def parse(cls, filename: str) -> List[dict]:
f = zipfile.ZipFile(filename)
context = get_context(f)
# 读取主文档 xml
document = context.get("officeDocument")
# 读取文档关联 xml
content_rels = (context.get("content_path2rels") or {}).get(document)
# 读取主文档
xml_content = f.read(document)
tree = etree.fromstring(xml_content)
section = []
for index, paragraph in enumerate(cls._iterparagraphs(tree)):
content = [
text.strip()
for text in cls._itertext(paragraph, f, content_rels)
if text.strip()
]
if content:
section.append({
"page": str(index + 1),
"context": " ".join(content)
})
return section
def docx2python(docx_filename: str,
image_folder: Optional[str] = None,
html: bool = False,
extract_image: bool = True) -> DocxContent:
"""Unzip a docx file and extract contents.
There's a bit of ugly try/except toward the bottom.
One file in 5300 had the headers and footers mislabeled in
``word/_rels.document.xml.rels``. Instead of ``header.xml``, this had the
header identified as ``word/header.xml``. After trying with
``content_dir/file``, try again with just ``file``.
"""
zipf = zipfile.ZipFile(docx_filename)
context = get_context(zipf)
context["do_html"] = html
def file_text(filename_):
context["rId2Target"] = {
x["Id"]: x["Target"]
for x in context["content_path2rels"][filename_]
}
try:
unzipped = zipf.read(filename_)
except KeyError:
# content dir specified twice
unzipped = zipf.read('/'.join(Path(filename_).parts[1:]))
return get_text(unzipped, context)
header = [file_text(filename) for filename in context["headers"]]
header = [x for y in header for x in y]
body = file_text(context["officeDocument"])
footer = [file_text(filename) for filename in context["footers"]]
footer = [x for y in footer for x in y]
footnotes = [file_text(filename) for filename in context["footnotes"]]
footnotes = [x for y in footnotes for x in y]
endnotes = [file_text(filename) for filename in context["endnotes"]]
endnotes = [x for y in endnotes for x in y]
if extract_image:
images = pull_image_files(zipf, context, image_folder)
else:
images = None
zipf.close()
return DocxContent(
header=header,
body=body,
footer=footer,
footnotes=footnotes,
endnotes=endnotes,
images=images,
properties=context["docProp2text"],
)
def test_no_image_files(self) -> None:
"""Pass silently when no image files."""
zipf = zipfile.ZipFile("resources/basic.docx")
context = get_context(zipf)
pull_image_files(zipf, context, "delete_this/path/to/images")
assert os.listdir("delete_this/path/to/images") == []
# clean up
shutil.rmtree("delete_this")
def test_pull_image_files(self) -> None:
"""Copy image files to output path."""
zipf = zipfile.ZipFile("resources/example.docx")
context = get_context(zipf)
pull_image_files(zipf, context, "delete_this/path/to/images")
assert os.listdir("delete_this/path/to/images") == [
"image1.png", "image2.jpg"
]
# clean up
shutil.rmtree("delete_this")
def test_lists(self) -> None:
"""Pass silently when no numbered or bulleted lists."""
zipf = zipfile.ZipFile("resources/basic.docx")
context = get_context(zipf)
assert "numId2numFmts" not in context
assert "numId2count" not in context
def docx_context() -> Dict[str, Any]:
"""result of running strip_text.get_context"""
zipf = zipfile.ZipFile("resources/example.docx")
return get_context(zipf)