			$result .= $this->buildErrorState();
210
211			return $result;
212			}
213
214			private function buildBeforeMatch(): string
215			{
216			return
217			$this->buildMethodPart("\$context = \$this->createContext(\$buffer, \$tokenFactory);") .
218			$this->buildMethodPart($this->spec->getBeforeMatch());
219			}
220
221			/**
222			* @param string $mode
223			* @param int $indent
224			* @return string
225			* @throws Exception
226			*/
227			private function buildFsmEntry(string $mode, int $indent = 2): string
228			{
229			$state = $this->getDfa($mode)->getStateMap()->getStartState();
230
231			return $this->buildMethodPart("goto {$this->buildStateLabel('state', $mode, $state)};", $indent);
232			}
233
234			private function buildStateLabel(string $prefix, string $mode, int $state): string
235			{
236			$contextSuffix = TokenMatcherInterface::DEFAULT_MODE == $mode
237			? ''
238			: ucfirst($mode);
239
240			return "{$prefix}{$contextSuffix}{$state}";
241			}
242
243			/**
244			* @param string $mode
245			* @return string
246			* @throws Exception
247			*/
248			private function buildFsmMoves(string $mode): string
249			{
250			$result = '';
251			foreach ($this->getDfa($mode)->getStateMap()->getStateList() as $stateIn) {
252			if ($this->isFinishStateWithSingleEnteringTransition($mode, $stateIn)) {
253			continue;
254			}
255			$result .=
256			$this->buildStateEntry($mode, $stateIn) .
257			$this->buildStateTransitionList($mode, $stateIn) .
258			$this->buildStateFinish($mode, $stateIn);
259			}
260
261			return $result;
262			}
263
264			/**
265			* @param string $mode
266			* @param int $stateIn
267			* @return string
268			* @throws Exception
269			*/
270			private function buildStateEntry(string $mode, int $stateIn): string
271			{
272			$result = '';
273			$result .= $this->buildMethodPart("{$this->buildStateLabel('state', $mode, $stateIn)}:");
274			$moves = $this->getDfa($mode)->getTransitionMap()->getExitList($stateIn);
275			if (empty($moves)) {
276			return $result;
277			}
278			$result .= $this->buildMethodPart("if (\$context->getBuffer()->isEnd()) {");
279			$result .= $this->getDfa($mode)->getStateMap()->isFinishState($stateIn)
280			? $this->buildMethodPart("goto {$this->buildStateLabel('finish', $mode, $stateIn)};", 3)
281			: $this->buildMethodPart("goto error;", 3);
282			$result .=
283			$this->buildMethodPart("}") .
284			$this->buildMethodPart("\$char = \$context->getBuffer()->getSymbol();");
285
286			return $result;
287			}
288
289			/**
290			* @param string $mode
291			* @param int $stateIn
292			* @return string
293			* @throws Exception
294			*/
295			private function buildStateTransitionList(string $mode, int $stateIn): string
296			{
297			$result = '';
298			foreach ($this->getDfa($mode)->getTransitionMap()->getExitList($stateIn) as $stateOut => $symbolList) {
299			foreach ($symbolList as $symbol) {
300			$result .=
301			$this->buildMethodPart("if ({$this->buildRangeSetCondition($mode, $symbol)}) {") .
302			$this->buildOnTransition() .
303			$this->buildMethodPart("\$context->getBuffer()->nextSymbol();", 3);
304			$result .= $this->isFinishStateWithSingleEnteringTransition($mode, $stateOut)
305			? $this->buildToken($mode, $stateOut, 3)
306			: $this->buildStateTransition($mode, $stateOut, 3);
307			$result .= $this->buildMethodPart("}");
308			}
309			}
310
311			return $result;
312			}
313
314			/**
315			* @param string $mode
316			* @param int $stateOut
317			* @param int $indent
318			* @return string
319			*/
320			private function buildStateTransition(string $mode, int $stateOut, int $indent = 3): string
321			{
322			return $this->buildMethodPart("goto {$this->buildStateLabel('state', $mode, $stateOut)};", $indent);
323			}
324
325			/**
326			* @param string $mode
327			* @param int $stateOut
328			* @return bool
329			* @throws Exception
330			*/
331			private function isFinishStateWithSingleEnteringTransition(string $mode, int $stateOut): bool
332			{
333			if (!$this->getDfa($mode)->getStateMap()->isFinishState($stateOut)) {
334			return false;
335			}
336			$enters = $this->getDfa($mode)->getTransitionMap()->getEnterList($stateOut);
337			$exits = $this->getDfa($mode)->getTransitionMap()->getExitList($stateOut);
338			if (!(count($enters) == 1 && count($exits) == 0)) {
339			return false;
340			}
341			$symbolList = array_pop($enters);
342
343			return count($symbolList) == 1;
344			}
345
346			private function buildHex(int $char): string
347			{
348			$hexChar = strtoupper(dechex($char));
349			if (strlen($hexChar) % 2 != 0) {
350			$hexChar = "0{$hexChar}";
351			}
352
353			return "0x{$hexChar}";
354			}
355
356			private function buildRangeCondition(Range $range): array
357			{
358			$startChar = $this->buildHex($range->getStart());
359			if ($range->getStart() == $range->getFinish()) {
360			return ["{$startChar} == \$char"];
361			}
362			$finishChar = $this->buildHex($range->getFinish());
363			if ($range->getStart() + 1 == $range->getFinish()) {
364			return [
365			"{$startChar} == \$char",
366			"{$finishChar} == \$char",
367			];
368			}
369
370			return ["{$startChar} <= \$char && \$char <= {$finishChar}"];
371			}
372
373			/**
374			* @param string $mode
375			* @param int $symbol
376			* @return string
377			* @throws Exception
378			*/
379			private function buildRangeSetCondition(string $mode, int $symbol): string
380			{
381			$rangeSet = $this->getDfa($mode)->getSymbolTable()->getRangeSet($symbol);
382
383			$conditionList = [];
384			foreach ($rangeSet->getRanges() as $range) {
385			$conditionList = array_merge($conditionList, $this->buildRangeCondition($range));
386			}
387			$result = implode(" \|\| ", $conditionList);
388			if (strlen($result) + 15 <= 120 \|\| count($conditionList) == 1) {
389			return ltrim($result);
390			}
391			$result = $this->buildMethodPart(implode(" \|\|\n", $conditionList), 1);
392			if (count($conditionList) > 10) {
393			$method = "isMode" . ucfirst($mode) . "Symbol{$symbol}";
394			$this->conditionFunctions[$method] = $result;
395
396			return "\$this->{$method}(\$char)";
397			}
398
399			return "\n " . ltrim($result);
400			}
401
402			private function buildConditionFunctions(): string
403			{
404			$result = '';
405
406			foreach ($this->conditionFunctions as $method => $conditionList) {
407			$result .=
408			"\n private function {$method}(int \$char): bool\n {\n" .
409			$this->buildMethodPart("return") .
410			$this->buildMethodPart(rtrim($conditionList) . ';') .
411			" }\n";
412			}
413
414			return $result;
415			}
416
417			/**
418			* @param string $mode
419			* @param int $stateIn
420			* @return string
421			* @throws Exception
422			*/
423			private function buildStateFinish(string $mode, int $stateIn): string
424			{
425			if (!$this->getDfa($mode)->getStateMap()->isFinishState($stateIn)) {
426			return $this->buildMethodPart("goto error;\n");
427			}
428			$result = '';
429			if (!empty($this->getDfa($mode)->getTransitionMap()->getExitList($stateIn))) {
430			$result .= $this->buildMethodPart("{$this->buildStateLabel('finish', $mode, $stateIn)}:");
431			}
432			$result .= "{$this->buildToken($mode, $stateIn)}\n";
433
434			return $result;
435			}
436
437			/**
438			* @param string $mode
439			* @param int $stateIn
440			* @param int $indent
441			* @return string
442			* @throws Exception
443			*/
444			private function buildToken(string $mode, int $stateIn, int $indent = 2): string
445			{
446			if (!isset($this->regExpFinishMap[$mode][$stateIn])) {
447			throw new Exception("No regular expressions found for state {$mode}:{$stateIn}");
448			}
449			$tokenSpec = $this->spec->getTokenSpec($mode, $this->regExpFinishMap[$mode][$stateIn]);
450
451			return
452			$this->buildMethodPart("// {$tokenSpec->getRegExp()}", $indent) .
