From 51fadf7435a8445272a5159202f6ecdb25d26f5c Mon Sep 17 00:00:00 2001
From: Maxence Lange <maxence@artificial-owl.com>
Date: Wed, 13 Feb 2019 08:47:13 -0100
Subject: [PATCH] adding settings to enable/disable pdf ocr

---
 js/admin.elements.js             |  3 +++
 js/admin.settings.js             |  4 +++-
 lib/Service/ConfigService.php    |  4 +++-
 lib/Service/TesseractService.php |  4 ++++
 templates/settings.admin.php     | 12 ++++++++++++
 5 files changed, 25 insertions(+), 2 deletions(-)

diff --git a/js/admin.elements.js b/js/admin.elements.js
index 9969895..aebfb1e 100644
--- a/js/admin.elements.js
+++ b/js/admin.elements.js
@@ -34,16 +34,19 @@ var fts_tesseract_elements = {
 	tesseract_ocr: null,
 	tesseract_psm: null,
 	tesseract_lang: null,
+	tesseract_pdf: null,
 
 	init: function () {
 		fts_tesseract_elements.tesseract_div = $('#files_ocr-tesseract');
 		fts_tesseract_elements.tesseract_psm = $('#tesseract_psm');
 		fts_tesseract_elements.tesseract_lang = $('#tesseract_lang');
 		fts_tesseract_elements.tesseract_ocr = $('#tesseract_ocr');
+		fts_tesseract_elements.tesseract_pdf = $('#tesseract_pdf');
 
 		fts_tesseract_elements.tesseract_ocr.on('change', fts_tesseract_elements.updateSettings);
 		fts_tesseract_elements.tesseract_psm.on('change', fts_tesseract_elements.updateSettings);
 		fts_tesseract_elements.tesseract_lang.on('change', fts_tesseract_elements.updateSettings);
+		fts_tesseract_elements.tesseract_pdf.on('change', fts_tesseract_elements.updateSettings);
 	},
 
 
diff --git a/js/admin.settings.js b/js/admin.settings.js
index d590817..911fe88 100644
--- a/js/admin.settings.js
+++ b/js/admin.settings.js
@@ -49,6 +49,7 @@ var fts_tesseract_settings = {
 		fts_tesseract_elements.tesseract_ocr.prop('checked', (result.tesseract_enabled === '1'));
 		fts_tesseract_elements.tesseract_psm.val(result.tesseract_psm);
 		fts_tesseract_elements.tesseract_lang.val(result.tesseract_lang);
+		fts_tesseract_elements.tesseract_pdf.prop('checked', (result.tesseract_pdf === '1'));
 
 		fts_admin_settings.tagSettingsAsSaved(fts_tesseract_elements.tesseract_div);
 
@@ -69,7 +70,8 @@ var fts_tesseract_settings = {
 		var data = {
 			tesseract_enabled: (fts_tesseract_elements.tesseract_ocr.is(':checked')) ? 1 : 0,
 			tesseract_psm: fts_tesseract_elements.tesseract_psm.val(),
-			tesseract_lang: fts_tesseract_elements.tesseract_lang.val()
+			tesseract_lang: fts_tesseract_elements.tesseract_lang.val(),
+			tesseract_pdf: (fts_tesseract_elements.tesseract_pdf.is(':checked')) ? 1 : 0
 		};
 
 		$.ajax({
diff --git a/lib/Service/ConfigService.php b/lib/Service/ConfigService.php
index b706a4d..c55fd08 100644
--- a/lib/Service/ConfigService.php
+++ b/lib/Service/ConfigService.php
@@ -46,11 +46,13 @@ class ConfigService {
 	const TESSERACT_ENABLED = 'tesseract_enabled';
 	const TESSERACT_PSM = 'tesseract_psm';
 	const TESSERACT_LANG = 'tesseract_lang';
+	const TESSERACT_PDF = 'tesseract_pdf';
 
 	public $defaults = [
 		self::TESSERACT_ENABLED => '0',
 		self::TESSERACT_PSM     => '4',
-		self::TESSERACT_LANG    => 'eng'
+		self::TESSERACT_LANG    => 'eng',
+		self::TESSERACT_PDF     => '0'
 	];
 
 
diff --git a/lib/Service/TesseractService.php b/lib/Service/TesseractService.php
index 16383b7..bf026ea 100644
--- a/lib/Service/TesseractService.php
+++ b/lib/Service/TesseractService.php
@@ -210,6 +210,10 @@ private function ocrPdf(AFilesDocument $document, File $file): bool {
 			return false;
 		}
 
+		if ($this->configService->getAppValue(ConfigService::TESSERACT_PDF) !== '1') {
+			return true;
+		}
+
 		try {
 			$path = $this->getAbsolutePath($file);
 			$pdf = new Pdf($path);
diff --git a/templates/settings.admin.php b/templates/settings.admin.php
index 8ec74f7..2f1717f 100644
--- a/templates/settings.admin.php
+++ b/templates/settings.admin.php
@@ -75,6 +75,18 @@
 				<input type="text" class="big" id="tesseract_lang" value=""/>
 			</div>
 		</div>
+
+		<div class="div-table-row tesseract_ocr_enabled">
+			<div class="div-table-col div-table-col-left">
+				<span class="leftcol">PDF</span>
+				<br/>
+				<em>enable the OCR of PDF (heavy on resource)</em>
+			</div>
+			<div class="div-table-col">
+				<input type="checkbox" id="tesseract_pdf" value="1"/>
+			</div>
+		</div>
+
 	</div>