def _clean_opf(self, opf):
nsmap = {}
for elem in opf.iter(tag=etree.Element):
nsmap.update(elem.nsmap)
for elem in opf.iter(tag=etree.Element):
if namespace(elem.tag) in ('', OPF1_NS) and ':' not in barename(elem.tag):
elem.tag = OPF(barename(elem.tag))
nsmap.update(OPF2_NSMAP)
attrib = dict(opf.attrib)
nroot = etree.Element(OPF('package'),
nsmap={None: OPF2_NS}, attrib=attrib)
metadata = etree.SubElement(nroot, OPF('metadata'), nsmap=nsmap)
ignored = (OPF('dc-metadata'), OPF('x-metadata'))
for elem in xpath(opf, 'o2:metadata//*'):
if elem.tag in ignored:
continue
if namespace(elem.tag) in DC_NSES:
tag = barename(elem.tag).lower()
elem.tag = '{%s}%s' % (DC11_NS, tag)
if elem.tag.startswith('dc:'):
tag = elem.tag.partition(':')[-1].lower()
elem.tag = '{%s}%s' % (DC11_NS, tag)
metadata.append(elem)
for element in xpath(opf, 'o2:metadata//o2:meta'):
metadata.append(element)
for tag in ('o2:manifest', 'o2:spine', 'o2:tours', 'o2:guide'):
for element in xpath(opf, tag):
nroot.append(element)
return nroot
def dump_text(self, elem, stylizer, page):
'''
@elem: The element in the etree that we are working on.
@stylizer: The style information attached to the element.
'''
# We can only processes tags. If there isn't a tag return any text.
if not isinstance(elem.tag, string_or_bytes) \
or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, string_or_bytes) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return ['']
# Setup our variables.
text = ['']
tags = []
tag = barename(elem.tag)
attribs = elem.attrib
if tag == 'body':
tag = 'div'
tags.append(tag)
# Remove attributes we won't want.
if 'style' in attribs:
del attribs['style']
# Turn the rest of the attributes into a string we can write with the tag.
at = ''
for k, v in attribs.items():
at += ' %s="%s"' % (k, prepare_string_for_xml(v, attribute=True))
# Write the tag.
text.append('<%s%s' % (tag, at))
if tag in SELF_CLOSING_TAGS:
text.append(' />')
else:
text.append('>')
# Process tags that contain text.
if hasattr(elem, 'text') and elem.text:
text.append(self.prepare_string_for_html(elem.text))
# Recurse down into tags within the tag we are in.
for item in elem:
text += self.dump_text(item, stylizer, page)
# Close all open tags.
tags.reverse()
for t in tags:
if t not in SELF_CLOSING_TAGS:
text.append('</%s>' % t)
# Add the text that is outside of the tag.
if hasattr(elem, 'tail') and elem.tail:
text.append(self.prepare_string_for_html(elem.tail))
return text
def serialize_elem(self, elem, item, nsrmap=NSRMAP):
buf = self.buf
if not isinstance(elem.tag, basestring) \
or namespace(elem.tag) not in nsrmap:
return
tag = 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 = 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 namespace(attr) not in nsrmap:
continue
attr = prefixname(attr, nsrmap)
buf.write(b' ')
if attr == 'href':
if self.serialize_href(val, item):
continue
elif attr == 'src':
href = 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(b'</%s>' % tag.encode('utf-8'))
def _read_opf(self):
data = self.oeb.container.read(None)
data = self.oeb.decode(data)
data = XMLDECL_RE.sub('', data)
data = re.sub(r'http://openebook.org/namespaces/oeb-package/1.0(/*)',
OPF1_NS, data)
try:
opf = etree.fromstring(data)
except etree.XMLSyntaxError:
data = xml_replace_entities(clean_xml_chars(data), encoding=None)
try:
opf = etree.fromstring(data)
self.logger.warn('OPF contains invalid HTML named entities')
except etree.XMLSyntaxError:
data = re.sub(r'(?is)<tours>.+</tours>', '', data)
data = data.replace('<dc-metadata>',
'<dc-metadata xmlns:dc="http://purl.org/metadata/dublin_core">')
try:
opf = etree.fromstring(data)
self.logger.warn('OPF contains invalid tours section')
except etree.XMLSyntaxError:
from calibre.ebooks.oeb.parse_utils import RECOVER_PARSER
opf = etree.fromstring(data, parser=RECOVER_PARSER)
self.logger.warn('OPF contains invalid markup, trying to parse it anyway')
ns = namespace(opf.tag)
if ns not in ('', OPF1_NS, OPF2_NS):
raise OEBError('Invalid namespace %r for OPF document' % ns)
opf = self._clean_opf(opf)
return opf
def mangle_elem(self, elem, stylizer):
if not isinstance(elem.tag, basestring) or \
namespace(elem.tag) != XHTML_NS:
return
children = list(elem)
style = stylizer.style(elem)
transform = style['text-transform']
variant = style['font-variant']
if elem.text:
if transform in TEXT_TRANSFORMS:
elem.text = self.text_transform(transform, elem.text)
if variant == 'small-caps':
self.smallcaps_elem(elem, 'text')
for child in children:
self.mangle_elem(child, stylizer)
if child.tail:
if transform in TEXT_TRANSFORMS:
child.tail = self.text_transform(transform, child.tail)
if variant == 'small-caps':
self.smallcaps_elem(child, 'tail')
def dump_text(self, subitems, elem, stylizer, end='', pre=False, li=''):
from calibre.ebooks.oeb.base import XHTML_NS, barename, namespace
if not isinstance(elem.tag, basestring) \
or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, basestring) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return ['']
text = ['']
style = stylizer.style(elem)
if elem.attrib.get('id') is not None and elem.attrib['id'] in [href for href, title in subitems]:
if self.curSubItem is not None and self.curSubItem != elem.attrib['id']:
self.curSubItem = elem.attrib['id']
text.append(u'\n\n%s%s\n\n' % (CALIBRE_SNB_BM_TAG, self.curSubItem))
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 ['']
tag = barename(elem.tag)
in_block = False
# Are we in a paragraph block?
if tag in BLOCK_TAGS or style['display'] in BLOCK_STYLES:
in_block = True
if not end.endswith(u'\n\n') and hasattr(elem, 'text') and elem.text:
text.append(u'\n\n')
if tag in SPACE_TAGS:
if not end.endswith('u ') and hasattr(elem, 'text') and elem.text:
text.append(u' ')
if tag == 'img':
text.append(u'\n\n%s%s\n\n' % (CALIBRE_SNB_IMG_TAG, ProcessFileName(elem.attrib['src'])))
if tag == 'br':
text.append(u'\n\n')
if tag == 'li':
li = '- '
pre = (tag == 'pre' or pre)
# Process tags that contain text.
if hasattr(elem, 'text') and elem.text:
if pre:
text.append((u'\n\n%s' % CALIBRE_SNB_PRE_TAG).join((li + elem.text).splitlines()))
else:
text.append(li + elem.text)
li = ''
for item in elem:
en = u''
if len(text) >= 2:
en = text[-1][-2:]
t = self.dump_text(subitems, item, stylizer, en, pre, li)[0]
text += t
if in_block:
text.append(u'\n\n')
if hasattr(elem, 'tail') and elem.tail:
if pre:
text.append((u'\n\n%s' % CALIBRE_SNB_PRE_TAG).join(elem.tail.splitlines()))
else:
text.append(li + elem.tail)
li = ''
return text, li
def dump_text(self, elem, stylizer):
'''
@elem: The element in the etree that we are working on.
@stylizer: The style information attached to the element.
'''
# We can only processes tags. If there isn't a tag return any text.
if not isinstance(elem.tag, basestring) \
or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, basestring) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return ['']
# Setup our variables.
text = []
style = stylizer.style(elem)
tags = []
tag = barename(elem.tag)
attribs = elem.attrib
# Ignore anything that is set to not be displayed.
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 ['']
# Soft scene breaks.
if 'margin-top' in style.cssdict() and style['margin-top'] != 'auto':
ems = int(round(float(style.marginTop) / style.fontSize) - 1)
if ems >= 1:
text.append(u'\n\n' * ems)
bq = '> ' * self.blockquotes
# Block level elements
if tag in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'p', 'div'):
h_tag = ''
if tag in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6'):
h_tag = '#' * int(tag[1]) + ' '
text.append('\n' + bq + h_tag)
tags.append('\n')
self.remove_space_after_newline = True
if style['font-style'] == 'italic' or tag in ('i', 'em'):
if tag not in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'cite'):
if self.style_italic == False: # noqa
text.append('*')
tags.append('*')
self.style_italic = True
if style['font-weight'] in ('bold', 'bolder') or tag in ('b',
'strong'):
if tag not in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'th'):
if self.style_bold == False: # noqa
text.append('**')
tags.append('**')
self.style_bold = True
if tag == 'br':
text.append(' \n')
self.remove_space_after_newline = True
if tag == 'blockquote':
self.blockquotes += 1
tags.append('>')
text.append('> ' * self.blockquotes)
elif tag == 'code':
if not self.in_pre and not self.in_code:
text.append('`')
tags.append('`')
self.in_code = True
elif tag == 'pre':
if not self.in_pre:
text.append('\n')
tags.append('pre')
self.in_pre = True
elif tag == 'hr':
text.append('\n* * *')
tags.append('\n')
elif tag == 'a':
# Only write links with absolute (external) urls.
if self.opts.keep_links and attribs.has_key(
'href') and '://' in attribs['href']: # noqa
title = ''
if attribs.has_key('title'): # noqa
title = ' "' + attribs['title'] + '"'
remove_space = self.remove_space_after_newline
title = self.remove_newlines(title)
self.remove_space_after_newline = remove_space
text.append('[')
tags.append('](' + attribs['href'] + title + ')')
elif tag == 'img':
if self.opts.keep_image_references:
txt = '!'
if attribs.has_key('alt'): # noqa
remove_space = self.remove_space_after_newline
txt += '[' + self.remove_newlines(attribs['alt']) + ']'
self.remove_space_after_newline = remove_space
txt += '(' + attribs['src'] + ')'
text.append(txt)
elif tag in ('ol', 'ul'):
tags.append(tag)
# Add the list to our lists of lists so we can track
# nested lists.
self.list.append({'name': tag, 'num': 0})
elif tag == 'li':
# Get the last list from our list of lists
if self.list:
li = self.list[-1]
else:
li = {'name': 'ul', 'num': 0}
# Add a new line to start the item
text.append('\n')
# Add indent if we have nested lists.
list_count = len(self.list)
# We only care about indenting nested lists.
if (list_count - 1) > 0:
text.append('\t' * (list_count - 1))
# Add blockquote if we have a blockquote in a list item.
text.append(bq)
# Write the proper sign for ordered and unorded lists.
if li['name'] == 'ul':
text.append('+ ')
elif li['name'] == 'ol':
li['num'] += 1
text.append(unicode(li['num']) + '. ')
# Process tags that contain text.
if hasattr(elem, 'text') and elem.text:
txt = elem.text
if self.in_pre:
txt = self.prepare_string_for_pre(txt)
elif self.in_code:
txt = self.remove_newlines(txt)
else:
txt = self.prepare_string_for_markdown(
self.remove_newlines(txt))
text.append(txt)
# Recurse down into tags within the tag we are in.
for item in elem:
text += self.dump_text(item, stylizer)
# Close all open tags.
tags.reverse()
for t in tags:
if t in ('pre', 'ul', 'ol', '>'):
if t == 'pre':
self.in_pre = False
text.append('\n')
elif t == '>':
self.blockquotes -= 1
elif t in ('ul', 'ol'):
if self.list:
self.list.pop()
text.append('\n')
else:
if t == '**':
self.style_bold = False
elif t == '*':
self.style_italic = False
elif t == '`':
self.in_code = False
text.append('%s' % t)
# Soft scene breaks.
if 'margin-bottom' in style.cssdict(
) and style['margin-bottom'] != 'auto':
ems = int(round((float(style.marginBottom) / style.fontSize) - 1))
if ems >= 1:
text.append(u'\n\n' * ems)
# Add the text that is outside of the tag.
if hasattr(elem, 'tail') and elem.tail:
tail = elem.tail
if self.in_pre:
tail = self.prepare_string_for_pre(tail)
elif self.in_code:
tail = self.remove_newlines(tail)
else:
tail = self.prepare_string_for_markdown(
self.remove_newlines(tail))
text.append(tail)
return text
def dump_text(self, elem, stylizer, page):
'''
@elem: The element in the etree that we are working on.
@stylizer: The style information attached to the element.