453			$this->buildSingleToken($tokenSpec, $indent);
454			}
455
456			private function buildSingleToken(TokenSpec $tokenSpec, int $indent): string
457			{
458			return
459			$this->buildMethodPart($tokenSpec->getCode(), $indent) .
460			$this->buildOnToken($indent) . "\n" .
461			$this->buildMethodPart("return true;", $indent);
462			}
463
464			private function buildErrorState(): string
465			{
466			$code = $this->spec->getOnError();
467
468			return
469			$this->buildMethodPart("error:") .
470			$this->buildMethodPart('' == $code ? "return false;" : $code);
471			}
472
473			private function buildMethodPart(string $code, int $indent = 2): string
474			{
475			if ('' == $code) {
476			return '';
477			}
478			$result = '';
479			$codeLineList = explode("\n", $code);
480			foreach ($codeLineList as $codeLine) {
481			$line = '';
482			for ($i = 0; $i < $indent; $i++) {
483			$line .= " ";
484			}
485			$result .= rtrim($line . $codeLine) . "\n";
486			}
487
488			return $result;
489			}
490
491			private function buildOnTransition(): string
492			{
493			return $this->buildMethodPart($this->spec->getOnTransition(), 3);
494			}
495
496			private function buildOnToken(int $indent = 2): string
497			{
498			return $this->buildMethodPart($this->spec->getOnToken(), $indent);
499			}
500
501			/**
502			* @param string $context
503			* @return Dfa
504			* @throws Exception
505			*/
506			private function getDfa(string $context): Dfa
507			{
508			if (!isset($this->dfa[$context])) {
509			$this->dfa[$context] = $this->buildDfa($context);
510			}
511
512			return $this->dfa[$context];
513			}
514
515			/**
516			* @param string $mode
517			* @return Dfa
518			* @throws Exception
519			*/
520			private function buildDfa(string $mode): Dfa
521			{
522			$nfa = new Nfa();
523			$startState = $nfa->getStateMap()->createState();
524			$nfa->getStateMap()->addStartState($startState);
525			$nfaRegExpMap = [];
526			/** @var Dfa[] $dfaList */
527			$dfaList = [];
528			foreach ($this->spec->getTokenSpecList($mode) as $tokenSpec) {
529			$existingStates = $nfa->getStateMap()->getStateList();
530			$regExpEntryState = $nfa->getStateMap()->createState();
531			$nfa
532			->getEpsilonTransitionMap()
533			->addTransition($startState, $regExpEntryState, true);
534			$this->buildRegExp($nfa, $regExpEntryState, $tokenSpec->getRegExp());
535			$regExpStates = array_diff($nfa->getStateMap()->getStateList(), $existingStates);
536			$nfaRegExpMap[$tokenSpec->getRegExp()] = $regExpStates;
537			$dfaList[$tokenSpec->getRegExp()] = $this->buildIndependentRegExp($tokenSpec->getRegExp());
538			}
539
540			$joinedNfa = new Nfa();
541			$startState = $joinedNfa->getStateMap()->createState();
542			$joinedNfa->getStateMap()->addStartState($startState);
543			$languageBuilder = LanguageBuilder::forNfa($joinedNfa);
544			$joinedNfaStates = [];
545			$nfaRegExpMap = [];
546			$regExpFinishMap = [];
547			foreach ($dfaList as $regExp => $dfa) {
548			$nfaRegExpMap[$regExp] = [];
549			foreach ($dfa->getStateMap()->getStateList() as $dfaState) {
550			$nfaState = $joinedNfa->getStateMap()->createState();
551			$nfaRegExpMap[$regExp][] = $nfaState;
552			$joinedNfaStates[$dfaState] = $nfaState;
553			if ($dfa->getStateMap()->isStartState($dfaState)) {
554			$joinedNfa->getEpsilonTransitionMap()->addTransition($startState, $nfaState, true);
