def fixEmDash(self):
if not self.gettingSize: self.progress = "Fixing em-dashes..."
document2 = self.document2
for paragraph in document2.paragraphs:
for run in paragraph.runs:
text = run.text
text = text.replace('---"', u'---”')
text = text.replace('"---', u'“---')
text = text.replace("---", u"—")
text = regexlib.replaceSub(text, " — ", "—")
text = regexlib.replaceSub(text, " —", "—")
text = regexlib.replaceSub(text, "— ", "—")
run.text = text
self.step += 1
def fixEmDash(self):
if not self.gettingSize: self.progress = "Fixing em-dashes..."
document2 = self.document2
for paragraph in document2.paragraphs:
for run in paragraph.runs:
text = run.text
text = text.replace('---"',u'---”')
text = text.replace('"---',u'“---')
text = text.replace("---",u"—")
text = regexlib.replaceSub(text," — ","—")
text = regexlib.replaceSub(text," —","—")
text = regexlib.replaceSub(text,"— ","—")
run.text = text
self.step += 1
def fixDoubleSpace(self):
if not self.gettingSize: self.progress = "Fixing double spaces..."
for paragraph in self.document2.paragraphs:
for run in paragraph.runs:
while regexlib.match(run.text, " ") != -1:
run.text = regexlib.replaceSub(run.text, " ", " ")
self.step += 1
def fixDoubleSpace(self):
if not self.gettingSize: self.progress = "Fixing double spaces..."
for paragraph in self.document2.paragraphs:
for run in paragraph.runs:
while regexlib.match(run.text, " ") != -1:
run.text = regexlib.replaceSub(run.text, " ", " ")
self.step += 1
def fixEllipses(self):
if not self.gettingSize: self.progress = "Fixing ellipses..."
for paragraph in self.document2.paragraphs:
for run in paragraph.runs:
text = run.text
text = regexlib.replaceSub(text, " . . . . .", "...")
text = regexlib.replaceSub(text, " . . . .", "...")
text = regexlib.replaceSub(text, " . . .", "...")
text = regexlib.replaceSub(text, ". . . . .", "...")
text = regexlib.replaceSub(text, ". . . .", "...")
text = regexlib.replaceSub(text, ". . .", "...")
text = regexlib.replaceSub(text, ".....", "...")
text = regexlib.replaceSub(text, "....", "...")
text = regexlib.replaceSub(text, " . .", u'...”')
text = regexlib.replaceSub(text, ". . ", u"“...")
run.text = text
self.step += 1
def fixEllipses(self):
if not self.gettingSize: self.progress = "Fixing ellipses..."
for paragraph in self.document2.paragraphs:
for run in paragraph.runs:
text = run.text
text = regexlib.replaceSub(text," . . . . .", "...")
text = regexlib.replaceSub(text," . . . .","...")
text = regexlib.replaceSub(text," . . .","...")
text = regexlib.replaceSub(text,". . . . .", "...")
text = regexlib.replaceSub(text,". . . .","...")
text = regexlib.replaceSub(text,". . .","...")
text = regexlib.replaceSub(text,".....","...")
text = regexlib.replaceSub(text,"....","...")
text = regexlib.replaceSub(text," . .",u'...”')
text = regexlib.replaceSub(text,". . ",u"“...")
run.text = text
self.step += 1
def fixWords(self):
if not self.gettingSize: self.progress = "Fixing words..."
for paragraph in self.document2.paragraphs:
for run in paragraph.runs:
text = run.text
text = regexlib.replaceWord(text, "comer", "corner")
text = regexlib.replaceWord(text, "bom", "born")
text = regexlib.replaceWord(text, "modem", "modern")
text = regexlib.replaceWord(text, "tiling", "thing")
text = regexlib.replaceWord(text, "diat", "that")
text = regexlib.replaceWord(text, "sec", "see")
text = regexlib.replaceWord(text, "secs", "sees")
text = regexlib.replaceWord(text, "Fd", "I'd")
text = regexlib.replaceWord(text, "diem", "them")
text = regexlib.replaceWord(text, "Modem", "Modern")
text = regexlib.replaceSub(text, "‘Tm", "“I’m")
text = regexlib.replaceSub(text, "Td", "“I’d")
text = regexlib.replaceWord(text, "tire", "the")
text = regexlib.replaceSub(text, "boy friend", "boyfriend")
text = regexlib.replaceSub(text, "girl friend", "girlfriend")
text = regexlib.replaceWord(text, "Pie", "He")
text = regexlib.replaceWord(text, "Fie", "He")
run.text = text
self.step += 1
def fixWords(self):
if not self.gettingSize: self.progress = "Fixing words..."