'''
# We can only processes tags. If there isn't a tag return any text.
if not isinstance(elem.tag, string_or_bytes) \
or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, string_or_bytes) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return ['']
# Setup our variables.
text = ['']
style = stylizer.style(elem)
tags = []
tag = barename(elem.tag)
attribs = elem.attrib
style_a = '%s' % style
style_a = style_a if style_a else ''
if tag == 'body':
# Change the body to a div so we can merge multiple files.
tag = 'div'
# Add page-break-brefore: always because renders typically treat a new file (we're merging files)
# as a page break and remove all other page break types that might be set.
style_a = 'page-break-before: always; %s' % re.sub('page-break-[^:]+:[^;]+;?', '', style_a)
# Remove unnecessary spaces.
style_a = re.sub(r'\s{2,}', ' ', style_a).strip()
tags.append(tag)
# Remove attributes we won't want.
if 'class' in attribs:
del attribs['class']
if 'style' in attribs:
del attribs['style']
# Turn the rest of the attributes into a string we can write with the tag.
at = ''
for k, v in attribs.items():
at += ' %s="%s"' % (k, prepare_string_for_xml(v, attribute=True))
# Turn style into strings for putting in the tag.
style_t = ''
if style_a:
style_t = ' style="%s"' % style_a.replace('"', "'")
# Write the tag.
text.append('<%s%s%s' % (tag, at, style_t))
if tag in SELF_CLOSING_TAGS:
text.append(' />')
else:
text.append('>')
# Process tags that contain text.
if hasattr(elem, 'text') and elem.text:
text.append(self.prepare_string_for_html(elem.text))
# Recurse down into tags within the tag we are in.
for item in elem:
text += self.dump_text(item, stylizer, page)
# Close all open tags.
tags.reverse()
for t in tags:
if t not in SELF_CLOSING_TAGS:
text.append('</%s>' % t)
# Add the text that is outside of the tag.
if hasattr(elem, 'tail') and elem.tail:
text.append(self.prepare_string_for_html(elem.tail))
return text
def dump_text(self, elem, stylizer, page, tag_stack=[]):
from calibre.ebooks.oeb.base import XHTML_NS, barename, namespace
if not isinstance(elem.tag, basestring) \
or namespace(elem.tag) != XHTML_NS:
return []
text = []
tags = []
style = stylizer.style(elem)
if style['display'] in ('none', 'oeb-page-head', 'oeb-page-foot') \
or style['visibility'] == 'hidden':
return []
tag = barename(elem.tag)
# Are we in a paragraph block?
if tag in BLOCK_TAGS or style['display'] in BLOCK_STYLES:
tags.append('block')
# Process tags that need special processing and that do not have inner
# text. Usually these require an argument.
if tag in IMAGE_TAGS:
if elem.attrib.get('src', None):
if page.abshref(elem.attrib['src']) not in self.image_hrefs.keys():
if len(self.image_hrefs.keys()) == 0:
self.image_hrefs[page.abshref(elem.attrib['src'])] = 'cover.png'
else:
self.image_hrefs[page.abshref(elem.attrib['src'])] = image_name('%s.png' % len(self.image_hrefs.keys()), self.image_hrefs.keys()).strip('\x00')
text.append('\\m="%s"' % self.image_hrefs[page.abshref(elem.attrib['src'])])
elif tag == 'hr':
w = '\\w'
width = elem.get('width')
if width:
if not width.endswith('%'):
width += '%'
w += '="%s"' % width
else:
w += '="50%"'
text.append(w)
elif tag == 'br':
text.append('\n\\c \n\\c\n')
# TOC markers.
toc_name = elem.attrib.get('name', None)
toc_id = elem.attrib.get('id', None)
# Only write the TOC marker if the tag isn't a heading and we aren't in one.
if (toc_id or toc_name) and tag not in ('h1', 'h2','h3','h4','h5','h6') and \
'x' not in tag_stack+tags and 'X0' not in tag_stack+tags and \
'X1' not in tag_stack+tags and 'X2' not in tag_stack+tags and \
'X3' not in tag_stack+tags and 'X4' not in tag_stack+tags:
toc_page = page.href
if self.toc.get(toc_page, None):
for toc_x in (toc_name, toc_id):
toc_title, toc_depth = self.toc[toc_page].get(toc_x, (None, 0))
if toc_title:
toc_depth = max(min(toc_depth, 4), 0)
text.append('\\C%s="%s"' % (toc_depth, toc_title))
# Process style information that needs holds a single tag.
# Commented out because every page in an OEB book starts with this style.
if style['page-break-before'] == 'always':
text.append('\\p')
# Process basic PML tags.
pml_tag = TAG_MAP.get(tag, None)
if pml_tag and pml_tag not in tag_stack+tags:
text.append('\\%s' % pml_tag)
tags.append(pml_tag)
# Special processing of tags that require an argument.
# Anchors links
if tag in LINK_TAGS and 'q' not in tag_stack+tags:
href = elem.get('href')
if href:
href = page.abshref(href)
if '://' not in href:
if '#' not in href:
href += '#'
if href not in self.link_hrefs.keys():
self.link_hrefs[href] = 'calibre_link-%s' % len(self.link_hrefs.keys())
href = '#%s' % self.link_hrefs[href]
text.append('\\q="%s"' % href)
tags.append('q')
# Anchor ids
id_name = elem.get('id')
name_name = elem.get('name')
for name_x in (id_name, name_name):
if name_x:
text.append(self.get_anchor(page, name_x))
# Processes style information
for s in STYLES:
style_tag = s[1].get(style[s[0]], None)
if style_tag and style_tag not in tag_stack+tags:
text.append('\\%s' % style_tag)
tags.append(style_tag)
# margin left
try:
mms = int(float(style['margin-left']) * 100 / style.height)
if mms:
text.append('\\T="%s%%"' % mms)
except:
pass
# Soft scene breaks.
try:
ems = int(round((float(style.marginTop) / style.fontSize) - 1))
if ems >= 1:
text.append('\n\\c \n\\c\n')
except:
pass
# Proccess text within this tag.
if hasattr(elem, 'text') and elem.text:
text.append(self.prepare_string_for_pml(elem.text))
# Process inner tags
for item in elem:
text += self.dump_text(item, stylizer, page, tag_stack+tags)
# Close opened tags.
tags.reverse()
text += self.close_tags(tags)
#if tag in SEPARATE_TAGS:
# text.append('\n\n')
if style['page-break-after'] == 'always':
text.append('\\p')
# Process text after this tag but not within another.
if hasattr(elem, 'tail') and elem.tail:
text.append(self.prepare_string_for_pml(elem.tail))
return text
def dump_text(self, elem, stylizer):
'''
@elem: The element in the etree that we are working on.
@stylizer: The style information attached to the element.
'''
# We can only processes tags. If there isn't a tag return any text.
if not isinstance(elem.tag, basestring) \
or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, basestring) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return ['']
# Setup our variables.
text = []
style = stylizer.style(elem)
tags = []
tag = barename(elem.tag)
attribs = elem.attrib
# Ignore anything that is set to not be displayed.
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 ['']
# Soft scene breaks.
if 'margin-top' in style.cssdict() and style['margin-top'] != 'auto':
ems = int(round(float(style.marginTop) / style.fontSize) - 1)
if ems >= 1:
text.append(u'\n\n' * ems)
bq = '> ' * self.blockquotes
# Block level elements
if tag in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'p', 'div'):
h_tag = ''
if tag in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6'):
h_tag = '#' * int(tag[1]) + ' '
text.append('\n' + bq + h_tag)
tags.append('\n')
self.remove_space_after_newline = True
if style['font-style'] == 'italic' or tag in ('i', 'em'):
if tag not in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'cite'):
if self.style_italic == False:
text.append('*')
tags.append('*')
self.style_italic = True
if style['font-weight'] in ('bold', 'bolder') or tag in ('b', 'strong'):
if tag not in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'th'):
if self.style_bold == False:
text.append('**')
tags.append('**')
self.style_bold = True
if tag == 'br':
text.append(' \n')
self.remove_space_after_newline = True
if tag == 'blockquote':
self.blockquotes += 1
tags.append('>')
text.append('> ' * self.blockquotes)
elif tag == 'code':
if not self.in_pre and not self.in_code:
text.append('`')
tags.append('`')
self.in_code = True
elif tag == 'pre':
if not self.in_pre:
text.append('\n')
tags.append('pre')
self.in_pre = True
elif tag == 'hr':
text.append('\n* * *')
tags.append('\n')
elif tag == 'a':
# Only write links with absolute (external) urls.
if self.opts.keep_links and attribs.has_key('href') and '://' in attribs['href']:
title = ''
if attribs.has_key('title'):
title = ' "' + attribs['title'] + '"'
remove_space = self.remove_space_after_newline
title = self.remove_newlines(title)
self.remove_space_after_newline = remove_space
text.append('[')
tags.append('](' + attribs['href'] + title + ')')
elif tag == 'img':
if self.opts.keep_image_references:
txt = '!'
if attribs.has_key('alt'):
remove_space = self.remove_space_after_newline
txt += '[' + self.remove_newlines(attribs['alt']) + ']'
self.remove_space_after_newline = remove_space
txt += '(' + attribs['src'] + ')'
text.append(txt)
elif tag in ('ol', 'ul'):
tags.append(tag)
# Add the list to our lists of lists so we can track
# nested lists.
self.list.append({'name': tag, 'num': 0})
elif tag == 'li':
# Get the last list from our list of lists
if self.list:
li = self.list[-1]
else:
li = {'name': 'ul', 'num': 0}
# Add a new line to start the item
text.append('\n')
# Add indent if we have nested lists.
list_count = len(self.list)
# We only care about indenting nested lists.
if (list_count - 1) > 0:
text.append('\t' * (list_count - 1))
# Add blockquote if we have a blockquote in a list item.
text.append(bq)
# Write the proper sign for ordered and unorded lists.
if li['name'] == 'ul':
text.append('+ ')
elif li['name'] == 'ol':
li['num'] += 1
text.append(unicode(li['num']) + '. ')
# Process tags that contain text.
if hasattr(elem, 'text') and elem.text:
txt = elem.text
if self.in_pre:
txt = self.prepare_string_for_pre(txt)
elif self.in_code:
txt = self.remove_newlines(txt)
else:
txt = self.prepare_string_for_markdown(self.remove_newlines(txt))
text.append(txt)
# Recurse down into tags within the tag we are in.
for item in elem:
text += self.dump_text(item, stylizer)
# Close all open tags.
tags.reverse()
for t in tags:
if t in ('pre', 'ul', 'ol', '>'):
if t == 'pre':
self.in_pre = False
text.append('\n')
elif t == '>':
self.blockquotes -= 1
elif t in ('ul', 'ol'):
if self.list:
self.list.pop()
text.append('\n')
else:
if t == '**':
self.style_bold = False
elif t == '*':
self.style_italic = False
elif t == '`':
self.in_code = False
text.append('%s' % t)
# Soft scene breaks.
if 'margin-bottom' in style.cssdict() and style['margin-bottom'] != 'auto':
ems = int(round((float(style.marginBottom) / style.fontSize) - 1))
if ems >= 1:
text.append(u'\n\n' * ems)
# Add the text that is outside of the tag.
if hasattr(elem, 'tail') and elem.tail:
tail = elem.tail
if self.in_pre:
tail = self.prepare_string_for_pre(tail)
elif self.in_code:
tail = self.remove_newlines(tail)
else:
tail = self.prepare_string_for_markdown(self.remove_newlines(tail))
text.append(tail)
return text
def mobimlize_elem(self, elem, stylizer, bstate, istates,
ignore_valign=False):
if not isinstance(elem.tag, basestring) \
or namespace(elem.tag) != 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':
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 = XHTML('a')
else:
return
tag = 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']
istate.indent = style['text-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 = (u'\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 '') + (u'\xa0' * spaces)
else:
last = elem[-1]
last.text = (last.text or '') + (u'\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'] in ('pre', '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 style['font-family'] 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[urlnormalize(href)]
except:
self.oeb.logger.warn('Failed to find image:',
href)
else:
try:
width, height = identify_data(item.data)[:2]
except:
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 = float(width)/float(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:
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 = u'\u201c' + t
t = elem.tail
if not t:
t = ''
elem.tail = u'\u201d' + t
text = None
if elem.text:
if istate.preserve:
text = elem.text
elif len(elem) > 0 and isspace(elem.text):
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, (float, int)) and abs(valign) != 0)
issup = valign in ('super', 'text-top', 'top') or (
isinstance(valign, (float, int)) 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(XHTML('html'), nsmap=MOBI_NSMAP)
vbstate = BlockState(etree.SubElement(nroot, XHTML('body')))
vbstate.para = etree.SubElement(vbstate.body, 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, XHTML(vtag))
vtag = etree.SubElement(vtag, 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:
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)):
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:
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 == u'\xa0' and len(para) < 1:
if style.height > 2:
para.getparent().replace(para, etree.Element(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 dump_text(self, elem, stylizer, tag_stack=[]):
from calibre.ebooks.oeb.base import (XHTML_NS, namespace, barename,
urlnormalize)
if not isinstance(elem.tag, string_or_bytes) \
or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, string_or_bytes) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return elem.tail
return ''
text = ''
style = stylizer.style(elem)
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 ''
tag = barename(elem.tag)
tag_count = 0
# Are we in a paragraph block?
if tag in BLOCK_TAGS or style['display'] in BLOCK_STYLES:
if 'block' not in tag_stack:
tag_count += 1
tag_stack.append('block')
# Process tags that need special processing and that do not have inner
# text. Usually these require an argument
if tag == 'img':
src = elem.get('src')
if src:
src = urlnormalize(self.currently_dumping_item.abshref(src))
block_start = ''
block_end = ''
if 'block' not in tag_stack:
block_start = r'{\par\pard\hyphpar '
block_end = '}'
text += '%s SPECIAL_IMAGE-%s-REPLACE_ME %s' % (block_start,
src, block_end)
single_tag = SINGLE_TAGS.get(tag, None)
if single_tag:
text += single_tag
rtf_tag = TAGS.get(tag, None)
if rtf_tag and rtf_tag not in tag_stack:
tag_count += 1
text += '{%s\n' % rtf_tag
tag_stack.append(rtf_tag)
# Processes style information
for s in STYLES:
style_tag = s[1].get(style[s[0]], None)
if style_tag and style_tag not in tag_stack:
tag_count += 1
text += '{%s\n' % style_tag
tag_stack.append(style_tag)
# Proccess tags that contain text.
if hasattr(elem, 'text') and elem.text:
text += txt2rtf(elem.text)
for item in elem:
text += self.dump_text(item, stylizer, tag_stack)
for i in range(0, tag_count):
end_tag = tag_stack.pop()
if end_tag != 'block':
if tag in BLOCK_TAGS:
text += r'\par\pard\plain\hyphpar}'
else:
text += '}'
if hasattr(elem, 'tail') and elem.tail:
if 'block' in tag_stack:
text += '%s' % txt2rtf(elem.tail)
else:
text += r'{\par\pard\hyphpar %s}' % txt2rtf(elem.tail)
return text
def mobimlize_elem(self, elem, stylizer, bstate, istates, ignore_valign=False):
if not isinstance(elem.tag, basestring) or namespace(elem.tag) != 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":
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 = XHTML("a")
else:
return
tag = 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 = (u"\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 "") + (u"\xa0" * spaces)
else:
last = elem[-1]
last.text = (last.text or "") + (u"\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) / (72.0 / 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[urlnormalize(href)]
except:
self.oeb.logger.warn("Failed to find image:", href)
else:
try:
width, height = identify_data(item.data)[:2]
except:
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 = float(width) / float(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:
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 = u"\u201c" + t
t = elem.tail
if not t:
t = ""
elem.tail = u"\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, (float, int)) and abs(valign) != 0
)
issup = valign in ("super", "text-top", "top") or (isinstance(valign, (float, int)) 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(XHTML("html"), nsmap=MOBI_NSMAP)
vbstate = BlockState(etree.SubElement(nroot, XHTML("body")))
vbstate.para = etree.SubElement(vbstate.body, 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, XHTML(vtag))
vtag = etree.SubElement(vtag, 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 == u"\xa0" and len(para) < 1:
if style.height > 2:
para.getparent().replace(para, etree.Element(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 dump_text(self, elem, stylizer):
'''
@elem: The element in the etree that we are working on.
