topicModelingTickets/preprocessing.py

460 lines
10 KiB
Python

# -*- coding: utf-8 -*-
from datetime import datetime
import csv
import sys
from miscellaneous import *
from datetime import datetime
import time
import textacy
from scipy import *
import os
csv.field_size_limit(sys.maxsize)
FILEPATH = os.path.dirname(os.path.realpath(__file__)) + "/"
# load config
config_ini = FILEPATH + "config.ini"
config = ConfigParser.ConfigParser()
with open(config_ini) as f:
config.read_file(f)
REGEX_SPECIALCHAR = r'[`\-=~%^&*()_+\[\]{};\'\\:"|</>]' #+r',.'
REGEX_TOPLVL = r'\.[a-z]{2,3}(\.[a-z]{2,3})?'
THESAURUS = {}
WORDS= {}
LEMMAS= {}
NOUNS= {}
VORNAMEN= {}
DE_STOP_WORDS= {}
EN_STOP_WORDS= {}
############# filter tokens
def filterTokens(tokens, funclist):
# in:tokenlist, funclist
# out: tokenlist
for f in funclist:
tokens = list(filter(f, tokens))
return tokens
def keepPOS(pos_list):
return lambda tok: tok.pos_ in pos_list
def keepNouns(noun_list=NOUNS):
return lambda tok: tok.lower_ in noun_list
def removePOS(pos_list):
return lambda tok: tok.pos_ not in pos_list
def removeWords(words, keep=None):
if hasattr(keep, '__iter__'):
for k in keep:
try:
words.remove(k)
except ValueError:
pass
return lambda tok: tok.lower_ not in words
def keepENT(ent_list):
return lambda tok: tok.ent_type_ in ent_list
def removeENT(ent_list):
return lambda tok: tok.ent_type_ not in ent_list
def remove_words_containing_Numbers():
return lambda tok: not bool(re.search('\d', tok.lower_))
def remove_words_containing_topLVL():
return lambda tok: not bool(re.search(REGEX_TOPLVL, tok.lower_))
def remove_words_containing_specialCharacters():
return lambda tok: not bool(re.search(REGEX_SPECIALCHAR, tok.lower_))
def remove_long_words():
return lambda tok: not len(tok.lower_) < 2
def remove_short_words():
return lambda tok: not len(tok.lower_) > 35
def remove_first_names():
return lambda tok: tok.lower_ not in [name.lower() for name in VORNAMEN]
############# strings
def remove_addresses(string):
pass # todo remove_addresses idee postal.parser und zu metadaten hinzufügen
def lemmatizeWord(word,lemma_dict=LEMMAS,n=3):
for i in range(n):
try:
word = lemma_dict[word.lower()] if word.lower() in lemma_dict.keys() else word.lower()
except:
print(word)
return word
def getFirstSynonym(word, thesaurus=THESAURUS):
if not isinstance(word, str):
return str(word)
word = word.lower()
if word in thesaurus.keys():
return thesaurus[word]
else:
return str(word)
########################## Spellchecking ##########################################
# http://norvig.com/spell-correct.html
# http://wortschatz.uni-leipzig.de/en/download
import re
def words(text): return re.findall(r'\w+', text.lower())
def P(word, N=sum(WORDS.values())):
"Probability of `word`."
return WORDS[word] / N
def correction(word):
"Most probable spelling correction for word."
return max(candidates(word), key=P)
def candidates(word):
"Generate possible spelling corrections for word."
return (known([word]) or known(edits1(word)) or known(edits2(word)) or [word])
def known(words):
"The subset of `words` that appear in the dictionary of WORDS."
return set(w for w in words if w in WORDS)
def edits1(word):
"All edits that are one edit away from `word`."