for paragraph in self.document2.paragraphs:
for run in paragraph.runs:
text = run.text
text = regexlib.replaceWord(text,"comer","corner")
text = regexlib.replaceWord(text,"bom","born")
text = regexlib.replaceWord(text,"modem","modern")
text = regexlib.replaceWord(text,"tiling","thing")
text = regexlib.replaceWord(text,"diat","that")
text = regexlib.replaceWord(text,"sec","see")
text = regexlib.replaceWord(text,"secs","sees")
text = regexlib.replaceWord(text,"Fd","I'd")
text = regexlib.replaceWord(text,"diem","them")
text = regexlib.replaceWord(text,"Modem","Modern")
text = regexlib.replaceSub(text,"‘Tm","“I’m")
text = regexlib.replaceSub(text,"Td","“I’d")
text = regexlib.replaceWord(text,"tire","the")
text = regexlib.replaceSub(text,"boy friend","boyfriend")
text = regexlib.replaceSub(text,"girl friend","girlfriend")
text = regexlib.replaceWord(text,"Pie", "He")
text = regexlib.replaceWord(text,"Fie", "He")
run.text = text
self.step += 1
def fixPunctuation(self):
if not self.gettingSize: self.progress = "Fixing punctuation..."
for paragraph in self.document2.paragraphs:
for run in paragraph.runs:
text = run.text
#Hyphens
text = regexlib.removeSub(text, " -")
text = regexlib.replaceSub(text, "- ", "-")
#Misc
text = regexlib.replaceSub(text, "/", "I")
text = regexlib.removeSub(text, "\\")
text = regexlib.removeSub(text, "^")
#Asterisks
text = regexlib.replaceSub(text, "* * * *", "& & & &")
text = regexlib.removeSub(text, "*")
text = regexlib.replaceSub(text, "& & & &", "* * * *")
# Periods and Spacing
text = regexlib.replaceSub(text, "# .#", "# #")
text = regexlib.replaceSub(text, " ", " ")
text = regexlib.replaceSub(text, "“ ‘", "“‘")
run.text = text
self.step += 1
def fixPunctuation(self):
if not self.gettingSize: self.progress = "Fixing punctuation..."
for paragraph in self.document2.paragraphs:
for run in paragraph.runs:
text = run.text
#Hyphens
text = regexlib.removeSub(text," -")
text = regexlib.replaceSub(text,"- ","-")
#Misc
text = regexlib.replaceSub(text,"/","I")
text = regexlib.removeSub(text,"\\")
text = regexlib.removeSub(text,"^")
#Asterisks
text = regexlib.replaceSub(text,"* * * *", "& & & &")
text = regexlib.removeSub(text,"*")
text = regexlib.replaceSub(text,"& & & &", "* * * *")
# Periods and Spacing
text = regexlib.replaceSub(text,"# .#","# #")
text = regexlib.replaceSub(text," "," ")
text = regexlib.replaceSub(text,"“ ‘","“‘")
run.text = text
self.step += 1
def fixApostrophes(self):
if not self.gettingSize: self.progress = "Fixing apostrophes..."
for paragraph in self.document2.paragraphs:
for run in paragraph.runs:
text = run.text
text = regexlib.replaceSub(text, "''", "\"")
text = regexlib.replaceSub(text, " '", u' ‘')
text = regexlib.replaceSub(text, "' ", u'’ ')
text = regexlib.replaceSub(text, ".'", u'.’')
text = regexlib.replaceSub(text, ",'", u',’')
text = regexlib.replaceSub(text, "!'", u'!’')
text = regexlib.replaceSub(text, "?'", u'?’')
text = regexlib.replaceSub(text, "...'", u'...’')
text = regexlib.replaceSub(text, "'.", u'’.')
text = regexlib.replaceSub(text, "#—'#", u'#—‘#')
text = regexlib.replaceSub(text, "#'—#", u'#’—#')
text = regexlib.replaceSub(text, "#—'", u'#—’')
text = regexlib.replaceSub(text, "'—#", u'‘—#')
text = regexlib.replaceSub(text, "',", u'’,')
text = regexlib.replaceSub(text, "'!", u'’!')
text = regexlib.replaceSub(text, "'?", u'’?')
text = regexlib.replaceSub(text, "'...", u'’...')