@stylizer: The style information attached to the element.
'''
# We can only processes tags. If there isn't a tag return any text.
if not isinstance(elem.tag, str) \
or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, str) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return ['']
# Setup our variables.
text = ['']
style = stylizer.style(elem)
tags = []
tag = barename(elem.tag)
attribs = elem.attrib
# Ignore anything that is set to not be displayed.
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 ['']
# Soft scene breaks.
if 'margin-top' in style.cssdict() and style['margin-top'] != 'auto':
ems = min(int(round(float(style.marginTop) / style.fontSize) - 1), self.MAX_EM)
if ems >= 1:
text.append('\n\n\xa0' * ems)
if tag in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'p', 'div'):
if tag == 'div':
tag = 'p'
text.append(self.build_block(tag, style, attribs, stylizer))
text.append('. ')
tags.append('\n')
if style['font-style'] == 'italic' or tag in ('i', 'em'):
if tag not in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'cite'):
if self.style_italic == False: # noqa
if self.in_a_link:
text.append('_')
tags.append('_')
else:
text.append('[_')
tags.append('_]')
self.style_embed.append('_')
self.style_italic = True
if style['font-weight'] in ('bold', 'bolder') or tag in ('b', 'strong'):
if tag not in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'th'):
if self.style_bold == False: # noqa
if self.in_a_link:
text.append('*')
tags.append('*')
else:
text.append('[*')
tags.append('*]')
self.style_embed.append('*')
self.style_bold = True
if style['text-decoration'] == 'underline' or tag in ('u', 'ins'):
if tag != 'a':
if self.style_under == False: # noqa
text.append('[+')
tags.append('+]')
self.style_embed.append('+')
self.style_under = True
if style['text-decoration'] == 'line-through' or tag in ('strike', 'del', 's'):
if self.style_strike == False: # noqa
text.append('[-')
tags.append('-]')
self.style_embed.append('-')
self.style_strike = True
if tag == 'br':
for i in reversed(self.style_embed):
text.append(i)
text.append('\n')
for i in self.style_embed:
text.append(i)
tags.append('')
self.remove_space_after_newline = True
if tag == 'blockquote':
text.append('\nbq. ')
tags.append('\n')
elif tag in ('abbr', 'acronym'):
text.append('')
txt = attribs['title']
tags.append('(' + txt + ')')
elif tag == 'sup':
text.append('^')
tags.append('^')
elif tag == 'sub':
text.append('~')
tags.append('~')
elif tag == 'code':
if self.in_pre:
text.append('\nbc. ')
tags.append('')
else:
text.append('@')
tags.append('@')
elif tag == 'cite':
text.append('??')
tags.append('??')
elif tag == 'hr':
text.append('\n***')
tags.append('\n')
elif tag == 'pre':
self.in_pre = True
text.append('\npre. ')
tags.append('pre\n')
elif tag == 'a':
if self.opts.keep_links:
if 'href' in attribs: # noqa
text.append('"')
tags.append('a')
tags.append('":' + attribs['href'])
self.our_links.append(attribs['href'])
if 'title' in attribs: # noqa
tags.append('(' + attribs['title'] + ')')
self.in_a_link = True
else:
text.append('%')
tags.append('%')
elif tag == 'img':
if self.opts.keep_image_references:
txt = '!' + self.check_halign(style)
txt += self.check_valign(style)
txt += attribs['src']
text.append(txt)
if 'alt' in attribs: # noqa
txt = attribs['alt']
if txt != '':
text.append('(' + txt + ')')
tags.append('!')
elif tag in ('ol', 'ul'):
self.list.append({'name': tag, 'num': 0})
text.append('')
tags.append(tag)
elif tag == 'li':
if self.list:
li = self.list[-1]
else:
li = {'name': 'ul', 'num': 0}
text.append('\n')
if li['name'] == 'ul':
text.append('*' * len(self.list) + ' ')
elif li['name'] == 'ol':
text.append('#' * len(self.list) + ' ')
tags.append('')
elif tag == 'dl':
text.append('\n')
tags.append('')
elif tag == 'dt':
text.append('')
tags.append('\n')
elif tag == 'dd':
text.append(' ')
tags.append('')
elif tag == 'dd':
text.append('')
tags.append('\n')
elif tag == 'table':
txt = self.build_block(tag, style, attribs, stylizer)
txt += '. \n'
if txt != '\ntable. \n':
text.append(txt)
else:
text.append('\n')
tags.append('')
elif tag == 'tr':
txt = self.build_block('', style, attribs, stylizer)
txt += '. '
if txt != '\n. ':
txt = re.sub('\n', '', txt)
text.append(txt)
tags.append('|\n')
elif tag == 'td':
text.append('|')
txt = ''
txt += self.check_halign(style)
txt += self.check_valign(style)
if 'colspan' in attribs: # noqa
txt += '\\' + attribs['colspan']
if 'rowspan' in attribs: # noqa
txt += '/' + attribs['rowspan']
txt += self.check_styles(style)
if txt != '':
text.append(txt + '. ')
tags.append('')
elif tag == 'th':
text.append('|_. ')
tags.append('')
elif tag == 'span':
if style['font-variant'] == 'small-caps':
if self.style_smallcap == False: # noqa
text.append('&')
tags.append('&')
self.style_smallcap = True
else:
if self.in_a_link == False: # noqa
txt = '%'
if self.opts.keep_links:
txt += self.check_id_tag(attribs)
txt += self.check_styles(style)
if txt != '%':
text.append(txt)
tags.append('%')
if self.opts.keep_links and 'id' in attribs: # noqa
if tag not in ('body', 'div', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'p', 'span', 'table'):
text.append(self.check_id_tag(attribs))
# Process the styles for any that we want to keep
if tag not in ('body', 'div', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'p', 'hr', 'a', 'img',
'span', 'table', 'tr', 'td'):
if not self.in_a_link:
text.append(self.check_styles(style))
# Process tags that contain text.
if hasattr(elem, 'text') and elem.text:
txt = elem.text
if not self.in_pre:
txt = self.prepare_string_for_textile(self.remove_newlines(txt))
text.append(txt)
self.id_no_text = ''
# Recurse down into tags within the tag we are in.
for item in elem:
text += self.dump_text(item, stylizer)
# Close all open tags.
tags.reverse()
for t in tags:
if t in ('pre', 'ul', 'ol', 'li', 'table'):
if t == 'pre':
self.in_pre = False
elif t in ('ul', 'ol'):
if self.list:
self.list.pop()
if not self.list:
text.append('\n')
else:
if t == 'a':
self.in_a_link = False
t = ''
text.append(self.id_no_text)
self.id_no_text = ''
if t in ('*]', '*'):
self.style_bold = False
elif t in ('_]', '_'):
self.style_italic = False
elif t == '+]':
self.style_under = False
elif t == '-]':
self.style_strike = False
elif t == '&':
self.style_smallcap = False
if t in ('*]', '_]', '+]', '-]', '*', '_'):
txt = self.style_embed.pop()
text.append('%s' % t)
# Soft scene breaks.
if 'margin-bottom' in style.cssdict() and style['margin-bottom'] != 'auto':
ems = min(int(round((float(style.marginBottom) / style.fontSize) - 1)), self.MAX_EM)
if ems >= 1:
text.append('\n\n\xa0' * ems)
# Add the text that is outside of the tag.
if hasattr(elem, 'tail') and elem.tail:
tail = elem.tail
if not self.in_pre:
tail = self.prepare_string_for_textile(self.remove_newlines(tail))
text.append(tail)
return text
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.
'''
from calibre.ebooks.oeb.base import XHTML_NS, barename, namespace
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, string_or_bytes) or namespace(
elem_tree.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, string_or_bytes) and namespace(p.tag) == 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 = 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:
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':
if elem_tree.attrib.get('src', None):
# Only write the image tag if it is in the manifest.
ihref = 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] = '_%s.jpg' % len(
self.image_hrefs)
p_txt, p_tag = self.ensure_p()
fb2_out += p_txt
tags += p_tag
fb2_out.append('<image xlink: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 == '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 dump_text(self, elem, stylizer, page):
'''
@elem: The element in the etree that we are working on.
@stylizer: The style information attached to the element.
'''
# We can only processes tags. If there isn't a tag return any text.
if not isinstance(elem.tag, string_or_bytes) \
or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, string_or_bytes) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return ['']
# Setup our variables.
text = ['']
style = stylizer.style(elem)
tags = []
tag = barename(elem.tag)
attribs = elem.attrib
style_a = '%s' % style
style_a = style_a if style_a else ''
if tag == 'body':
# Change the body to a div so we can merge multiple files.
tag = 'div'
# Add page-break-brefore: always because renders typically treat a new file (we're merging files)
# as a page break and remove all other page break types that might be set.
style_a = 'page-break-before: always; %s' % re.sub('page-break-[^:]+:[^;]+;?', '', style_a)
# Remove unnecessary spaces.
style_a = re.sub(r'\s{2,}', ' ', style_a).strip()
tags.append(tag)
# Remove attributes we won't want.
if 'class' in attribs:
del attribs['class']
if 'style' in attribs:
del attribs['style']
# Turn the rest of the attributes into a string we can write with the tag.
at = ''
for k, v in attribs.items():
at += ' %s="%s"' % (k, prepare_string_for_xml(v, attribute=True))
# Turn style into strings for putting in the tag.
style_t = ''
if style_a:
style_t = ' style="%s"' % style_a.replace('"', "'")
# Write the tag.
text.append('<%s%s%s' % (tag, at, style_t))
if tag in SELF_CLOSING_TAGS:
text.append(' />')
else:
text.append('>')
# Process tags that contain text.
if hasattr(elem, 'text') and elem.text:
text.append(self.prepare_string_for_html(elem.text))
# Recurse down into tags within the tag we are in.
for item in elem:
text += self.dump_text(item, stylizer, page)
# Close all open tags.
tags.reverse()
for t in tags:
if t not in SELF_CLOSING_TAGS:
text.append('</%s>' % t)
# Add the text that is outside of the tag.
if hasattr(elem, 'tail') and elem.tail:
text.append(self.prepare_string_for_html(elem.tail))
return text
def dump_text(self, elem, stylizer, page):
'''
@elem: The element in the etree that we are working on.
@stylizer: The style information attached to the element.
@page: OEB page used to determine absolute urls.
'''
from calibre.ebooks.oeb.base import XHTML_NS, barename, namespace
if not isinstance(elem.tag, string_or_bytes) \
or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, string_or_bytes) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return ['']
text = ['']
style = stylizer.style(elem)
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 ['']
tag = barename(elem.tag)
tag_id = elem.attrib.get('id', None)
in_block = False
in_heading = False
# Are we in a heading?
# This can either be a heading tag or a TOC item.
if tag in HEADING_TAGS or '%s#%s' % (page.href,
tag_id) in self.toc_ids:
in_heading = True
if not self.last_was_heading:
text.append('\n\n\n\n\n\n')
# Are we in a paragraph block?
if tag in BLOCK_TAGS or style['display'] in BLOCK_STYLES:
if self.opts.remove_paragraph_spacing and not in_heading:
text.append('\t')
in_block = True
if tag in SPACE_TAGS:
text.append(' ')
# Hard scene breaks.
if tag == 'hr':
text.append('\n\n* * *\n\n')
# Soft scene breaks.
try:
ems = int(round((float(style.marginTop) / style.fontSize) - 1))
if ems >= 1:
text.append('\n' * ems)
except:
pass
# Process tags that contain text.
if hasattr(elem, 'text') and elem.text:
text.append(elem.text)
# Recurse down into tags within the tag we are in.
for item in elem:
text += self.dump_text(item, stylizer, page)
if in_block:
text.append('\n\n')
if in_heading:
text.append('\n')
self.last_was_heading = True
else:
self.last_was_heading = False
if hasattr(elem, 'tail') and elem.tail:
text.append(elem.tail)
return text
def dump_text(self, subitems, elem, stylizer, end='', pre=False, li=''):
from calibre.ebooks.oeb.base import XHTML_NS, barename, namespace
if not isinstance(elem.tag, string_or_bytes) \
or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, string_or_bytes) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return ['']
text = ['']
style = stylizer.style(elem)
if elem.attrib.get('id') is not None and elem.attrib['id'] in [href for href, title in subitems]:
if self.curSubItem is not None and self.curSubItem != elem.attrib['id']:
self.curSubItem = elem.attrib['id']
text.append(f'\n\n{CALIBRE_SNB_BM_TAG}{self.curSubItem}\n\n')
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 ['']
tag = barename(elem.tag)
in_block = False
# Are we in a paragraph block?
if tag in BLOCK_TAGS or style['display'] in BLOCK_STYLES:
in_block = True
if not end.endswith('\n\n') and hasattr(elem, 'text') and elem.text:
text.append('\n\n')
if tag in SPACE_TAGS:
if not end.endswith('u ') and hasattr(elem, 'text') and elem.text:
text.append(' ')
if tag == 'img':
text.append('\n\n{}{}\n\n'.format(CALIBRE_SNB_IMG_TAG, ProcessFileName(elem.attrib['src'])))
if tag == 'br':
text.append('\n\n')
if tag == 'li':
li = '- '
pre = (tag == 'pre' or pre)
# Process tags that contain text.
if hasattr(elem, 'text') and elem.text:
if pre:
text.append(('\n\n%s' % CALIBRE_SNB_PRE_TAG).join((li + elem.text).splitlines()))
else:
text.append(li + elem.text)
li = ''
for item in elem:
en = ''
if len(text) >= 2:
en = text[-1][-2:]
t = self.dump_text(subitems, item, stylizer, en, pre, li)[0]
text += t
if in_block:
text.append('\n\n')
if hasattr(elem, 'tail') and elem.tail:
if pre:
text.append(('\n\n%s' % CALIBRE_SNB_PRE_TAG).join(elem.tail.splitlines()))
else:
text.append(li + elem.tail)
li = ''
return text, li
def dump_text(self, elem, stylizer):
'''
@elem: The element in the etree that we are working on.
