lxbfYeaa/BookConverter
Archived
Template
1
0
Fork 0
forked from LiveCarta/BookConverter
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add lines with style 'border-bottom'

Browse Source
This commit is contained in:
Kiryl
2021-10-26 17:47:51 +03:00
parent 24210c9999
commit 8c37482616
2 changed files with 25 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files

32
src/epub_converter/html_epub_preprocessor.py
View File

@@ -144,7 +144,7 @@ def clean_headings_content(content: Tag, title: str):
break
def _heading_tag2p_tag(body_tag):
def heading_tag_to_p_tag(body_tag):
"""
Function to convert all lower level headings to p tags
"""
@@ -267,7 +267,7 @@ def unwrap_structural_tags(body_tag):
if not tag_.parent.attrs.get('class'):
tag_.parent.attrs['class'] = tag_class
def _preserve_class_in_section_tag(tag_) -> bool:
def preserve_class_in_section_tag(tag_) -> bool:
# to save css style inherited from class, copy class to child <p>
# this is for Wiley books with boxes
# returns True, if <section> could be unwrapped
@@ -288,10 +288,10 @@ def unwrap_structural_tags(body_tag):
else:
return True
def _add_table_to_abc_books(tag_, border, bg_color):
def add_table_to_abc_books(tag_, border, bg_color):
wrap_block_tag_with_table(body_tag, old_tag=tag_, width='100', border=border, bg_color=bg_color)
def _add_span_to_save_ids_for_links(tag_to_be_removed):
def add_span_to_save_ids_for_links(tag_to_be_removed):
if tag_to_be_removed.attrs.get('id'):
insert_span_with_attrs_before_tag(main_tag=body_tag, tag=tag_to_be_removed,
id_=tag_to_be_removed.attrs['id'],
@@ -311,17 +311,17 @@ def unwrap_structural_tags(body_tag):
if div.attrs.get('class'):
div_class = div.attrs['class'] if not isinstance(div.attrs['class'], list) else div.attrs['class'][0]
if div_class in ['C409', 'C409a']:
_add_table_to_abc_books(div, border='solid 3px', bg_color='#e7e7e9')
add_table_to_abc_books(div, border='solid 3px', bg_color='#e7e7e9')
elif div_class in ['C441', 'C816']:
_add_table_to_abc_books(div, border='solid #6e6e70 1px', bg_color='#e7e7e8')
add_table_to_abc_books(div, border='solid #6e6e70 1px', bg_color='#e7e7e8')
if div.attrs.get('style'):
if 'background-color' in div.attrs['style']:
end_index = div.attrs['style'].find('background-color') + len('background-color')
start_index_of_color = end_index + 2
bg_color = div.attrs['style'][start_index_of_color:start_index_of_color+7]
_add_table_to_abc_books(div, border='', bg_color=bg_color)
add_table_to_abc_books(div, border='', bg_color=bg_color)
if div.attrs.get('style') == '':
del div.attrs['style']
@@ -331,19 +331,19 @@ def unwrap_structural_tags(body_tag):
div.name = 'p'
continue
_add_span_to_save_ids_for_links(div)
add_span_to_save_ids_for_links(div)
div.unwrap()
for s in body_tag.find_all("section"):
could_be_unwrapped = True
if s.attrs.get('class'):
could_be_unwrapped = _preserve_class_in_section_tag(s)
_add_span_to_save_ids_for_links(s)
could_be_unwrapped = preserve_class_in_section_tag(s)
add_span_to_save_ids_for_links(s)
if could_be_unwrapped:
s.unwrap()
for s in body_tag.find_all("article"):
_add_span_to_save_ids_for_links(s)
add_span_to_save_ids_for_links(s)
s.unwrap()
for s in body_tag.find_all("figure"):
@@ -351,22 +351,22 @@ def unwrap_structural_tags(body_tag):
s.attrs['style'] = "text-align: center;" # to center image inside this tag
for s in body_tag.find_all("figcaption"):
_add_span_to_save_ids_for_links(s)
add_span_to_save_ids_for_links(s)
s.unwrap()
for s in body_tag.find_all("aside"):
s.name = 'blockquote'
for s in body_tag.find_all("main"):
_add_span_to_save_ids_for_links(s)
add_span_to_save_ids_for_links(s)
s.unwrap()
for s in body_tag.find_all("body"):
_add_span_to_save_ids_for_links(s)
add_span_to_save_ids_for_links(s)
s.unwrap()
for s in body_tag.find_all("html"):
_add_span_to_save_ids_for_links(s)
add_span_to_save_ids_for_links(s)
s.unwrap()
for s in body_tag.find_all("header"):
@@ -385,7 +385,7 @@ def unwrap_structural_tags(body_tag):
parents_marks_are_body = [x.parent == body_tag for x in marks]
assert all(parents_marks_are_body), 'Anchor for chapter is deeper than 2 level. Chapters can not be parsed.'
_heading_tag2p_tag(body_tag)
heading_tag_to_p_tag(body_tag)
# wrap NavigableString with <p>
for node in body_tag: