def serialize_elem(self, elem, item, nsrmap=NSRMAP):
buf = self.buf
if not isinstance(elem.tag, (str, bytes)) \
or parse_utils.namespace(elem.tag) not in nsrmap:
return
tag = base.prefixname(elem.tag, nsrmap)
# Previous layers take care of @name
id_ = elem.attrib.pop('id', None)
if id_:
href = '#'.join((item.href, id_))
offset = self.anchor_offset or buf.tell()
key = base.urlnormalize(href)
# Only set this id_offset if it wasn't previously seen
self.id_offsets[key] = self.id_offsets.get(key, offset)
if self.anchor_offset is not None and \
tag == 'a' and not elem.attrib and \
not len(elem) and not elem.text:
return
self.anchor_offset = buf.tell()
buf.write(b'<')
buf.write(tag.encode('utf-8'))
if elem.attrib:
for attr, val in elem.attrib.items():
if parse_utils.namespace(attr) not in nsrmap:
continue
attr = base.prefixname(attr, nsrmap)
buf.write(b' ')
if attr == 'href':
if self.serialize_href(val, item):
continue
elif attr == 'src':
href = base.urlnormalize(item.abshref(val))
if href in self.images:
index = self.images[href]
self.used_images.add(href)
buf.write(b'recindex="%05d"' % index)
continue
buf.write(attr.encode('utf-8'))
buf.write(b'="')
self.serialize_text(val, quot=True)
buf.write(b'"')
buf.write(b'>')
if elem.text or len(elem) > 0:
if elem.text:
self.anchor_offset = None
self.serialize_text(elem.text)
for child in elem:
self.serialize_elem(child, item)
if child.tail:
self.anchor_offset = None
self.serialize_text(child.tail)
buf.write(('</%s>' % tag).encode('utf-8'))
def __call__(self, oeb, context):
import css_parser
oeb.logger.info('Trimming unused files from manifest...')
self.opts = context
used = set()
for term in oeb.metadata:
for item in oeb.metadata[term]:
if item.value in oeb.manifest.hrefs:
used.add(oeb.manifest.hrefs[item.value])
elif item.value in oeb.manifest.ids:
used.add(oeb.manifest.ids[item.value])
for ref in oeb.guide.values():
path, _ = urllib.parse.urldefrag(ref.href)
if path in oeb.manifest.hrefs:
used.add(oeb.manifest.hrefs[path])
# TOC items are required to be in the spine
for item in oeb.spine:
used.add(item)
unchecked = used
while unchecked:
new = set()
for item in unchecked:
if (item.media_type in OEB_DOCS or
item.media_type[-4:] in ('/xml', '+xml')) and \
item.data is not None:
hrefs = [r[2] for r in iterlinks(item.data)]
for href in hrefs:
if isinstance(href, bytes):
href = href.decode('utf-8')
try:
href = item.abshref(urlnormalize(href))
except:
continue
if href in oeb.manifest.hrefs:
found = oeb.manifest.hrefs[href]
if found not in used:
new.add(found)
elif item.media_type == CSS_MIME:
for href in css_parser.getUrls(item.data):
href = item.abshref(urlnormalize(href))
if href in oeb.manifest.hrefs:
found = oeb.manifest.hrefs[href]
if found not in used:
new.add(found)
used.update(new)
unchecked = new
for item in oeb.manifest.values():
if item not in used:
oeb.logger.info('Trimming %r from manifest' % item.href)
oeb.manifest.remove(item)
def _toc_from_html(self, opf):
if 'toc' not in self.oeb.guide:
return False
self.log.debug('Reading TOC from HTML...')
itempath, frag = urllib.parse.urldefrag(self.oeb.guide['toc'].href)
item = self.oeb.manifest.hrefs[itempath]
html = item.data
if frag:
elems = base.xpath(html, './/*[@id="%s"]' % frag)
if not elems:
elems = base.xpath(html, './/*[@name="%s"]' % frag)
elem = elems[0] if elems else html
while elem != html and not base.xpath(elem, './/h:a[@href]'):
elem = elem.getparent()
html = elem
titles = collections.defaultdict(list)
order = []
for anchor in base.xpath(html, './/h:a[@href]'):
href = anchor.attrib['href']
href = item.abshref(base.urlnormalize(href))
path, frag = urllib.parse.urldefrag(href)
if path not in self.oeb.manifest.hrefs:
continue
title = base.xml2text(anchor)
title = base.COLLAPSE_RE.sub(' ', title.strip())
if href not in titles:
order.append(href)
titles[href].append(title)
toc = self.oeb.toc
for href in order:
toc.add(' '.join(titles[href]), href)
return True
def process_fonts(self):
''' Make sure all fonts are embeddable '''
from ebook_converter.ebooks.oeb.base import urlnormalize
from ebook_converter.utils.fonts.utils import remove_embed_restriction
processed = set()
for item in list(self.oeb.manifest):
if not hasattr(item.data, 'cssRules'):
continue
for i, rule in enumerate(item.data.cssRules):
if rule.type == rule.FONT_FACE_RULE:
try:
s = rule.style
src = s.getProperty('src').propertyValue[0].uri
except:
continue
path = item.abshref(src)
ff = self.oeb.manifest.hrefs.get(urlnormalize(path), None)
if ff is None:
continue
raw = nraw = ff.data
if path not in processed:
processed.add(path)
try:
nraw = remove_embed_restriction(raw)
except:
continue
if nraw != raw:
ff.data = nraw
self.oeb.container.write(path, nraw)
def __call__(self, oeb, opts):
import css_parser
self.log = oeb.logger
self.opts = opts
self.oeb = oeb
for item in oeb.manifest.items:
self.current_item = item
if etree.iselement(item.data):
rewrite_links(self.current_item.data, self.url_replacer)
elif hasattr(item.data, 'cssText'):
css_parser.replaceUrls(item.data, self.url_replacer)
if self.oeb.guide:
for ref in self.oeb.guide.values():
href = urlnormalize(ref.href)
href, frag = urllib.parse.urldefrag(href)
replacement = self.rename_map.get(href, None)
if replacement is not None:
nhref = replacement
if frag:
nhref += '#' + frag
ref.href = nhref
if self.oeb.toc:
self.fix_toc_entry(self.oeb.toc)
def serialize_href(self, href, _base=None):
"""
Serialize the href attribute of an <a> or <reference> tag. It is
serialized as filepos="000000000" and a pointer to its location is
stored in self.href_offsets so that the correct value can be filled in
at the end.
"""
hrefs = self.oeb.manifest.hrefs
try:
path, frag = urllib.parse.urldefrag(base.urlnormalize(href))
except ValueError:
# Unparseable URL
return False
if path and _base:
path = _base.abshref(path)
if path and path not in hrefs:
return False
buf = self.buf
item = hrefs[path] if path else None
if item and item.spine_position is None:
return False
path = item.href if item else _base.href
href = '#'.join((path, frag)) if frag else path
buf.write(b'filepos=')
self.href_offsets[href].append(buf.tell())
buf.write(b'0000000000')
return True
def find_font_face_rules(sheet, oeb):
'''
Find all @font-face rules in the given sheet and extract the relevant info from them.
sheet can be either a ManifestItem or a CSSStyleSheet.
'''
ans = []
try:
rules = sheet.data.cssRules
except AttributeError:
rules = sheet.cssRules
for i, rule in enumerate(rules):
if rule.type != rule.FONT_FACE_RULE:
continue
props = get_font_properties(rule, default='normal')
if not props['font-family'] or not props['src']:
continue
try:
path = sheet.abshref(props['src'])
except AttributeError:
path = props['src']
ff = oeb.manifest.hrefs.get(urlnormalize(path), None)
if not ff:
continue
props['item'] = ff
if props['font-weight'] in {'bolder', 'lighter'}:
props['font-weight'] = '400'
props['weight'] = int(props['font-weight'])
props['rule'] = rule
props['chars'] = set()
ans.append(props)
return ans
def rewrite_link(self, url, page=None):
if not page:
return url
abs_url = page.abshref(base.urlnormalize(url))
if abs_url in self.images:
return 'images/%s' % self.images[abs_url]
if abs_url in self.links:
return self.links[abs_url]
return url
def rasterize_item(self, item):
html = item.data
hrefs = self.oeb.manifest.hrefs
for elem in xpath(html, '//h:img[@src]'):
src = urlnormalize(elem.attrib['src'])
image = hrefs.get(item.abshref(src), None)
if image and image.media_type == SVG_MIME:
style = self.stylizer(item).style(elem)
self.rasterize_external(elem, style, item, image)
for elem in xpath(html, '//h:object[@type="%s" and @data]' % SVG_MIME):
data = urlnormalize(elem.attrib['data'])
image = hrefs.get(item.abshref(data), None)
if image and image.media_type == SVG_MIME:
style = self.stylizer(item).style(elem)
self.rasterize_external(elem, style, item, image)
for elem in xpath(html, '//svg:svg'):
style = self.stylizer(item).style(elem)
self.rasterize_inline(elem, style, item)
def inspect_cover(self, href):
from ebook_converter.ebooks.oeb.base import urlnormalize
for x in self.oeb.manifest:
if x.href == urlnormalize(href):
try:
raw = x.data
return identify(raw)[1:]
except Exception:
self.log.exception('Failed to read cover image dimensions')
return -1, -1
def __call__(self, oeb, context):
has_toc = getattr(getattr(oeb, 'toc', False), 'nodes', False)
if 'toc' in oeb.guide:
# Ensure toc pointed to in <guide> is in spine
from ebook_converter.ebooks.oeb.base import urlnormalize
href = urlnormalize(oeb.guide['toc'].href)
if href in oeb.manifest.hrefs:
item = oeb.manifest.hrefs[href]
if (hasattr(item.data, 'xpath') and
base.XPath('//h:a[@href]')(item.data)):
if oeb.spine.index(item) < 0:
if self.position == 'end':
oeb.spine.add(item, linear=False)
else:
oeb.spine.insert(0, item, linear=True)
return
elif has_toc:
oeb.guide.remove('toc')
else:
oeb.guide.remove('toc')
if not has_toc:
return
oeb.logger.info('Generating in-line TOC...')
title = self.title or oeb.translate(DEFAULT_TITLE)
style = self.style
if style not in STYLE_CSS:
oeb.logger.error('Unknown TOC style %r', style)
style = 'nested'
id, css_href = oeb.manifest.generate('tocstyle', 'tocstyle.css')
oeb.manifest.add(id, css_href, base.CSS_MIME, data=STYLE_CSS[style])
language = str(oeb.metadata.language[0])
contents = base.element(None, base.tag('xhtml', 'html'),
nsmap={None: const.XHTML_NS},
attrib={base.tag('xml', 'lang'): language})
head = base.element(contents, base.tag('xhtml', 'head'))
htitle = base.element(head, base.tag('xhtml', 'title'))
htitle.text = title
base.element(head, base.tag('xhtml', 'link'), rel='stylesheet',
type=base.CSS_MIME, href=css_href)
body = base.element(contents, base.tag('xhtml', 'body'),
attrib={'class': 'calibre_toc'})
h1 = base.element(body, base.tag('xhtml', 'h2'),
attrib={'class': 'calibre_toc_header'})
h1.text = title
self.add_toc_level(body, oeb.toc)
id, href = oeb.manifest.generate('contents', 'contents.xhtml')
item = oeb.manifest.add(id, href, base.XHTML_MIME, data=contents)
if self.position == 'end':
oeb.spine.add(item, linear=False)
else:
oeb.spine.insert(0, item, linear=True)
oeb.guide.add('toc', 'Table of Contents', href)
def fix_toc_entry(self, toc):
if toc.href:
href = urlnormalize(toc.href)
href, frag = urllib.parse.urldefrag(href)
replacement = self.rename_map.get(href, None)
if replacement is not None:
nhref = replacement
if frag:
nhref = '#'.join((nhref, frag))
toc.href = nhref
for x in toc:
self.fix_toc_entry(x)
def remove_images(self, item, limit=1):
path = XPath('//h:img[@src]')
removed = 0
for img in path(item.data):
if removed >= limit:
break
href = item.abshref(img.get('src'))
image = self.oeb.manifest.hrefs.get(href)
if image is None:
href = urlnormalize(href)
image = self.oeb.manifest.hrefs.get(href)
if image is not None:
self.oeb.manifest.remove(image)
self.oeb.guide.remove_by_href(href)
img.getparent().remove(img)
removed += 1
return removed
def _guide_from_opf(self, opf):
guide = self.oeb.guide
manifest = self.oeb.manifest
for elem in base.xpath(opf, '/o2:package/o2:guide/o2:reference'):
ref_href = elem.get('href')
path = base.urlnormalize(urllib.parse.urldefrag(ref_href)[0])
if path not in manifest.hrefs:
corrected_href = None
for href in manifest.hrefs:
if href.lower() == path.lower():
corrected_href = href
break
if corrected_href is None:
self.logger.warn('Guide reference %r not found' % ref_href)
continue
ref_href = corrected_href
typ = elem.get('type')
if typ not in guide:
guide.add(typ, elem.get('title'), ref_href)
def url_replacer(self, orig_url):
url = urlnormalize(orig_url)
parts = urllib.parse.urlparse(url)
if parts.scheme:
# Only rewrite local URLs
return orig_url
path, frag = urllib.parse.urldefrag(url)
if self.renamed_items_map:
orig_item = self.renamed_items_map.get(self.current_item.href,
self.current_item)
else:
orig_item = self.current_item
href = orig_item.abshref(path)
replacement = self.current_item.relhref(self.rename_map.get(
href, href))
if frag:
replacement += '#' + frag
return replacement
def dataize_svg(self, item, svg=None):
if svg is None:
svg = item.data
hrefs = self.oeb.manifest.hrefs
for elem in xpath(svg, '//svg:*[@xl:href]'):
href = urlnormalize(elem.attrib[base.tag('xlink', 'href')])
path = urllib.parse.urldefrag(href)[0]
if not path:
continue
abshref = item.abshref(path)
if abshref not in hrefs:
continue
linkee = hrefs[abshref]
data = linkee.bytes_representation
ext = what(None, data) or 'jpg'
with PersistentTemporaryFile(suffix='.' + ext) as pt:
pt.write(data)
self.temp_files.append(pt.name)
elem.attrib[base.tag('xlink', 'href')] = pt.name
return svg
def _pages_from_page_map(self, opf):
item = self._find_page_map(opf)
if item is None:
return False
pmap = item.data
pages = self.oeb.pages
for page in base.xpath(pmap, 'o2:page'):
name = page.get('name', '')
href = page.get('href')
if not href:
continue
name = base.COLLAPSE_RE.sub(' ', name.strip())
href = item.abshref(base.urlnormalize(href))
type = 'normal'
if not name:
type = 'special'
elif name.lower().strip('ivxlcdm') == '':
type = 'front'
pages.add(name, href, type=type)
return True
def _toc_from_tour(self, opf):
result = base.xpath(opf, 'o2:tours/o2:tour')
if not result:
return False
self.log.debug('Reading TOC from tour...')
