From 8717dd906667c5bf7a6c11f202ab475b14456990 Mon Sep 17 00:00:00 2001 From: Michael Williamson Date: Sun, 1 Dec 2013 21:33:53 +0000 Subject: Add support for reading hyperlinks --- mammoth/documents.py | 3 ++- mammoth/docx/document_xml.py | 14 +++++++++++++- 2 files changed, 15 insertions(+), 2 deletions(-) (limited to 'mammoth') diff --git a/mammoth/documents.py b/mammoth/documents.py index 851755e..6ce4582 100644 --- a/mammoth/documents.py +++ b/mammoth/documents.py @@ -4,7 +4,7 @@ Document = collections.namedtuple("Document", ["children"]) Paragraph = collections.namedtuple("Paragraph", ["children", "style_name", "numbering"]) Run = collections.namedtuple("Run", ["children", "style_name", "is_bold", "is_italic"]) Text = collections.namedtuple("Text", ["value"]) - +Hyperlink = collections.namedtuple("Hyperlink", ["href", "children"]) class Tab(object): pass @@ -34,3 +34,4 @@ def tab(): image = Image +hyperlink = Hyperlink diff --git a/mammoth/docx/document_xml.py b/mammoth/docx/document_xml.py index 9c895d4..fad4f87 100644 --- a/mammoth/docx/document_xml.py +++ b/mammoth/docx/document_xml.py @@ -96,12 +96,24 @@ def _create_reader(numbering, content_types, relationships, docx_file): @handler("w:tab") def tab(element): return results.success(documents.tab()) - + + @handler("w:ins") def ins(element): return _read_xml_elements(element.children) + @handler("w:hyperlink") + def hyperlink(element): + relationship_id = element.attributes.get("r:id") + children_result = _read_xml_elements(element.children) + if relationship_id is None: + return children_result + else: + href = relationships[relationship_id].target + return children_result.map(lambda children: documents.hyperlink(href, children)) + + @handler("w:drawing") def drawing(element): return _read_xml_elements(element.children) -- cgit v1.2.3