@stylizer: The style information attached to the element.
'''
# We can only processes tags. If there isn't a tag return any text.
if not isinstance(elem.tag, basestring) \
or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, basestring) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return ['']
# Setup our variables.
text = ['']
style = stylizer.style(elem)
tags = []
tag = barename(elem.tag)
attribs = elem.attrib
# Ignore anything that is set to not be displayed.
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 ['']
# Soft scene breaks.
if 'margin-top' in style.cssdict() and style['margin-top'] != 'auto':
ems = min(int(round(float(style.marginTop) / style.fontSize) - 1), self.MAX_EM)
if ems >= 1:
text.append(u'\n\n\xa0' * ems)
if tag in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'p', 'div'):
if tag == 'div':
tag = 'p'
text.append(self.build_block(tag, style, attribs, stylizer))
text.append('. ')
tags.append('\n')
if style['font-style'] == 'italic' or tag in ('i', 'em'):
if tag not in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'cite'):
if self.style_italic == False:
if self.in_a_link:
text.append('_')
tags.append('_')
else:
text.append('[_')
tags.append('_]')
self.style_embed.append('_')
self.style_italic = True
if style['font-weight'] in ('bold', 'bolder') or tag in ('b', 'strong'):
if tag not in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'th'):
if self.style_bold == False:
if self.in_a_link:
text.append('*')
tags.append('*')
else:
text.append('[*')
tags.append('*]')
self.style_embed.append('*')
self.style_bold = True
if style['text-decoration'] == 'underline' or tag in ('u', 'ins'):
if tag != 'a':
if self.style_under == False:
text.append('[+')
tags.append('+]')
self.style_embed.append('+')
self.style_under = True
if style['text-decoration'] == 'line-through' or tag in ('strike', 'del', 's'):
if self.style_strike == False:
text.append('[-')
tags.append('-]')
self.style_embed.append('-')
self.style_strike = True
if tag == 'br':
for i in reversed(self.style_embed):
text.append(i)
text.append('\n')
for i in self.style_embed:
text.append(i)
tags.append('')
self.remove_space_after_newline = True
if tag == 'blockquote':
text.append('\nbq. ')
tags.append('\n')
elif tag in ('abbr', 'acronym'):
text.append('')
txt = attribs['title']
tags.append('(' + txt + ')')
elif tag == 'sup':
text.append('^')
tags.append('^')
elif tag == 'sub':
text.append('~')
tags.append('~')
elif tag == 'code':
if self.in_pre:
text.append('\nbc. ')
tags.append('')
else:
text.append('@')
tags.append('@')
elif tag == 'cite':
text.append('??')
tags.append('??')
elif tag == 'hr':
text.append('\n***')
tags.append('\n')
elif tag == 'pre':
self.in_pre = True
text.append('\npre. ')
tags.append('pre\n')
elif tag == 'a':
if self.opts.keep_links:
if attribs.has_key('href'):
text.append('"')
tags.append('a')
tags.append('":' + attribs['href'])
self.our_links.append(attribs['href'])
if attribs.has_key('title'):
tags.append('(' + attribs['title'] + ')')
self.in_a_link = True
else:
text.append('%')
tags.append('%')
elif tag == 'img':
if self.opts.keep_image_references:
txt = '!' + self.check_halign(style)
txt += self.check_valign(style)
txt += attribs['src']
text.append(txt)
if attribs.has_key('alt'):
txt = attribs['alt']
if txt != '':
text.append('(' + txt + ')')
tags.append('!')
elif tag in ('ol', 'ul'):
self.list.append({'name': tag, 'num': 0})
text.append('')
tags.append(tag)
elif tag == 'li':
if self.list: li = self.list[-1]
else: li = {'name': 'ul', 'num': 0}
text.append('\n')
if li['name'] == 'ul':
text.append('*' * len(self.list) + ' ')
elif li['name'] == 'ol':
text.append('#' * len(self.list) + ' ')
tags.append('')
elif tag == 'dl':
text.append('\n')
tags.append('')
elif tag == 'dt':
text.append('')
tags.append('\n')
elif tag == 'dd':
text.append(' ')
tags.append('')
elif tag == 'dd':
text.append('')
tags.append('\n')
elif tag == 'table':
txt = self.build_block(tag, style, attribs, stylizer)
txt += '. \n'
if txt != '\ntable. \n':
text.append(txt)
else:
text.append('\n')
tags.append('')
elif tag == 'tr':
txt = self.build_block('', style, attribs, stylizer)
txt += '. '
if txt != '\n. ':
txt = re.sub ('\n', '', txt)
text.append(txt)
tags.append('|\n')
elif tag == 'td':
text.append('|')
txt = ''
txt += self.check_halign(style)
txt += self.check_valign(style)
if attribs.has_key ('colspan'):
txt += '\\' + attribs['colspan']
if attribs.has_key ('rowspan'):
txt += '/' + attribs['rowspan']
txt += self.check_styles(style)
if txt != '':
text.append(txt + '. ')
tags.append('')
elif tag == 'th':
text.append('|_. ')
tags.append('')
elif tag == 'span':
if style['font-variant'] == 'small-caps':
if self.style_smallcap == False:
text.append('&')
tags.append('&')
self.style_smallcap = True
else:
if self.in_a_link == False:
txt = '%'
if self.opts.keep_links:
txt += self.check_id_tag(attribs)
txt += self.check_styles(style)
if txt != '%':
text.append(txt)
tags.append('%')
if self.opts.keep_links and attribs.has_key('id'):
if tag not in ('body', 'div', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'p', 'span', 'table'):
text.append(self.check_id_tag(attribs))
# Process the styles for any that we want to keep
if tag not in ('body', 'div', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'p', 'hr', 'a', 'img', \
'span', 'table', 'tr', 'td'):
if not self.in_a_link:
text.append(self.check_styles(style))
# Process tags that contain text.
if hasattr(elem, 'text') and elem.text:
txt = elem.text
if not self.in_pre:
txt = self.prepare_string_for_textile(self.remove_newlines(txt))
text.append(txt)
self.id_no_text = u''
# Recurse down into tags within the tag we are in.
for item in elem:
text += self.dump_text(item, stylizer)
# Close all open tags.
tags.reverse()
for t in tags:
if t in ('pre', 'ul', 'ol', 'li', 'table'):
if t == 'pre':
self.in_pre = False
elif t in ('ul', 'ol'):
if self.list: self.list.pop()
if not self.list: text.append('\n')
else:
if t == 'a':
self.in_a_link = False
t = ''
text.append(self.id_no_text)
self.id_no_text = u''
if t in ('*]', '*'):
self.style_bold = False
elif t in ('_]', '_'):
self.style_italic = False
elif t == '+]':
self.style_under = False
elif t == '-]':
self.style_strike = False
elif t == '&':
self.style_smallcap = False
if t in ('*]', '_]', '+]', '-]', '*', '_'):
txt = self.style_embed.pop()
text.append('%s' % t)
# Soft scene breaks.
if 'margin-bottom' in style.cssdict() and style['margin-bottom'] != 'auto':
ems = min(int(round((float(style.marginBottom) / style.fontSize) - 1)), self.MAX_EM)
if ems >= 1:
text.append(u'\n\n\xa0' * ems)
# Add the text that is outside of the tag.
if hasattr(elem, 'tail') and elem.tail:
tail = elem.tail
if not self.in_pre:
tail = self.prepare_string_for_textile(self.remove_newlines(tail))
text.append(tail)
return text
def dump_text(self, elem, stylizer, page, tag_stack=[]):
from calibre.ebooks.oeb.base import XHTML_NS, barename, namespace
if not isinstance(elem.tag,
string_or_bytes) or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, string_or_bytes) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return []
text = []
tags = []
style = stylizer.style(elem)
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 []
tag = barename(elem.tag)
# Are we in a paragraph block?
if tag in BLOCK_TAGS or style['display'] in BLOCK_STYLES:
tags.append('block')
# Process tags that need special processing and that do not have inner
# text. Usually these require an argument.
if tag in IMAGE_TAGS:
if elem.attrib.get('src', None):
if page.abshref(
elem.attrib['src']) not in self.image_hrefs.keys():
if len(self.image_hrefs.keys()) == 0:
self.image_hrefs[page.abshref(
elem.attrib['src'])] = 'cover.png'
else:
self.image_hrefs[page.abshref(
elem.attrib['src'])] = image_name(
'%s.png' % len(self.image_hrefs.keys()),
self.image_hrefs.keys()).strip('\x00')
text.append('\\m="%s"' %
self.image_hrefs[page.abshref(elem.attrib['src'])])
elif tag == 'hr':
w = r'\w'
width = elem.get('width')
if width:
if not width.endswith('%'):
width += '%'
w += '="%s"' % width
else:
w += '="50%"'
text.append(w)
elif tag == 'br':
text.append('\n\\c \n\\c\n')
# TOC markers.
toc_name = elem.attrib.get('name', None)
toc_id = elem.attrib.get('id', None)
# Only write the TOC marker if the tag isn't a heading and we aren't in one.
if (toc_id or toc_name) and tag not in ('h1', 'h2','h3','h4','h5','h6') and \
'x' not in tag_stack+tags and 'X0' not in tag_stack+tags and \
'X1' not in tag_stack+tags and 'X2' not in tag_stack+tags and \
'X3' not in tag_stack+tags and 'X4' not in tag_stack+tags:
toc_page = page.href
if self.toc.get(toc_page, None):
for toc_x in (toc_name, toc_id):
toc_title, toc_depth = self.toc[toc_page].get(
toc_x, (None, 0))
if toc_title:
toc_depth = max(min(toc_depth, 4), 0)
text.append(r'\C{}="{}"'.format(toc_depth, toc_title))
# Process style information that needs holds a single tag.
# Commented out because every page in an OEB book starts with this style.
if style['page-break-before'] == 'always':
text.append(r'\p')
# Process basic PML tags.
pml_tag = TAG_MAP.get(tag, None)
if pml_tag and pml_tag not in tag_stack + tags:
text.append(r'\%s' % pml_tag)
tags.append(pml_tag)
# Special processing of tags that require an argument.
# Anchors links
if tag in LINK_TAGS and 'q' not in tag_stack + tags:
href = elem.get('href')
if href:
href = page.abshref(href)
if '://' not in href:
if '#' not in href:
href += '#'
if href not in self.link_hrefs.keys():
self.link_hrefs[href] = 'calibre_link-%s' % len(
self.link_hrefs.keys())
href = '#%s' % self.link_hrefs[href]
text.append(r'\q="%s"' % href)
tags.append('q')
# Anchor ids
id_name = elem.get('id')
name_name = elem.get('name')
for name_x in (id_name, name_name):
if name_x:
text.append(self.get_anchor(page, name_x))
# Processes style information
for s in STYLES:
style_tag = s[1].get(style[s[0]], None)
if style_tag and style_tag not in tag_stack + tags:
text.append(r'\%s' % style_tag)
tags.append(style_tag)
# margin left
try:
mms = int(float(style['margin-left']) * 100 / style.height)
if mms:
text.append(r'\T="%s%%"' % mms)
except:
pass
# Soft scene breaks.
try:
ems = int(round((float(style.marginTop) / style.fontSize) - 1))
if ems >= 1:
text.append('\n\\c \n\\c\n')
except:
pass
# Proccess text within this tag.
if hasattr(elem, 'text') and elem.text:
text.append(self.prepare_string_for_pml(elem.text))
# Process inner tags
for item in elem:
text += self.dump_text(item, stylizer, page, tag_stack + tags)
# Close opened tags.
tags.reverse()
text += self.close_tags(tags)
# if tag in SEPARATE_TAGS:
# text.append('\n\n')
if style['page-break-after'] == 'always':
text.append(r'\p')
# Process text after this tag but not within another.
if hasattr(elem, 'tail') and elem.tail:
text.append(self.prepare_string_for_pml(elem.tail))
return text
def dump_text(self, elem, stylizer, page):
'''
@elem: The element in the etree that we are working on.
@stylizer: The style information attached to the element.