tour = result[0]
toc = self.oeb.toc
toc.title = tour.get('title')
sites = base.xpath(tour, 'o2:site')
for site in sites:
title = site.get('title')
href = site.get('href')
if not title or not href:
continue
path, _ = urllib.parse.urldefrag(base.urlnormalize(href))
if path not in self.oeb.manifest.hrefs:
self.logger.warn('TOC reference %r not found' % href)
continue
id = site.get('id')
toc.add(title, href, id=id)
return True
def rewrite_links(self, url):
href, frag = urllib.parse.urldefrag(url)
try:
href = self.current_item.abshref(href)
except ValueError:
# Unparseable URL
return url
try:
href = base.urlnormalize(href)
except ValueError:
# href has non utf-8 quoting
return url
if href in self.map:
anchor_map = self.map[href]
nhref = anchor_map[frag if frag else None]
nhref = self.current_item.relhref(nhref)
if frag:
nhref = '#'.join((polyglot.unquote(nhref), frag))
return nhref
return url
def _pages_from_ncx(self, opf, item):
if item is None:
return False
ncx = item.data
if ncx is None:
return False
ptargets = base.xpath(ncx, 'ncx:pageList/ncx:pageTarget')
if not ptargets:
return False
pages = self.oeb.pages
for ptarget in ptargets:
name = ''.join(base.xpath(ptarget, 'ncx:navLabel/ncx:text/text()'))
name = base.COLLAPSE_RE.sub(' ', name.strip())
href = base.xpath(ptarget, 'ncx:content/@src')
if not href:
continue
href = item.abshref(base.urlnormalize(href[0]))
id = ptarget.get('id')
type = ptarget.get('type', 'normal')
klass = ptarget.get('class')
pages.add(name, href, type=type, id=id, klass=klass)
return True
def _spine_add_extra(self):
manifest = self.oeb.manifest
spine = self.oeb.spine
unchecked = set(spine)
selector = base.XPath('h:body//h:a/@href')
extras = set()
while unchecked:
new = set()
for item in unchecked:
if item.media_type not in base.OEB_DOCS:
# TODO: handle fallback chains
continue
for href in selector(item.data):
href, _ = urllib.parse.urldefrag(href)
if not href:
continue
try:
href = item.abshref(base.urlnormalize(href))
except ValueError: # Malformed URL
continue
if href not in manifest.hrefs:
continue
found = manifest.hrefs[href]
if found.media_type not in base.OEB_DOCS or \
found in spine or found in extras:
continue
new.add(found)
extras.update(new)
unchecked = new
version = int(self.oeb.version[0])
removed_items_to_ignore = getattr(self.oeb, 'removed_items_to_ignore',
())
for item in sorted(extras):
if item.href in removed_items_to_ignore:
continue
if version >= 2:
self.logger.warn('Spine-referenced file %r not in spine' %
item.href)
spine.add(item, linear=False)
def serialize_toc_level(tocref, href=None):
# add the provided toc level to the output stream
# if href is provided add a link ref to the toc level output (e.g. feed_0/index.html)
if href is not None:
# resolve the section url in id_offsets
buf.write(b'<mbp:pagebreak />')
self.id_offsets[base.urlnormalize(href)] = buf.tell()
if tocref.klass == "periodical":
buf.write(b'<div> <div height="1em"></div>')
else:
t = tocref.title
if isinstance(t, str):
t = t.encode('utf-8')
buf.write(
b'<div></div> <div> <h2 height="1em"><font size="+2"><b>' +
t + b'</b></font></h2> <div height="1em"></div>')
buf.write(b'<ul>')
for tocitem in tocref.nodes:
buf.write(b'<li><a filepos=')
itemhref = tocitem.href
if tocref.klass == 'periodical':
# This is a section node.
# For periodical tocs, the section urls are like r'feed_\d+/index.html'
# We dont want to point to the start of the first article
# so we change the href.
itemhref = re.sub(r'article_\d+/', '', itemhref)
self.href_offsets[itemhref].append(buf.tell())
buf.write(b'0000000000')
buf.write(b' ><font size="+1"><b><u>')
t = tocitem.title
if isinstance(t, str):
t = t.encode('utf-8')
buf.write(t)
buf.write(b'</u></b></font></a></li>')
buf.write(b'</ul><div height="1em"></div></div><mbp:pagebreak />')
def serialize_item(self, item):
'''
Serialize an individual item from the spine of the input document.
A reference to this item is stored in self.href_offsets
'''
buf = self.buf
if not item.linear:
self.breaks.append(buf.tell() - 1)
self.id_offsets[base.urlnormalize(item.href)] = buf.tell()
if item.is_section_start:
buf.write(b'<a ></a> ')
if item.is_article_start:
buf.write(b'<a ></a> <a ></a>')
for elem in item.data.find(base.tag('xhtml', 'body')):
self.serialize_elem(elem, item)
if self.write_page_breaks_after_item:
buf.write(b'<mbp:pagebreak/>')
if item.is_article_end:
# Kindle periodical article end marker
buf.write(b'<a ></a> <a ></a>')
if item.is_section_end:
buf.write(b' <a ></a>')
self.anchor_offset = None
def dump_text(self, elem_tree, stylizer, page, tag_stack=[]):
"""
This function is intended to be used in a recursive manner. dump_text
will run though all elements in the elem_tree and call itself on each
element.
self.image_hrefs will be populated by calling this function.
@param elem_tree: etree representation of XHTML content to be
transformed.
@param stylizer: Used to track the style of elements within the tree.
@param page: OEB page used to determine absolute urls.
@param tag_stack: List of open FB2 tags to take into account.
@return: List of string representing the XHTML converted to FB2 markup.
