Adding VTT support

README.md

@@ -39,6 +39,11 @@ Unit tests are in the `tests` folder and can be run with:
 python -m unittest discover -s tests
 ```
 
+Run single test:
+```
+python -m unittest tests.test_video.TestVideo.test_vtt_autogenerated
+```
+
 Retrieve large.jpg as 800px wide JPEG
 * http://127.0.0.1:8080/iiif/large.jpg/full/800,/0/default.jpg 
 

iiify/resolver.py

@@ -506,6 +506,7 @@ def create_manifest3(identifier, domain=None, page=None):
         # sort the files into originals and derivatives, splitting the derivatives into buckets based on the original
         originals = []
         derivatives = {}
+        vttfiles = {}
         for f in metadata['files']:
             if f['source'] == 'derivative':
                 if f['original'] in derivatives:
@@ -514,6 +515,14 @@ def create_manifest3(identifier, domain=None, page=None):
                     derivatives[f['original']] = {f['format']: f}
             elif f['source'] == 'original':
                 originals.append(f)
+
+            if f['format'] == 'Web Video Text Tracks':
+                # Example: cruz-test.en.vtt and 34C3_-_International_Image_Interoperability_Framework_IIIF_Kulturinstitutionen_schaffen_interop-SvH4fbjOT0A.autogenerated.vtt
+                sourceFilename = re.sub('\.[a-zA-H-]*\.vtt', '', f['name'])
+                if sourceFilename not in vttfiles:
+                    vttfiles[sourceFilename] = []    
+
+                vttfiles[sourceFilename].append(f)    
 
         # create the canvases for each original
         for file in [f for f in originals if f['format'] in ['MPEG4', 'h.264 MPEG4', '512Kb MPEG4', 'HiRes MPEG4', 'MPEG2', 'h.264', 'Matroska', 'Ogg Video', 'Ogg Theora', 'WebM', 'Windows Media', 'Cinepack']]:
@@ -522,6 +531,32 @@ def create_manifest3(identifier, domain=None, page=None):
             c_id = f"{URI_PRIFIX}/{identifier}/{slugged_id}/canvas"
             c = Canvas(id=c_id, label=normalised_id, duration=float(file['length']), height=int(file['height']), width=int(file['width']))
 
+            # Add vtt if present
+            if vttfiles and normalised_id in vttfiles:
+                vttAPId = f"{URI_PRIFIX}/{identifier}/{slugged_id}/vtt"
+
+                vttNo = 1
+                for vttFile in vttfiles[normalised_id]:
+                    vtAnno = c.make_annotation(id=f"{URI_PRIFIX}/{identifier}/{slugged_id}/annotation/vtt/{vttNo}", 
+                                               motivation="supplementing", 
+                                               target=c.id, 
+                                               anno_page_id=vttAPId,
+                                               body={"id": f"{domain}resource/{identifier}/{vttFile['name']}",
+                                                     "type": "Text",
+                                                    "format": "text/vtt",
+                                                    })
+                    # add label and language
+                    if vttFile['name'].endswith("autogenerated.vtt"):
+                        vtAnno.body.label = { 'en': ['autogenerated']}
+                    else:
+                        # Assume language
+                        splitName = vttFile['name'].split(".")
