Merge pull request dwhly-proj#106 from dwhly-proj/paul/ocr_force_update

Paul/ocr force update
volt4ire · Aug 21, 2021 · 2520c2b · 2520c2b
2 parents f6bdd0a + 805ca05
commit 2520c2b
Show file tree

Hide file tree

Showing 5 changed files with 46 additions and 14 deletions.
diff --git a/droppdf/apps/_static/apps/js/ocr_pdf_upload.js b/droppdf/apps/_static/apps/js/ocr_pdf_upload.js
@@ -79,10 +79,17 @@
                         $('#run-button')
                             .addClass('disabled')
                             .attr('disabled', true)
+
+                    } else if (response.file_info.has_text) {
+
+                        $('#run-button').hide();
+                        $('#upload-has-text').show();
+
                     } else {
                         $('#run-button')
                             .removeClass('disabled')
                             .attr('disabled', false)
+                            .show()
                     }
                 };
             },
@@ -164,8 +171,13 @@
     $(document).ready(function() {
         upload_in_progress = false;
         $('.button-box').hide();
+        $('#upload-has-text').hide();
         $('#in-progress').hide();
-        $('#pdf-file').attr('disabled', false);
+
+        $('#pdf-file')
+            .attr('disabled', false)
+            .val('')
+
         $('#progress-bar-inner').css('width', '0%');
         $('#pdf-file')
             .attr('disabled', false)

diff --git a/droppdf/apps/_templates/ocr_pdf.html b/droppdf/apps/_templates/ocr_pdf.html
@@ -80,6 +80,15 @@
             color: red;
         }
 
+        .success-msg {
+            padding-top: 1em;
+            color: green;
+        }
+
+        .black-text {
+            color: #000;
+        }
+
         .button-box {
             display: flex;
             flex: 1;
@@ -245,7 +254,7 @@
             Add text to an image PDF.
         </div>
 
-        <form id="param-form" action="/ocr/upload/" method="POST" enctype="multipart/form-data">
+        <form id="param-form" action="/ocr/upload/" method="POST" autocomplete="off" enctype="multipart/form-data">
 
             <div class="option-box">
             </div>
@@ -288,19 +297,27 @@ <h3 id="filename"></h3>
                         <i class="fa fa-upload"></i>
                         Run OCR
                     </div>
-
-                    <div id="run-force" class="button" style="width: 8em" onclick="runOCR(true)">
-                        <i class="fa fa-upload"></i>
-                        Force OCR
-                    </div>
-                    <div class="help-label">
-                        <i class="fa fa-question-circle" title="Forcing OCR will replace any existing text with OCRd text. If there is a question, it's likely better to try 'Run OCR' first."></i>
-                    </div>
                 </div>
 
                 <div id="upload-error" class="small error" style="display: none;">
                 </div>
 
+                <div id="upload-has-text" class="small success-msg" style="display: none;">
+                    <span style="font-size: 1.2em">
+                        <i class="fa fa-check"></i>
+                    </span>
+                    This document already has selectable text and is most likely ready to use for annotations.
+                    <br><br>
+                    If the quality of the current OCR'd text is not acceptable you can redo the OCR process which may give better results.
+                    <br>
+                    (This process will take some time and in rare instances will not be successful, depending on the qualities of the uploaded document). 
+                    <br><br>
+                    <div id="run-force" class="button black-text" style="width: 8em" onclick="runOCR(true)">
+                        <i class="fa fa-upload"></i>
+                        Redo OCR
+                    </div>
+                </div>
+
             </div>
 
         </form>

diff --git a/droppdf/apps/_templates/ocr_pdf_result.html b/droppdf/apps/_templates/ocr_pdf_result.html
@@ -162,7 +162,7 @@
                 OCR in progress. Started at <span id="time-start"></span>
             </p>
             <p>
-                Large files can sometimes take several minutes, or up to five or ten if the Force OCR option is used. 
+                Large files can sometimes take several minutes, or up to five or ten if the OCR is a redo with existing text. 
             </p>
             <p>
             A clickable download link will appear when processing is complete.

diff --git a/droppdf/apps/apps/ocr/views.py b/droppdf/apps/apps/ocr/views.py
@@ -70,7 +70,10 @@ def upload(request):
 
         #already_has_text?
         if check_pdf_has_text(new_filename):
-            processing_error = 'This PDF already has text. Use the "Force OCR" button to overwrite text with a fresh OCR if desired. If file was OCRd on previous upload those results will be provided'
+            #processing_error = 'This PDF already has text. Use the "Force OCR" button to overwrite text with a fresh OCR if desired. If file was OCRd on previous upload those results will be provided'
+            has_text = True
+        else:
+            has_text = False
 
 
         if not existing_name:
@@ -114,7 +117,7 @@ def upload(request):
         data = {'file_info': {'filename': filename, 'size': file_.size,
                     'new_filename': new_filename, 'processing_error': processing_error,
                     'tempfile_path': tempfile_path, 'already_exists': already_exists,
-                    'md5_hash': md5_hash}}
+                    'md5_hash': md5_hash, 'has_text': has_text}}
 
         return JsonResponse(data)
 

diff --git a/droppdf/droppdf/celery.py b/droppdf/droppdf/celery.py
@@ -5,7 +5,7 @@
 
 from django.conf import settings
 
-app = Celery('financial_planning_app', broker=settings.BROKER_URL)
+app = Celery('droppdf_app', broker=settings.BROKER_URL)
 
 app.config_from_object('django.conf:settings', namespace='CELERY')
 app.autodiscover_tasks(lambda: settings.INSTALLED_APPS)