From 7d8e8197b03f87a8cac2eed8f75e8a41136cc760 Mon Sep 17 00:00:00 2001 From: Michael Williamson Date: Wed, 15 Jul 2015 16:19:37 +0100 Subject: Split off raw_text module --- mammoth/__init__.py | 16 +++------------- 1 file changed, 3 insertions(+), 13 deletions(-) (limited to 'mammoth/__init__.py') diff --git a/mammoth/__init__.py b/mammoth/__init__.py index 8ec2400..f67f890 100644 --- a/mammoth/__init__.py +++ b/mammoth/__init__.py @@ -1,4 +1,5 @@ -from . import docx, conversion, options, documents, images, underline +from . import docx, conversion, options, images, underline +from .raw_text import extract_raw_text_from_element __all__ = ["convert_to_html", "extract_raw_text", "images", "underline"] @@ -24,15 +25,4 @@ def convert(fileobj, transform_document=None, id_prefix=None, **kwargs): def extract_raw_text(fileobj): - return docx.read(fileobj).map(_extract_raw_text_from_element) - - -def _extract_raw_text_from_element(element): - if isinstance(element, documents.Text): - return element.value - else: - text = "".join(map(_extract_raw_text_from_element, element.children)) - if isinstance(element, documents.Paragraph): - return text + "\n\n" - else: - return text + return docx.read(fileobj).map(extract_raw_text_from_element) -- cgit v1.2.3