diff options
| -rw-r--r-- | mammoth/docx/__init__.py | 31 | ||||
| -rw-r--r-- | mammoth/docx/body_xml.py | 2 | ||||
| -rw-r--r-- | mammoth/docx/document_xml.py | 14 | ||||
| -rw-r--r-- | mammoth/options.py | 2 | ||||
| -rw-r--r-- | tests/mammoth_tests.py | 16 | ||||
| -rwxr-xr-x | tests/test-data/comments.docx | bin | 0 -> 15526 bytes |
6 files changed, 54 insertions, 11 deletions
diff --git a/mammoth/docx/__init__.py b/mammoth/docx/__init__.py index 47e7a21..8fb150a 100644 --- a/mammoth/docx/__init__.py +++ b/mammoth/docx/__init__.py @@ -7,39 +7,56 @@ from .relationships_xml import read_relationships_xml_element, Relationships from .numbering_xml import read_numbering_xml_element, Numbering from .styles_xml import read_styles_xml_element from .notes_xml import create_footnotes_reader, create_endnotes_reader +from .comments_xml import create_comments_reader from .files import Files from . import body_xml, office_xml from ..zips import open_zip +_empty_result = results.success([]) + + def read(fileobj): zip_file = open_zip(fileobj, "r") body_readers = _body_readers(getattr(fileobj, "name", None), zip_file) - return _read_notes(zip_file, body_readers).bind(lambda notes: - _read_document(zip_file, body_readers, notes)) + return results.combine([ + _read_notes(zip_file, body_readers), + _read_comments(zip_file, body_readers), + ]).bind(lambda referents: + _read_document(zip_file, body_readers, notes=referents[0], comments=referents[1]) + ) def _read_notes(zip_file, body_readers): - empty_result = results.success([]) - read_footnotes_xml = create_footnotes_reader(body_readers("footnotes")) footnotes = _try_read_entry_or_default( - zip_file, "word/footnotes.xml", read_footnotes_xml, default=empty_result) + zip_file, "word/footnotes.xml", read_footnotes_xml, default=_empty_result) read_endnotes_xml = create_endnotes_reader(body_readers("endnotes")) endnotes = _try_read_entry_or_default( - zip_file, "word/endnotes.xml", read_endnotes_xml, default=empty_result) + zip_file, "word/endnotes.xml", read_endnotes_xml, default=_empty_result) return results.combine([footnotes, endnotes]).map(lists.flatten) + + +def _read_comments(zip_file, body_readers): + return _try_read_entry_or_default( + zip_file, + "word/comments.xml", + create_comments_reader(body_readers("comments")), + default=_empty_result, + ) + -def _read_document(zip_file, body_readers, notes): +def _read_document(zip_file, body_readers, notes, comments): with zip_file.open("word/document.xml") as document_fileobj: document_xml = office_xml.read(document_fileobj) return read_document_xml_element( document_xml, body_reader=body_readers("document"), notes=notes, + comments=comments, ) diff --git a/mammoth/docx/body_xml.py b/mammoth/docx/body_xml.py index fe75f7c..b1be2e5 100644 --- a/mammoth/docx/body_xml.py +++ b/mammoth/docx/body_xml.py @@ -52,13 +52,13 @@ def _create_reader(numbering, content_types, relationships, styles, docx_file, f "office-word:wrap", "v:shadow", "v:shapetype", + "w:annotationRef", "w:bookmarkEnd", "w:sectPr", "w:proofErr", "w:lastRenderedPageBreak", "w:commentRangeStart", "w:commentRangeEnd", - "w:commentReference", "w:del", "w:footnoteRef", "w:endnoteRef", diff --git a/mammoth/docx/document_xml.py b/mammoth/docx/document_xml.py index 1c15001..e0d19a4 100644 --- a/mammoth/docx/document_xml.py +++ b/mammoth/docx/document_xml.py @@ -1,13 +1,21 @@ from .. import documents -def read_document_xml_element(element, +def read_document_xml_element( + element, body_reader, - notes=None): + notes=None, + comments=None): if notes is None: notes = [] + if comments is None: + comments = [] body_element = element.find_child("w:body") return body_reader.read_all(body_element.children) \ - .map(lambda children: documents.document(children, documents.notes(notes))) + .map(lambda children: documents.document( + children, + notes=documents.notes(notes), + comments=comments + )) diff --git a/mammoth/options.py b/mammoth/options.py index 3bc6eb7..d337ec2 100644 --- a/mammoth/options.py +++ b/mammoth/options.py @@ -46,6 +46,8 @@ p[style-name='footnote text'] => p r[style-name='footnote reference'] => p[style-name='endnote text'] => p r[style-name='endnote reference'] => +p[style-name='annotation text'] => p +r[style-name='annotation reference'] => # LibreOffice p[style-name='Footnote'] => p diff --git a/tests/mammoth_tests.py b/tests/mammoth_tests.py index 201df70..8ad9b33 100644 --- a/tests/mammoth_tests.py +++ b/tests/mammoth_tests.py @@ -185,6 +185,22 @@ def relationships_are_handled_properly_in_footnotes(): @istest +def when_style_mapping_is_defined_for_comment_references_then_comments_are_included(): + expected_html = ( + '<p>Ouch' + + '<sup><a href="#doc-42-comment-0" id="doc-42-comment-ref-0">[MW 1]</a></sup>.' + + '<sup><a href="#doc-42-comment-2" id="doc-42-comment-ref-2">[MW 2]</a></sup></p>' + + '<dl><dt id="doc-42-comment-0">MW 1</dt><dd><p>A tachyon walks into a bar. <a href="#doc-42-comment-ref-0">↑</a></p></dd>' + + '<dt id="doc-42-comment-2">MW 2</dt><dd><p>Fin. <a href="#doc-42-comment-ref-2">↑</a></p></dd></dl>' + ) + + with open(test_path("comments.docx"), "rb") as fileobj: + result = mammoth.convert_to_html(fileobj=fileobj, id_prefix="doc-42-", style_map="comment-reference => sup") + assert_equal([], result.messages) + assert_equal(expected_html, result.value) + + +@istest def text_boxes_are_read(): with open(test_path("text-box.docx"), "rb") as fileobj: result = mammoth.convert_to_html(fileobj=fileobj) diff --git a/tests/test-data/comments.docx b/tests/test-data/comments.docx Binary files differnew file mode 100755 index 0000000..32cdae1 --- /dev/null +++ b/tests/test-data/comments.docx |