"""
elem = elem_tree
# Ensure what we are converting is not a string and that the fist tag
# is part of the XHTML namespace.
if (not isinstance(elem_tree.tag, (str, bytes))
or parse_utils.namespace(elem_tree.tag) != const.XHTML_NS):
p = elem.getparent()
if (p is not None and isinstance(p.tag, (str, bytes))
and parse_utils.namespace(p.tag) == const.XHTML_NS
and elem.tail):
return [elem.tail]
return []
style = stylizer.style(elem_tree)
if style['display'] in ('none', 'oeb-page-head', 'oeb-page-foot') \
or style['visibility'] == 'hidden':
if hasattr(elem, 'tail') and elem.tail:
return [elem.tail]
return []
# FB2 generated output.
fb2_out = []
# FB2 tags in the order they are opened. This will be used to close
# the tags.
tags = []
# First tag in tree
tag = parse_utils.barename(elem_tree.tag)
# Number of blank lines above tag
try:
ems = int(round((float(style.marginTop) / style.fontSize) - 1))
if ems < 0:
ems = 0
except Exception:
ems = 0
# Convert TOC entries to <title>s and add <section>s
if self.opts.sectionize == 'toc':
# A section cannot be a child of any other element than another
# section, so leave the tag alone if there are parents
if not tag_stack:
# There are two reasons to start a new section here: the TOC
# pointed to this page (then we use the first non-<body> on
# the page as a <title>), or the TOC pointed to a specific
# element
newlevel = 0
toc_entry = self.toc.get(page.href, None)
if toc_entry is not None:
if None in toc_entry:
if (tag != 'body' and hasattr(elem_tree, 'text')
and elem_tree.text):
newlevel = 1
self.toc[page.href] = None
if (not newlevel
and elem_tree.attrib.get('id', None) is not None):
newlevel = toc_entry.get(
elem_tree.attrib.get('id', None), None)
# Start a new section if necessary
if newlevel:
while newlevel <= self.section_level:
fb2_out.append('</section>')
self.section_level -= 1
fb2_out.append('<section>')
self.section_level += 1
fb2_out.append('<title>')
tags.append('title')
if self.section_level == 0:
# If none of the prior processing made a section, make one now
# to be FB2 spec compliant
fb2_out.append('<section>')
self.section_level += 1
# Process the XHTML tag and styles. Converted to an FB2 tag.
# Use individual if statement not if else. There can be only one XHTML
# tag but it can have multiple styles.
if tag == 'img' and elem_tree.attrib.get('src', None):
# Only write the image tag if it is in the manifest.
ihref = base.urlnormalize(page.abshref(elem_tree.attrib['src']))
if ihref in self.oeb_book.manifest.hrefs:
if ihref not in self.image_hrefs:
self.image_hrefs[ihref] = 'img_%s' % len(self.image_hrefs)
p_txt, p_tag = self.ensure_p()
fb2_out += p_txt
tags += p_tag
fb2_out.append('<image l:href="#%s"/>' %
self.image_hrefs[ihref])
else:
self.log.warn(u'Ignoring image not in manifest: %s' % ihref)
if tag in ('br', 'hr') or ems >= 1:
if ems < 1:
multiplier = 1
else:
multiplier = ems
if self.in_p:
closed_tags = []
open_tags = tag_stack + tags
open_tags.reverse()
for t in open_tags:
fb2_out.append('</%s>' % t)
closed_tags.append(t)
if t == 'p':
break
fb2_out.append('<empty-line/>' * multiplier)
closed_tags.reverse()
for t in closed_tags:
fb2_out.append('<%s>' % t)
else:
fb2_out.append('<empty-line/>' * multiplier)
if tag in ('div', 'li', 'p'):
p_text, added_p = self.close_open_p(tag_stack + tags)
fb2_out += p_text
if added_p:
tags.append('p')
if tag == 'a' and elem_tree.attrib.get('href', None):
# Handle only external links for now
if urllib.parse.urlparse(elem_tree.attrib['href']).netloc:
p_txt, p_tag = self.ensure_p()
fb2_out += p_txt
tags += p_tag
fb2_out.append('<a l:href="%s">' %
base.urlnormalize(elem_tree.attrib['href']))
tags.append('a')
if tag == 'b' or style['font-weight'] in ('bold', 'bolder'):
s_out, s_tags = self.handle_simple_tag('strong', tag_stack + tags)
fb2_out += s_out
tags += s_tags
if tag == 'i' or style['font-style'] == 'italic':
s_out, s_tags = self.handle_simple_tag('emphasis',
tag_stack + tags)
fb2_out += s_out
tags += s_tags
if (tag in ('del', 'strike')
or style['text-decoration'] == 'line-through'):
s_out, s_tags = self.handle_simple_tag('strikethrough',
tag_stack + tags)
fb2_out += s_out
tags += s_tags
if tag == 'sub':
s_out, s_tags = self.handle_simple_tag('sub', tag_stack + tags)
fb2_out += s_out
tags += s_tags
if tag == 'sup':
s_out, s_tags = self.handle_simple_tag('sup', tag_stack + tags)
fb2_out += s_out
tags += s_tags
# Process element text.
if hasattr(elem_tree, 'text') and elem_tree.text:
if not self.in_p:
fb2_out.append('<p>')
fb2_out.append(prepare_string_for_xml(elem_tree.text))
if not self.in_p:
fb2_out.append('</p>')
# Process sub-elements.
for item in elem_tree:
fb2_out += self.dump_text(item, stylizer, page, tag_stack + tags)
# Close open FB2 tags.
tags.reverse()
fb2_out += self.close_tags(tags)
# Process element text that comes after the close of the XHTML tag but
# before the next XHTML tag.
if hasattr(elem_tree, 'tail') and elem_tree.tail:
if not self.in_p:
fb2_out.append('<p>')
fb2_out.append(prepare_string_for_xml(elem_tree.tail))
if not self.in_p:
fb2_out.append('</p>')
return fb2_out
def serialize_body(self):
'''
Serialize all items in the spine of the document. Non linear items are
moved to the end.
'''
buf = self.buf
def serialize_toc_level(tocref, href=None):
# add the provided toc level to the output stream
# if href is provided add a link ref to the toc level output (e.g. feed_0/index.html)
if href is not None:
# resolve the section url in id_offsets
buf.write(b'<mbp:pagebreak />')
self.id_offsets[base.urlnormalize(href)] = buf.tell()
if tocref.klass == "periodical":
buf.write(b'<div> <div height="1em"></div>')
else:
t = tocref.title
if isinstance(t, str):
t = t.encode('utf-8')
buf.write(
b'<div></div> <div> <h2 height="1em"><font size="+2"><b>' +
t + b'</b></font></h2> <div height="1em"></div>')
buf.write(b'<ul>')
for tocitem in tocref.nodes:
buf.write(b'<li><a filepos=')
itemhref = tocitem.href
if tocref.klass == 'periodical':
