From 4ffe36e4d07575c312d8e76cbe41c08c8f136a4f Mon Sep 17 00:00:00 2001
From: zadam <zadam.apps@gmail.com>
Date: Wed, 15 Feb 2023 16:08:00 +0100
Subject: [PATCH] added options for OCR and extract text from PDF

---
 .../widgets/type_widgets/content_widget.js    |  6 ++--
 .../options/images/extract_text_from_pdf.js   | 28 +++++++++++++++++++
 .../options/{ => images}/images.js            |  2 +-
 .../type_widgets/options/images/ocr.js        | 28 +++++++++++++++++++
 src/routes/api/options.js                     |  4 ++-
 5 files changed, 64 insertions(+), 4 deletions(-)
 create mode 100644 src/public/app/widgets/type_widgets/options/images/extract_text_from_pdf.js
 rename src/public/app/widgets/type_widgets/options/{ => images}/images.js (98%)
 create mode 100644 src/public/app/widgets/type_widgets/options/images/ocr.js

diff --git a/src/public/app/widgets/type_widgets/content_widget.js b/src/public/app/widgets/type_widgets/content_widget.js
index 7b6660158..34577b385 100644
--- a/src/public/app/widgets/type_widgets/content_widget.js
+++ b/src/public/app/widgets/type_widgets/content_widget.js
@@ -12,7 +12,7 @@ import VimKeyBindingsOptions from "./options/code_notes/vim_key_bindings.js";
 import WrapLinesOptions from "./options/code_notes/wrap_lines.js";
 import CodeAutoReadOnlySizeOptions from "./options/code_notes/code_auto_read_only_size.js";
 import CodeMimeTypesOptions from "./options/code_notes/code_mime_types.js";
-import ImageOptions from "./options/images.js";
+import ImageOptions from "./options/images/images.js";
 import SpellcheckOptions from "./options/spellcheck.js";
 import PasswordOptions from "./options/password.js";
 import EtapiOptions from "./options/etapi.js";
@@ -28,6 +28,8 @@ import ConsistencyChecksOptions from "./options/advanced/consistency_checks.js";
 import VacuumDatabaseOptions from "./options/advanced/vacuum_database.js";
 import DatabaseAnonymizationOptions from "./options/advanced/database_anonymization.js";
 import BackendLogWidget from "./content/backend_log.js";
+import OcrOptions from "./options/images/ocr.js";
+import ExtractTextFromPdfOptions from "./options/images/extract_text_from_pdf.js";
 
 const TPL = `<div class="note-detail-content-widget note-detail-printable">
     <style>
@@ -68,7 +70,7 @@ const CONTENT_WIDGETS = {
         CodeAutoReadOnlySizeOptions,
         CodeMimeTypesOptions
     ],
-    _optionsImages: [ ImageOptions ],
+    _optionsImages: [ ImageOptions, OcrOptions, ExtractTextFromPdfOptions ],
     _optionsSpellcheck: [ SpellcheckOptions ],
     _optionsPassword: [ PasswordOptions ],
     _optionsEtapi: [ EtapiOptions ],
diff --git a/src/public/app/widgets/type_widgets/options/images/extract_text_from_pdf.js b/src/public/app/widgets/type_widgets/options/images/extract_text_from_pdf.js
new file mode 100644
index 000000000..6d046c799
--- /dev/null
+++ b/src/public/app/widgets/type_widgets/options/images/extract_text_from_pdf.js
@@ -0,0 +1,28 @@
+import OptionsWidget from "../options_widget.js";
+
+const TPL = `
+<div class="options-section">
+    <h4>Extract text from PDF files</h4>
+    
+    <label>
+        <input class="extract-text-from-pdf" type="checkbox">
+        Extract text from PDF
+    </label>
+    
+    <p>Text extracted from PDFs will be considered when fulltext searching.</p>
+</div>
+`;
+
+export default class ExtractTextFromPdfOptions extends OptionsWidget {
+    doRender() {
+        this.$widget = $(TPL);
+
+        this.$extractTextFromPdf = this.$widget.find(".extract-text-from-pdf");
+        this.$extractTextFromPdf.on("change", () =>
+            this.updateCheckboxOption('extractTextFromPdf', this.$extractTextFromPdf));
+    }
+
+    optionsLoaded(options) {
+        this.setCheckboxState(this.$extractTextFromPdf, options.extractTextFromPdf);
+    }
+}
diff --git a/src/public/app/widgets/type_widgets/options/images.js b/src/public/app/widgets/type_widgets/options/images/images.js
similarity index 98%
rename from src/public/app/widgets/type_widgets/options/images.js
rename to src/public/app/widgets/type_widgets/options/images/images.js
index fdc69df1f..38eae0db8 100644
--- a/src/public/app/widgets/type_widgets/options/images.js
+++ b/src/public/app/widgets/type_widgets/options/images/images.js
@@ -1,4 +1,4 @@
-import OptionsWidget from "./options_widget.js";
+import OptionsWidget from "../options_widget.js";
 
 const TPL = `
 <div class="options-section">
diff --git a/src/public/app/widgets/type_widgets/options/images/ocr.js b/src/public/app/widgets/type_widgets/options/images/ocr.js
new file mode 100644
index 000000000..649243696
--- /dev/null
+++ b/src/public/app/widgets/type_widgets/options/images/ocr.js
@@ -0,0 +1,28 @@
+import OptionsWidget from "../options_widget.js";
+
+const TPL = `
+<div class="options-section">
+    <h4>OCR</h4>
+    
+    <label>
+        <input class="ocr-images" type="checkbox">
+        Extract text from images using OCR
+    </label>
+    
+    <p>Text extracted from images will be considered when fulltext searching.</p>
+</div>
+`;
+
+export default class OcrOptions extends OptionsWidget {
+    doRender() {
+        this.$widget = $(TPL);
+
+        this.$ocrImages = this.$widget.find(".ocr-images");
+        this.$ocrImages.on("change", () =>
+            this.updateCheckboxOption('ocrImages', this.$ocrImages));
+    }
+
+    optionsLoaded(options) {
+        this.setCheckboxState(this.$ocrImages, options.ocrImages);
+    }
+}
diff --git a/src/routes/api/options.js b/src/routes/api/options.js
index ddb4321ec..dc1007be2 100644
--- a/src/routes/api/options.js
+++ b/src/routes/api/options.js
@@ -61,7 +61,9 @@ const ALLOWED_OPTIONS = new Set([
     'downloadImagesAutomatically',
     'minTocHeadings',
     'checkForUpdates',
-    'disableTray'
+    'disableTray',
+    'ocrImages',
+    'extractTextFromPdf'
 ]);
 
 function getOptions() {