555			}
556			if ($dfa->getStateMap()->isFinishState($dfaState)) {
557			$regExpFinishMap[$regExp][] = $nfaState;
558			$joinedNfa->getStateMap()->addFinishState($nfaState);
559			}
560			}
561			foreach ($dfa->getTransitionMap()->getTransitionList() as $dfaStateIn => $transitions) {
562			foreach ($transitions as $dfaStateOut => $symbols) {
563			foreach ($symbols as $symbol) {
564			$rangeSet = $dfa->getSymbolTable()->getRangeSet($symbol);
565			$newSymbols = $languageBuilder->getSymbolList(...$rangeSet->getRanges());
566			$oldSymbols = $joinedNfa
567			->getSymbolTransitionMap()
568			->transitionExists($joinedNfaStates[$dfaStateIn], $joinedNfaStates[$dfaStateOut])
569			? $joinedNfa
570			->getSymbolTransitionMap()
571			->getTransition($joinedNfaStates[$dfaStateIn], $joinedNfaStates[$dfaStateOut])
572			: [];
573			$joinedNfa->getSymbolTransitionMap()->replaceTransition(
574			$joinedNfaStates[$dfaStateIn],
575			$joinedNfaStates[$dfaStateOut],
576			array_unique(array_merge($oldSymbols, $newSymbols))
577			);
578			}
579			}
580			}
581			}
582
583			$dfa = new Dfa();
584			(new DfaBuilder($dfa, $joinedNfa))->run();
585
586			$dfaRegExpFinishMap = [];
587			foreach ($dfa->getStateMap()->getFinishStateList() as $dfaFinishState) {
588			$nfaFinishStates = array_intersect(
589			$dfa->getStateMap()->getStateValue($dfaFinishState),
590			$joinedNfa->getStateMap()->getFinishStateList()
591			);
592			foreach ($regExpFinishMap as $regExp => $regExpFinishStates) {
593			foreach ($nfaFinishStates as $nfaFinishState) {
594			if (in_array($nfaFinishState, $regExpFinishStates)) {
595			$dfaRegExpFinishMap[$dfaFinishState] = (string) $regExp;
596			break 2;
597			}
598			}
599			}
600			}
601			foreach ($this->spec->getTokenSpecList($mode) as $tokenSpec) {
602			if (!in_array($tokenSpec->getRegExp(), $dfaRegExpFinishMap)) {
603			throw new Exception("Token not reachable for regular expression: {$tokenSpec->getRegExp()} ");
604			}
605			}
606			$this->regExpFinishMap[$mode] = $dfaRegExpFinishMap;
607
608			return $dfa;
609			}
610
611			/**
612			* @param Nfa $nfa
613			* @param int $entryState
614			* @param string $regExp
615			* @throws Exception
616			*/
617			private function buildRegExp(Nfa $nfa, int $entryState, string $regExp): void
618			{
619			$buffer = CharBufferFactory::createFromString($regExp);
620			$tree = new Tree();
621			ParserFactory::createFromBuffer($tree, $buffer)->run();
622			$nfaBuilder = new NfaBuilder($nfa, PropertyLoader::create());
623			$nfaBuilder->setStartState($entryState);
624			(new Translator($tree, $nfaBuilder))->run();
625			}
626
627			/**
628			* @param string $regExp
629			* @return Dfa
630			* @throws Exception
631			*/
632			private function buildIndependentRegExp(string $regExp): Dfa
633			{
634			$buffer = CharBufferFactory::createFromString($regExp);
635			$tree = new Tree();
636			ParserFactory::createFromBuffer($tree, $buffer)->run();
637			$nfa = new Nfa();
638			$nfaBuilder = new NfaBuilder($nfa, PropertyLoader::create());
639			(new Translator($tree, $nfaBuilder))->run();
640
641			return DfaBuilder::fromNfa($nfa);
642			}
643			}
644

remorhaz / php-unilex

Push — master ( 5848ce...73ad9b )

TokenMatcherGenerator::buildOnTransition() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like