From 22472473056161bf84a2d80f576e116b66fb5633 Mon Sep 17 00:00:00 2001
From: Lester Caine <lester@lsces.co.uk>
Date: Tue, 2 Jun 2026 13:48:57 +0100
Subject: PDF edit: protect text layer, add read-only view and reload option

- Preserve liberty_content.data on save: copy existing data into edit
  before store() so the format verify plugin does not wipe it with null
  when the textarea is absent from a PDF form submission
- Re-extract text layer on demand: reload_pdf checkbox calls
  mime_pdf_text_extract() via mStorage source_file, mirrors the existing
  generate_thumbnails pattern in edit_image.php
- Template: show extracted text in a read-only scrollable textarea for
  PDFs (no name attr so not submitted); separate Re-extract checkbox;
  editable description textarea shown only for non-PDF attachments

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---
 edit_image.php           | 25 +++++++++++++++++++++++++
 templates/edit_image.tpl | 16 ++++++++++++++++
 2 files changed, 41 insertions(+)

diff --git a/edit_image.php b/edit_image.php
index 22ae0cb..aaa00e6 100755
--- a/edit_image.php
+++ b/edit_image.php
@@ -64,6 +64,12 @@ if( !empty($_REQUEST['saveImage']) || !empty($_REQUEST['regenerateThumbnails'] )
 		}
 	}
 
+	$isPdf = !empty( $gContent->mInfo['mime_type'] ) && $gContent->mInfo['mime_type'] === 'application/pdf';
+	if( $isPdf && empty( $_REQUEST['reload_pdf'] ) ) {
+		// Preserve the extracted text layer — don't let an empty edit field wipe liberty_content.data
+		$_REQUEST['edit'] = $gContent->mInfo['data'];
+	}
+
 	$_REQUEST['purge_from_galleries'] = true;
 	if( $gContent->store($_REQUEST) ) {
 		// refresh all hashes
@@ -93,6 +99,25 @@ if( !empty($_REQUEST['saveImage']) || !empty($_REQUEST['regenerateThumbnails'] )
 		if( !empty( $_REQUEST['generate_thumbnails'] ) ) {
 			$gContent->generateThumbnails();
 		}
+		if( $isPdf && !empty( $_REQUEST['reload_pdf'] ) ) {
+			foreach( $gContent->mStorage as $attachmentId => $attachment ) {
+				if( !empty( $attachment['source_file'] ) ) {
+					$reloadHash = [
+						'attachment_id' => $attachmentId,
+						'content_id'    => $gContent->mContentId,
+						'upload'        => ['source_file' => $attachment['source_file']],
+					];
+					\Bitweaver\Liberty\mime_pdf_text_extract( $reloadHash );
+					if( !empty( $reloadHash['data'] ) ) {
+						$gBitSystem->mDb->query(
+							"UPDATE `".BIT_DB_PREFIX."liberty_content` SET `data` = ? WHERE `content_id` = ?",
+							[$reloadHash['data'], $gContent->mContentId]
+						);
+					}
+					break;
+				}
+			}
+		}
 		if( empty( $gContent->mErrors ) ) {
 			// add a refresh parameter to the URL so the thumbnails will properly refresh first go reload
 			header( 'Location: '.$gContent->getDisplayUrl().($gBitSystem->isFeatureActive( 'pretty_urls' ) ? '?' : '&' ).'refresh=1' );
diff --git a/templates/edit_image.tpl b/templates/edit_image.tpl
index c2e1913..7ad5c9c 100755
--- a/templates/edit_image.tpl
+++ b/templates/edit_image.tpl
@@ -40,12 +40,28 @@
 							{/forminput}
 						</div>
 
+						{if $gContent->mInfo.mime_type == 'application/pdf'}
+						<div class="form-group">
+							{formlabel label="Extracted Text"}
+							{forminput}
+								<textarea class="input-xlarge" rows="6" cols="50" readonly="readonly">{$gContent->mInfo.data|escape}</textarea>
+							{/forminput}
+						</div>
+						<div class="form-group">
+							{forminput}
+								{forminput label="checkbox"}
+									<input type="checkbox" name="reload_pdf" value="1"/> {tr}Re-extract text layer from PDF{/tr}
+								{/forminput}
+							{/forminput}
+						</div>
+						{else}
 						<div class="form-group">
 							{formlabel label="Description" for="image-desc"}
 							{forminput}
 								<textarea name="edit" class="input-xlarge" id="image-desc" rows="4" cols="50">{$gContent->mInfo.data|escape}</textarea>
 							{/forminput}
 						</div>
+						{/if}
 
 						<div class="form-group">
 							{if $gContent->getDownloadUrl()}
-- 
cgit v1.3