def __compile_expressions(self):
SIMPLE_RPL = {
"\\\\": "\\backslash ",
"\\~": "\\~ ",
"\\;": "\\; ",
"&": "&",
"<": "<",
">": ">",
"\\~": "\\~ ",
"\\_": "\\_ ",
"\\:": "\\: ",
"\\-": "\\- ",
# turn into a generic token to eliminate special
# cases and make processing easier
"\\{": "\\ob ",
# turn into a generic token to eliminate special
# cases and make processing easier
"\\}": "\\cb ",
# put a backslash in front of to eliminate special cases and
# make processing easier
"{": "\\{",
# put a backslash in front of to eliminate special cases and
# make processing easier
"}": "\\}",
}
self.__replace_spchar = MReplace(SIMPLE_RPL)
# add ;? in case of char following \u
self.__ms_hex_exp = re.compile(r"\\\'([0-9a-fA-F]{2})")
self.__utf_exp = re.compile(r"\\u(-?\d{3,6}) ?")
self.__bin_exp = re.compile(r"(?:\\bin(-?\d{0,10})[\n ]+)[01\n]+")
# manage upr/ud situations
self.__utf_ud = re.compile(
r"\\{[\n ]?\\upr[\n ]?(?:\\{.*?\\})[\n ]?" +
r"\\{[\n ]?\\*[\n ]?\\ud[\n ]?(\\{.*?\\})[\n ]?\\}[\n ]?\\}")
# add \n in split for whole file reading
# why keep backslash whereas \is replaced before?
# remove \n from endline char
self.__splitexp = re.compile(
r"(\\[{}]|\n|\\[^\s\\{}&]+(?:[ \t\r\f\v])?)")
# this is for old RTF
self.__par_exp = re.compile(r'(\\\n+|\\ )')
# handle improper cs char-style with \* before without {
self.__cs_ast = re.compile(r'\\\*([\n ]*\\cs\d+[\n \\]+)')
# handle cw using a digit as argument and without space as delimiter
self.__cwdigit_exp = re.compile(r"(\\[a-zA-Z]+[\-0-9]+)([^0-9 \\]+)")
_mreplace = MReplace({
'–': '--',
'–': '--',
'–': '--',
'—': '---',
'—': '---',
'—': '---',
'…': '...',
'…': '...',
'…': '...',
'“': '"',
'”': '"',
'„': '"',
'″': '"',
'“': '"',
'”': '"',
'„': '"',
'″': '"',
'“':'"',
'”':'"',
'„':'"',
'″':'"',
'‘':"'",
'’':"'",
'′':"'",
'‘':"'",
'’':"'",
'′':"'",
'‘':"'",
'’':"'",
'′':"'",
})
def __call__(self, html, remove_special_chars=None,
get_preprocess_html=False):
if remove_special_chars is not None:
html = remove_special_chars.sub('', html)
html = html.replace('\0', '')
is_pdftohtml = self.is_pdftohtml(html)
if self.is_baen(html):
rules = []
elif self.is_book_designer(html):
rules = self.BOOK_DESIGNER
elif is_pdftohtml:
rules = self.PDFTOHTML
else:
rules = []
start_rules = []
if not getattr(self.extra_opts, 'keep_ligatures', False):
html = _ligpat.sub(lambda m:LIGATURES[m.group()], html)
user_sr_rules = {}
# Function for processing search and replace
def do_search_replace(search_pattern, replace_txt):
from calibre.ebooks.conversion.search_replace import compile_regular_expression
try:
search_re = compile_regular_expression(search_pattern)
if not replace_txt:
replace_txt = ''
rules.insert(0, (search_re, replace_txt))
user_sr_rules[(search_re, replace_txt)] = search_pattern
except Exception as e:
self.log.error('Failed to parse %r regexp because %s' %
(search, as_unicode(e)))
# search / replace using the sr?_search / sr?_replace options
for i in range(1, 4):
search, replace = 'sr%d_search'%i, 'sr%d_replace'%i
search_pattern = getattr(self.extra_opts, search, '')
replace_txt = getattr(self.extra_opts, replace, '')
if search_pattern:
do_search_replace(search_pattern, replace_txt)
# multi-search / replace using the search_replace option
search_replace = getattr(self.extra_opts, 'search_replace', None)
if search_replace:
search_replace = json.loads(search_replace)
for search_pattern, replace_txt in reversed(search_replace):
do_search_replace(search_pattern, replace_txt)
end_rules = []
# delete soft hyphens - moved here so it's executed after header/footer removal
if is_pdftohtml:
