Fail meaningful when image based subtitles run off the screen

.github/workflows/unit_tests.yml

@@ -20,17 +20,25 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ["py38", "py39", "py310", "py311", "py312"]
+        python-version: [ "py312"]
 
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v4
 
       - name: Run Test
         id: tests
         run: |
-          ./run_tests.sh test_${{ matrix.python-version }} 
+          ./run_tests.sh test_${{ matrix.python-version }}
         continue-on-error: true
 
+      - name: Test Report
+        uses: dorny/test-reporter@v1
+        if: always()
+        with:
+          name: Test Results (${{ matrix.python-version }})
+          path: junit.xml
+          reporter: java-junit
+
       - name: Archive production artifacts
         uses: actions/upload-artifact@v4
         with:

docker-compose.yml

@@ -2,7 +2,7 @@ version: '3.8'
 
 services:
   test_py38:
-    image: python:3.8-slim-bullseye
+    image: python:3.8-slim-bookworm
     command: sh -c "
       cd pycaption;
       pip install --upgrade pip;
@@ -14,7 +14,7 @@ services:
       - .:/pycaption
 
   test_py39:
-    image: python:3.9-slim-bullseye
+    image: python:3.9-slim-bookworm
     command: sh -c "
       cd pycaption;
       pip install --upgrade pip;
@@ -26,7 +26,7 @@ services:
       - .:/pycaption
 
   test_py310:
-    image: python:3.10-slim-bullseye
+    image: python:3.10-slim-bookworm
     command: sh -c "
       cd pycaption;
       pip install --upgrade pip;
@@ -38,7 +38,7 @@ services:
       - .:/pycaption
 
   test_py311:
-    image: python:3.11-slim-bullseye
+    image: python:3.11-slim-bookworm
     command: sh -c "
       cd pycaption;
       pip install --upgrade pip;
@@ -50,7 +50,7 @@ services:
       - .:/pycaption
 
   test_py312:
-    image: python:3.12-slim-bullseye
+    image: python:3.12-slim-bookworm
     command: sh -c "
       cd pycaption;
       pip install --upgrade pip;

pycaption/exceptions.py

@@ -41,3 +41,10 @@ class CaptionLineLengthError(CaptionReadError):
     """
     Error raised when a Caption has a line longer than 32 characters.
     """
+
+
+class CaptionRendererError(Exception):
+    """
+    Error raised when caption content cannot be rendered correctly,
+    e.g. text runs off screen or required glyphs are missing from the font.
+    """

pycaption/srt.py

@@ -1,4 +1,3 @@
-import os
 from copy import deepcopy
 
 from .base import (
@@ -7,7 +6,6 @@
 from .exceptions import CaptionReadNoCaptions, InvalidInputError
 
 import re
-from PIL import Image, ImageFont, ImageDraw
 import warnings
 
 warnings.simplefilter('once', DeprecationWarning)
@@ -162,14 +160,6 @@ def _recreate_lang(self, captions, position='bottom'):
         srt = ''
         count = 1
 
-        fnt = ImageFont.truetype(os.path.dirname(__file__) + '/NotoSansDisplay-Regular-Note-Math.ttf', 30)
-
-        img = None
-        draw = None
-        if position == 'top':
-            img = Image.new('RGB', (self.video_width, self.video_height), (0, 255, 0))
-            draw = ImageDraw.Draw(img)
-
         for caption in captions:
             # Generate the text
             new_content = ''
@@ -190,13 +180,19 @@ def _recreate_lang(self, captions, position='bottom'):
                 # Use the old behavior, output just the timestamp, no coordinates.
                 timestamp = '%s --> %s' % (start[:12], end[:12])
             elif position == 'top':
+                # Approximate character dimensions for coordinate estimation.
+                # The player uses its own font so these are just reasonable hints.
+                char_width = round(self.video_width * 0.018)
+                line_height = round(self.video_height * 0.05)
                 padding_top = 10
-                l, t, r, b = draw.textbbox((0, 0), new_content, font=fnt)
-                l, t, r, b = draw.textbbox((self.video_width / 2 - r / 2, padding_top), new_content, font=fnt)
-                x1 = str(round(l)).zfill(3)
-                x2 = str(round(r)).zfill(3)
-                y1 = str(round(t)).zfill(3)
-                y2 = str(round(b)).zfill(3)
+                lines = new_content.split('\n')
+                max_line_len = max(len(line) for line in lines)
+                text_width = max_line_len * char_width
+                text_height = len(lines) * line_height
+                x1 = str(round(self.video_width / 2 - text_width / 2)).zfill(3)
+                x2 = str(round(self.video_width / 2 + text_width / 2)).zfill(3)
+                y1 = str(padding_top).zfill(3)
+                y2 = str(padding_top + text_height).zfill(3)
                 timestamp = '%s --> %s X1:%s X2:%s Y1:%s Y2:%s' % (start[:12], end[:12], x1, x2, y1, y2)
             else:
                 raise ValueError('Unsupported position: %s' % position)

pycaption/subtitler_image_based.py

@@ -10,6 +10,7 @@
 from langcodes import Language, tag_distance
 
 from pycaption.base import BaseWriter, CaptionSet, Caption, CaptionNode, CaptionList
+from pycaption.exceptions import CaptionRendererError
 from pycaption.geometry import UnitEnum, Size
 