text = regexlib.replaceSub(text, "#'#", u'#’#')
text = regexlib.replaceSub(text, "'", u'‘')
run.text = text
self.step += 1
def fixQuotations(self):
if not self.gettingSize: self.progress = "Fixing quotations..."
for paragraph in self.document2.paragraphs:
for run in paragraph.runs:
text = run.text
text = regexlib.replaceSub(text, ' "', u' “')
text = regexlib.replaceSub(text, '" ', u'” ')
text = regexlib.replaceSub(text, '."', u'.”')
text = regexlib.replaceSub(text, ',"', u',”')
text = regexlib.replaceSub(text, '!"', u'!”')
text = regexlib.replaceSub(text, '?"', u'?”')
text = regexlib.replaceSub(text, '..."', u'...”')
text = regexlib.replaceSub(text, '#—"#', u'#—“#')
text = regexlib.replaceSub(text, '#"—#', u'#”—#')
text = regexlib.replaceSub(text, '#—"', u'#—”')
text = regexlib.replaceSub(text, '"—#', u'“—#')
text = regexlib.replaceSub(text, '’"', u'’”')
text = regexlib.replaceSub(text, '\'"', u'’”')
text = regexlib.replaceSub(text, '";', u'”;')
text = regexlib.replaceSub(text, '":', u'”:')
text = regexlib.replaceSub(text, "' \"", u'’”')
text = regexlib.replaceSub(text, "’ \"", u'’”')
text = regexlib.replaceSub(text, '"', u'“')
text = regexlib.replaceSub(text, "”’", u'’”')
run.text = text
self.step += 1
def fixApostrophes(self):
if not self.gettingSize: self.progress = "Fixing apostrophes..."
for paragraph in self.document2.paragraphs:
for run in paragraph.runs:
text = run.text
text = regexlib.replaceSub(text,"''","\"")
text = regexlib.replaceSub(text," '",u' ‘')
text = regexlib.replaceSub(text,"' ",u'’ ')
text = regexlib.replaceSub(text,".'",u'.’')
text = regexlib.replaceSub(text,",'",u',’')
text = regexlib.replaceSub(text,"!'",u'!’')
text = regexlib.replaceSub(text,"?'",u'?’')
text = regexlib.replaceSub(text,"...'",u'...’')
text = regexlib.replaceSub(text,"'.",u'’.')
text = regexlib.replaceSub(text,"#—'#",u'#—‘#')
text = regexlib.replaceSub(text,"#'—#",u'#’—#')
text = regexlib.replaceSub(text,"#—'",u'#—’')
text = regexlib.replaceSub(text,"'—#",u'‘—#')
text = regexlib.replaceSub(text,"',",u'’,')
text = regexlib.replaceSub(text,"'!",u'’!')
text = regexlib.replaceSub(text,"'?",u'’?')
text = regexlib.replaceSub(text,"'...",u'’...')
text = regexlib.replaceSub(text,"#'#",u'#’#')
text = regexlib.replaceSub(text,"'",u'‘')
run.text = text
self.step += 1
def fixQuotations(self):
if not self.gettingSize: self.progress = "Fixing quotations..."
for paragraph in self.document2.paragraphs:
for run in paragraph.runs:
text = run.text
text = regexlib.replaceSub(text,' "',u' “')
text = regexlib.replaceSub(text,'" ',u'” ')
text = regexlib.replaceSub(text,'."',u'.”')
text = regexlib.replaceSub(text,',"',u',”')
text = regexlib.replaceSub(text,'!"',u'!”')
text = regexlib.replaceSub(text,'?"',u'?”')
text = regexlib.replaceSub(text,'..."',u'...”')
text = regexlib.replaceSub(text,'#—"#',u'#—“#')
text = regexlib.replaceSub(text,'#"—#',u'#”—#')
text = regexlib.replaceSub(text,'#—"',u'#—”')
text = regexlib.replaceSub(text,'"—#',u'“—#')
text = regexlib.replaceSub(text,'’"',u'’”')
text = regexlib.replaceSub(text,'\'"',u'’”')
text = regexlib.replaceSub(text,'";',u'”;')
text = regexlib.replaceSub(text,'":',u'”:')
text = regexlib.replaceSub(text,"' \"",u'’”')
text = regexlib.replaceSub(text,"’ \"",u'’”')
text = regexlib.replaceSub(text,'"',u'“')
text = regexlib.replaceSub(text,"”’",u'’”')
run.text = text
self.step += 1