'''
# We can only processes tags. If there isn't a tag return any text.
if not isinstance(elem.tag, basestring) \
or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, basestring) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return ['']
# Setup our variables.
text = ['']
style = stylizer.style(elem)
tags = []
tag = barename(elem.tag)
attribs = elem.attrib
style_a = '%s' % style
if tag == 'body':
tag = 'div'
if not style['page-break-before'] == 'always':
style_a = 'page-break-before: always;' + ' ' if style_a else '' + style_a
tags.append(tag)
# Remove attributes we won't want.
if 'class' in attribs:
del attribs['class']
if 'style' in attribs:
del attribs['style']
# Turn the rest of the attributes into a string we can write with the tag.
at = ''
for k, v in attribs.items():
at += ' %s="%s"' % (k, prepare_string_for_xml(v, attribute=True))
# Turn style into strings for putting in the tag.
style_t = ''
if style_a:
style_t = ' style="%s"' % style_a
# Write the tag.
text.append('<%s%s%s>' % (tag, at, style_t))
# Process tags that contain text.
if hasattr(elem, 'text') and elem.text:
text.append(self.prepare_string_for_html(elem.text))
# Recurse down into tags within the tag we are in.
for item in elem:
text += self.dump_text(item, stylizer, page)
# Close all open tags.
tags.reverse()
for t in tags:
text.append('</%s>' % t)
# Add the text that is outside of the tag.
if hasattr(elem, 'tail') and elem.tail:
text.append(self.prepare_string_for_html(elem.tail))
return text
def dump_text(self, elem, stylizer, page):
'''
@elem: The element in the etree that we are working on.
@stylizer: The style information attached to the element.
'''
# We can only processes tags. If there isn't a tag return any text.
if not isinstance(elem.tag, basestring) \
or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, basestring) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return ['']
# Setup our variables.
text = ['']
style = stylizer.style(elem)
tags = []
tag = barename(elem.tag)
attribs = elem.attrib
if tag == 'body':
tag = 'div'
tags.append(tag)
# Ignore anything that is set to not be displayed.
if style['display'] in ('none', 'oeb-page-head', 'oeb-page-foot') \
or style['visibility'] == 'hidden':
return ['']
# Remove attributes we won't want.
if 'class' in attribs:
del attribs['class']
if 'style' in attribs:
del attribs['style']
# Turn the rest of the attributes into a string we can write with the tag.
at = ''
for k, v in attribs.items():
at += ' %s="%s"' % (k, prepare_string_for_xml(v, attribute=True))
# Write the tag.
text.append('<%s%s' % (tag, at))
if tag in SELF_CLOSING_TAGS:
text.append(' />')
else:
text.append('>')
# Turn styles into tags.
if style['font-weight'] in ('bold', 'bolder'):
text.append('<b>')
tags.append('b')
if style['font-style'] == 'italic':
text.append('<i>')
tags.append('i')
if style['text-decoration'] == 'underline':
text.append('<u>')
tags.append('u')
if style['text-decoration'] == 'line-through':
text.append('<s>')
tags.append('s')
# Process tags that contain text.
if hasattr(elem, 'text') and elem.text:
text.append(self.prepare_string_for_html(elem.text))
# Recurse down into tags within the tag we are in.
for item in elem:
text += self.dump_text(item, stylizer, page)
# Close all open tags.
tags.reverse()
for t in tags:
if t not in SELF_CLOSING_TAGS:
text.append('</%s>' % t)
# Add the text that is outside of the tag.
if hasattr(elem, 'tail') and elem.tail:
text.append(self.prepare_string_for_html(elem.tail))
return text
def flatten_node(self, node, stylizer, names, styles, psize, item_id):
if not isinstance(node.tag, basestring) \
or namespace(node.tag) != XHTML_NS:
return
tag = barename(node.tag)
style = stylizer.style(node)
cssdict = style.cssdict()
try:
font_size = style['font-size']
except:
font_size = self.sbase if self.sbase is not None else \
self.context.source.fbase
if 'align' in node.attrib:
if tag != 'img':
cssdict['text-align'] = node.attrib['align']
else:
val = node.attrib['align']
if val in ('middle', 'bottom', 'top'):
cssdict['vertical-align'] = val
elif val in ('left', 'right'):
cssdict['text-align'] = val
del node.attrib['align']
if node.tag == XHTML('font'):
tags = ['descendant::h:%s'%x for x in ('p', 'div', 'table', 'h1',
'h2', 'h3', 'h4', 'h5', 'h6', 'ol', 'ul', 'dl', 'blockquote')]
tag = 'div' if XPath('|'.join(tags))(node) else 'span'
node.tag = XHTML(tag)
if 'size' in node.attrib:
def force_int(raw):
return int(re.search(r'([0-9+-]+)', raw).group(1))
size = node.attrib['size'].strip()
if size:
fnums = self.context.source.fnums
if size[0] in ('+', '-'):
# Oh, the warcrimes
try:
esize = 3 + force_int(size)
except:
esize = 3
if esize < 1:
esize = 1
if esize > 7:
esize = 7
font_size = fnums[esize]
else:
try:
font_size = fnums[force_int(size)]
except:
font_size = fnums[3]
cssdict['font-size'] = '%.1fpt'%font_size
del node.attrib['size']
if 'face' in node.attrib:
cssdict['font-family'] = node.attrib['face']
del node.attrib['face']
if 'color' in node.attrib:
cssdict['color'] = node.attrib['color']
del node.attrib['color']
if 'bgcolor' in node.attrib:
cssdict['background-color'] = node.attrib['bgcolor']
del node.attrib['bgcolor']
if cssdict.get('font-weight', '').lower() == 'medium':
cssdict['font-weight'] = 'normal' # ADE chokes on font-weight medium
fsize = font_size
if not self.context.disable_font_rescaling:
_sbase = self.sbase if self.sbase is not None else \
self.context.source.fbase
dyn_rescale = dynamic_rescale_factor(node)
if dyn_rescale is not None:
fsize = self.fmap[_sbase]
fsize *= dyn_rescale
cssdict['font-size'] = '%0.5fem'%(fsize/psize)
psize = fsize
elif 'font-size' in cssdict or tag == 'body':
fsize = self.fmap[font_size]
try:
cssdict['font-size'] = "%0.5fem" % (fsize / psize)
except ZeroDivisionError:
cssdict['font-size'] = '%.1fpt'%fsize
psize = fsize
try:
minlh = self.context.minimum_line_height / 100.
if style['line-height'] < minlh * fsize:
cssdict['line-height'] = str(minlh)
except:
self.oeb.logger.exception('Failed to set minimum line-height')
if cssdict:
for x in self.filter_css:
cssdict.pop(x, None)
if cssdict:
if self.lineh and self.fbase and tag != 'body':
self.clean_edges(cssdict, style, psize)
if 'display' in cssdict and cssdict['display'] == 'in-line':
cssdict['display'] = 'inline'
if self.unfloat and 'float' in cssdict \
and cssdict.get('display', 'none') != 'none':
del cssdict['display']
if self.untable and 'display' in cssdict \
and cssdict['display'].startswith('table'):
display = cssdict['display']
if display == 'table-cell':
cssdict['display'] = 'inline'
else:
cssdict['display'] = 'block'
if 'vertical-align' in cssdict \
and cssdict['vertical-align'] == 'sup':
cssdict['vertical-align'] = 'super'
if self.lineh and 'line-height' not in cssdict:
lineh = self.lineh / psize
cssdict['line-height'] = "%0.5fem" % lineh
if (self.context.remove_paragraph_spacing or
self.context.insert_blank_line) and tag in ('p', 'div'):
if item_id != 'calibre_jacket' or self.context.output_profile.name == 'Kindle':
for prop in ('margin', 'padding', 'border'):
for edge in ('top', 'bottom'):
cssdict['%s-%s'%(prop, edge)] = '0pt'
if self.context.insert_blank_line:
cssdict['margin-top'] = cssdict['margin-bottom'] = \
'%fem'%self.context.insert_blank_line_size
indent_size = self.context.remove_paragraph_spacing_indent_size
keep_indents = indent_size < 0.0
if (self.context.remove_paragraph_spacing and not keep_indents and
cssdict.get('text-align', None) not in ('center', 'right')):
cssdict['text-indent'] = "%1.1fem" % indent_size
if cssdict:
items = cssdict.items()
items.sort()
css = u';\n'.join(u'%s: %s' % (key, val) for key, val in items)
classes = node.get('class', '').strip() or 'calibre'
klass = STRIPNUM.sub('', classes.split()[0].replace('_', ''))
if css in styles:
match = styles[css]
else:
match = klass + str(names[klass] or '')
styles[css] = match
names[klass] += 1
node.attrib['class'] = match
elif 'class' in node.attrib:
del node.attrib['class']
if 'style' in node.attrib:
del node.attrib['style']
for child in node:
self.flatten_node(child, stylizer, names, styles, psize, item_id)
def dump_text(self, elem, stylizer, tag_stack=[]):
from calibre.ebooks.oeb.base import XHTML_NS, namespace, barename, urlnormalize
if not isinstance(elem.tag, basestring) or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, basestring) and namespace(p.tag) == XHTML_NS and elem.tail:
return elem.tail
return u""
text = u""
style = stylizer.style(elem)
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 u""
tag = barename(elem.tag)
tag_count = 0
# Are we in a paragraph block?
if tag in BLOCK_TAGS or style["display"] in BLOCK_STYLES:
if "block" not in tag_stack:
tag_count += 1
tag_stack.append("block")
# Process tags that need special processing and that do not have inner
# text. Usually these require an argument
if tag == "img":
src = elem.get("src")
if src:
src = urlnormalize(self.currently_dumping_item.abshref(src))
block_start = ""
block_end = ""
if "block" not in tag_stack:
block_start = "{\\par\\pard\\hyphpar "
block_end = "}"
text += "%s SPECIAL_IMAGE-%s-REPLACE_ME %s" % (block_start, src, block_end)
single_tag = SINGLE_TAGS.get(tag, None)
if single_tag:
text += single_tag
rtf_tag = TAGS.get(tag, None)
if rtf_tag and rtf_tag not in tag_stack:
tag_count += 1
text += "{%s\n" % rtf_tag
tag_stack.append(rtf_tag)
# Processes style information
for s in STYLES:
style_tag = s[1].get(style[s[0]], None)
if style_tag and style_tag not in tag_stack:
tag_count += 1
text += "{%s\n" % style_tag
tag_stack.append(style_tag)
# Proccess tags that contain text.
if hasattr(elem, "text") and elem.text:
text += txt2rtf(elem.text)
for item in elem:
text += self.dump_text(item, stylizer, tag_stack)
for i in range(0, tag_count):
end_tag = tag_stack.pop()
if end_tag != "block":
if tag in BLOCK_TAGS:
text += u"\\par\\pard\\plain\\hyphpar}"
else:
text += u"}"
if hasattr(elem, "tail") and elem.tail:
if "block" in tag_stack:
text += "%s" % txt2rtf(elem.tail)
else:
text += "{\\par\\pard\\hyphpar %s}" % txt2rtf(elem.tail)
return text
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.
'''
from calibre.ebooks.oeb.base import XHTML_NS, barename, namespace
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, basestring) or namespace(elem_tree.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, basestring) and namespace(p.tag) == 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 = 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:
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':
if elem_tree.attrib.get('src', None):
# Only write the image tag if it is in the manifest.
ihref = 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] = '_%s.jpg' % len(self.image_hrefs)
p_txt, p_tag = self.ensure_p()
fb2_out += p_txt
tags += p_tag
fb2_out.append('<image xlink: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 == '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 dump_text(self, elem, stylizer, tag_stack=[]):
from calibre.ebooks.oeb.base import (XHTML_NS, namespace, barename,
urlnormalize)
if not isinstance(elem.tag, basestring) \
or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, basestring) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return elem.tail
return u''
text = u''
style = stylizer.style(elem)
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 u''
tag = barename(elem.tag)
tag_count = 0
# Are we in a paragraph block?
if tag in BLOCK_TAGS or style['display'] in BLOCK_STYLES:
if 'block' not in tag_stack:
tag_count += 1
tag_stack.append('block')
# Process tags that need special processing and that do not have inner
# text. Usually these require an argument
if tag == 'img':
src = elem.get('src')
if src:
src = urlnormalize(self.currently_dumping_item.abshref(src))
block_start = ''
block_end = ''
if 'block' not in tag_stack:
block_start = '{\\par\\pard\\hyphpar '
block_end = '}'
text += '%s SPECIAL_IMAGE-%s-REPLACE_ME %s' % (block_start, src, block_end)
single_tag = SINGLE_TAGS.get(tag, None)
if single_tag:
text += single_tag
rtf_tag = TAGS.get(tag, None)
if rtf_tag and rtf_tag not in tag_stack:
tag_count += 1
text += '{%s\n' % rtf_tag
tag_stack.append(rtf_tag)
# Processes style information
for s in STYLES:
style_tag = s[1].get(style[s[0]], None)
if style_tag and style_tag not in tag_stack:
tag_count += 1
text += '{%s\n' % style_tag
tag_stack.append(style_tag)
