Add support for preserving base64 encoded images (#1140)

* optional reserve base64 string in markdown _CustomMarkdownify and pptx
* add other converter para support
* fix linter
* Use *kwarg to pass keep_data_uri para.
* Add module cli vector tests
* Fixed formatting, and adjusted tests.
This commit is contained in:
Yuzhong Zhang
2025-03-21 09:50:23 +08:00
committed by GitHub
parent c0a511ecff
commit 52432bd228
12 changed files with 214 additions and 26 deletions

View File

@@ -25,8 +25,11 @@ GENERAL_TEST_VECTORS = [
"# Abstract",
"# Introduction",
"AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation",
"data:image/png;base64...",
],
must_not_include=[
"data:image/png;base64,iVBORw0KGgoAAAANSU",
],
must_not_include=[],
),
FileTestVector(
filename="test.xlsx",
@@ -65,8 +68,9 @@ GENERAL_TEST_VECTORS = [
"AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation",
"a3f6004b-6f4f-4ea8-bee3-3741f4dc385f", # chart title
"2003", # chart value
"![This phrase of the caption is Human-written.](Picture4.jpg)",
],
must_not_include=[],
must_not_include=["data:image/jpeg;base64,/9j/4AAQSkZJRgABAQE"],
),
FileTestVector(
filename="test_outlook_msg.msg",
@@ -230,3 +234,45 @@ GENERAL_TEST_VECTORS = [
must_not_include=[],
),
]
DATA_URI_TEST_VECTORS = [
FileTestVector(
filename="test.docx",
mimetype="application/vnd.openxmlformats-officedocument.wordprocessingml.document",
charset=None,
url=None,
must_include=[
"314b0a30-5b04-470b-b9f7-eed2c2bec74a",
"49e168b7-d2ae-407f-a055-2167576f39a1",
"## d666f1f7-46cb-42bd-9a39-9a39cf2a509f",
"# Abstract",
"# Introduction",
"AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation",
"data:image/png;base64,iVBORw0KGgoAAAANSU",
],
must_not_include=[
"data:image/png;base64...",
],
),
FileTestVector(
filename="test.pptx",
mimetype="application/vnd.openxmlformats-officedocument.presentationml.presentation",
charset=None,
url=None,
must_include=[
"2cdda5c8-e50e-4db4-b5f0-9722a649f455",
"04191ea8-5c73-4215-a1d3-1cfb43aaaf12",
"44bf7d06-5e7a-4a40-a2e1-a2e42ef28c8a",
"1b92870d-e3b5-4e65-8153-919f4ff45592",
"AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation",
"a3f6004b-6f4f-4ea8-bee3-3741f4dc385f", # chart title
"2003", # chart value
"![This phrase of the caption is Human-written.]", # image caption
"data:image/jpeg;base64,/9j/4AAQSkZJRgABAQE",
],
must_not_include=[
"![This phrase of the caption is Human-written.](Picture4.jpg)",
],
),
]

