def smarten_punctuation(container, report):
from ebook_converter.ebooks.conversion.preprocess import smarten_punctuation
smartened = False
for path in container.spine_items:
name = container.abspath_to_name(path)
changed = False
with container.open(name, 'r+b') as f:
html = container.decode(f.read())
newhtml = smarten_punctuation(html, container.log)
if newhtml != html:
changed = True
report('Smartened punctuation in: %s' % name)
newhtml = strip_encoding_declarations(newhtml)
f.seek(0)
f.truncate()
f.write(codecs.BOM_UTF8 + newhtml.encode('utf-8'))
if changed:
# Add an encoding declaration (it will be added automatically when
# serialized)
root = container.parsed(name)
for m in root.xpath(
'descendant::*[local-name()="meta" and @http-equiv]'):
m.getparent().remove(m)
container.dirty(name)
smartened = True
if not smartened:
report('No punctuation that could be smartened found')
return smartened
def parse_html(markup):
if isinstance(markup, str):
markup = chardet.strip_encoding_declarations(markup)
markup = chardet.substitute_entites(markup)
else:
markup = chardet.xml_to_unicode(markup,
strip_encoding_pats=True,
resolve_entities=True)[0]
markup = cleantext.clean_xml_chars(markup)
return html5_soup.parse(markup, return_root=False)
def parse(raw,
decoder=None,
log=None,
line_numbers=True,
linenumber_attribute=None,
replace_entities=True,
force_html5_parse=False):
if isinstance(raw, bytes):
raw = xml_to_unicode(raw)[0] if decoder is None else decoder(raw)
raw = handle_private_entities(raw)
if replace_entities:
# Handle �
raw = entities.xml_replace_entities(raw).replace('\0', '')
raw = raw.replace('\r\n', '\n').replace('\r', '\n')
# Remove any preamble before the opening html tag as it can cause problems,
# especially doctypes, preserve the original linenumbers by inserting
# newlines at the start
pre = raw[:2048]
for match in re.finditer(r'<\s*html', pre, flags=re.I):
newlines = raw.count('\n', 0, match.start())
raw = ('\n' * newlines) + raw[match.start():]
break
raw = strip_encoding_declarations(raw,
limit=10 * 1024,
preserve_newlines=True)
if force_html5_parse:
return parse_html5(raw,
log=log,
line_numbers=line_numbers,
linenumber_attribute=linenumber_attribute,
replace_entities=False,
fix_newlines=False)
try:
ans = etree.fromstring(raw)
if ans.tag != '{%s}html' % const.XHTML_NS:
raise ValueError('Root tag is not <html> in the XHTML namespace')
if linenumber_attribute:
for elem in ans.iter(etree.element):
if elem.sourceline is not None:
elem.set(linenumber_attribute, str(elem.sourceline))
return ans
except Exception:
if log is not None:
log.exception('Failed to parse as XML, parsing as tag soup')
return parse_html5(raw,
log=log,
line_numbers=line_numbers,
linenumber_attribute=linenumber_attribute,
replace_entities=False,
fix_newlines=False)
def expand_mobi8_markup(mobi8_reader, resource_map, log):
# First update all internal links that are based on offsets
parts = update_internal_links(mobi8_reader, log)
# Remove pointless markup inserted by kindlegen
remove_kindlegen_markup(parts, mobi8_reader.aid_anchor_suffix,
mobi8_reader.linked_aids)
# Handle substitutions for the flows pieces first as they may
# be inlined into the xhtml text
flows = update_flow_links(mobi8_reader, resource_map, log)
# Insert inline flows into the markup
insert_flows_into_markup(parts, flows, mobi8_reader, log)
# Insert raster images into markup
insert_images_into_markup(parts, resource_map, log)
# Perform general markup cleanups
upshift_markup(parts)
# Update the parts and flows stored in the reader
mobi8_reader.parts = parts
mobi8_reader.flows = flows
# write out the parts and file flows
os.mkdir('text') # directory containing all parts
spine = []
for i, part in enumerate(parts):
pi = mobi8_reader.partinfo[i]
with open(os.path.join(pi.type, pi.filename), 'wb') as f:
part = strip_encoding_declarations(part)
part = part.replace('<head>', '<head><meta charset="UTF-8"/>', 1)
f.write(part.encode('utf-8'))
spine.append(f.name)
for i, flow in enumerate(flows):
fi = mobi8_reader.flowinfo[i]
if fi.format == 'file':
if not os.path.exists(fi.dir):
os.mkdir(fi.dir)
with open(os.path.join(fi.dir, fi.fname), 'wb') as f:
f.write(flow.encode('utf-8'))
return spine
def parse_html(data,
log=None,
decoder=None,
preprocessor=None,
filename='<string>',
non_html_file_tags=frozenset()):
if log is None:
log = LOG
filename = force_unicode(filename, enc=filesystem_encoding)
if not isinstance(data, str):
if decoder is not None:
data = decoder(data)
else:
data = xml_to_unicode(data)[0]
data = strip_encoding_declarations(data)
# Remove DOCTYPE declaration as it messes up parsing
# In particular, it causes tostring to insert xmlns
# declarations, which messes up the coercing logic
pre = ''
idx = data.find('<html')
if idx == -1:
idx = data.find('<HTML')
has_html4_doctype = False
if idx > -1:
pre = data[:idx]
data = data[idx:]
if '<!DOCTYPE' in pre: # Handle user defined entities
# kindlegen produces invalid xhtml with uppercase attribute names
# if fed HTML 4 with uppercase attribute names, so try to detect
# and compensate for that.
has_html4_doctype = re.search(r'<!DOCTYPE\s+[^>]+HTML\s+4.0[^.]+>',
pre) is not None
# Process private entities
user_entities = {}
for match in re.finditer(r'<!ENTITY\s+(\S+)\s+([^>]+)', pre):
val = match.group(2)
if val.startswith('"') and val.endswith('"'):
val = val[1:-1]
user_entities[match.group(1)] = val
if user_entities:
pat = re.compile(r'&(%s);' %
('|'.join(list(user_entities.keys()))))
data = pat.sub(lambda m: user_entities[m.group(1)], data)
if preprocessor is not None:
data = preprocessor(data)
# There could be null bytes in data if it had � entities in it
data = data.replace('\0', '')
data = raw = clean_word_doc(data, log)
# Try with more & more drastic measures to parse
try:
data = etree.fromstring(data)
check_for_html5(pre, data)
except (HTML5Doc, etree.XMLSyntaxError):
log.debug('Initial parse failed, using more' ' forgiving parsers')
raw = data = xml_replace_entities(raw)
try:
data = etree.fromstring(data)
check_for_html5(pre, data)
except (HTML5Doc, etree.XMLSyntaxError):
log.debug('Parsing %s as HTML' % filename)
data = raw
try:
data = html5_parse(data)
except Exception:
log.exception(
'HTML 5 parsing failed, falling back to older parsers')
data = _html4_parse(data)
if has_html4_doctype or data.tag == 'HTML' or (
len(data) and (data[-1].get('LANG') or data[-1].get('DIR'))):
# Lower case all tag and attribute names
data.tag = data.tag.lower()
for x in data.iterdescendants():
try:
x.tag = x.tag.lower()
for key, val in tuple(x.attrib.items()):
del x.attrib[key]
key = key.lower()
x.attrib[key] = val
except:
pass
if barename(data.tag) != 'html':
if barename(data.tag) in non_html_file_tags:
raise NotHTML(data.tag)
log.warn('File %r does not appear to be (X)HTML' % filename)
nroot = etree.fromstring('<html></html>')
has_body = False
for child in list(data):
if isinstance(child.tag,
(str, bytes)) and barename(child.tag) == 'body':
has_body = True
break
parent = nroot
if not has_body:
log.warn('File %r appears to be a HTML fragment' % filename)
nroot = etree.fromstring('<html><body/></html>')
parent = nroot[0]
for child in list(data.iter()):
oparent = child.getparent()
if oparent is not None:
oparent.remove(child)
parent.append(child)
data = nroot
# Force into the XHTML namespace
if not namespace(data.tag):
log.warn('Forcing', filename, 'into XHTML namespace')
data.attrib['xmlns'] = const.XHTML_NS
data = etree.tostring(data, encoding='unicode')
try:
data = etree.fromstring(data)
except:
data = data.replace(':=', '=').replace(':>', '>')
