subtitles: support srt files and convert to vtt

Author: zubeydecivelek

cds/modules/deposit/ext.py

@@ -26,14 +26,18 @@
 
 import re
 import mimetypes
+import tempfile
+import os
+from io import BytesIO
 
 from invenio_base.signals import app_loaded
 from invenio_db import db
-from invenio_files_rest.models import ObjectVersionTag
+from invenio_files_rest.models import ObjectVersion, ObjectVersionTag
 from invenio_files_rest.signals import file_uploaded
 from invenio_files_rest.errors import InvalidKeyError
 from invenio_indexer.signals import before_record_index
 from invenio_records_files.utils import sorted_files_from_bucket
+from srt_to_vtt import srt_to_vtt
 
 from ..invenio_deposit.signals import post_action
 from .indexer import cdsdeposit_indexer_receiver
@@ -45,6 +49,59 @@
 )
 
 
+def _create_vtt_from_srt(srt_obj):
+    """Create a VTT file from an SRT file.
+    
+    :param srt_obj: ObjectVersion of the SRT file
+    :returns: ObjectVersion of the created VTT file or None
+    """
+    # Generate VTT filename from SRT filename
+    vtt_key = srt_obj.key.rsplit('.', 1)[0] + '.vtt'
+    
+    # Check if VTT file already exists
+    existing_vtt = ObjectVersion.get(srt_obj.bucket_id, vtt_key)
+    if existing_vtt:
+        # If it exists, skip
+        return existing_vtt
+
+    # Ensure the SRT file has a file instance
+    if not srt_obj.file or not srt_obj.file.uri:
+        return None
+    
+    srt_path = srt_obj.file.uri
+    vtt_path = None
+    try:
+        # Create temporary VTT file for output
+        with tempfile.NamedTemporaryFile(mode='w', suffix='.vtt', delete=False, encoding='utf-8') as vtt_file:
+            vtt_path = vtt_file.name
+        
+        # Convert using srt-to-vtt library
+        srt_to_vtt(srt_path, vtt_path)
+        
+        # Read the converted VTT content
+        with open(vtt_path, 'rb') as f:
+            vtt_content = f.read()
+        
+        # Create VTT ObjectVersion
+        vtt_obj = ObjectVersion.create(
+            bucket=srt_obj.bucket,
+            key=vtt_key,
+            stream=BytesIO(vtt_content),
+            size=len(vtt_content)
+        )
+        _create_tags(vtt_obj)
+        return vtt_obj
+    except (OSError, IOError, AttributeError, Exception):
+        return None
+    finally:
+        # Clean up temporary VTT file
+        if vtt_path and os.path.exists(vtt_path):
+            try:
+                os.unlink(vtt_path)
+            except OSError:
+                pass
+
+
 def _create_tags(obj):
     """Create additional tags for file."""
     pattern_subtitle = re.compile(r".*_([a-zA-Z]{2})\.vtt$")
@@ -53,6 +110,7 @@ def _create_tags(obj):
     # Get the media_type and content_type(file ext)
     file_name = obj.key
     mimetypes.add_type("subtitle/vtt", ".vtt")
+    mimetypes.add_type("text/srt", ".srt")
     guessed_type = mimetypes.guess_type(file_name)[0]
     if guessed_type is None:        
         raise InvalidKeyError(description=f"Unsupported File: {file_name}")
@@ -73,6 +131,12 @@ def _create_tags(obj):
             # other tags
             ObjectVersionTag.create_or_update(obj, "content_type", "vtt")
             ObjectVersionTag.create_or_update(obj, "context_type", "subtitle")
+        elif file_ext == "srt":
+            # Create VTT version from SRT
+            try:
+                _create_vtt_from_srt(obj)
+            except Exception:
+                pass
         # poster tag 
         elif pattern_poster.match(file_name):
             ObjectVersionTag.create_or_update(obj, "context_type", "poster")