# Proccess tags that contain text.
if hasattr(elem, 'text') and elem.text:
text += txt2rtf(elem.text)
for item in elem:
text += self.dump_text(item, stylizer, tag_stack)
for i in range(0, tag_count):
end_tag = tag_stack.pop()
if end_tag != 'block':
if tag in BLOCK_TAGS:
text += u'\\par\\pard\\plain\\hyphpar}'
else:
text += u'}'
if hasattr(elem, 'tail') and elem.tail:
if 'block' in tag_stack:
text += '%s' % txt2rtf(elem.tail)
else:
text += '{\\par\\pard\\hyphpar %s}' % txt2rtf(elem.tail)
return text
def mobimlize_elem(self,
elem,
stylizer,
bstate,
istates,
ignore_valign=False):
if not isinstance(elem.tag, string_or_bytes) \
or namespace(elem.tag) != 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 = XHTML('a')
else:
return
tag = 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')
try:
istate.indent = style['text-indent']
except Exception:
istate.indent = 0
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 = convert_margin(style, 'margin-left')
padding = asfloat(style['padding-left'])
if tag != 'body':
left = margin + padding
istate.left += left
vmargin = convert_margin(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 = convert_margin(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 = convert_margin(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[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(XHTML('html'), nsmap=MOBI_NSMAP)
vbstate = BlockState(etree.SubElement(nroot, XHTML('body')))
vbstate.para = etree.SubElement(vbstate.body, 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, XHTML(vtag))
vtag = etree.SubElement(vtag, 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(XHTML('br')))
else:
# This is too small to be rendered effectively, drop it
para.getparent().remove(para)
bstate.para = None
bstate.istate = None
vmargin = convert_margin(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 dump_text(self, elem, stylizer, page, tag_stack=[]):
from calibre.ebooks.oeb.base import XHTML_NS, barename, namespace
if not isinstance(elem.tag,
string_or_bytes) or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, string_or_bytes) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return ['']
text = ['']
style = stylizer.style(elem)
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 ['']
tag = barename(elem.tag)
tag_count = 0
# Process tags that need special processing and that do not have inner
# text. Usually these require an argument
if tag in IMAGE_TAGS:
if elem.attrib.get('src', None):
if page.abshref(
elem.attrib['src']) not in self.name_map.keys():
self.name_map[page.abshref(
elem.attrib['src'])] = unique_name(
'%s' % len(self.name_map.keys()),
self.name_map.keys())
text.append('<IMG SRC="%s">' %
self.name_map[page.abshref(elem.attrib['src'])])
rb_tag = tag.upper() if tag in TAGS else None
if rb_tag:
tag_count += 1
text.append('<%s>' % rb_tag)
tag_stack.append(rb_tag)
# Anchors links
if tag in LINK_TAGS:
href = elem.get('href')
if href:
href = page.abshref(href)
if '://' not in href:
if '#' not in href:
href += '#'
if href not in self.link_hrefs.keys():
self.link_hrefs[href] = 'calibre_link-%s' % len(
self.link_hrefs.keys())
href = self.link_hrefs[href]
text.append('<A HREF="#%s">' % href)
tag_count += 1
tag_stack.append('A')
# Anchor ids
id_name = elem.get('id')
if id_name:
text.append(self.get_anchor(page, id_name))
# Processes style information
for s in STYLES:
style_tag = s[1].get(style[s[0]], None)
if style_tag:
style_tag = style_tag.upper()
tag_count += 1
text.append('<%s>' % style_tag)
tag_stack.append(style_tag)
# Proccess tags that contain text.
if hasattr(elem, 'text') and elem.text:
text.append(prepare_string_for_xml(elem.text))
for item in elem:
text += self.dump_text(item, stylizer, page, tag_stack)
close_tag_list = []
for i in range(0, tag_count):
close_tag_list.insert(0, tag_stack.pop())
text += self.close_tags(close_tag_list)
if hasattr(elem, 'tail') and elem.tail:
text.append(prepare_string_for_xml(elem.tail))
return text
def dump_text(self, elem, stylizer, page):
'''
@elem: The element in the etree that we are working on.
@stylizer: The style information attached to the element.
'''
# We can only processes tags. If there isn't a tag return any text.
if not isinstance(elem.tag, string_or_bytes) \
or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, string_or_bytes) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return ['']
# Setup our variables.
text = ['']
style = stylizer.style(elem)
tags = []
tag = barename(elem.tag)
attribs = elem.attrib
if tag == 'body':
tag = 'div'
tags.append(tag)
# Ignore anything that is set to not be displayed.
if style['display'] in ('none', 'oeb-page-head', 'oeb-page-foot') \
or style['visibility'] == 'hidden':
return ['']
# Remove attributes we won't want.
if 'class' in attribs:
del attribs['class']
if 'style' in attribs:
del attribs['style']
# Turn the rest of the attributes into a string we can write with the tag.
at = ''
for k, v in attribs.items():
at += ' %s="%s"' % (k, prepare_string_for_xml(v, attribute=True))
# Write the tag.
text.append('<%s%s' % (tag, at))
if tag in SELF_CLOSING_TAGS:
text.append(' />')
else:
text.append('>')
# Turn styles into tags.
if style['font-weight'] in ('bold', 'bolder'):
text.append('<b>')
tags.append('b')
if style['font-style'] == 'italic':
text.append('<i>')
tags.append('i')
if style['text-decoration'] == 'underline':
text.append('<u>')
tags.append('u')
if style['text-decoration'] == 'line-through':
text.append('<s>')
tags.append('s')
# Process tags that contain text.
if hasattr(elem, 'text') and elem.text:
text.append(self.prepare_string_for_html(elem.text))
# Recurse down into tags within the tag we are in.
for item in elem:
text += self.dump_text(item, stylizer, page)
# Close all open tags.
tags.reverse()
for t in tags:
if t not in SELF_CLOSING_TAGS:
text.append('</%s>' % t)
# Add the text that is outside of the tag.
if hasattr(elem, 'tail') and elem.tail:
text.append(self.prepare_string_for_html(elem.tail))
return text
def dump_text(self, elem, stylizer, page):
'''
@elem: The element in the etree that we are working on.
@stylizer: The style information attached to the element.
'''
# We can only processes tags. If there isn't a tag return any text.
if not isinstance(elem.tag, string_or_bytes) \
or namespace(elem.tag) not in (XHTML_NS, SVG_NS):
p = elem.getparent()
if p is not None and isinstance(p.tag, string_or_bytes) and namespace(p.tag) in (XHTML_NS, SVG_NS) \
and elem.tail:
return [elem.tail]
return ['']
# Setup our variables.
text = ['']
tags = []
tag = barename(elem.tag)
attribs = elem.attrib
if tag == 'body':
tag = 'div'
tags.append(tag)
# Remove attributes we won't want.
if 'style' in attribs:
del attribs['style']
# Turn the rest of the attributes into a string we can write with the tag.
at = ''
for k, v in attribs.items():
k = k.split('}')[-1]
at += ' %s="%s"' % (k, prepare_string_for_xml(v, attribute=True))
# Write the tag.
text.append('<%s%s' % (tag, at))
if tag in SELF_CLOSING_TAGS:
text.append(' />')
else:
text.append('>')
# Process tags that contain text.
if hasattr(elem, 'text') and elem.text:
text.append(self.prepare_string_for_html(elem.text))
# Recurse down into tags within the tag we are in.
for item in elem:
text += self.dump_text(item, stylizer, page)
# Close all open tags.
tags.reverse()
for t in tags:
if t not in SELF_CLOSING_TAGS:
text.append('</%s>' % t)
# Add the text that is outside of the tag.
if hasattr(elem, 'tail') and elem.tail:
text.append(self.prepare_string_for_html(elem.tail))
return text
def dump_text(self, elem, stylizer, page):
'''
@elem: The element in the etree that we are working on.
@stylizer: The style information attached to the element.
@page: OEB page used to determine absolute urls.
'''
from calibre.ebooks.oeb.base import XHTML_NS, barename, namespace
if not isinstance(elem.tag, basestring) \
or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, basestring) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return ['']
text = ['']
style = stylizer.style(elem)
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 ['']
tag = barename(elem.tag)
tag_id = elem.attrib.get('id', None)
in_block = False
in_heading = False
# Are we in a heading?
# This can either be a heading tag or a TOC item.
if tag in HEADING_TAGS or '%s#%s' % (page.href, tag_id) in self.toc_ids:
in_heading = True
if not self.last_was_heading:
text.append('\n\n\n\n\n\n')
# Are we in a paragraph block?
if tag in BLOCK_TAGS or style['display'] in BLOCK_STYLES:
if self.opts.remove_paragraph_spacing and not in_heading:
text.append(u'\t')
in_block = True
if tag in SPACE_TAGS:
text.append(u' ')
# Hard scene breaks.
if tag == 'hr':
text.append('\n\n* * *\n\n')
# Soft scene breaks.
try:
ems = int(round((float(style.marginTop) / style.fontSize) - 1))
if ems >= 1:
text.append('\n' * ems)
except:
pass
# Process tags that contain text.
if hasattr(elem, 'text') and elem.text:
text.append(elem.text)
# Recurse down into tags within the tag we are in.
for item in elem:
text += self.dump_text(item, stylizer, page)
if in_block:
text.append(u'\n\n')
if in_heading:
text.append(u'\n')
self.last_was_heading = True
else:
self.last_was_heading = False
if hasattr(elem, 'tail') and elem.tail:
text.append(elem.tail)
return text
def flatten_node(self, node, stylizer, names, styles, pseudo_styles, psize,
item_id):
if not isinstance(node.tag, basestring) \
or namespace(node.tag) != XHTML_NS:
return
tag = barename(node.tag)
style = stylizer.style(node)
cssdict = style.cssdict()
try:
font_size = style['font-size']
except:
font_size = self.sbase if self.sbase is not None else \
self.context.source.fbase
if tag == 'body' and isinstance(font_size, (int, float)):
stylizer.body_font_size = font_size
if 'align' in node.attrib:
if tag != 'img':
cssdict['text-align'] = node.attrib['align']
else:
val = node.attrib['align']
if val in ('middle', 'bottom', 'top'):
cssdict['vertical-align'] = val
elif val in ('left', 'right'):
cssdict['float'] = val
del node.attrib['align']
if node.tag == XHTML('font'):
tags = [
'descendant::h:%s' % x
for x in ('p', 'div', 'table', 'h1', 'h2', 'h3', 'h4', 'h5',
'h6', 'ol', 'ul', 'dl', 'blockquote')
]
tag = 'div' if XPath('|'.join(tags))(node) else 'span'
node.tag = XHTML(tag)
if 'size' in node.attrib:
def force_int(raw):
return int(re.search(r'([0-9+-]+)', raw).group(1))
size = node.attrib['size'].strip()
if size:
fnums = self.context.source.fnums
if size[0] in ('+', '-'):
# Oh, the warcrimes
try:
esize = 3 + force_int(size)
except:
esize = 3
if esize < 1:
esize = 1
if esize > 7:
esize = 7
font_size = fnums[esize]
else:
try:
font_size = fnums[force_int(size)]
except:
font_size = fnums[3]
cssdict['font-size'] = '%.1fpt' % font_size
del node.attrib['size']
if 'face' in node.attrib:
cssdict['font-family'] = node.attrib['face']
del node.attrib['face']
if 'color' in node.attrib:
try:
cssdict['color'] = Property('color',
node.attrib['color']).value
except (ValueError, SyntaxErr):
pass
del node.attrib['color']
if 'bgcolor' in node.attrib:
try:
cssdict['background-color'] = Property(
'background-color', node.attrib['bgcolor']).value
except (ValueError, SyntaxErr):
pass
del node.attrib['bgcolor']
if cssdict.get('font-weight', '').lower() == 'medium':
cssdict[
'font-weight'] = 'normal' # ADE chokes on font-weight medium
fsize = font_size
is_drop_cap = (cssdict.get('float', None) == 'left'
and 'font-size' in cssdict and len(node) == 0
and node.text and len(node.text) == 1)
# Detect drop caps generated by the docx input plugin
if (node.tag and node.tag.endswith('}p') and len(node) == 0
and node.text and len(node.text.strip()) == 1 and not node.tail
and 'line-height' in cssdict and 'font-size' in cssdict):
dp = node.getparent()
if dp.tag and dp.tag.endswith('}div') and len(
dp) == 1 and not dp.text:
if stylizer.style(dp).cssdict().get('float', None) == 'left':
is_drop_cap = True
if not self.context.disable_font_rescaling and not is_drop_cap:
_sbase = self.sbase if self.sbase is not None else \
self.context.source.fbase
dyn_rescale = dynamic_rescale_factor(node)
if dyn_rescale is not None:
fsize = self.fmap[_sbase]
fsize *= dyn_rescale
cssdict['font-size'] = '%0.5fem' % (fsize / psize)
psize = fsize
elif 'font-size' in cssdict or tag == 'body':
fsize = self.fmap[font_size]
try:
cssdict['font-size'] = "%0.5fem" % (fsize / psize)
except ZeroDivisionError:
cssdict['font-size'] = '%.1fpt' % fsize
psize = fsize
try:
minlh = self.context.minimum_line_height / 100.
if not is_drop_cap and style['line-height'] < minlh * fsize:
cssdict['line-height'] = str(minlh)
except:
self.oeb.logger.exception('Failed to set minimum line-height')
if cssdict:
for x in self.filter_css:
cssdict.pop(x, None)
if cssdict:
if self.lineh and self.fbase and tag != 'body':
self.clean_edges(cssdict, style, psize)
if 'display' in cssdict and cssdict['display'] == 'in-line':
cssdict['display'] = 'inline'
if self.unfloat and 'float' in cssdict \
and cssdict.get('display', 'none') != 'none':
del cssdict['display']
if self.untable and 'display' in cssdict \
and cssdict['display'].startswith('table'):
display = cssdict['display']
if display == 'table-cell':
cssdict['display'] = 'inline'
else:
cssdict['display'] = 'block'
if 'vertical-align' in cssdict \
and cssdict['vertical-align'] == 'sup':
cssdict['vertical-align'] = 'super'
if self.lineh and 'line-height' not in cssdict:
lineh = self.lineh / psize
cssdict['line-height'] = "%0.5fem" % lineh
if (self.context.remove_paragraph_spacing
or self.context.insert_blank_line) and tag in ('p', 'div'):
if item_id != 'calibre_jacket' or self.context.output_profile.name == 'Kindle':
for prop in ('margin', 'padding', 'border'):
for edge in ('top', 'bottom'):
cssdict['%s-%s' % (prop, edge)] = '0pt'
if self.context.insert_blank_line:
cssdict['margin-top'] = cssdict['margin-bottom'] = \
'%fem'%self.context.insert_blank_line_size
indent_size = self.context.remove_paragraph_spacing_indent_size
keep_indents = indent_size < 0.0
if (self.context.remove_paragraph_spacing and not keep_indents
and cssdict.get('text-align',
None) not in ('center', 'right')):
cssdict['text-indent'] = "%1.1fem" % indent_size
pseudo_classes = style.pseudo_classes(self.filter_css)
if cssdict or pseudo_classes:
keep_classes = set()
if cssdict:
items = sorted(cssdict.items())
css = u';\n'.join(u'%s: %s' % (key, val) for key, val in items)
classes = node.get('class', '').strip() or 'calibre'
klass = ascii_text(
STRIPNUM.sub('',
classes.split()[0].replace('_', '')))
if css in styles:
match = styles[css]
else:
match = klass + str(names[klass] or '')
styles[css] = match
names[klass] += 1
node.attrib['class'] = match
keep_classes.add(match)
for psel, cssdict in pseudo_classes.iteritems():
items = sorted(cssdict.iteritems())
css = u';\n'.join(u'%s: %s' % (key, val) for key, val in items)
pstyles = pseudo_styles[psel]
if css in pstyles:
match = pstyles[css]
else:
# We have to use a different class for each psel as
# otherwise you can have incorrect styles for a situation
# like: a:hover { color: red } a:link { color: blue } a.x:hover { color: green }
# If the pcalibre class for a:hover and a:link is the same,
# then the class attribute for a.x tags will contain both
# that class and the class for a.x:hover, which is wrong.
klass = 'pcalibre'
match = klass + str(names[klass] or '')
pstyles[css] = match
names[klass] += 1
keep_classes.add(match)
node.attrib['class'] = ' '.join(keep_classes)
elif 'class' in node.attrib:
del node.attrib['class']
if 'style' in node.attrib:
del node.attrib['style']
for child in node:
self.flatten_node(child, stylizer, names, styles, pseudo_styles,
psize, item_id)
def flatten_node(self, node, stylizer, names, styles, pseudo_styles, psize, item_id):
if not isinstance(node.tag, basestring) or namespace(node.tag) != XHTML_NS:
return
tag = barename(node.tag)
style = stylizer.style(node)
cssdict = style.cssdict()
try:
font_size = style["font-size"]
except:
font_size = self.sbase if self.sbase is not None else self.context.source.fbase
if "align" in node.attrib:
if tag != "img":
cssdict["text-align"] = node.attrib["align"]
else:
val = node.attrib["align"]
if val in ("middle", "bottom", "top"):
cssdict["vertical-align"] = val
elif val in ("left", "right"):
cssdict["float"] = val
del node.attrib["align"]
if node.tag == XHTML("font"):
tags = [
"descendant::h:%s" % x
for x in ("p", "div", "table", "h1", "h2", "h3", "h4", "h5", "h6", "ol", "ul", "dl", "blockquote")
]
tag = "div" if XPath("|".join(tags))(node) else "span"
node.tag = XHTML(tag)
if "size" in node.attrib:
def force_int(raw):
return int(re.search(r"([0-9+-]+)", raw).group(1))
size = node.attrib["size"].strip()
if size:
fnums = self.context.source.fnums
if size[0] in ("+", "-"):
# Oh, the warcrimes
try:
esize = 3 + force_int(size)
except:
esize = 3
if esize < 1:
esize = 1
if esize > 7:
esize = 7
font_size = fnums[esize]
else:
try:
font_size = fnums[force_int(size)]
except:
font_size = fnums[3]
cssdict["font-size"] = "%.1fpt" % font_size
del node.attrib["size"]
if "face" in node.attrib:
cssdict["font-family"] = node.attrib["face"]
del node.attrib["face"]
if "color" in node.attrib:
try:
cssdict["color"] = Property("color", node.attrib["color"]).value
except (ValueError, SyntaxErr):
pass
del node.attrib["color"]
if "bgcolor" in node.attrib:
try:
cssdict["background-color"] = Property("background-color", node.attrib["bgcolor"]).value
except (ValueError, SyntaxErr):
pass
del node.attrib["bgcolor"]
if cssdict.get("font-weight", "").lower() == "medium":
cssdict["font-weight"] = "normal" # ADE chokes on font-weight medium
fsize = font_size
is_drop_cap = (
cssdict.get("float", None) == "left"
and "font-size" in cssdict
and len(node) == 0
and node.text
and len(node.text) == 1
)
is_drop_cap = is_drop_cap or (
# The docx input plugin generates drop caps that look like this
len(node) == 1
and not node.text
and len(node[0]) == 0
and node[0].text
and not node[0].tail
and len(node[0].text) == 1
and "line-height" in cssdict
and "font-size" in cssdict
)
if not self.context.disable_font_rescaling and not is_drop_cap:
_sbase = self.sbase if self.sbase is not None else self.context.source.fbase
dyn_rescale = dynamic_rescale_factor(node)
if dyn_rescale is not None:
fsize = self.fmap[_sbase]
fsize *= dyn_rescale
cssdict["font-size"] = "%0.5fem" % (fsize / psize)
psize = fsize
elif "font-size" in cssdict or tag == "body":
fsize = self.fmap[font_size]
try:
cssdict["font-size"] = "%0.5fem" % (fsize / psize)
except ZeroDivisionError:
cssdict["font-size"] = "%.1fpt" % fsize
psize = fsize
try:
minlh = self.context.minimum_line_height / 100.0
if not is_drop_cap and style["line-height"] < minlh * fsize:
cssdict["line-height"] = str(minlh)
except:
self.oeb.logger.exception("Failed to set minimum line-height")
if cssdict:
for x in self.filter_css:
cssdict.pop(x, None)
if cssdict:
if self.lineh and self.fbase and tag != "body":
self.clean_edges(cssdict, style, psize)
if "display" in cssdict and cssdict["display"] == "in-line":
cssdict["display"] = "inline"
if self.unfloat and "float" in cssdict and cssdict.get("display", "none") != "none":
del cssdict["display"]
if self.untable and "display" in cssdict and cssdict["display"].startswith("table"):
display = cssdict["display"]
if display == "table-cell":
cssdict["display"] = "inline"
else:
cssdict["display"] = "block"
if "vertical-align" in cssdict and cssdict["vertical-align"] == "sup":
cssdict["vertical-align"] = "super"
if self.lineh and "line-height" not in cssdict:
lineh = self.lineh / psize
cssdict["line-height"] = "%0.5fem" % lineh
if (self.context.remove_paragraph_spacing or self.context.insert_blank_line) and tag in ("p", "div"):
if item_id != "calibre_jacket" or self.context.output_profile.name == "Kindle":
for prop in ("margin", "padding", "border"):
for edge in ("top", "bottom"):
cssdict["%s-%s" % (prop, edge)] = "0pt"
if self.context.insert_blank_line:
cssdict["margin-top"] = cssdict["margin-bottom"] = "%fem" % self.context.insert_blank_line_size
indent_size = self.context.remove_paragraph_spacing_indent_size
keep_indents = indent_size < 0.0
if (
self.context.remove_paragraph_spacing
and not keep_indents
and cssdict.get("text-align", None) not in ("center", "right")
):
cssdict["text-indent"] = "%1.1fem" % indent_size
pseudo_classes = style.pseudo_classes(self.filter_css)
if cssdict or pseudo_classes:
keep_classes = set()
if cssdict:
items = sorted(cssdict.items())
css = u";\n".join(u"%s: %s" % (key, val) for key, val in items)
classes = node.get("class", "").strip() or "calibre"
klass = ascii_text(STRIPNUM.sub("", classes.split()[0].replace("_", "")))
if css in styles:
match = styles[css]
else:
match = klass + str(names[klass] or "")
styles[css] = match
names[klass] += 1
node.attrib["class"] = match
keep_classes.add(match)
for psel, cssdict in pseudo_classes.iteritems():
items = sorted(cssdict.iteritems())
css = u";\n".join(u"%s: %s" % (key, val) for key, val in items)
pstyles = pseudo_styles[psel]
if css in pstyles:
match = pstyles[css]
else:
# We have to use a different class for each psel as
# otherwise you can have incorrect styles for a situation
# like: a:hover { color: red } a:link { color: blue } a.x:hover { color: green }
# If the pcalibre class for a:hover and a:link is the same,
# then the class attribute for a.x tags will contain both
# that class and the class for a.x:hover, which is wrong.
klass = "pcalibre"
match = klass + str(names[klass] or "")
pstyles[css] = match
names[klass] += 1
keep_classes.add(match)
node.attrib["class"] = " ".join(keep_classes)
elif "class" in node.attrib:
del node.attrib["class"]
if "style" in node.attrib:
del node.attrib["style"]
for child in node:
self.flatten_node(child, stylizer, names, styles, pseudo_styles, psize, item_id)
def dump_text(self, elem, stylizer, page, tag_stack=[]):
from calibre.ebooks.oeb.base import XHTML_NS, barename, namespace
if not isinstance(elem.tag, basestring) or namespace(elem.tag) != XHTML_NS:
p = elem.getparent()
if p is not None and isinstance(p.tag, basestring) and namespace(p.tag) == XHTML_NS \
and elem.tail:
return [elem.tail]
return [u'']
text = [u'']
style = stylizer.style(elem)
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 [u'']
tag = barename(elem.tag)
tag_count = 0
# Process tags that need special processing and that do not have inner
# text. Usually these require an argument
if tag in IMAGE_TAGS:
if elem.attrib.get('src', None):
if page.abshref(elem.attrib['src']) not in self.name_map.keys():
self.name_map[page.abshref(elem.attrib['src'])] = unique_name('%s' % len(self.name_map.keys()), self.name_map.keys())
text.append('<IMG SRC="%s">' % self.name_map[page.abshref(elem.attrib['src'])])
rb_tag = tag.upper() if tag in TAGS else None
if rb_tag:
tag_count += 1
text.append('<%s>' % rb_tag)
tag_stack.append(rb_tag)
# Anchors links
if tag in LINK_TAGS:
href = elem.get('href')
if href:
href = page.abshref(href)
if '://' not in href:
if '#' not in href:
href += '#'
if href not in self.link_hrefs.keys():
self.link_hrefs[href] = 'calibre_link-%s' % len(self.link_hrefs.keys())
href = self.link_hrefs[href]
text.append('<A HREF="#%s">' % href)
tag_count += 1
tag_stack.append('A')
# Anchor ids
id_name = elem.get('id')
if id_name:
text.append(self.get_anchor(page, id_name))
# Processes style information
for s in STYLES:
style_tag = s[1].get(style[s[0]], None)
if style_tag:
style_tag = style_tag.upper()
tag_count += 1
text.append('<%s>' % style_tag)
tag_stack.append(style_tag)
# Proccess tags that contain text.
if hasattr(elem, 'text') and elem.text:
text.append(prepare_string_for_xml(elem.text))
for item in elem:
text += self.dump_text(item, stylizer, page, tag_stack)
close_tag_list = []
for i in range(0, tag_count):
close_tag_list.insert(0, tag_stack.pop())
text += self.close_tags(close_tag_list)
if hasattr(elem, 'tail') and elem.tail:
text.append(prepare_string_for_xml(elem.tail))
return text
def flatten_node(self, node, stylizer, names, styles, pseudo_styles, psize, item_id):
if not isinstance(node.tag, string_or_bytes) \
or namespace(node.tag) != XHTML_NS:
return
tag = barename(node.tag)
style = stylizer.style(node)
cssdict = style.cssdict()
try:
font_size = style['font-size']
except:
font_size = self.sbase if self.sbase is not None else \
self.context.source.fbase
if tag == 'body' and isinstance(font_size, numbers.Number):
stylizer.body_font_size = font_size
if 'align' in node.attrib:
if tag != 'img':
cssdict['text-align'] = node.attrib['align']
if cssdict['text-align'] == 'center':
# align=center causes tables to be center aligned,
# which text-align does not. And the ever trustworthy Word
# uses this construct in its HTML output. See
# https://bugs.launchpad.net/bugs/1569583
if tag == 'table':
if 'margin-left' not in cssdict and 'margin-right' not in cssdict:
cssdict['margin-left'] = cssdict['margin-right'] = 'auto'
else:
for table in node.iterchildren(XHTML("table")):
ts = stylizer.style(table)
if ts.get('margin-left') is None and ts.get('margin-right') is None:
ts.set('margin-left', 'auto')
ts.set('margin-right', 'auto')
else:
val = node.attrib['align']
if val in ('middle', 'bottom', 'top'):
cssdict['vertical-align'] = val
elif val in ('left', 'right'):
cssdict['float'] = val
del node.attrib['align']
if 'valign' in node.attrib and tag == 'td':
if cssdict.get('vertical-align') == 'inherit':
cssdict['vertical-align'] = node.attrib['valign']
del node.attrib['valign']
if node.tag == XHTML('font'):
tags = ['descendant::h:%s'%x for x in ('p', 'div', 'table', 'h1',
'h2', 'h3', 'h4', 'h5', 'h6', 'ol', 'ul', 'dl', 'blockquote')]
tag = 'div' if XPath('|'.join(tags))(node) else 'span'
node.tag = XHTML(tag)
if 'size' in node.attrib:
def force_int(raw):
return int(re.search(r'([0-9+-]+)', raw).group(1))
size = node.attrib['size'].strip()
if size:
fnums = self.context.source.fnums
if size[0] in ('+', '-'):
# Oh, the warcrimes
try:
esize = 3 + force_int(size)
except:
esize = 3
if esize < 1:
esize = 1
if esize > 7:
esize = 7
font_size = fnums[esize]
else:
try:
font_size = fnums[force_int(size)]
except:
font_size = fnums[3]
cssdict['font-size'] = '%.1fpt'%font_size
del node.attrib['size']
if 'face' in node.attrib:
cssdict['font-family'] = node.attrib['face']
del node.attrib['face']
if 'color' in node.attrib:
try:
cssdict['color'] = Property('color', node.attrib['color']).value
except (ValueError, SyntaxErr):
pass
del node.attrib['color']
if 'bgcolor' in node.attrib:
try:
cssdict['background-color'] = Property('background-color', node.attrib['bgcolor']).value
except (ValueError, SyntaxErr):
pass
del node.attrib['bgcolor']
if tag == 'ol' and 'type' in node.attrib:
del node.attrib['type']
if cssdict.get('font-weight', '').lower() == 'medium':
cssdict['font-weight'] = 'normal' # ADE chokes on font-weight medium
fsize = font_size
is_drop_cap = (cssdict.get('float', None) == 'left' and 'font-size' in cssdict and len(node) == 0 and node.text and (
len(node.text) == 1 or (len(node.text) == 2 and 0x2000 <= ord(node.text[0]) <= 0x206f)))
# Detect drop caps generated by the docx input plugin
if node.tag and node.tag.endswith('}p') and len(node) == 0 and node.text and len(node.text.strip()) == 1 and \
not node.tail and 'line-height' in cssdict and 'font-size' in cssdict:
dp = node.getparent()
if dp.tag and dp.tag.endswith('}div') and len(dp) == 1 and not dp.text:
if stylizer.style(dp).cssdict().get('float', None) == 'left':
is_drop_cap = True
if not self.context.disable_font_rescaling and not is_drop_cap:
_sbase = self.sbase if self.sbase is not None else \
self.context.source.fbase
dyn_rescale = dynamic_rescale_factor(node)
if dyn_rescale is not None:
fsize = self.fmap[_sbase]
fsize *= dyn_rescale
cssdict['font-size'] = '%0.5fem'%(fsize/psize)
psize = fsize
elif 'font-size' in cssdict or tag == 'body':
fsize = self.fmap[font_size]
try:
cssdict['font-size'] = "%0.5fem" % (fsize / psize)
except ZeroDivisionError:
cssdict['font-size'] = '%.1fpt'%fsize
psize = fsize
try:
minlh = self.context.minimum_line_height / 100.