# This is a section node.
# For periodical tocs, the section urls are like r'feed_\d+/index.html'
# We dont want to point to the start of the first article
# so we change the href.
itemhref = re.sub(r'article_\d+/', '', itemhref)
self.href_offsets[itemhref].append(buf.tell())
buf.write(b'0000000000')
buf.write(b' ><font size="+1"><b><u>')
t = tocitem.title
if isinstance(t, str):
t = t.encode('utf-8')
buf.write(t)
buf.write(b'</u></b></font></a></li>')
buf.write(b'</ul><div height="1em"></div></div><mbp:pagebreak />')
self.anchor_offset = buf.tell()
buf.write(b'<body>')
self.body_start_offset = buf.tell()
if self.is_periodical:
top_toc = self.oeb.toc.nodes[0]
serialize_toc_level(top_toc)
spine = [item for item in self.oeb.spine if item.linear]
spine.extend([item for item in self.oeb.spine if not item.linear])
for item in spine:
if self.is_periodical and item.is_section_start:
for section_toc in top_toc.nodes:
if base.urlnormalize(item.href) == section_toc.href:
# create section url of the form r'feed_\d+/index.html'
section_url = re.sub(r'article_\d+/', '',
section_toc.href)
serialize_toc_level(section_toc, section_url)
section_toc.href = section_url
break
self.serialize_item(item)
self.body_end_offset = buf.tell()
buf.write(b'</body>')
def _toc_from_navpoint(self, item, toc, navpoint):
children = base.xpath(navpoint, 'ncx:navPoint')
for child in children:
title = ''.join(base.xpath(child, 'ncx:navLabel/ncx:text/text()'))
title = base.COLLAPSE_RE.sub(' ', title.strip())
href = base.xpath(child, 'ncx:content/@src')
if not title:
self._toc_from_navpoint(item, toc, child)
continue
if (not href
or not href[0]) and not base.xpath(child, 'ncx:navPoint'):
# This node is useless
continue
if href and href[0]:
href = item.abshref(base.urlnormalize(href[0]))
else:
href = ''
path, _ = urllib.parse.urldefrag(href)
if path and path not in self.oeb.manifest.hrefs:
path = base.urlnormalize(path)
if href and path not in self.oeb.manifest.hrefs:
self.logger.warn('TOC reference %r not found' % href)
gc = base.xpath(child, 'ncx:navPoint')
if not gc:
# This node is useless
continue
id = child.get('id')
klass = child.get('class', 'chapter')
try:
po = int(child.get('playOrder',
self.oeb.toc.next_play_order()))
except Exception:
po = self.oeb.toc.next_play_order()
authorElement = base.xpath(
child, 'descendant::calibre:meta[@name = "author"]')
if authorElement:
author = authorElement[0].text
else:
author = None
descriptionElement = base.xpath(
child, 'descendant::calibre:meta[@name = '
'"description"]')
if descriptionElement:
description = etree.tostring(descriptionElement[0],
method='text',
encoding='unicode').strip()
if not description:
description = None
else:
description = None
index_image = base.xpath(
child, 'descendant::calibre:meta[@name = '
'"toc_thumbnail"]')
toc_thumbnail = (index_image[0].text if index_image else None)
if not toc_thumbnail or not toc_thumbnail.strip():
toc_thumbnail = None
node = toc.add(title,
href,
id=id,
klass=klass,
play_order=po,
description=description,
author=author,
toc_thumbnail=toc_thumbnail)
self._toc_from_navpoint(item, node, child)
def create_oebbook(self, htmlpath, basedir, opts, log, mi):
import uuid
from ebook_converter.ebooks.conversion.plumber import create_oebbook
from ebook_converter.ebooks.oeb.base import (DirContainer,
rewrite_links, urlnormalize, BINARY_MIME, OEB_STYLES,
xpath, urlquote)
from ebook_converter.ebooks.oeb.transforms.metadata import \
meta_info_to_oeb_metadata
from ebook_converter.ebooks.html.input import get_filelist
from ebook_converter.ebooks.metadata import string_to_authors
from ebook_converter.utils.localization import canonicalize_lang
import css_parser, logging
css_parser.log.setLevel(logging.WARN)
self.OEB_STYLES = OEB_STYLES
oeb = create_oebbook(log, None, opts, self,
encoding=opts.input_encoding, populate=False)
self.oeb = oeb
metadata = oeb.metadata
meta_info_to_oeb_metadata(mi, metadata, log)
if not metadata.language:
l = canonicalize_lang(getattr(opts, 'language', None))
if not l:
oeb.logger.warn('Language not specified')
l = get_lang().replace('_', '-')
metadata.add('language', l)
if not metadata.creator:
a = getattr(opts, 'authors', None)
if a:
a = string_to_authors(a)
if not a:
oeb.logger.warn('Creator not specified')
a = [self.oeb.translate('Unknown')]
for aut in a:
metadata.add('creator', aut)
if not metadata.title:
oeb.logger.warn('Title not specified')
metadata.add('title', self.oeb.translate('Unknown'))
bookid = str(uuid.uuid4())
metadata.add('identifier', bookid, id='uuid_id', scheme='uuid')
for ident in metadata.identifier:
if 'id' in ident.attrib:
self.oeb.uid = metadata.identifier[0]
break
filelist = get_filelist(htmlpath, basedir, opts, log)
filelist = [f for f in filelist if not f.is_binary]
htmlfile_map = {}
for f in filelist:
path = f.path
oeb.container = DirContainer(os.path.dirname(path), log,
ignore_opf=True)
bname = os.path.basename(path)
id, href = oeb.manifest.generate(id='html', href=sanitize_file_name(bname))
htmlfile_map[path] = href
item = oeb.manifest.add(id, href, 'text/html')
if path == htmlpath and '%' in path:
bname = urlquote(bname)
item.html_input_href = bname
oeb.spine.add(item, True)
self.added_resources = {}
self.log = log
self.log('Normalizing filename cases')
for path, href in htmlfile_map.items():
self.added_resources[path] = href
self.urlnormalize, self.DirContainer = urlnormalize, DirContainer
self.urldefrag = urllib.parse.urldefrag
self.BINARY_MIME = BINARY_MIME
self.log('Rewriting HTML links')
for f in filelist:
path = f.path
dpath = os.path.dirname(path)
oeb.container = DirContainer(dpath, log, ignore_opf=True)
href = htmlfile_map[path]
try:
item = oeb.manifest.hrefs[href]
except KeyError:
item = oeb.manifest.hrefs[urlnormalize(href)]
rewrite_links(item.data,
functools.partial(self.resource_adder, base=dpath))
for item in oeb.manifest.values():
if item.media_type in self.OEB_STYLES:
dpath = None
for path, href in self.added_resources.items():
if href == item.href:
dpath = os.path.dirname(path)
break
css_parser.replaceUrls(item.data,
functools.partial(self.resource_adder, base=dpath))
toc = self.oeb.toc
self.oeb.auto_generated_toc = True
titles = []
headers = []
for item in self.oeb.spine:
if not item.linear:
continue
html = item.data
title = ''.join(xpath(html, '/h:html/h:head/h:title/text()'))
title = re.sub(r'\s+', ' ', title.strip())
if title:
titles.append(title)
headers.append('(unlabled)')