data = data.replace('<http:/>', '')
try:
data = etree.fromstring(data)
except etree.XMLSyntaxError:
log.warn('Stripping comments from %s' % filename)
data = re.compile(r'<!--.*?-->', re.DOTALL).sub('', data)
data = data.replace(
"<?xml version='1.0' encoding='utf-8'?><o:p></o:p>", '')
data = data.replace("<?xml version='1.0' encoding='utf-8'??>",
'')
try:
data = etree.fromstring(data)
except etree.XMLSyntaxError:
log.warn('Stripping meta tags from %s' % filename)
data = re.sub(r'<meta\s+[^>]+?>', '', data)
data = etree.fromstring(data)
elif namespace(data.tag) != const.XHTML_NS:
# OEB_DOC_NS, but possibly others
ns = namespace(data.tag)
attrib = dict(data.attrib)
nroot = etree.Element(XHTML('html'),
nsmap={None: const.XHTML_NS},
attrib=attrib)
for elem in data.iterdescendants():
if isinstance(elem.tag, (str, bytes)) and \
namespace(elem.tag) == ns:
elem.tag = XHTML(barename(elem.tag))
for elem in data:
nroot.append(elem)
data = nroot
# Remove non default prefixes referring to the XHTML namespace
data = ensure_namespace_prefixes(data, {None: const.XHTML_NS})
data = merge_multiple_html_heads_and_bodies(data, log)
# Ensure has a <head/>
head = xpath(data, '/h:html/h:head')
head = head[0] if head else None
if head is None:
log.warn('File %s missing <head/> element' % filename)
head = etree.Element(XHTML('head'))
data.insert(0, head)
title = etree.SubElement(head, XHTML('title'))
title.text = 'Unknown'
elif not xpath(data, '/h:html/h:head/h:title'):
title = etree.SubElement(head, XHTML('title'))
title.text = 'Unknown'
# Ensure <title> is not empty
title = xpath(data, '/h:html/h:head/h:title')[0]
if not title.text or not title.text.strip():
title.text = 'Unknown'
# Remove any encoding-specifying <meta/> elements
for meta in META_XP(data):
meta.getparent().remove(meta)
meta = etree.SubElement(head,
XHTML('meta'),
attrib={'http-equiv': 'Content-Type'})
meta.set('content',
'text/html; charset=utf-8') # Ensure content is second attribute
# Ensure has a <body/>
if not xpath(data, '/h:html/h:body'):
body = xpath(data, '//h:body')
if body:
body = body[0]
body.getparent().remove(body)
data.append(body)
else:
log.warn('File %s missing <body/> element' % filename)
etree.SubElement(data, XHTML('body'))
# Remove microsoft office markup
r = [
x for x in data.iterdescendants(etree.Element)
if 'microsoft-com' in x.tag
]
for x in r:
x.tag = XHTML('span')
def remove_elem(a):
p = a.getparent()
idx = p.index(a) - 1
p.remove(a)
if a.tail:
if idx < 0:
if p.text is None:
p.text = ''
p.text += a.tail
else:
if p[idx].tail is None:
p[idx].tail = ''
p[idx].tail += a.tail
# Remove hyperlinks with no content as they cause rendering
# artifacts in browser based renderers
# Also remove empty <b>, <u> and <i> tags
for a in xpath(data, '//h:a[@href]|//h:i|//h:b|//h:u'):
if a.get('id', None) is None and a.get('name', None) is None \
and len(a) == 0 and not a.text:
remove_elem(a)
# Convert <br>s with content into paragraphs as ADE can't handle
# them
for br in xpath(data, '//h:br'):
if len(br) > 0 or br.text:
br.tag = XHTML('div')
# Remove any stray text in the <head> section and format it nicely
data.text = '\n '
head = xpath(data, '//h:head')
if head:
head = head[0]
head.text = '\n '
head.tail = '\n '
for child in head:
child.tail = '\n '
child.tail = '\n '
return data
def generate_html(comments):
args = {
'author':
author,
'comments':
comments,
'css':
css,
'footer':
'',
'pubdate':
pubdate,
'pubdate_label':
'Published',
'publisher':
publisher,
'rating':
rating,
'rating_label':
'Rating',
'searchable_tags':
' '.join(saxutils.escape(t) + 'ttt' for t in tags.tags_list),
'series':
series,
'series_label':
'Series',
'tags':
tags,
'tags_label':
'Tags',
'title':
title,
'title_str':
title_str,
'xmlns':
const.XHTML_NS
}
for key in mi.custom_field_keys():
