Fix the block count and test for images in test_chunk_renderer.
Browse files
tests/renderers/test_chunk_renderer.py
CHANGED
|
@@ -4,13 +4,26 @@ from marker.renderers.chunk import ChunkRenderer
|
|
| 4 |
|
| 5 |
|
| 6 |
@pytest.mark.config({"page_range": [0]})
|
| 7 |
-
def
|
| 8 |
renderer = ChunkRenderer()
|
| 9 |
chunk_output = renderer(pdf_document)
|
| 10 |
blocks = chunk_output.blocks
|
| 11 |
page_info = chunk_output.page_info
|
| 12 |
|
| 13 |
-
assert len(blocks) ==
|
| 14 |
assert blocks[0].block_type == "SectionHeader"
|
| 15 |
assert page_info[0]["bbox"] is not None
|
| 16 |
assert page_info[0]["polygon"] is not None
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4 |
|
| 5 |
|
| 6 |
@pytest.mark.config({"page_range": [0]})
|
| 7 |
+
def test_chunk_renderer(pdf_document):
|
| 8 |
renderer = ChunkRenderer()
|
| 9 |
chunk_output = renderer(pdf_document)
|
| 10 |
blocks = chunk_output.blocks
|
| 11 |
page_info = chunk_output.page_info
|
| 12 |
|
| 13 |
+
assert len(blocks) == 14
|
| 14 |
assert blocks[0].block_type == "SectionHeader"
|
| 15 |
assert page_info[0]["bbox"] is not None
|
| 16 |
assert page_info[0]["polygon"] is not None
|
| 17 |
+
|
| 18 |
+
figure_groups = [block for block in blocks if block.block_type == "FigureGroup"]
|
| 19 |
+
figures = [block for block in blocks if block.block_type == "Figure"]
|
| 20 |
+
captions = [block for block in blocks if block.block_type == "Caption"]
|
| 21 |
+
|
| 22 |
+
assert len(figure_groups) == 1
|
| 23 |
+
assert len(figures) == 0
|
| 24 |
+
assert len(captions) == 0
|
| 25 |
+
|
| 26 |
+
figure_group = figure_groups[0]
|
| 27 |
+
assert figure_group.images is not None
|
| 28 |
+
assert len(figure_group.images) == 1
|
| 29 |
+
assert "<img src='/page/0/Figure/9'>" in figure_group.html
|