gitlab_medici/app/utils/TextParser.php @ 67ebe0dc

<?php


namespace App\Utils;

class TextParser
{
    private $textFile;
    private $objectTypes;
    private $surfaceTypes;

    public function __construct($file)
    {
        $this->textFile = $file;
    }

    /** Rozparsuje importovaný text */
    public function parseText()
    {
        $array = explode("\n", file_get_contents($this->textFile));
        $array = array_map('trim', $array);

        $nextBook = true;
        $output = array();
        $currentBook = "";
        $currentChapter = "";
        $currentSurface = $this->transformAbbrev("obverse");

        //TODO kontrola a parsovani object types (tablet, envelope)
        foreach ($array as $value) {
            if (strlen($value) == 0) {
                $nextBook = true;
            } elseif ($value[0] == '|' && $nextBook) {
                $currentBook = substr($value, 1);
                $currentChapter = "";
                $currentSurface = $this->transformAbbrev("obverse");
                $nextBook = false;
            } elseif ($value[0] == '|') {
                $currentChapter = substr($value, 1);
                $currentSurface = $this->transformAbbrev("obverse");
            } else {
                if (strchr($value, '$')) {
                    $currentSurface = $this->transformAbbrev(substr(strchr($value, "&"), 1, -1));
                    $output[$currentBook][$currentChapter][$currentSurface][] = trim(substr($value, 0, stripos($value, "&")));
                } else {
                    $output[$currentBook][$currentChapter][$currentSurface][] = $value;
                }
            }
        }

        return $output;
    }

    /** Nejvíc kontrola souboru, kontroluje zda soubor existuje a zda aspoň začátek souboru vyhovuje formátu */
    public function isFileParsable()
    {
        if (file_exists($this->textFile)) {
            $checker = fopen($this->textFile, "r");
            if (fgets($checker)[0] == '|') {
                return true;
            } else {
                return false;
            }
        } else {
            return false;
        }
    }

    public function setTypes($objects, $surfaces)
    {
        $this->objectTypes = $objects;
        $this->surfaceTypes = $surfaces;
    }

    /** Funkce pro transformaci zkratek surfaceType na jejich příslušná idčka v db */
    private function transformAbbrev($abbrev)
    {
        switch ($abbrev) {
            case "rev.":
                return array_search("reverse", $this->surfaceTypes);
            case "u.e.":
                return array_search("upper edge", $this->surfaceTypes);
            case "lo.e.":
                return array_search("lower edge", $this->surfaceTypes);
            case "le.e.":
                return array_search("left edge", $this->surfaceTypes);
            // Domnenky, v test souborech se nenachazi -> udelat podle toho export
            case "r.e.":
                return array_search("right edge", $this->surfaceTypes);
            case "s.i.1":
                return array_search("seal impression 1", $this->surfaceTypes);
            case "s.i.2":
                return array_search("seal impression 2", $this->surfaceTypes);
            case "s.i.3":
                return array_search("seal impression 3", $this->surfaceTypes);
            case "s.i.4":
                return array_search("seal impression 4", $this->surfaceTypes);
            case "s.i.5":
                return array_search("seal impression 5", $this->surfaceTypes);
            case "s.i.6":
                return array_search("seal impression 6", $this->surfaceTypes);
            case "s.i.7":
                return array_search("seal impression 7", $this->surfaceTypes);
            case "s.i.8":
                return array_search("seal impression 8", $this->surfaceTypes);
            default:
                return array_search("obverse", $this->surfaceTypes);
        }
    }

    public function getTextFile()
    {
        return $this->textFile;
    }
}