+                        lang = splitName[-2]
+                        vtAnno.body.add_label(lang, language="none")
+                        vtAnno.body.language = lang
+
+                    vttNo += 1
+
             # create intermediary objects
             ap = AnnotationPage(id=f"{URI_PRIFIX}/{identifier}/{slugged_id}/page")
             anno = Annotation(id=f"{URI_PRIFIX}/{identifier}/{slugged_id}/annotation", motivation="painting", target=c.id)

nginx-vhost.conf

@@ -37,4 +37,11 @@ server {
         # Reverse proxy with the variables captured above
         proxy_pass https://cantaloupe.prod.archive.org/iiif/$1/$2;
     }
+
+    location /iiif/resource/ {
+        add_header 'Access-Control-Allow-Origin' '*' always;
+        add_header 'Access-Control-Allow-Methods' 'GET, HEAD, POST, PUT, PATCH, DELETE' always;
+        # https://archive.org/download/cruz-test/cruz-test.af.vtt
+        proxy_pass https://archive.org/download/;
+    }
 }

tests/test_manifests.py

@@ -42,21 +42,13 @@ def test_v3_single_text_manifest(self):
         self.assertEqual(manifest['type'], "Manifest", f"Unexpected type. Expected Manifest go {manifest['type']}")
         self.assertEqual(len(manifest['items']),1,f"Expected 1 canvas but got: {len(manifest['items'])}")
 
-
     def test_v3_vermont_Life_Magazine(self):
         resp = self.test_app.get("/iiif/3/rbmsbk_ap2-v4_2001_V55N4/manifest.json")
         self.assertEqual(resp.status_code, 200)
         manifest = resp.json
 
         self.assertEqual(len(manifest['items']),116,f"Expected 116 canvas but got: {len(manifest['items'])}")
-
-    def test_v3_single_video_manifest(self):
-        resp = self.test_app.get("/iiif/3/youtube-7w8F2Xi3vFw/manifest.json")
-        self.assertEqual(resp.status_code, 200)
-        manifest = resp.json
-
-        self.assertEqual(len(manifest['items']),1,f"Expected 1 canvas but got: {len(manifest['items'])}")
-
+
     #logic to cover etree mediatype github issue #123
     def test_v3_etree_mediatype(self):
         resp = self.test_app.get("/iiif/3/gd72-04-14.aud.vernon.23662.sbeok.shnf/manifest.json")
@@ -66,30 +58,20 @@ def test_v3_etree_mediatype(self):
         self.assertEqual(len(manifest['items']),36,f"Expected 36 canvases but got: {len(manifest['items'])}")
         self.assertEqual(manifest['items'][0]['items'][0]['items'][0]['body']['items'][0]['type'],"Sound",f"Expected 'Sound' but got: {manifest['items'][0]['items'][0]['items'][0]['body']['items'][0]['type']}")
 
-
     def test_v3_64Kbps_MP3(self):
         resp = self.test_app.get("/iiif/3/TvQuran.com__Alafasi/manifest.json")
         self.assertEqual(resp.status_code, 200)
         manifest = resp.json
         self.assertEqual(len(manifest['items']),114,f"Expected 114 canvases but got: {len(manifest['items'])}")
         self.assertEqual("64Kbps MP3".lower() in resp.text.lower(), True, f"Expected the string '64Kbps MP3'")
 
-
     def test_v3_128Kbps_MP3(self):
         resp = self.test_app.get("/iiif/3/alice_in_wonderland_librivox/manifest.json")
         self.assertEqual(resp.status_code, 200)
         manifest = resp.json
         self.assertEqual(len(manifest['items']),12,f"Expected 12 canvases but got: {len(manifest['items'])}")
         self.assertEqual("128kbps mp3".lower() in resp.text.lower(), True, f"Expected the string '128kbps mp3'")
 
-    def test_v3_h264_MPEG4_OGG_Theora(self):
-        resp = self.test_app.get("/iiif/3/taboca_201002_03/manifest.json")
-        self.assertEqual(resp.status_code, 200)
-        manifest = resp.json
-        self.assertEqual(len(manifest['items']),251,f"Expected 251 canvases but got: {len(manifest['items'])}")
-        self.assertEqual("h.264 MPEG4".lower() in resp.text.lower(), True, f"Expected the string 'h.264 MPEG4'")
-        self.assertEqual("OGG Theora".lower() in resp.text.lower(), True, f"Expected the string 'OGG Theora'")
-
     def test_v3_aiff(self):
         resp = self.test_app.get("/iiif/3/PDextend_AIFF/manifest.json")
         self.assertEqual(resp.status_code, 200)

tests/test_video.py

@@ -0,0 +1,68 @@
+import unittest
+from flask.testing import FlaskClient