 
@@ -207,7 +208,7 @@ def write_images(
         missing_glyphs = self.get_missing_glyphs(fnt, self.get_characters(caps_final))
 
         if missing_glyphs:
-            raise ValueError(f'Selected font was missing glyphs: {" ".join(missing_glyphs.keys())}')
+            raise CaptionRendererError(f'Selected font was missing glyphs: {" ".join(missing_glyphs.keys())}')
 
         min_font_px = 16
         font_size = max(min_font_px, int(self.video_width * 0.05 * 0.6))  # rough estimate but should work
@@ -277,6 +278,18 @@ def printLine(self, draw: ImageDraw, caption_list: Caption, fnt: ImageFont, posi
                 else:
                     raise ValueError('Unknown "position": {}'.format(position))
 
+            # Check that text (including border outline) fits within the screen
+            border_offset = 1  # max adj value from border drawing loop
+            text_left = x - border_offset + l
+            text_top = y - border_offset + t
+            text_right = x + border_offset + r
+            text_bottom = y + border_offset + b
+            if text_left < 0 or text_top < 0 or text_right > self.video_width or text_bottom > self.video_height:
+                raise CaptionRendererError(
+                    f'Text runs off screen: text="{text}"'
+                )
+
+
             border = (*self.borderColor, 255)  # Add alpha for RGBA
             font = (*self.fontColor, 255)  # Add alpha for RGBA
             for adj in range(2):

setup.py

@@ -12,7 +12,7 @@
     "beautifulsoup4>=4.12.1",
     "lxml>=4.9.1",
     "cssutils>=2.0.0",
-    "Pillow~=11.1.0",
+    "Pillow~=12.1.0",
     "fonttools~=4.56.0",
     "langcodes~=3.5.0",
     "striprtf~=0.0.28"

tests/conftest.py

@@ -102,5 +102,5 @@
     sample_webvtt_last_cue_zero_start, sample_webvtt_empty_cue,
     sample_webvtt_multi_lang_en, sample_webvtt_multi_lang_de,
     sample_webvtt_empty_cue_output, sample_webvtt_timestamps,
-    sample_srt_top
+    sample_srt_top,
 )

tests/fixtures/webvtt.py

@@ -3,37 +3,37 @@
 @pytest.fixture(scope="session")
 def sample_srt_top():
     return """1
-00:00:09,209 --> 00:00:12,312 X1:262 X2:458 Y1:020 Y2:051
+00:00:09,209 --> 00:00:12,312 X1:250 X2:470 Y1:010 Y2:034
 ( clock ticking )
 
 2
-00:00:14,848 --> 00:00:17,000 X1:233 X2:487 Y1:021 Y2:125
-MAN: 
-When we think 
+00:00:14,848 --> 00:00:17,000 X1:230 X2:490 Y1:010 Y2:082
+MAN:
+When we think
 ♪ ...say bow, wow, ♪
 
 3
-00:00:17,000 --> 00:00:18,752 X1:162 X2:558 Y1:020 Y2:044
+00:00:17,000 --> 00:00:18,752 X1:158 X2:562 Y1:010 Y2:034
 we have this vision of Einstein
 
 4
-00:00:18,752 --> 00:00:20,887 X1:208 X2:512 Y1:020 Y2:081
-as an old, wrinkly man 
+00:00:18,752 --> 00:00:20,887 X1:210 X2:510 Y1:010 Y2:058
+as an old, wrinkly man
 with white hair.
 
 5
-00:00:20,887 --> 00:00:26,760 X1:190 X2:530 Y1:021 Y2:118
-MAN 2: 
-E equals m c-squared is 
+00:00:20,887 --> 00:00:26,760 X1:191 X2:529 Y1:010 Y2:082
+MAN 2:
+E equals m c-squared is
 not about an old Einstein.
 
 6
-00:00:26,760 --> 00:00:32,200 X1:147 X2:573 Y1:021 Y2:081
-MAN 2: 
+00:00:26,760 --> 00:00:32,200 X1:132 X2:588 Y1:010 Y2:058
+MAN 2:
 It's all about an eternal Einstein.
 
 7
-00:00:32,200 --> 00:00:36,200 X1:187 X2:533 Y1:021 Y2:044
+00:00:32,200 --> 00:00:36,200 X1:230 X2:490 Y1:010 Y2:034
 <LAUGHING & WHOOPS!>
 """
 

tests/test_srt_conversion.py

@@ -67,7 +67,7 @@ def test_webvtt_to_srt_conversion(self, sample_srt, sample_webvtt):
 
     def test_webvtt_to_srt_conversion_pos_top(self, sample_srt_top, sample_webvtt):
         caption_set = WebVTTReader().read(sample_webvtt)
-        results = SRTWriter(video_width=720, video_height=480 ).write(caption_set, position='top')
+        results = SRTWriter(video_width=720, video_height=480).write(caption_set, position='top')
 