if not is_drop_cap and style['line-height'] < minlh * fsize:
cssdict['line-height'] = str(minlh)
except:
self.oeb.logger.exception('Failed to set minimum line-height')
if cssdict:
for x in self.filter_css:
popval = cssdict.pop(x, None)
if self.body_font_family and popval and x == 'font-family' \
and popval.partition(',')[0][1:-1] == self.body_font_family.partition(',')[0][1:-1]:
cssdict[x] = popval
if cssdict:
if self.lineh and self.fbase and tag != 'body':
self.clean_edges(cssdict, style, psize)
if 'display' in cssdict and cssdict['display'] == 'in-line':
cssdict['display'] = 'inline'
if self.unfloat and 'float' in cssdict \
and cssdict.get('display', 'none') != 'none':
del cssdict['display']
if self.untable and 'display' in cssdict \
and cssdict['display'].startswith('table'):
display = cssdict['display']
if display == 'table-cell':
cssdict['display'] = 'inline'
else:
cssdict['display'] = 'block'
if 'vertical-align' in cssdict \
and cssdict['vertical-align'] == 'sup':
cssdict['vertical-align'] = 'super'
if self.lineh and 'line-height' not in cssdict:
lineh = self.lineh / psize
cssdict['line-height'] = "%0.5fem" % lineh
if (self.context.remove_paragraph_spacing or self.context.insert_blank_line) and tag in ('p', 'div'):
if item_id != 'calibre_jacket' or self.context.output_profile.name == 'Kindle':
for prop in ('margin', 'padding', 'border'):
for edge in ('top', 'bottom'):
cssdict['%s-%s'%(prop, edge)] = '0pt'
if self.context.insert_blank_line:
cssdict['margin-top'] = cssdict['margin-bottom'] = \
'%fem'%self.context.insert_blank_line_size
indent_size = self.context.remove_paragraph_spacing_indent_size
keep_indents = indent_size < 0.0
if (self.context.remove_paragraph_spacing and not keep_indents and cssdict.get('text-align', None) not in ('center', 'right')):
cssdict['text-indent'] = "%1.1fem" % indent_size
pseudo_classes = style.pseudo_classes(self.filter_css)
if cssdict or pseudo_classes:
keep_classes = set()
if cssdict:
items = sorted(iteritems(cssdict))
css = u';\n'.join(u'%s: %s' % (key, val) for key, val in items)
classes = node.get('class', '').strip() or 'calibre'
# lower() because otherwise if the document uses the same class
# name with different case, both cases will apply, leading
# to incorrect results.
klass = ascii_text(STRIPNUM.sub('', classes.split()[0])).lower().strip().replace(' ', '_')
if css in styles:
match = styles[css]
else:
match = klass + str(names[klass] or '')
styles[css] = match
names[klass] += 1
node.attrib['class'] = match
keep_classes.add(match)
for psel, cssdict in iteritems(pseudo_classes):
items = sorted(iteritems(cssdict))
css = u';\n'.join(u'%s: %s' % (key, val) for key, val in items)
pstyles = pseudo_styles[psel]
if css in pstyles:
match = pstyles[css]
else:
# We have to use a different class for each psel as
# otherwise you can have incorrect styles for a situation
# like: a:hover { color: red } a:link { color: blue } a.x:hover { color: green }
# If the pcalibre class for a:hover and a:link is the same,
# then the class attribute for a.x tags will contain both
# that class and the class for a.x:hover, which is wrong.
klass = 'pcalibre'
match = klass + str(names[klass] or '')
pstyles[css] = match
names[klass] += 1
keep_classes.add(match)
node.attrib['class'] = ' '.join(keep_classes)
elif 'class' in node.attrib:
del node.attrib['class']
if 'style' in node.attrib:
del node.attrib['style']
for child in node:
self.flatten_node(child, stylizer, names, styles, pseudo_styles, psize, item_id)
def flatten_node(self,
node,
stylizer,
names,
styles,
pseudo_styles,
psize,
item_id,
recurse=True):
if not isinstance(node.tag, string_or_bytes) \
or namespace(node.tag) != XHTML_NS:
return
tag = barename(node.tag)
style = stylizer.style(node)
cssdict = style.cssdict()
try:
font_size = style['font-size']
except:
font_size = self.sbase if self.sbase is not None else \
self.context.source.fbase
if tag == 'body' and isinstance(font_size, numbers.Number):
stylizer.body_font_size = font_size
if 'align' in node.attrib:
if tag != 'img':
cssdict['text-align'] = node.attrib['align']
if cssdict['text-align'] == 'center':
# align=center causes tables to be center aligned,
# which text-align does not. And the ever trustworthy Word
# uses this construct in its HTML output. See
# https://bugs.launchpad.net/bugs/1569583
if tag == 'table':
if 'margin-left' not in cssdict and 'margin-right' not in cssdict:
cssdict['margin-left'] = cssdict[
'margin-right'] = 'auto'
else:
for table in node.iterchildren(XHTML("table")):
ts = stylizer.style(table)
if ts.get('margin-left') is None and ts.get(
'margin-right') is None:
ts.set('margin-left', 'auto')
ts.set('margin-right', 'auto')
else:
val = node.attrib['align']
if val in ('middle', 'bottom', 'top'):
cssdict['vertical-align'] = val
elif val in ('left', 'right'):
cssdict['float'] = val
del node.attrib['align']
if 'valign' in node.attrib and tag == 'td':
if cssdict.get('vertical-align') == 'inherit':
cssdict['vertical-align'] = node.attrib['valign']
del node.attrib['valign']
if node.tag == XHTML('font'):
tags = [
'descendant::h:%s' % x
for x in ('p', 'div', 'table', 'h1', 'h2', 'h3', 'h4', 'h5',
'h6', 'ol', 'ul', 'dl', 'blockquote')
]
tag = 'div' if XPath('|'.join(tags))(node) else 'span'
node.tag = XHTML(tag)
if 'size' in node.attrib:
def force_int(raw):
return int(re.search(r'([0-9+-]+)', raw).group(1))
size = node.attrib['size'].strip()
if size:
fnums = self.context.source.fnums
if size[0] in ('+', '-'):
# Oh, the warcrimes
try:
esize = 3 + force_int(size)
except:
esize = 3
if esize < 1:
esize = 1
if esize > 7:
esize = 7
font_size = fnums[esize]
else:
try:
font_size = fnums[force_int(size)]
except:
font_size = fnums[3]
cssdict['font-size'] = '%.1fpt' % font_size
del node.attrib['size']
if 'face' in node.attrib:
cssdict['font-family'] = node.attrib['face']
del node.attrib['face']
if 'color' in node.attrib:
try:
cssdict['color'] = Property('color',
node.attrib['color']).value
except (ValueError, SyntaxErr):
pass
del node.attrib['color']
if 'bgcolor' in node.attrib:
try:
cssdict['background-color'] = Property(
'background-color', node.attrib['bgcolor']).value
except (ValueError, SyntaxErr):
pass
del node.attrib['bgcolor']
if tag == 'ol' and 'type' in node.attrib:
del node.attrib['type']
if cssdict.get('font-weight', '').lower() == 'medium':
cssdict[
'font-weight'] = 'normal' # ADE chokes on font-weight medium
fsize = font_size
is_drop_cap = (
cssdict.get('float', None) == 'left' and 'font-size' in cssdict
and len(node) == 0 and node.text and
(len(node.text) == 1 or
(len(node.text) == 2 and 0x2000 <= ord(node.text[0]) <= 0x206f)))
# Detect drop caps generated by the docx input plugin
if node.tag and node.tag.endswith('}p') and len(node) == 0 and node.text and len(node.text.strip()) == 1 and \
not node.tail and 'line-height' in cssdict and 'font-size' in cssdict:
dp = node.getparent()
if dp.tag and dp.tag.endswith('}div') and len(
dp) == 1 and not dp.text:
if stylizer.style(dp).cssdict().get('float', None) == 'left':
is_drop_cap = True
if not self.context.disable_font_rescaling and not is_drop_cap:
_sbase = self.sbase if self.sbase is not None else \
self.context.source.fbase
dyn_rescale = node.attrib.pop('data-calibre-rescale', None)
if dyn_rescale is not None:
try:
dyn_rescale = float(dyn_rescale) / 100
except Exception:
dyn_rescale = 1
fsize = self.fmap[_sbase]
fsize *= dyn_rescale
cssdict['font-size'] = '%0.5fem' % (fsize / psize)
psize = fsize
elif 'font-size' in cssdict or tag == 'body':
fsize = self.fmap[font_size]
try:
cssdict['font-size'] = "%0.5fem" % (fsize / psize)
except ZeroDivisionError:
cssdict['font-size'] = '%.1fpt' % fsize
psize = fsize
try:
minlh = self.context.minimum_line_height / 100.
slh = style['line-height']
if not is_drop_cap and isinstance(
slh, numbers.Number) and slh < minlh * fsize:
cssdict['line-height'] = unicode_type(minlh)
except Exception:
self.oeb.logger.exception('Failed to set minimum line-height')
if cssdict:
for x in self.filter_css:
popval = cssdict.pop(x, None)
if self.body_font_family and popval and x == 'font-family' \
and popval.partition(',')[0][1:-1] == self.body_font_family.partition(',')[0][1:-1]:
cssdict[x] = popval
if cssdict:
if self.lineh and self.fbase and tag != 'body':
self.clean_edges(cssdict, style, psize)
if 'display' in cssdict and cssdict['display'] == 'in-line':
cssdict['display'] = 'inline'
if self.unfloat and 'float' in cssdict \
and cssdict.get('display', 'none') != 'none':
del cssdict['display']
if self.untable and 'display' in cssdict \
and cssdict['display'].startswith('table'):
display = cssdict['display']
if display == 'table-cell':
cssdict['display'] = 'inline'
else:
cssdict['display'] = 'block'
if 'vertical-align' in cssdict \
and cssdict['vertical-align'] == 'sup':
cssdict['vertical-align'] = 'super'
if self.lineh and 'line-height' not in cssdict:
lineh = self.lineh / psize
cssdict['line-height'] = "%0.5fem" % lineh
if (self.context.remove_paragraph_spacing
or self.context.insert_blank_line) and tag in ('p', 'div'):
if item_id != 'calibre_jacket' or self.context.output_profile.name == 'Kindle':
for prop in ('margin', 'padding', 'border'):
for edge in ('top', 'bottom'):
cssdict['%s-%s' % (prop, edge)] = '0pt'
if self.context.insert_blank_line:
cssdict['margin-top'] = cssdict['margin-bottom'] = \
'%fem'%self.context.insert_blank_line_size
indent_size = self.context.remove_paragraph_spacing_indent_size
keep_indents = indent_size < 0.0
if (self.context.remove_paragraph_spacing and not keep_indents
and cssdict.get('text-align',
None) not in ('center', 'right')):
cssdict['text-indent'] = "%1.1fem" % indent_size
pseudo_classes = style.pseudo_classes(self.filter_css)
if cssdict or pseudo_classes:
keep_classes = set()
if cssdict:
items = sorted(iteritems(cssdict))
css = ';\n'.join(u'%s: %s' % (key, val) for key, val in items)
classes = node.get('class', '').strip() or 'calibre'
classes_list = classes.split()
# lower() because otherwise if the document uses the same class
# name with different case, both cases will apply, leading
# to incorrect results.
klass = ascii_text(STRIPNUM.sub(
'', classes_list[0])).lower().strip().replace(' ', '_')
if css in styles:
match = styles[css]
else:
match = klass + unicode_type(names[klass] or '')
styles[css] = match
names[klass] += 1
node.attrib['class'] = match
keep_classes.add(match)
for psel, cssdict in iteritems(pseudo_classes):
items = sorted(iteritems(cssdict))
css = ';\n'.join('%s: %s' % (key, val) for key, val in items)
pstyles = pseudo_styles[psel]
if css in pstyles:
match = pstyles[css]
else:
# We have to use a different class for each psel as
# otherwise you can have incorrect styles for a situation
# like: a:hover { color: red } a:link { color: blue } a.x:hover { color: green }
# If the pcalibre class for a:hover and a:link is the same,
# then the class attribute for a.x tags will contain both
# that class and the class for a.x:hover, which is wrong.
klass = 'pcalibre'
match = klass + unicode_type(names[klass] or '')
pstyles[css] = match
names[klass] += 1
keep_classes.add(match)
node.attrib['class'] = ' '.join(keep_classes)
elif 'class' in node.attrib:
del node.attrib['class']
if 'style' in node.attrib:
del node.attrib['style']
if recurse:
for child in node:
self.flatten_node(child, stylizer, names, styles,
pseudo_styles, psize, item_id)