+from iiify.app import app
+
+class TestVideo(unittest.TestCase):
+
+    def setUp(self) -> None:
+        self.test_app = FlaskClient(app)
+
+    def test_v3_single_video_manifest(self):
+        resp = self.test_app.get("/iiif/3/youtube-7w8F2Xi3vFw/manifest.json")
+        self.assertEqual(resp.status_code, 200)
+        manifest = resp.json
+
+        self.assertEqual(len(manifest['items']),1,f"Expected 1 canvas but got: {len(manifest['items'])}")   
+
+    def test_v3_h264_MPEG4_OGG_Theora(self):
+        resp = self.test_app.get("/iiif/3/taboca_201002_03/manifest.json")
+        self.assertEqual(resp.status_code, 200)
+        manifest = resp.json
+        self.assertEqual(len(manifest['items']),251,f"Expected 251 canvases but got: {len(manifest['items'])}")
+        self.assertEqual("h.264 MPEG4".lower() in resp.text.lower(), True, f"Expected the string 'h.264 MPEG4'")
+        self.assertEqual("OGG Theora".lower() in resp.text.lower(), True, f"Expected the string 'OGG Theora'")
+
+    def test_vtt_autogenerated(self):
+        resp = self.test_app.get("/iiif/3/youtube-SvH4fbjOT0A/manifest.json?recache=true")
+        self.assertEqual(resp.status_code, 200)
+        manifest = resp.json
+
+        self.assertEqual(len(manifest['items']),1,f"Expected 1 canvas but got: {len(manifest['items'])}")
+        self.assertTrue('annotations' in manifest['items'][0], "Expected annotations in manifest")
+        self.assertTrue(isinstance(manifest['items'][0]['annotations'], list), "Expected annotations to be a list")
+        self.assertEqual(len(manifest['items'][0]['annotations']), 1, "Expected 1 item in annotations")
+        annotationPage = manifest['items'][0]['annotations'][0]
+        self.assertEqual(annotationPage['type'], 'AnnotationPage', "Expected annotations to contain annotation page")
+
+        self.assertTrue('items' in annotationPage and isinstance(annotationPage['items'],list) and len(annotationPage['items']) == 1, f"Expected annotation page to contain a list of items which contains 1 item. Found {annotationPage['items']}")
+        annotation = annotationPage['items'][0]
+        self.assertEqual(annotation['type'], 'Annotation', "Expected annotationPage to contain annotations")
+        self.assertEqual(annotation['motivation'], 'supplementing', "Expected annotation to have the supplementing annotation")
+        self.assertTrue('body' in annotation, "Expected annotation to have a body")
+        body = annotation['body']
+        self.assertEqual(body['type'],'Text', "Expected body to have a type text")
+        self.assertEqual(body['format'],'text/vtt', "Expected body to have a type text")
+        self.assertEqual(body['label']['en'][0], "autogenerated", "Expected VTT file to have the label autogenerated")
+        self.assertFalse("language" in body, "We don't know the language for this item so there shouldn't be a language specified")
+        self.assertEqual(body['id'], "https://localhost/iiif/resource/youtube-SvH4fbjOT0A/34C3_-_International_Image_Interoperability_Framework_IIIF_Kulturinstitutionen_schaffen_interop-SvH4fbjOT0A.autogenerated.vtt","Unexpected URL for the VTT file")
+
+    def test_vtt_multilingual(self):
+        resp = self.test_app.get("/iiif/3/cruz-test/manifest.json?recache=true")
+        self.assertEqual(resp.status_code, 200)
+        manifest = resp.json
+
+        canvas = manifest['items'][0]
+        self.assertTrue('annotations' in canvas, 'Expected annotations in Canvas')
+        self.assertEqual(len(canvas['annotations']), 1, 'Expected one AnnotationPage')
+        annotations = canvas['annotations'][0]['items']
+        self.assertEqual(len(annotations), 104, 'Expected all 104 langues')
+
+        # Check welsh
+        for item in annotations:
+            self.assertTrue('language' in item['body'], f"All vtt files should have a language: {item}")
+            if item['body']['language'] == 'cy':
+                self.assertEqual(item['body']['id'], 'https://localhost/iiif/resource/cruz-test/cruz-test.cy.vtt', 'Unexpected link for the Welsh vtt file')
+
+
+if __name__ == '__main__':
+    unittest.main()