         assert isinstance(results, str)
         self.assert_srt_equals(sample_srt_top, results)

tests/test_subtitler_image_based.py

@@ -0,0 +1,106 @@
+
+import os
+
+import pytest
+from PIL import Image, ImageDraw, ImageFont
+
+from pycaption.base import Caption, CaptionNode
+from pycaption.exceptions import CaptionRendererError
+from pycaption.geometry import Layout, Point, Size, UnitEnum
+from pycaption.subtitler_image_based import SubtitleImageBasedWriter
+
+
+FONT_PATH = os.path.join(
+    os.path.dirname(__file__),
+    '..', 'pycaption', 'NotoSansDisplay-Regular-Note-Math.ttf'
+)
+
+
+def make_caption(text, layout_info=None):
+    nodes = [CaptionNode.create_text(text)]
+    return Caption(0, 1000000, nodes, layout_info=layout_info)
+
+
+def make_source_layout(x_pct, y_pct):
+    origin = Point(Size(x_pct, UnitEnum.PERCENT), Size(y_pct, UnitEnum.PERCENT))
+    return Layout(origin=origin)
+
+
+def make_writer_and_draw(width, height):
+    writer = SubtitleImageBasedWriter(video_width=width, video_height=height)
+    img = Image.new('RGBA', (width, height), (0, 0, 0, 0))
+    draw = ImageDraw.Draw(img)
+    return writer, draw
+
+
+class TestTextOffScreenCentered:
+    def test_short_text_fits(self):
+        writer, draw = make_writer_and_draw(720, 480)
+        fnt = ImageFont.truetype(FONT_PATH, 20)
+        caption = make_caption("Hello")
+        writer.printLine(draw, [caption], fnt, position='bottom', align='center')
+
+    def test_long_text_runs_off(self):
+        writer, draw = make_writer_and_draw(200, 100)
+        fnt = ImageFont.truetype(FONT_PATH, 20)
+        caption = make_caption("This text is way too long to fit on a tiny screen")
+        with pytest.raises(CaptionRendererError, match="Text runs off screen"):
+            writer.printLine(draw, [caption], fnt, position='bottom', align='center')
+
+
+class TestTextOffScreenLeft:
+    def test_short_text_fits(self):
+        writer, draw = make_writer_and_draw(720, 480)
+        fnt = ImageFont.truetype(FONT_PATH, 20)
+        caption = make_caption("Hello")
+        writer.printLine(draw, [caption], fnt, position='bottom', align='left')
+
+    def test_long_text_runs_off(self):
+        writer, draw = make_writer_and_draw(200, 100)
+        fnt = ImageFont.truetype(FONT_PATH, 20)
+        caption = make_caption("This text is way too long to fit on a tiny screen")
+        with pytest.raises(CaptionRendererError, match="Text runs off screen"):
+            writer.printLine(draw, [caption], fnt, position='bottom', align='left')
+
+
+class TestTextOffScreenRight:
+    def test_short_text_fits(self):
+        writer, draw = make_writer_and_draw(720, 480)
+        fnt = ImageFont.truetype(FONT_PATH, 20)
+        caption = make_caption("Hello")
+        writer.printLine(draw, [caption], fnt, position='bottom', align='right')
+
+    def test_long_text_runs_off(self):
+        writer, draw = make_writer_and_draw(200, 100)
+        fnt = ImageFont.truetype(FONT_PATH, 20)
+        caption = make_caption("This text is way too long to fit on a tiny screen")
+        with pytest.raises(CaptionRendererError, match="Text runs off screen"):
+            writer.printLine(draw, [caption], fnt, position='bottom', align='right')
+
+
+class TestTextOffScreenSourcePosition:
+    def test_centered_fits(self):
+        writer, draw = make_writer_and_draw(720, 480)
+        fnt = ImageFont.truetype(FONT_PATH, 20)
+        layout = make_source_layout(x_pct=30, y_pct=50)
+        caption = make_caption("Hello", layout_info=layout)
+        writer.printLine(draw, [caption], fnt, position='source', align='left')
+
+    def test_right_sticks_out(self):
+        # 200px wide screen, text at x=80% (x=160), text wider than remaining space
+        writer, draw = make_writer_and_draw(200, 200)
+        fnt = ImageFont.truetype(FONT_PATH, 20)
+        layout = make_source_layout(x_pct=80, y_pct=50)
+        caption = make_caption("This text sticks out on the right", layout_info=layout)
+        with pytest.raises(CaptionRendererError, match="Text runs off screen"):
+            writer.printLine(draw, [caption], fnt, position='source', align='left')
+
+    def test_left_sticks_out(self):
+        # 200px wide screen, text at x=10% (x=20), text wider than screen
+        # so repositioning pushes x past the left edge
+        writer, draw = make_writer_and_draw(200, 200)
+        fnt = ImageFont.truetype(FONT_PATH, 20)
+        layout = make_source_layout(x_pct=10, y_pct=50)
+        caption = make_caption("This text sticks out on the left", layout_info=layout)
+        with pytest.raises(CaptionRendererError, match="Text runs off screen"):
+            writer.printLine(draw, [caption], fnt, position='source', align='left')