m = mi.get_user_metadata(key, False) or {}
try:
display_name, val = mi.format_field_extended(key)[:2]
dkey = key.replace('#', '_')
dt = m.get('datatype')
if dt == 'series':
args[dkey] = Series(mi.get(key), mi.get(key + '_index'))
elif dt == 'rating':
args[dkey] = rating_to_stars(
mi.get(key),
m.get('display', {}).get('allow_half_stars', False))
elif dt == 'comments':
val = val or ''
display = m.get('display', {})
ctype = display.get('interpret_as') or 'html'
if ctype == 'long-text':
val = ('<pre style="white-space:pre-wrap">%s</pre>' %
saxutils.escape(val))
elif ctype == 'short-text':
val = '<span>%s</span>' % saxutils.escape(val)
elif ctype == 'markdown':
val = markdown(val)
else:
val = comments_to_html(val)
args[dkey] = val
else:
args[dkey] = saxutils.escape(val)
args[dkey + '_label'] = saxutils.escape(display_name)
except Exception:
# if the val (custom column contents) is None, don't add to
# args
pass
if False:
print("Custom column values available in jacket template:")
for key in args.keys():
if key.startswith('_') and not key.endswith('_label'):
print(" %s: %s" % ('#' + key[1:], args[key]))
# Used in the comment describing use of custom columns in templates
# Don't change this unless you also change it in template.xhtml
args['_genre_label'] = args.get('_genre_label', '{_genre_label}')
args['_genre'] = args.get('_genre', '{_genre}')
formatter = SafeFormatter()
generated_html = formatter.format(template, **args)
has_data['series'] = bool(series)
has_data['tags'] = bool(tags)
has_data['rating'] = bool(rating)
has_data['pubdate'] = bool(pubdate)
return strip_encoding_declarations(generated_html)
def extract_content(self, output_dir, parse_cache):
output_dir = os.path.abspath(output_dir)
self.check_for_drm()
processed_records = self.extract_text()
if self.debug is not None:
parse_cache['calibre_raw_mobi_markup'] = self.mobi_html
self.add_anchors()
self.processed_html = self.processed_html.decode(
self.book_header.codec, 'ignore')
self.processed_html = self.processed_html.replace('</</', '</')
self.processed_html = re.sub(r'</([a-zA-Z]+)<', r'</\1><',
self.processed_html)
self.processed_html = self.processed_html.replace('\ufeff', '')
# Remove tags of the form <xyz: ...> as they can cause issues further
# along the pipeline
self.processed_html = re.sub(r'</{0,1}[a-zA-Z]+:\s+[^>]*>', '',
self.processed_html)
self.processed_html = strip_encoding_declarations(self.processed_html)
self.processed_html = re.sub(r'&(\S+?);',
entities.xml_entity_to_unicode,
self.processed_html)
image_name_map = self.extract_images(processed_records, output_dir)
self.replace_page_breaks()
self.cleanup_html()
self.log.debug('Parsing HTML...')
self.processed_html = clean_xml_chars(self.processed_html)
try:
root = html.fromstring(self.processed_html)
if len(root.xpath('//html')) > 5:
root = html.fromstring(self.processed_html
.replace('\x0c', '')
.replace('\x14', ''))
except Exception:
self.log.warning('MOBI markup appears to contain random bytes. '
'Stripping.')
self.processed_html = self.remove_random_bytes(self.processed_html)
root = html.fromstring(self.processed_html)
if root.xpath('descendant::p/descendant::p'):
from html5_parser import parse
self.log.warning('Malformed markup, parsing using html5-parser')
self.processed_html = strip_encoding_declarations(
self.processed_html)
# These trip up the html5 parser causing all content to be placed
# under the <guide> tag
self.processed_html = re.sub(r'<metadata>.+?</metadata>', '',
self.processed_html, flags=re.I)
self.processed_html = re.sub(r'<guide>.+?</guide>', '',
self.processed_html, flags=re.I)
try:
root = parse(self.processed_html, maybe_xhtml=False,
keep_doctype=False, sanitize_names=True)
except Exception:
self.log.warning('MOBI markup appears to contain random '
'bytes. Stripping.')
