en_btr_backend/app/Handlers/UlbXmlImportHandler.php

<?php
namespace App\Handlers;

use App\Book;
use App\Chapter;
use App\Word;
use App\Verse;
use DOMDocument;
use DOMElement;
/**
 * UlbXmlImportHandler.php
 *
 * @author: Leonard Smith <leonard@acornwebconsultants.com>
 * Date: 10/2/20
 * Time: 8:04 AM
 */
class UlbXmlImportHandler
{
    use BookXmlFilesTrait;

    const REPO_FOLDER = '/ulb/';

    /**
     * @var array
     */
    protected $availableFiles = [];

    public function run() : void
    {
        // Collect available xml files
        foreach (self::$bookXmlFiles as $name => $filename)
        {
            $filepath = $this->getBookXmlFilePath(self::REPO_FOLDER, $name);

            if (file_exists($filepath)) {
                $this->availableFiles[$name] = $filepath;
            }
        }

        $this->loopOverXmlFiles();
    }

    public function loopOverXmlFiles() : void
    {
        foreach ($this->availableFiles as $name => $filepath) {
            $this->importBook($name, $filepath);
        }
    }

    /**
     * @param string $bookTitle
     * @param string $filepath
     * @return Book
     */
    public function importBook(string $bookTitle, string $filepath) : Book
    {
        /** @var DOMDocument $document */
        $document = $this->openBook($filepath);

        $book = Book::create([
            'id' => strtolower($bookTitle),
            'name' => ucfirst($bookTitle),
        ]);

        $chapterCollection = $document->getElementsByTagName('chapter');

        foreach ($chapterCollection as $chapterElem) {
            $chapter = $this->importChapter($chapterElem, $book->id);
            $book->chapters()->save($chapter);
        }

        return $book;
    }

    /**
     * @param DOMElement $chapterElem
     * @return Chapter
     */
    public function importChapter(DOMElement $chapterElem, string $book_id) : Chapter
    {
        $chapterName = $this->parseChapterName($chapterElem);

        $chapter = Chapter::create([
            'id' => implode('-', [$book_id, $chapterName]),
            'name' => $chapterName,
        ]);

        $verseCollection = $chapterElem->getElementsByTagName('verse');

        foreach ($verseCollection as $verseElem) {
            $verse = $this->importVerse($verseElem, $chapter->id);
            $chapter->verses()->save($verse);
        }

        return $chapter;
    }

    /**
     * @param DOMElement $verseElem
     * @return Verse
     */
    public function importVerse(DOMElement $verseElem, string $chapter_id) : Verse
    {
        $verseNumber = $this->parseVerseNumber($verseElem);

        $verse = Verse::create([
            'id' => implode('-', [$chapter_id, $verseNumber]),
            'name' => $verseNumber,
            'greek_text' => $verseElem->getElementsByTagName('Greek')[0]->nodeValue,
            'ulb_text' => $verseElem->getElementsByTagName('ULB')[0]->nodeValue,
        ]);

        $wordCollection = $verseElem->getElementsByTagName('w');

        foreach ($wordCollection as $wordElem) {
            $word = $this->importWord($wordElem, $verse->id);
            $verse->words()->save($word);
        }

        return $verse;
    }

    /**
     * @param DOMElement $wordElem
     * @return Word
     */
    public function importWord(DOMElement $wordElem, string $verse_id) : Word
    {
        $ognt_sort = $wordElem->getAttribute('OGNTsort');

        // NOTE: We have to switch things around a bit as the incoming XML file
        // uses lexeme for lemma and lemma for the strongs number
        $word = Word::create([
            'id' => implode('-', [$verse_id, $ognt_sort]),
            'ulb' => $wordElem->nodeValue,
            'greek' => $wordElem->getAttribute('text'),
            'lemma' =>  $wordElem->getAttribute('lexeme'),
            'morph' => $wordElem->getAttribute('morph'),
            'ognt_sort' => $ognt_sort,
            'strongs_number' => $this->formatStrongsNumber($wordElem->getAttribute('lemma')),
        ]);

        return $word;
    }

    /**
     * Not all of the strongs numbers coming from the XML files are formatted the same. Let's fix that here.
     *
     * @param string $strongsNumber
     * @return string
     */
    protected function formatStrongsNumber(string $strongsNumber)
    {
        return 'G' . ltrim($strongsNumber, "Gg");
    }

    /**
     * @param DOMElement $chapterElem
     * @return string
     */
    protected function parseChapterName(DOMElement $chapterElem) : string
    {
        $osisId = $chapterElem->getAttribute('osisID');

        preg_match("|^[A-Za-z1-3]*.([0-9]*)$|", $osisId, $matches);

        return $matches[1];
    }

    /**
     * @param DOMElement $verseElem
     * @return string
     */
    protected function parseVerseNumber(DOMElement $verseElem) : string
    {
        $string = $verseElem->getAttribute('name');
        preg_match("|^[A-Za-z1-3]*\s*[0-9]*:([0-9]*)$|", $string, $matches);

        return $matches[1];
    }

    /**
     * @param $filepath
     * @return DOMDocument
     */
    protected function openBook($filepath) : DOMDocument
    {
        $document = new \DOMDocument;
        $document->load($filepath);

        return $document;
    }
}