for tag in ('h1', 'h2', 'h3', 'h4', 'h5', 'strong'):
expr = '/h:html/h:body//h:%s[position()=1]/text()'
header = ''.join(xpath(html, expr % tag))
header = re.sub(r'\s+', ' ', header.strip())
if header:
headers[-1] = header
break
use = titles
if len(titles) > len(set(titles)):
use = headers
for title, item in zip(use, self.oeb.spine):
if not item.linear:
continue
toc.add(title, item.href)
oeb.container = DirContainer(os.getcwd(), oeb.log, ignore_opf=True)
return oeb
def convert_epub3_nav(self, nav_path, opf, log, opts):
from lxml import etree
from ebook_converter.ebooks.chardet import xml_to_unicode
from ebook_converter.ebooks.oeb.polish.parsing import parse
from ebook_converter.ebooks.oeb.base import \
serialize
from ebook_converter.ebooks.oeb.polish.toc import first_child
from tempfile import NamedTemporaryFile
with open(nav_path, 'rb') as f:
raw = f.read()
raw = xml_to_unicode(raw, strip_encoding_pats=True,
assume_utf8=True)[0]
root = parse(raw, log=log)
ncx = etree.fromstring('<ncx xmlns="http://www.daisy.org/z3986/2005/'
'ncx/" version="2005-1" xml:lang="eng">'
'<navMap/></ncx>')
navmap = ncx[0]
et = '{%s}type' % const.EPUB_NS
bn = os.path.basename(nav_path)
def add_from_li(li, parent):
href = text = None
for x in li.iterchildren(base.tag('xhtml', 'a'),
base.tag('xhtml', 'span')):
text = etree.tostring(
x, method='text', encoding='unicode',
with_tail=False).strip() or ' '.join(
x.xpath('descendant-or-self::*/@title')).strip()
href = x.get('href')
if href:
if href.startswith('#'):
href = bn + href
break
np = parent.makeelement(base.tag('ncx', 'navPoint'))
parent.append(np)
np.append(np.makeelement(base.tag('ncx', 'navLabel')))
np[0].append(np.makeelement(base.tag('ncx', 'text')))
np[0][0].text = text
if href:
np.append(
np.makeelement(base.tag('ncx', 'content'),
attrib={'src': href}))
return np
def process_nav_node(node, toc_parent):
for li in node.iterchildren(base.tag('xhtml', 'li')):
child = add_from_li(li, toc_parent)
ol = first_child(li, base.tag('xhtml', 'ol'))
if child is not None and ol is not None:
process_nav_node(ol, child)
for nav in root.iterdescendants(base.tag('xhtml', 'nav')):
if nav.get(et) == 'toc':
ol = first_child(nav, base.tag('xhtml', 'ol'))
if ol is not None:
process_nav_node(ol, navmap)
break
else:
return
with NamedTemporaryFile(suffix='.ncx',
dir=os.path.dirname(nav_path),
delete=False) as f:
f.write(etree.tostring(ncx, encoding='utf-8'))
ncx_href = os.path.relpath(f.name, os.getcwd()).replace(os.sep, '/')
ncx_id = opf.create_manifest_item(ncx_href, base.NCX_MIME,
append=True).get('id')
for spine in opf.root.xpath('//*[local-name()="spine"]'):
spine.set('toc', ncx_id)
url = os.path.relpath(nav_path).replace(os.sep, '/')
opts.epub3_nav_href = base.urlnormalize(url)
opts.epub3_nav_parsed = root
if getattr(self, 'removed_cover', None):
changed = False
base_path = os.path.dirname(nav_path)
for elem in root.xpath('//*[@href]'):
href, frag = elem.get('href').partition('#')[::2]
link_path = (os.path.relpath(
os.path.join(base_path, urllib.parse.unquote(href)),
base_path))
abs_href = base.urlnormalize(link_path)
if abs_href == self.removed_cover:
changed = True
elem.set('data-calibre-removed-titlepage', '1')
if changed:
with open(nav_path, 'wb') as f:
f.write(base.serialize(root, 'application/xhtml+xml'))
def mobimlize_elem(self, elem, stylizer, bstate, istates,
ignore_valign=False):
if not isinstance(elem.tag, (str, bytes)) \
or parse_utils.namespace(elem.tag) != const.XHTML_NS:
return
style = stylizer.style(elem)
# <mbp:frame-set/> does not exist lalalala
if ((style['display'] in ('none', 'oeb-page-head', 'oeb-page-foot') or style['visibility'] == 'hidden') and
elem.get('data-calibre-jacket-searchable-tags', None) != '1'):
id_ = elem.get('id', None)
if id_:
# Keep anchors so people can use display:none
# to generate hidden TOCs
tail = elem.tail
elem.clear()
elem.text = None
elem.set('id', id_)
elem.tail = tail
elem.tag = base.tag('xhtml', 'a')
else:
return
tag = parse_utils.barename(elem.tag)
istate = copy.copy(istates[-1])
istate.rendered = False
istate.list_num = 0
if tag == 'ol' and 'start' in elem.attrib:
try:
istate.list_num = int(elem.attrib['start'])-1
except:
pass
istates.append(istate)
left = 0
display = style['display']
if display == 'table-cell':
display = 'inline'
elif display.startswith('table'):
display = 'block'
isblock = (not display.startswith('inline') and style['display'] !=
'none')
isblock = isblock and style['float'] == 'none'
isblock = isblock and tag != 'br'
if isblock:
bstate.para = None
istate.halign = style['text-align']
rawti = style._get('text-indent')
istate.indent = style['text-indent']
if hasattr(rawti, 'strip') and '%' in rawti:
# We have a percentage text indent, these can come out looking
# too large if the user chooses a wide output profile like
# tablet
istate.indent = min(style._unit_convert(rawti, base=500), istate.indent)
if style['margin-left'] == 'auto' \
and style['margin-right'] == 'auto':
istate.halign = 'center'
margin = asfloat(style['margin-left'])
padding = asfloat(style['padding-left'])
if tag != 'body':
left = margin + padding
istate.left += left
vmargin = asfloat(style['margin-top'])
bstate.vmargin = max((bstate.vmargin, vmargin))
vpadding = asfloat(style['padding-top'])
if vpadding > 0:
bstate.vpadding += bstate.vmargin
bstate.vmargin = 0
bstate.vpadding += vpadding
elif not istate.href:
margin = asfloat(style['margin-left'])
padding = asfloat(style['padding-left'])
lspace = margin + padding
if lspace > 0:
spaces = int(round((lspace * 3) / style['font-size']))
elem.text = ('\xa0' * spaces) + (elem.text or '')
margin = asfloat(style['margin-right'])
padding = asfloat(style['padding-right'])
rspace = margin + padding
if rspace > 0:
spaces = int(round((rspace * 3) / style['font-size']))
if len(elem) == 0:
elem.text = (elem.text or '') + ('\xa0' * spaces)
else:
last = elem[-1]
last.text = (last.text or '') + ('\xa0' * spaces)
if bstate.content and style['page-break-before'] in PAGE_BREAKS:
bstate.pbreak = True
istate.fsize = self.mobimlize_font(style['font-size'])
istate.italic = True if style['font-style'] == 'italic' else False
weight = style['font-weight']
istate.bold = weight in ('bold', 'bolder') or asfloat(weight) > 400
istate.preserve = style['white-space'] == 'pre'
istate.pre_wrap = style['white-space'] == 'pre-wrap'
istate.bgcolor = style['background-color']
istate.fgcolor = style['color']
istate.strikethrough = style.effective_text_decoration == 'line-through'
