voussoir/sigilplugins
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Add remove_header_br.

Browse source
This commit is contained in:
Ethan Dalool 2020-01-07 18:23:44 -08:00
parent 8026bb55c4
commit 205bbf61a2
1 changed files with 10 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
cleanerupper/plugin.py
View file

@ -64,6 +64,16 @@ def bring_punctuation_into_italics(html):
html = re.sub('</{tag}>\\{punct}'.format(**locals()), '{punct}</{tag}>'.format(**locals()), html) html = re.sub('</{tag}>\\{punct}'.format(**locals()), '{punct}</{tag}>'.format(**locals()), html)
return html return html
@html_cleaner
def remove_header_br(html):
html = re.sub(r'<h1>([^\n]+?)\s*<br/>\s*([^\n]+?)</h1>', r'<h1>\1 \2</h1>', html)
html = re.sub(r'<h2>([^\n]+?)\s*<br/>\s*([^\n]+?)</h2>', r'<h2>\1 \2</h2>', html)
html = re.sub(r'<h3>([^\n]+?)\s*<br/>\s*([^\n]+?)</h3>', r'<h3>\1 \2</h3>', html)
html = re.sub(r'<h4>([^\n]+?)\s*<br/>\s*([^\n]+?)</h4>', r'<h4>\1 \2</h4>', html)
html = re.sub(r'<h5>([^\n]+?)\s*<br/>\s*([^\n]+?)</h5>', r'<h5>\1 \2</h5>', html)
html = re.sub(r'<h6>([^\n]+?)\s*<br/>\s*([^\n]+?)</h6>', r'<h6>\1 \2</h6>', html)
return html
@html_cleaner @html_cleaner
def remove_misc_strings(html): def remove_misc_strings(html):
html = html.replace('epub:type="pagebreak"', '') html = html.replace('epub:type="pagebreak"', '')