View File

@@ -7,9 +7,17 @@ import locale
from typing import List
if __name__ == "__main__":
from _test_vectors import GENERAL_TEST_VECTORS, FileTestVector
from _test_vectors import (
GENERAL_TEST_VECTORS,
DATA_URI_TEST_VECTORS,
FileTestVector,
)
else:
from ._test_vectors import GENERAL_TEST_VECTORS, FileTestVector
from ._test_vectors import (
GENERAL_TEST_VECTORS,
DATA_URI_TEST_VECTORS,
FileTestVector,
)
from markitdown import (
MarkItDown,
@@ -149,6 +157,39 @@ def test_convert_url(shared_tmp_dir, test_vector):
assert test_string not in stdout
@pytest.mark.parametrize("test_vector", DATA_URI_TEST_VECTORS)
def test_output_to_file_with_data_uris(shared_tmp_dir, test_vector) -> None:
"""Test CLI functionality when keep_data_uris is enabled"""
output_file = os.path.join(shared_tmp_dir, test_vector.filename + ".output")
result = subprocess.run(
[
"python",
"-m",
"markitdown",
"--keep-data-uris",
"-o",
output_file,
os.path.join(TEST_FILES_DIR, test_vector.filename),
],
capture_output=True,
text=True,
)
assert result.returncode == 0, f"CLI exited with error: {result.stderr}"
assert os.path.exists(output_file), f"Output file not created: {output_file}"
with open(output_file, "r") as f:
output_data = f.read()
for test_string in test_vector.must_include:
assert test_string in output_data
for test_string in test_vector.must_not_include:
assert test_string not in output_data
os.remove(output_file)
assert not os.path.exists(output_file), f"Output file not deleted: {output_file}"
if __name__ == "__main__":
import sys
import tempfile
@@ -156,6 +197,7 @@ if __name__ == "__main__":
"""Runs this file's tests from the command line."""
with tempfile.TemporaryDirectory() as tmp_dir:
# General tests
for test_function in [
test_output_to_stdout,
test_output_to_file,
@@ -169,4 +211,17 @@ if __name__ == "__main__":
)
test_function(tmp_dir, test_vector)
print("OK")
# Data URI tests
for test_function in [
test_output_to_file_with_data_uris,
]:
for test_vector in DATA_URI_TEST_VECTORS:
print(
f"Running {test_function.__name__} on {test_vector.filename}...",
end="",
)
test_function(tmp_dir, test_vector)
print("OK")
print("All tests passed!")

View File

@@ -6,9 +6,9 @@ import codecs
if __name__ == "__main__":
from _test_vectors import GENERAL_TEST_VECTORS
from _test_vectors import GENERAL_TEST_VECTORS, DATA_URI_TEST_VECTORS
else:
from ._test_vectors import GENERAL_TEST_VECTORS
from ._test_vectors import GENERAL_TEST_VECTORS, DATA_URI_TEST_VECTORS
from markitdown import (
MarkItDown,
@@ -124,10 +124,52 @@ def test_convert_url(test_vector):
assert string not in result.markdown
@pytest.mark.parametrize("test_vector", DATA_URI_TEST_VECTORS)
def test_convert_with_data_uris(test_vector):
"""Test API functionality when keep_data_uris is enabled"""
markitdown = MarkItDown()
# Test local file conversion
result = markitdown.convert(
os.path.join(TEST_FILES_DIR, test_vector.filename),
keep_data_uris=True,
url=test_vector.url,
)
for string in test_vector.must_include:
assert string in result.markdown
for string in test_vector.must_not_include:
assert string not in result.markdown
@pytest.mark.parametrize("test_vector", DATA_URI_TEST_VECTORS)
def test_convert_stream_with_data_uris(test_vector):
"""Test the conversion of a stream with no stream info."""
markitdown = MarkItDown()
stream_info = StreamInfo(
extension=os.path.splitext(test_vector.filename)[1],
mimetype=test_vector.mimetype,
charset=test_vector.charset,
)
with open(os.path.join(TEST_FILES_DIR, test_vector.filename), "rb") as stream:
result = markitdown.convert(
stream, stream_info=stream_info, keep_data_uris=True, url=test_vector.url
)
for string in test_vector.must_include:
assert string in result.markdown
for string in test_vector.must_not_include:
assert string not in result.markdown
if __name__ == "__main__":
import sys
"""Runs this file's tests from the command line."""
# General tests
for test_function in [
test_guess_stream_info,
test_convert_local,
@@ -141,4 +183,17 @@ if __name__ == "__main__":
)
test_function(test_vector)
print("OK")
# Data URI tests
for test_function in [
test_convert_with_data_uris,
test_convert_stream_with_data_uris,
]:
for test_vector in DATA_URI_TEST_VECTORS:
print(
f"Running {test_function.__name__} on {test_vector.filename}...", end=""
)
test_function(test_vector)
print("OK")
print("All tests passed!")