istate.underline = style.effective_text_decoration == 'underline'
ff = style['font-family'].lower() if hasattr(style['font-family'], 'lower') else ''
if 'monospace' in ff or 'courier' in ff or ff.endswith(' mono'):
istate.family = 'monospace'
elif ('sans-serif' in ff or 'sansserif' in ff or 'verdana' in ff or
'arial' in ff or 'helvetica' in ff):
istate.family = 'sans-serif'
else:
istate.family = 'serif'
if 'id' in elem.attrib:
istate.ids.add(elem.attrib['id'])
if 'name' in elem.attrib:
istate.ids.add(elem.attrib['name'])
if tag == 'a' and 'href' in elem.attrib:
istate.href = elem.attrib['href']
istate.attrib.clear()
if tag == 'img' and 'src' in elem.attrib:
istate.attrib['src'] = elem.attrib['src']
istate.attrib['align'] = 'baseline'
cssdict = style.cssdict()
valign = cssdict.get('vertical-align', None)
if valign in ('top', 'bottom', 'middle'):
istate.attrib['align'] = valign
for prop in ('width', 'height'):
if cssdict[prop] != 'auto':
value = style[prop]
if value == getattr(self.profile, prop):
result = '100%'
else:
# Amazon's renderer does not support
# img sizes in units other than px
# See #7520 for test case
try:
pixs = int(round(float(value) /
(72/self.profile.dpi)))
except:
continue
result = str(pixs)
istate.attrib[prop] = result
if 'width' not in istate.attrib or 'height' not in istate.attrib:
href = self.current_spine_item.abshref(elem.attrib['src'])
try:
item = self.oeb.manifest.hrefs[base.urlnormalize(href)]
except:
self.oeb.logger.warn('Failed to find image:',
href)
else:
try:
width, height = identify(item.data)[1:]
except Exception:
self.oeb.logger.warn('Invalid image:', href)
else:
if 'width' not in istate.attrib and 'height' not in \
istate.attrib:
istate.attrib['width'] = str(width)
istate.attrib['height'] = str(height)
else:
ar = width / height
if 'width' not in istate.attrib:
try:
width = int(istate.attrib['height'])*ar
except:
pass
istate.attrib['width'] = str(int(width))
else:
try:
height = int(istate.attrib['width'])/ar
except:
pass
istate.attrib['height'] = str(int(height))
item.unload_data_from_memory()
elif tag == 'hr' and asfloat(style['width']) > 0 and style._get('width') not in {'100%', 'auto'}:
raww = style._get('width')
if hasattr(raww, 'strip') and '%' in raww:
istate.attrib['width'] = raww
else:
prop = style['width'] / self.profile.width
istate.attrib['width'] = "%d%%" % int(round(prop * 100))
elif display == 'table':
tag = 'table'
elif display == 'table-row':
tag = 'tr'
elif display == 'table-cell':
tag = 'td'
if tag in TABLE_TAGS and self.ignore_tables:
tag = 'span' if tag == 'td' else 'div'
if tag in ('table', 'td', 'tr'):
col = style.backgroundColor
if col:
elem.set('bgcolor', col)
css = style.cssdict()
if 'border' in css or 'border-width' in css:
elem.set('border', '1')
if tag in TABLE_TAGS:
for attr in ('rowspan', 'colspan', 'width', 'border', 'scope',
'bgcolor'):
if attr in elem.attrib:
istate.attrib[attr] = elem.attrib[attr]
if tag == 'q':
t = elem.text
if not t:
t = ''
elem.text = '\u201c' + t
t = elem.tail
if not t:
t = ''
elem.tail = '\u201d' + t
text = None
if elem.text:
if istate.preserve or istate.pre_wrap:
text = elem.text
elif (len(elem) > 0 and isspace(elem.text) and hasattr(elem[0].tag, 'rpartition') and
elem[0].tag.rpartition('}')[-1] not in INLINE_TAGS):
text = None
else:
text = COLLAPSE.sub(' ', elem.text)
valign = style['vertical-align']
not_baseline = valign in ('super', 'sub', 'text-top',
'text-bottom', 'top', 'bottom') or (
isinstance(valign, numbers.Number) and abs(valign) != 0)
issup = valign in ('super', 'text-top', 'top') or (
isinstance(valign, numbers.Number) and valign > 0)
vtag = 'sup' if issup else 'sub'
if not_baseline and not ignore_valign and tag not in NOT_VTAGS and not isblock:
nroot = etree.Element(base.tag('xhtml', 'html'), nsmap=MOBI_NSMAP)
vbstate = BlockState(etree.SubElement(nroot, base.tag('xhtml', 'body')))
vbstate.para = etree.SubElement(vbstate.body, base.tag('xhtml', 'p'))
self.mobimlize_elem(elem, stylizer, vbstate, istates,
ignore_valign=True)
if len(istates) > 0:
istates.pop()
if len(istates) == 0:
istates.append(FormatState())
at_start = bstate.para is None
if at_start:
self.mobimlize_content('span', '', bstate, istates)
parent = bstate.para if bstate.inline is None else bstate.inline
if parent is not None:
vtag = etree.SubElement(parent, base.tag('xhtml', vtag))
vtag = etree.SubElement(vtag, base.tag('xhtml', 'small'))
# Add anchors
for child in vbstate.body:
if child is not vbstate.para:
vtag.append(child)
else:
break
if vbstate.para is not None:
if vbstate.para.text:
vtag.text = vbstate.para.text
for child in vbstate.para:
vtag.append(child)
return
if tag == 'blockquote':
old_mim = self.opts.mobi_ignore_margins
self.opts.mobi_ignore_margins = False
if (text or tag in CONTENT_TAGS or tag in NESTABLE_TAGS or (
# We have an id but no text and no children, the id should still
# be added.
istate.ids and tag in ('a', 'span', 'i', 'b', 'u') and
len(elem)==0)):
if tag == 'li' and len(istates) > 1 and 'value' in elem.attrib:
try:
value = int(elem.attrib['value'])
istates[-2].list_num = value - 1
except:
pass
self.mobimlize_content(tag, text, bstate, istates)
for child in elem:
self.mobimlize_elem(child, stylizer, bstate, istates)
tail = None
if child.tail:
if istate.preserve or istate.pre_wrap:
tail = child.tail
elif bstate.para is None and isspace(child.tail):
tail = None
else:
tail = COLLAPSE.sub(' ', child.tail)
if tail:
self.mobimlize_content(tag, tail, bstate, istates)
if tag == 'blockquote':
self.opts.mobi_ignore_margins = old_mim
if bstate.content and style['page-break-after'] in PAGE_BREAKS:
bstate.pbreak = True
if isblock:
para = bstate.para
if para is not None and para.text == '\xa0' and len(para) < 1:
if style.height > 2:
para.getparent().replace(para, etree.Element(base.tag('xhtml', 'br')))
else:
# This is too small to be rendered effectively, drop it
para.getparent().remove(para)
bstate.para = None
bstate.istate = None
vmargin = asfloat(style['margin-bottom'])
bstate.vmargin = max((bstate.vmargin, vmargin))
vpadding = asfloat(style['padding-bottom'])
if vpadding > 0:
bstate.vpadding += bstate.vmargin
bstate.vmargin = 0
bstate.vpadding += vpadding
if bstate.nested and bstate.nested[-1].tag == elem.tag:
bstate.nested.pop()
istates.pop()
def __init__(self, tree, path, oeb, opts, profile=None,
extra_css='', user_css='', base_css=''):
self.oeb, self.opts = oeb, opts
self.profile = profile
if self.profile is None:
