fix(subtitle): join cues that a segment boundary splits in stpp and wvtt

Share one append_cue join across DASH fTTML, ISM and fVTT parsing. Also map "geo-block" and "geolocation" errors to GEOFENCE in the API.

docs/guide/subtitles.md

@@ -52,7 +52,7 @@ format-appropriate clean-up to make a single, valid file:
   `‎` / `‏` bidirectional entities.
 - **Segment merging**: when a WebVTT or boxed subtitle arrives as many DASH/HLS
   segments, unshackle merges the segments back into one continuous document with
-  correct timing.
+  correct timing. A cue that a segment boundary cuts in two comes out as one cue.
 - **WebVTT sanitization**: merged WebVTT runs through a repair pipeline that clamps
   negative timestamps to `00:00:00.000`, removes stray cue identifiers (like `Q0`, `Q1`)
   that confuse downstream parsers, and merges multi-line cues that a service split into

tests/remote/unit/test_errors.py

@@ -114,6 +114,8 @@ def test_build_error_response_debug_mode_includes_traceback() -> None:
         (TimeoutError("read timeout"), APIErrorCode.NETWORK_ERROR),
         (Exception("Not available in your region"), APIErrorCode.GEOFENCE),
         (Exception("Service is geoblocked in GB"), APIErrorCode.GEOFENCE),
+        (Exception("Content is geo-blocked."), APIErrorCode.GEOFENCE),
+        (Exception("Manifest request failed: GeoLocationBlocked"), APIErrorCode.GEOFENCE),
         (
             Exception("Service is not available in your region (GB)"),
             APIErrorCode.GEOFENCE,

tests/tracks/test_subtitle_convert.py

@@ -351,6 +351,20 @@ def test_segmented_fvtt_parses_and_converts(tmp_path, monkeypatch):
     assert cue_count(out) == 2
 
 
+def test_segmented_fttml_joins_a_cue_split_at_a_segment_boundary():
+    data = build_stpp_mp4(
+        ttml_fragment(("00:00:01.000", "00:00:02.000", "Carried")),
+        ttml_fragment(("00:00:02.000", "00:00:03.500", "Carried"), ("00:00:03.500", "00:00:04.000", "Next")),
+    )
+
+    captions = Subtitle.parse(data, Codec.fTTML).get_captions("en")
+
+    assert [(c.start / 1e6, c.end / 1e6, c.get_text()) for c in captions] == [
+        (1.0, 3.5, "Carried"),
+        (3.5, 4.0, "Next"),
+    ]
+
+
 def test_segmented_fttml_parses_and_converts(tmp_path, monkeypatch):
     monkeypatch.setattr("unshackle.core.config.config.subtitle", {}, raising=False)
     frag = (
@@ -448,8 +462,8 @@ def test_dash_fvtt_cues_lose_the_presentation_time_offset(tmp_path, monkeypatch)
     assert "00:00:02.000 --> 00:00:03.000" in text
 
 
-def ttml_fragment(*cues: tuple[str, str]) -> str:
+def ttml_fragment(*cues: tuple[str, ...]) -> str:
-    body = "".join(f'<p begin="{begin}" end="{end}">Line</p>' for begin, end in cues)
+    body = "".join(f'<p begin="{cue[0]}" end="{cue[1]}">{cue[2] if len(cue) > 2 else "Line"}</p>' for cue in cues)
     return f'<tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en"><body><div>{body}</div></body></tt>'
 
 

unshackle/core/api/errors.py

@@ -208,7 +208,10 @@ def categorize_exception(
     exc_str = f"{exc} {root}".lower()
     exc_type = type(root).__name__
 
-    if any(keyword in exc_str for keyword in ["geofence", "geoblock", "region", "not available in", "territory"]):
+    if any(
+        keyword in exc_str
+        for keyword in ["geofence", "geoblock", "geo-block", "geolocation", "region", "not available in", "territory"]
+    ):
         return APIError(
             error_code=APIErrorCode.GEOFENCE,
             message=f"Content not available in your region: {redact_secrets(str(exc))}",

unshackle/core/tracks/subtitle.py

@@ -376,6 +376,14 @@ class Subtitle(Track):
             boxes = MP4.parse_stream(BytesIO(data))
         return [box.data for box in boxes if box.type == b"mdat"]
 
+    @staticmethod
+    def append_cue(kept: CaptionList, caption: Caption) -> None:
+        """Append caption to kept, or extend the last cue that a segment boundary cut and caption carries on."""
+        if kept and kept[-1].end == caption.start and kept[-1].get_text() == caption.get_text():
+            kept[-1].end = caption.end
+        else:
+            kept.append(caption)
+
     @staticmethod
     def first_cue_start(caption_set: pycaption.CaptionSet) -> Optional[int]:
         """The start of the first cue, in microseconds, looking past empty languages; None with no cues."""
@@ -413,10 +421,8 @@ class Subtitle(Track):
                     for caption in fragment.get_captions(language):
                         caption.start = max(caption.start + shift, 0)
                         caption.end += shift
-                        if kept and kept[-1].end == caption.start and kept[-1].get_text() == caption.get_text():
+                        if caption.end > 0:
-                            kept[-1].end = caption.end  # a cue carried into the next fragment
+                            Subtitle.append_cue(kept, caption)
-                        elif caption.end > 0:
-                            kept.append(caption)
         if not any(captions.values()):
             logging.getLogger("Subtitle").debug(f"ISM text: {len(payloads)} fragments, {len(times)} fragment times")
             self.convert(Subtitle.Codec.TimedTextMarkupLang)
@@ -837,7 +843,8 @@ class Subtitle(Track):
                     for payload in Subtitle.mdat_payloads(data)
                 ):
                     for lang in segment.get_languages():
-                        caption_lists[lang].extend(segment.get_captions(lang))
+                        for caption in segment.get_captions(lang):
+                            Subtitle.append_cue(caption_lists[lang], caption)
                 caption_set = pycaption.CaptionSet(caption_lists)
             elif codec == Subtitle.Codec.TimedTextMarkupLang:
                 text = try_ensure_utf8(data).decode("utf8")
@@ -1109,12 +1116,7 @@ class Subtitle(Track):
                             nodes=nodes,
                             layout_info=layout,
                         )
-                        p_caption = captions[-1] if captions else None
+                        Subtitle.append_cue(captions, caption)
-                        if p_caption and caption.start == p_caption.end and str(caption.nodes) == str(p_caption.nodes):
-                            # it's a duplicate, but lets take its end time
-                            p_caption.end = caption.end
-                            continue
-                        captions.append(caption)
 
         return captions, language