luciano/markitdown
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

finished webp support

Browse source
This commit is contained in:
rong-xyz 2025-04-23 06:39:24 +00:00
parent 46b44d3ebd
commit bc67a318a1
8 changed files with 57 additions and 10 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

43
packages/markitup/src/markitup/_base_converter.py
View file

@ -2,9 +2,11 @@ import os
import tempfile
from warnings import warn
from typing import Any, Union, BinaryIO, Optional, List, Dict
from ._schemas import StreamInfo
from ._schemas import StreamInfo, Config
import re
import base64
from PIL import Image
from io import BytesIO
class DocumentConverterResult:
@ -13,6 +15,7 @@ class DocumentConverterResult:
def __init__(
self,
markdown: str = "",
config: Optional[Config] = None,
*,
title: Optional[str] = None,
audio_stream: Optional[BinaryIO] = None,
@ -26,12 +29,16 @@ class DocumentConverterResult:
Parameters:
- markdown: The converted Markdown text.
- config: Optional configuration settings.
- title: Optional title of the document.
- audio_stream: Optional audio data.
- stream_info: Optional stream information.
"""
self.markdown = markdown
self.audio_stream = audio_stream
self.title = title
self.stream_info = stream_info
self.config = config
def to_llm(self) -> List[Dict[str, Any]]:
"""
@ -65,6 +72,18 @@ class DocumentConverterResult:
# Extract image data
alt_text, content_type, b64_data = match.groups()
if self.config.image_use_webp:
# Decode base64 data
img_data = base64.b64decode(b64_data)
# Check if it's already a WebP image
if "webp" not in content_type.lower():
# Convert to WebP
webp_data = self._convert_image_to_webp(img_data)
# Replace with WebP data
b64_data = base64.b64encode(webp_data).decode('utf-8')
content_type = "image/webp"
# Add the image
content.append({
"type": "image",
@ -94,6 +113,28 @@ class DocumentConverterResult:
})
return content
def _convert_image_to_webp(self, image_data: bytes, quality: int = 80) -> bytes:
"""
Convert image data to WebP format.
Parameters:
- image_data: The original image data as bytes.
- quality: The quality setting (0-100) for WebP conversion.
Returns:
- WebP converted image data as bytes.
"""
img = Image.open(BytesIO(image_data))
# Convert to RGB if image has alpha channel or is not in RGB mode
if img.mode in ('RGBA', 'LA') or (img.mode != 'RGB' and img.mode != 'L'):
img = img.convert('RGB')
# Save as WebP to a BytesIO object
webp_buffer = BytesIO()
img.save(webp_buffer, format="WEBP", quality=quality)
webp_buffer.seek(0)
return webp_buffer.read()
@property
def text_content(self) -> str:
"""Soft-deprecated alias for `markdown`. New code should migrate to using `markdown` or __str__."""

2
packages/markitup/src/markitup/_markitup.py
View file

@ -56,7 +56,7 @@ class MarkItUp:
case "xls":
return XlsConverter(config=self.config).convert(stream, stream_info), stream_info
case "csv":
return CsvConverter().convert(stream, stream_info), stream_info
return CsvConverter(config=self.config).convert(stream, stream_info), stream_info
case "docx":
return DocxConverter(config=self.config).convert(stream, stream_info), stream_info
case "image":

4
packages/markitup/src/markitup/converters/_audio_converter.py
View file

@ -29,8 +29,8 @@ class AudioConverter(DocumentConverter):
file_stream, magic_type=stream_info.magic_type)
if transcript:
md_content += "\n\n### Audio Transcript:\n" + transcript
return DocumentConverterResult(markdown=md_content.strip())
return DocumentConverterResult(markdown=md_content.strip(), config=self.config)
else:
return DocumentConverterResult(audio_stream=file_stream, stream_info=stream_info)
return DocumentConverterResult(audio_stream=file_stream, stream_info=stream_info, config=self.config)
# Return the result

6
packages/markitup/src/markitup/converters/_csv_converter.py
View file

@ -4,13 +4,15 @@ import io
from typing import BinaryIO, Any
from charset_normalizer import from_bytes
from .._base_converter import DocumentConverter, DocumentConverterResult
from .._schemas import StreamInfo
from .._schemas import StreamInfo, Config
class CsvConverter(DocumentConverter):
"""
Converts CSV files to Markdown tables.
"""
def __init__(self, config: Config):
self.config = config
def convert(
self,
@ -48,4 +50,4 @@ class CsvConverter(DocumentConverter):
result = "\n".join(markdown_table)
return DocumentConverterResult(markdown=result)
return DocumentConverterResult(markdown=result, config=self.config)

3
packages/markitup/src/markitup/converters/_html_converter.py
View file

@ -52,10 +52,11 @@ class HtmlConverter(DocumentConverter):
# remove leading and trailing \n
webpage_text = webpage_text.strip()
print(webpage_text)
return DocumentConverterResult(
markdown=webpage_text,
title=None if soup.title is None else soup.title.string,
config=self.config,
)
def convert_string(

2
packages/markitup/src/markitup/converters/_image_converter.py
View file

@ -39,8 +39,10 @@ class ImageConverter(DocumentConverter):
return DocumentConverterResult(
markdown=markdown_content,
config=self.config,
)
else:
return DocumentConverterResult(
markdown="No Image read as the supported modalities do not include 'image'",
config=self.config,
)

1
packages/markitup/src/markitup/converters/_pdf_converter.py
View file

@ -63,4 +63,5 @@ class PdfConverter(DocumentConverter):
doc.close()
return DocumentConverterResult(
markdown=markdown_content,
config=self.config,
)

2
packages/markitup/src/markitup/converters/_pptx_converter.py
View file

@ -116,7 +116,7 @@ class PptxConverter(DocumentConverter):
md_content += notes_frame.text
md_content = md_content.strip()
return DocumentConverterResult(markdown=md_content.strip())
return DocumentConverterResult(markdown=md_content.strip(), config=self.config)
def _is_picture(self, shape):
if shape.shape_type == pptx.enum.shapes.MSO_SHAPE_TYPE.PICTURE: