Backup_Repos/calibre
1
0
Fork 0
mirror of git://github.com/kovidgoyal/calibre.git synced 2026-05-08 21:34:56 +02:00
Code Issues Projects Releases Wiki Activity

Added support for <img>, <link> and <style> tags to html2lrf

Browse source
This commit is contained in:
Kovid Goyal 2007-04-19 03:13:02 +00:00
parent d69fad53f4
commit 22872ee668
1 changed files with 28 additions and 6 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

34
src/libprs500/lrf/html/convert_from.py
View file

@ -20,14 +20,16 @@
import os, re, sys
from htmlentitydefs import name2codepoint
from optparse import OptionParser
from urllib import urlopen
from libprs500.lrf.html.BeautifulSoup import BeautifulSoup, Comment, Tag, NavigableString
from libprs500.lrf.pylrs.pylrs import Book, Page, Paragraph, TextBlock, CR, Italic
from libprs500.lrf.pylrs.pylrs import Book, Page, Paragraph, TextBlock, \
CR, Italic, ImageStream, ImageBlock
from libprs500.lrf.pylrs.pylrs import Span as _Span
from libprs500.lrf import ConversionError
class Span(_Span):
replaced_entities = [ 'amp', 'lt', 'gt' , 'ldquo', 'rdquo', 'lsquo', 'rsquo' ]
replaced_entities = [ 'amp', 'lt', 'gt' , 'ldquo', 'rdquo', 'lsquo', 'rsquo', 'nbsp' ]
patterns = [ re.compile('&'+i+';') for i in replaced_entities ]
targets = [ unichr(name2codepoint[i]) for i in replaced_entities ]
rules = zip(patterns, targets)
@ -283,8 +285,8 @@ def sanctify_css(css):
for key in css.keys():
test = key.lower()
if test.startswith('margin') or 'indent' in test or \
'padding' in test or 'border' in test or test in \
['color', 'display', 'text-decoration', 'letter-spacing']:
'padding' in test or 'border' in test or 'page-break' in test \
or test in ['color', 'display', 'text-decoration', 'letter-spacing']:
css.pop(key)
return css
@ -321,9 +323,29 @@ def process_text_tag(tag, pcss):
return
if tagname in ["title", "script", "meta"]:
pass
elif tagname == 'img':
if tag.has_key('src'):
if os.access(tag['src'], os.R_OK):
self.current_block.append(self.current_para)
self.current_page.append(self.current_block)
ib = ImageBlock(ImageStream(tag['src']))
self.current_page.append(ib)
self.current_block = TextBlock()
self.current_para = Paragraph()
elif tagname in ['style', 'link']:
# TODO: Append CSS to self.css
pass
if tagname == 'style':
for c in tag.contents:
if isinstance(c,NavigableString):
self.css.update(self.parse_css(str(c)))
elif tag.has_key('type') and tag['type'] == "text/css" \
and tag.has_key('href'):
url = tag['href']
if url.startswith('http://'):
f = urlopen(url)
else:
f = open(url)
self.css.update(f.read())
f.close()
elif tagname == 'p':
css = self.tag_css(tag, parent_css=parent_css)
indent = css.pop('text-indent', '')