self.processed_html = self.remove_random_bytes(
self.processed_html)
root = parse(self.processed_html, maybe_xhtml=False,
keep_doctype=False, sanitize_names=True)
if len(root.xpath('body/descendant::*')) < 1:
# There are probably stray </html>s in the markup
self.processed_html = self.processed_html.replace('</html>',
'')
root = parse(self.processed_html, maybe_xhtml=False,
keep_doctype=False, sanitize_names=True)
if root.tag != 'html':
self.log.warning('File does not have opening <html> tag')
nroot = html.fromstring('<html><head></head><body></body></html>')
bod = nroot.find('body')
for child in list(root):
child.getparent().remove(child)
bod.append(child)
root = nroot
htmls = list(root.xpath('//html'))
if len(htmls) > 1:
self.log.warning('Markup contains multiple <html> tags, merging.')
# Merge all <head> and <body> sections
for h in htmls:
p = h.getparent()
if hasattr(p, 'remove'):
p.remove(h)
bodies, heads = root.xpath('//body'), root.xpath('//head')
for x in root:
root.remove(x)
head, body = map(root.makeelement, ('head', 'body'))
for h in heads:
for x in h:
h.remove(x)
head.append(x)
for b in bodies:
for x in b:
b.remove(x)
body.append(x)
root.append(head), root.append(body)
for x in root.xpath('//script'):
x.getparent().remove(x)
head = root.xpath('//head')
if head:
head = head[0]
else:
head = root.makeelement('head', {})
root.insert(0, head)
head.text = '\n\t'
link = head.makeelement('link', {'type': 'text/css',
'href': 'styles.css',
'rel': 'stylesheet'})
head.insert(0, link)
link.tail = '\n\t'
title = head.xpath('descendant::title')
m = head.makeelement('meta', {'http-equiv': 'Content-Type',
'content': 'text/html; charset=utf-8'})
head.insert(0, m)
if not title:
title = head.makeelement('title', {})
try:
title.text = self.book_header.title
except ValueError:
title.text = clean_ascii_chars(self.book_header.title)
title.tail = '\n\t'
head.insert(0, title)
head.text = '\n\t'
self.upshift_markup(root, image_name_map)
guides = root.xpath('//guide')
guide = guides[0] if guides else None
metadata_elems = root.xpath('//metadata')
if metadata_elems and self.book_header.exth is None:
self.read_embedded_metadata(root, metadata_elems[0], guide)
for elem in guides + metadata_elems:
elem.getparent().remove(elem)
htmlfile = os.path.join(output_dir, 'index.html')
try:
for ref in guide.xpath('descendant::reference'):
if 'href' in ref.attrib:
ref.attrib['href'] = (os.path.basename(htmlfile) +
ref.attrib['href'])
except AttributeError:
pass
def write_as_utf8(path, data):
if isinstance(data, str):
data = data.encode('utf-8')
with open(path, 'wb') as f:
f.write(data)
parse_cache[htmlfile] = root
self.htmlfile = htmlfile
ncx = io.BytesIO()
opf, ncx_manifest_entry = self.create_opf(htmlfile, guide, root)
self.created_opf_path = os.path.splitext(htmlfile)[0] + '.opf'
opf.render(open(self.created_opf_path, 'wb'), ncx,
ncx_manifest_entry=ncx_manifest_entry)
ncx = ncx.getvalue()
if ncx:
ncx_path = os.path.join(os.path.dirname(htmlfile), 'toc.ncx')
write_as_utf8(ncx_path, ncx)
css = [self.base_css_rules, '\n\n']
for cls, rule in self.tag_css_rules.items():
css.append('.%s { %s }\n\n' % (cls, rule))
write_as_utf8('styles.css', ''.join(css))
if self.book_header.exth is not None or self.embedded_mi is not None:
self.log.debug('Creating OPF...')
ncx = io.BytesIO()
opf, ncx_manifest_entry = self.create_opf(htmlfile, guide, root)
opf.render(open(os.path.splitext(htmlfile)[0] + '.opf', 'wb'), ncx,
ncx_manifest_entry)
ncx = ncx.getvalue()
if ncx:
write_as_utf8(os.path.splitext(htmlfile)[0] + '.ncx', ncx)
