add CSV input format support

rBiblia · Nov 24, 2023 · ee36fed · ee36fed
1 parent 42b5e67
commit ee36fed
Show file tree

Hide file tree

Showing 5 changed files with 140 additions and 89 deletions.
diff --git a/README.md b/README.md
@@ -5,13 +5,14 @@ This conversion tool allows you to convert Bible translation in one of the suppo
 
 Supported input formats are:
 
-Name               | Id      | File extension
--------------------|---------|---------------
-eSword             | sword   | *.bblx, *.mybible
-Zefania            | zefania | *.xml
-theWord            | word    | *.ot, *.nt, *.ont
-tab as a separator | tsv     | *.tsv
-MyBible (Android)  | mybible | *.SQLite3 
+Name                 | Id      | File extension
+---------------------|---------|---------------
+eSword               | sword   | *.bblx, *.mybible
+Zefania              | zefania | *.xml
+theWord              | word    | *.ot, *.nt, *.ont
+tab as a separator   | tsv     | *.tsv
+comma as a separator | csv     | *.csv  
+MyBible (Android)    | mybible | *.SQLite3 
 
 Compilation steps:
 -------------------
@@ -46,6 +47,9 @@ php bib2xml.phar convert input_translation.bblx
 Changelog:
 ----------
 
+* v0.6.0.0 (2023-11-24):
+    - added `csv` input format support
+
 * v0.5.0.0 (2022-03-17):
     - upgraded PHP to v8
     - added `mybible` input format support
@@ -75,7 +79,7 @@ Changelog:
 About
 -----
 
-Original author: [Rafał Toborek](https://kontakt.toborek.info)
+Author: [Rafał Toborek](https://kontakt.toborek.info)
 
 Feel free to contribute and [support my work](https://rbiblia.toborek.info/donation/).
 

diff --git a/project/src/Command/ConvertCommand.php b/project/src/Command/ConvertCommand.php
@@ -34,6 +34,10 @@ class ConvertCommand extends Command
             'name' => 'TSV (tab as a separator)',
             'ext' => ['tsv'],
         ],
+        'csv' => [
+            'name' => 'CSV (comma as a separator)',
+            'ext' => ['csv'],
+        ],
         'mybible' => [
             'name' => 'MyBible',
             'ext' => ['sqlite3'],

diff --git a/project/src/Input/CsvInput.php b/project/src/Input/CsvInput.php
@@ -0,0 +1,14 @@
+<?php
+
+declare(strict_types=1);
+
+namespace rBibliaBibleConverter\Input;
+
+use rBibliaBibleConverter\Input\Traits\CharSeparatedTrait;
+
+class CsvInput implements InputConverter
+{
+    use CharSeparatedTrait;
+
+    private string $separator = ';';
+}
diff --git a/project/src/Input/Traits/CharSeparatedTrait.php b/project/src/Input/Traits/CharSeparatedTrait.php
@@ -0,0 +1,107 @@
+<?php
+
+declare(strict_types=1);
+
+namespace rBibliaBibleConverter\Input\Traits;
+
+use rBibliaBibleConverter\Bible\Books;
+use rBibliaBibleConverter\Reader\InputReader;
+use rBibliaBibleConverter\Text\Sanitizer;
+use rBibliaBibleConverter\Value\About;
+use rBibliaBibleConverter\Value\Book;
+use rBibliaBibleConverter\Value\Chapter;
+use rBibliaBibleConverter\Value\Translation;
+use rBibliaBibleConverter\Value\Verse;
+
+trait CharSeparatedTrait
+{
+    /** @var InputReader */
+    private $input;
+
+    /** @var Sanitizer */
+    private $sanitizer;
+
+    /** @var Books */
+    private $books;
+
+    public function __construct(InputReader $input, Sanitizer $sanitizer, Books $books)
+    {
+        $this->input = $input;
+        $this->sanitizer = $sanitizer;
+        $this->books = $books;
+    }
+
+    public function getTranslation(): Translation
+    {
+        $translationArray = [];
+
+        $content = explode(\PHP_EOL, $this->input->getContent());
+        foreach ($content as $line) {
+            if (empty(trim($line))) {
+                continue;
+            }
+
+            $verseAsArray = explode($this->separator, $line);
+
+            $bookId = $verseAsArray[0];
+            $chapterId = $verseAsArray[1];
+            $verseId = $verseAsArray[2];
+
+            // more than one separator could appear in verse body
+            $verse = '';
+            $count = count($verseAsArray);
+            for ($i = 3; $i < $count; $i++) {
+                $verse .= $verseAsArray[$i];
+
+                if ($i < $count-1) {
+                    $verse .= $this->separator;
+                }
+            }
+
+            if (!isset($translationArray[$bookId])) {
+                $translationArray[$bookId] = [];
+            }
+
+            if (!isset($translationArray[$bookId][$chapterId])) {
+                $translationArray[$bookId][$chapterId] = [];
+            }
+
+            if (!isset($translationArray[$bookId][$chapterId][$verseId])) {
+                $translationArray[$bookId][$chapterId][$verseId] = trim(str_replace(['\f6 ', '&ndash;', '&mdash;'], ['', '-', '-'], $verse));
+            }
+        }
+
+        $translation = new Translation($this->getAbout());
+
+        foreach ($translationArray as $bookId => $books) {
+            $book = new Book($this->books->convertNameToOneOfTheSupportedBooksId((string) $bookId));
+
+            foreach ($books as $chapterId => $chapters) {
+                $chapter = new Chapter($chapterId);
+
+                foreach ($chapters as $verseId => $content) {
+                    $chapter->addVerse(new Verse($verseId, $content));
+                }
+
+                $book->addChapter($chapter);
+            }
+
+            $translation->addBook($book);
+        }
+
+        return $translation;
+    }
+
+    private function getAbout(): About
+    {
+        return new About(
+            '',
+            '',
+            '',
+            '',
+            '',
+            '',
+            ''
+        );
+    }
+}
diff --git a/project/src/Input/TsvInput.php b/project/src/Input/TsvInput.php
@@ -4,89 +4,11 @@
 
