Remorhaz\UniLex\RegExp\PropertyBuilder::buildScripts()

Complexity

Conditions	11
Paths	38

Size

Total Lines	70
Code Lines	48

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

<?php
    public function buildScripts(array $index): array
    {
        /** @var RangeSet[] $ranges */
        $ranges = [];
        $source = new SplFileObject(__DIR__ . '/../../data/Scripts.txt');
        $lastKnownCode = null;
        $unknownRanges = [];

        echo "Parsing: ";
        $rangeCount = 0;

        while (!$source->eof()) {
            $line = $source->fgets();
            if (false === $line) {
                throw new RuntimeException("Error reading line from scripts file");
            }
            $dataWithComment = explode('#', $line, 2);
            $data = trim($dataWithComment[0] ?? '');
            if ('' == $data) {
                continue;
            }
            $rangeWithProp = explode(';', $data);
            $unsplittedRange = trim($rangeWithProp[0] ?? null);
            $prop = trim($rangeWithProp[1] ?? null);
            if (!isset($unsplittedRange, $prop)) {
                throw new RuntimeException("Invalid range or property");
            }
            $splittedRange = explode('..', $unsplittedRange);
            $start = hexdec($splittedRange[0]);
            $finish = isset($splittedRange[1])
                ? hexdec($splittedRange[1])
                : $start;
            if (!isset($lastKnownCode)) {
                if ($start > 0) {
                    $unknownRanges[] = new Range(0, $start - 1);

$start - 1 of type double is incompatible with the type integer|null expected by parameter $finish of Remorhaz\UniLex\RegExp\FSM\Range::__construct().

                }
            } elseif ($start - $lastKnownCode > 1) {
                $unknownRanges[] = new Range($lastKnownCode + 1, $start - 1);
            }
            $lastKnownCode = $finish;

            if (!isset($ranges[$prop])) {
                $ranges[$prop] = [];
            }
            $range = new Range($start, $finish);

It seems like $start can also be of type double; however, parameter $start of Remorhaz\UniLex\RegExp\FSM\Range::__construct() does only seem to accept integer, maybe add an additional type check?

            $ranges[$prop][] = $range;
            echo ".";
            $rangeCount++;
        }
        $ranges['Unknown'] = $unknownRanges;
        echo ". {$rangeCount} ranges\n";
        $source = null;

The assignment to $source is dead and can be removed.

        echo "Building range sets: ";
        $rangeSetCount = 0;
        $rangeSets = [];
        foreach ($ranges as $prop => $rangeList) {
            $rangeSets[$prop] = new RangeSet(...$rangeList);
            echo ".";
            $rangeSetCount++;
        }
        echo " {$rangeSetCount} range sets\n";

        /*echo "Build Unknown property...";
        $fullRangeSet = new RangeSet(new Range(0x0, 0x10FFFF));
        $knownRangeSet = new RangeSet(...$knownRanges);
        $unknownRangeSet = $rangeSetCalc->xor($fullRangeSet, $knownRangeSet);
        $ranges['Unknown'] = $unknownRangeSet;*/

        return $this->dumpProps($index, $rangeSets);
    }