cds/modules/deposit/static/templates/cds_deposit/types/video/uploader.html

@@ -148,8 +148,8 @@ <h5 class="text-muted"><strong>Tips and suggestions</strong></h5>
         ngf-model-options="{allowInvalid: false}"
         ngf-change="$ctrl.addFiles($newFiles, $invalidFiles)"
         ngf-select=""
-        ngf-pattern="'.vtt'"
-        ngf-accept="'.vtt'"
+        ngf-pattern="'.vtt,.srt'"
+        ngf-accept="'.vtt,.srt'"
         ngf-validate-fn="$ctrl.validateSubtitles($file)"
         ngf-max-size="500GB"
       ><i class="fa fa-plus-square"></i></a>
@@ -167,8 +167,8 @@ <h5 class="text-muted"><strong>Tips and suggestions</strong></h5>
   ng-if="!$ctrl.cdsDepositCtrl.isPublished()"
   ngf-select=""
   ngf-change="$ctrl.addFiles($newFiles, $invalidFiles)"
-  ngf-pattern="'.vtt'"
-  ngf-accept="'text/vtt'"
+  ngf-pattern="'.vtt,.srt'"
+  ngf-accept="'text/vtt,.vtt,.srt'"
   ngf-validate-fn="$ctrl.validateSubtitles($file)"
   ngf-max-size="500GB"
   ngf-multiple="true"
@@ -183,15 +183,15 @@ <h5 class="text-muted"><strong>Tips and suggestions</strong></h5>
         ngf-model-options="{allowInvalid: false}"
         ngf-change="$ctrl.addFiles($newFiles, $invalidFiles)"
         ngf-select=""
-        ngf-pattern="'.vtt'"
-        ngf-accept="'.vtt'"
+        ngf-pattern="'.vtt,.srt'"
+        ngf-accept="'.vtt,.srt'"
         ngf-validate-fn="$ctrl.validateSubtitles($file)"
-        ngf-max-size="500GB">select</a> <mark>.vtt</mark> files.
+        ngf-max-size="500GB">select</a> <mark>.vtt</mark> or <mark>.srt</mark> files.
         <hr class="my-10" />
         <div class="text-muted text-left">
           <h5 class="text-muted"><strong>Tips and suggestions</strong></h5>
           <ul>
-            <li>Subtitle filename should have a valid ISO language code. Example: <mark>subtitles_fr.vtt</mark> </li>
+            <li>Subtitle filename should have a valid ISO language code. Example: <mark>subtitles_fr.vtt</mark> or <mark>subtitles_fr.srt</mark> </li>
           </ul>
         </div>
     </p>

cds/modules/theme/assets/bootstrap3/js/cds_deposit/avc/components/cdsUploader.js

@@ -284,7 +284,12 @@ function cdsUploaderCtrl(
 
       // Filter out files without a valid MIME type or with zero size
       _files = _files.filter((file) => {
-        if (!file.type || file.type.trim() === "") {
+        // Allow SRT and VTT files even if they don't have a MIME type
+        var fileName = file.name.toLowerCase();
+        var isSubtitleFile =
+          fileName.endsWith(".vtt") || fileName.endsWith(".srt");
+
+        if ((!file.type || file.type.trim() === "") && !isSubtitleFile) {
           toaster.pop(
             "warning",
             "Invalid File Type",
@@ -544,13 +549,14 @@ function cdsUploaderCtrl(
   this.validateSubtitles = function (_file) {
     // Check if the filename matches the pattern and is a valid ISO language
     // i.e. jessica_jones-en.vtt
-    var match = _file.name.match(/(?:.+)[_|-]([a-zA-Z]{2}).vtt/) || [];
+    var match = _file.name.match(/(?:.+)[_|-]([a-zA-Z]{2})\.(vtt|srt)/) || [];
     return match.length > 1 && match[1] in isoLanguages;
   };
 
   this.validateAdditionalFiles = function (_file) {
-    // If it's a .vtt file, validate as subtitle
-    if (_file.name.toLowerCase().endsWith(".vtt")) {
+    // If it's a .vtt or .srt file, validate as subtitle
+    var fileName = _file.name.toLowerCase();
+    if (fileName.endsWith(".vtt") || fileName.endsWith(".srt")) {
       return this.validateSubtitles(_file);
     }
     // Accept other types

requirements.txt

@@ -184,6 +184,7 @@ simplekv==0.14.1
 six==1.17.0
 soupsieve==2.6
 speaklater==1.3
+srt-to-vtt==1.0.0
 SQLAlchemy==1.4.54
 SQLAlchemy-Continuum==1.4.1
 SQLAlchemy-Utils==0.38.3