 namespace rBibliaBibleConverter\Input;
 
-use rBibliaBibleConverter\Bible\Books;
-use rBibliaBibleConverter\Reader\InputReader;
-use rBibliaBibleConverter\Text\Sanitizer;
-use rBibliaBibleConverter\Value\About;
-use rBibliaBibleConverter\Value\Book;
-use rBibliaBibleConverter\Value\Chapter;
-use rBibliaBibleConverter\Value\Translation;
-use rBibliaBibleConverter\Value\Verse;
+use rBibliaBibleConverter\Input\Traits\CharSeparatedTrait;
 
 class TsvInput implements InputConverter
 {
-    /** @var InputReader */
-    private $input;
+    use CharSeparatedTrait;
 
-    /** @var Sanitizer */
-    private $sanitizer;
-
-    /** @var Books */
-    private $books;
-
-    public function __construct(InputReader $input, Sanitizer $sanitizer, Books $books)
-    {
-        $this->input = $input;
-        $this->sanitizer = $sanitizer;
-        $this->books = $books;
-    }
-
-    public function getTranslation(): Translation
-    {
-        $translationArray = [];
-
-        $content = explode(\PHP_EOL, $this->input->getContent());
-        foreach ($content as $line) {
-            if (empty(trim($line))) {
-                continue;
-            }
-
-            [$bookId, $chapterId, $verseId, $verse] = explode("\t", $line);
-
-            if (!isset($translationArray[$bookId])) {
-                $translationArray[$bookId] = [];
-            }
-
-            if (!isset($translationArray[$bookId][$chapterId])) {
-                $translationArray[$bookId][$chapterId] = [];
-            }
-
-            if (!isset($translationArray[$bookId][$chapterId][$verseId])) {
-                $translationArray[$bookId][$chapterId][$verseId] = trim(str_replace(['\f6 ', '&ndash;', '&mdash;'], ['', '-', '-'], $verse));
-            }
-        }
-
-        $translation = new Translation($this->getAbout());
-
-        foreach ($translationArray as $bookId => $books) {
-            $book = new Book($this->books->convertNameToOneOfTheSupportedBooksId((string) $bookId));
-
-            foreach ($books as $chapterId => $chapters) {
-                $chapter = new Chapter($chapterId);
-
-                foreach ($chapters as $verseId => $content) {
-                    $chapter->addVerse(new Verse($verseId, $content));
-                }
-
-                $book->addChapter($chapter);
-            }
-
-            $translation->addBook($book);
-        }
-
-        return $translation;
-    }
-
-    private function getAbout(): About
-    {
-        return new About(
-            '',
-            '',
-            '',
-            '',
-            '',
-            '',
-            ''
-        );
-    }
+    private string $separator = "\t";
 }