# unwrap/delete soft hyphens
end_rules.append((re.compile(u'[](</p>\s*<p>\s*)+\s*(?=[[a-z\d])'), lambda match: ''))
# unwrap/delete soft hyphens with formatting
end_rules.append((re.compile(u'[]\s*(</(i|u|b)>)+(</p>\s*<p>\s*)+\s*(<(i|u|b)>)+\s*(?=[[a-z\d])'), lambda match: ''))
length = -1
if getattr(self.extra_opts, 'unwrap_factor', 0.0) > 0.01:
docanalysis = DocAnalysis('pdf', html)
length = docanalysis.line_length(getattr(self.extra_opts, 'unwrap_factor'))
if length:
# print "The pdf line length returned is " + str(length)
# unwrap em/en dashes
end_rules.append((re.compile(u'(?<=.{%i}[–—])\s*<p>\s*(?=[[a-z\d])' % length), lambda match: ''))
end_rules.append(
# Un wrap using punctuation
(re.compile(u'(?<=.{%i}([a-zäëïöüàèìòùáćéíĺóŕńśúýâêîôûçąężıãõñæøþðßěľščťžňďřů,:)\IA\u00DF]|(?<!\&\w{4});))\s*(?P<ital></(i|b|u)>)?\s*(</p>\s*<p>\s*)+\s*(?=(<(i|b|u)>)?\s*[\w\d$(])' % length, re.UNICODE), wrap_lines), # noqa
)
for rule in self.PREPROCESS + start_rules:
html = rule[0].sub(rule[1], html)
if self.regex_wizard_callback is not None:
self.regex_wizard_callback(self.current_href, html)
if get_preprocess_html:
return html
def dump(raw, where):
import os
dp = getattr(self.extra_opts, 'debug_pipeline', None)
if dp and os.path.exists(dp):
odir = os.path.join(dp, 'input')
if os.path.exists(odir):
odir = os.path.join(odir, where)
if not os.path.exists(odir):
os.makedirs(odir)
name, i = None, 0
while not name or os.path.exists(os.path.join(odir, name)):
i += 1
name = '%04d.html'%i
with open(os.path.join(odir, name), 'wb') as f:
f.write(raw.encode('utf-8'))
# dump(html, 'pre-preprocess')
for rule in rules + end_rules:
try:
html = rule[0].sub(rule[1], html)
except Exception as e:
if rule in user_sr_rules:
self.log.error(
'User supplied search & replace rule: %s -> %s '
'failed with error: %s, ignoring.'%(
user_sr_rules[rule], rule[1], e))
else:
raise
if is_pdftohtml and length > -1:
# Dehyphenate
dehyphenator = Dehyphenator(self.extra_opts.verbose, self.log)
html = dehyphenator(html,'html', length)
if is_pdftohtml:
from calibre.ebooks.conversion.utils import HeuristicProcessor
pdf_markup = HeuristicProcessor(self.extra_opts, None)
totalwords = 0
if pdf_markup.get_word_count(html) > 7000:
html = pdf_markup.markup_chapters(html, totalwords, True)
# dump(html, 'post-preprocess')
# Handle broken XHTML w/ SVG (ugh)
if 'svg:' in html and SVG_NS not in html:
html = html.replace(
'<html', '<html xmlns:svg="%s"' % SVG_NS, 1)
if 'xlink:' in html and XLINK_NS not in html:
html = html.replace(
'<html', '<html xmlns:xlink="%s"' % XLINK_NS, 1)
html = XMLDECL_RE.sub('', html)
if getattr(self.extra_opts, 'asciiize', False):
from calibre.utils.localization import get_udc
from calibre.utils.mreplace import MReplace
unihandecoder = get_udc()
mr = MReplace(data={u'«':u'<'*3, u'»':u'>'*3})
html = mr.mreplace(html)
html = unihandecoder.decode(html)
if getattr(self.extra_opts, 'enable_heuristics', False):
from calibre.ebooks.conversion.utils import HeuristicProcessor
preprocessor = HeuristicProcessor(self.extra_opts, self.log)
html = preprocessor(html)
if is_pdftohtml:
html = html.replace('<!-- created by calibre\'s pdftohtml -->', '')
if getattr(self.extra_opts, 'smarten_punctuation', False):
html = smarten_punctuation(html, self.log)
try:
unsupported_unicode_chars = self.extra_opts.output_profile.unsupported_unicode_chars
except AttributeError:
unsupported_unicode_chars = u''
if unsupported_unicode_chars:
from calibre.utils.localization import get_udc
unihandecoder = get_udc()
for char in unsupported_unicode_chars:
asciichar = unihandecoder.decode(char)
html = html.replace(char, asciichar)
return html