# Use the default profile. This should really be using
# opts.output_profile, but I don't want to risk changing it, as
# doing so might well have hard to debug font size effects.
from ebook_converter.customize.ui import output_profiles
for x in output_profiles():
if x.short_name == 'default':
self.profile = x
break
if self.profile is None:
# Just in case the default profile is removed in the future :)
self.profile = opts.output_profile
self.body_font_size = self.profile.fbase
self.logger = oeb.logger
item = oeb.manifest.hrefs[path]
basename = os.path.basename(path)
cssname = os.path.splitext(basename)[0] + '.css'
stylesheets = [html_css_stylesheet()]
if base_css:
stylesheets.append(parseString(base_css, validate=False))
style_tags = base.xpath(tree, '//*[local-name()="style" or local-name()="link"]')
# Add css_parser parsing profiles from output_profile
for profile in self.opts.output_profile.extra_css_modules:
cssprofiles.addProfile(profile['name'],
profile['props'],
profile['macros'])
parser = CSSParser(fetcher=self._fetch_css_file,
log=logging.getLogger('calibre.css'))
for elem in style_tags:
if (elem.tag == base.tag('xhtml', 'style') and elem.get('type', base.CSS_MIME) in base.OEB_STYLES and media_ok(elem.get('media'))):
text = elem.text if elem.text else ''
for x in elem:
t = getattr(x, 'text', None)
if t:
text += '\n\n' + uenc.force_unicode(t, 'utf-8')
t = getattr(x, 'tail', None)
if t:
text += '\n\n' + uenc.force_unicode(t, 'utf-8')
if text:
text = oeb.css_preprocessor(text)
# We handle @import rules separately
parser.setFetcher(lambda x: ('utf-8', b''))
stylesheet = parser.parseString(text, href=cssname,
validate=False)
parser.setFetcher(self._fetch_css_file)
for rule in stylesheet.cssRules:
if rule.type == rule.IMPORT_RULE:
ihref = item.abshref(rule.href)
if not media_ok(rule.media.mediaText):
continue
hrefs = self.oeb.manifest.hrefs
if ihref not in hrefs:
self.logger.warning('Ignoring missing '
'stylesheet in @import '
'rule: %s', rule.href)
continue
sitem = hrefs[ihref]
if sitem.media_type not in base.OEB_STYLES:
self.logger.warning('CSS @import of non-CSS '
'file %r', rule.href)
continue
stylesheets.append(sitem.data)
# Make links to resources absolute, since these rules will
# be folded into a stylesheet at the root
replaceUrls(stylesheet, item.abshref,
ignoreImportRules=True)
stylesheets.append(stylesheet)
elif (elem.tag == base.tag('xhtml', 'link') and elem.get('href') and elem.get(
'rel', 'stylesheet').lower() == 'stylesheet' and elem.get(
'type', base.CSS_MIME).lower() in base.OEB_STYLES and media_ok(elem.get('media'))
):
href = base.urlnormalize(elem.attrib['href'])
path = item.abshref(href)
sitem = oeb.manifest.hrefs.get(path, None)
if sitem is None:
self.logger.warning('Stylesheet %r referenced by file %r '
'not in manifest', path, item.href)
continue
if not hasattr(sitem.data, 'cssRules'):
self.logger.warning('Stylesheet %r referenced by file %r '
'is not CSS', path, item.href)
continue
stylesheets.append(sitem.data)
csses = {'extra_css':extra_css, 'user_css':user_css}
for w, x in csses.items():
if x:
try:
text = x
stylesheet = parser.parseString(text, href=cssname,
validate=False)
stylesheets.append(stylesheet)
except Exception:
self.logger.exception('Failed to parse %s, ignoring.', w)
self.logger.debug('Bad css: %s', x)
# using oeb to store the rules, page rule and font face rules
# and generating them again if opts, profile or stylesheets are different
if (not hasattr(self.oeb, 'stylizer_rules')) \
or not self.oeb.stylizer_rules.same_rules(self.opts, self.profile, stylesheets):
self.oeb.stylizer_rules = StylizerRules(self.opts, self.profile, stylesheets)
self.rules = self.oeb.stylizer_rules.rules
self.page_rule = self.oeb.stylizer_rules.page_rule
self.font_face_rules = self.oeb.stylizer_rules.font_face_rules
self.flatten_style = self.oeb.stylizer_rules.flatten_style
self._styles = {}
pseudo_pat = re.compile(':{1,2}(%s)' % ('|'.join(INAPPROPRIATE_PSEUDO_CLASSES)), re.I)
select = Select(tree, ignore_inappropriate_pseudo_classes=True)
for _, _, cssdict, text, _ in self.rules:
fl = pseudo_pat.search(text)
try:
matches = tuple(select(text))
except SelectorError as err:
self.logger.error('Ignoring CSS rule with invalid selector: '
'%r (%s)', text, err)
continue
if fl is not None:
fl = fl.group(1)
if fl == 'first-letter' and getattr(self.oeb,
'plumber_output_format', '').lower() in {'mobi', 'docx'}:
# Fake first-letter
for elem in matches:
for x in elem.iter('*'):
if x.text:
punctuation_chars = []
text = str(x.text)
while text:
category = unicodedata.category(text[0])
if category[0] not in {'P', 'Z'}:
break
punctuation_chars.append(text[0])
text = text[1:]
special_text = ''.join(punctuation_chars) + \
(text[0] if text else '')
span = x.makeelement('{%s}span' %
const.XHTML_NS)
span.text = special_text
span.set('data-fake-first-letter', '1')
span.tail = text[1:]
x.text = None
x.insert(0, span)
self.style(span)._update_cssdict(cssdict)
break
else: # Element pseudo-class
for elem in matches:
self.style(elem)._update_pseudo_class(fl, cssdict)
else:
for elem in matches:
self.style(elem)._update_cssdict(cssdict)
for elem in base.xpath(tree, '//h:*[@style]'):
self.style(elem)._apply_style_attr(url_replacer=item.abshref)
num_pat = re.compile(r'[0-9.]+$')
for elem in base.xpath(tree, '//h:img[@width or @height]'):
style = self.style(elem)
# Check if either height or width is not default
is_styled = style._style.get('width', 'auto') != 'auto' or \
style._style.get('height', 'auto') != 'auto'
if not is_styled:
# Update img style dimension using width and height
upd = {}
for prop in ('width', 'height'):
val = elem.get(prop, '').strip()
try:
del elem.attrib[prop]
except:
pass
if val:
if num_pat.match(val) is not None:
val += 'px'
upd[prop] = val
if upd:
style._update_cssdict(upd)