letters = 'abcdefghijklmnopqrstuvwxyz'
splits = [(word[:i], word[i:]) for i in range(len(word) + 1)]
deletes = [L + R[1:] for L, R in splits if R]
transposes = [L + R[1] + R[0] + R[2:] for L, R in splits if len(R) > 1]
replaces = [L + c + R[1:] for L, R in splits if R for c in letters]
inserts = [L + c + R for L, R in splits for c in letters]
return set(deletes + transposes + replaces + inserts)
def edits2(word):
"All edits that are two edits away from `word`."
return (e2 for e1 in edits1(word) for e2 in edits1(e1))
def autocorrectWord(word):
try:
return correction(word)
except:
return word
############# stringcleaning
def processContentstream(textstream, parser, token_filterlist=None):
#pre parse
textstream = preparse(textstream)
pipe = parser.pipe(textstream)
for doc in pipe:
tokens = [tok for tok in doc]
# in parse
if token_filterlist is not None:
tokens = filterTokens(tokens, token_filterlist)
# post parse
tokens = [postparse(tok) for tok in tokens] #todo: informationsverlust von pos,tag etc.!
yield " ".join(tokens)
def preparse(stringstream):
for string in stringstream:
# cut_after
words = ["gruss", "grusse","gruesse","gruessen","grusses"]
for gr in words:
if gr in string:
string = string.rpartition(gr)[0]
break
yield string
def postparse(toktext):
"""
:param toktext: spacy.token
:return: string
"""
toktext = toktext.lower_
# remove_words_containing_topLVL
toktext = toktext if not re.search(REGEX_TOPLVL, toktext) else ""
# lemmatize
toktext = lemmatizeWord(toktext)
# synonyme normalisieren
toktext = getFirstSynonym(toktext)
# autocorrect
toktext = autocorrectWord(toktext)
return toktext
def processDictstream(dictstream, funcdict, parser):
"""
:param dictstream: dict-gen
:param funcdict:
clean_in_meta = {
"Solution":funclist,
...
}
:param parser: spacy-parser
:return: dict-gen
"""
for dic in dictstream:
result = {}
for key, value in dic.items():
if key in funcdict:
doc = parser(value)
tokens = [tok for tok in doc]
funclist = funcdict[key]
tokens = filterTokens(tokens, funclist)
result[key] = " ".join([tok.lower_ for tok in tokens])
else:
result[key] = value
yield result
##################################################################################################
ressources_path = FILEPATH + "ressources/"
path2thesaurus_dict = ressources_path + config.get("thesaurus","pickle_file")
path2wordsdict = ressources_path + config.get("spellchecking", "pickle_file")
path2lemmadict = ressources_path + config.get("lemmatization","pickle_file")
path2nouns_list = ressources_path + config.get("nouns","pickle_file")
path2firstnameslist = ressources_path + config.get("firstnames","pickle_file")
path2DEstopwordlist = ressources_path + config.get("de_stopwords", "pickle_file")
path2ENstopwordlist = ressources_path + config.get("en_stopwords", "pickle_file")
custom_words = get_list_from_config("preprocessing", "custom_words")
corpus_de_path = FILEPATH + config.get("de_corpus", "path")
de_plainpath = FILEPATH + config.get("de_corpus", "path") + "pre_labled_lines.txt"
corpus_en_path = FILEPATH + config.get("en_corpus", "path")
def preprocessCorpus(corpus_path, filter_tokens, clean_in_meta, lang="de", printrandom=10):
logprint("Preprocess {0}_corpus at {1}".format(lang, datetime.now()))
cleanCorpus_name = lang + "_clean_ticket"
preCorpus_name = lang + "_pre_ticket"
#load raw corpus and create new one
logprint("Load {0}_raw".format(lang))
clean_corpus, parser = load_corpus(corpus_name=cleanCorpus_name, corpus_path=corpus_path)
corpus = textacy.Corpus(parser)
## process and add files to textacy-corpi,
corpus.add_texts(
processContentstream(corpus2Text(clean_corpus), token_filterlist=filter_tokens, parser=parser),
processDictstream(corpus2Meta(clean_corpus), clean_in_meta,parser=parser)
)
# leere docs aus corpi kicken
corpus.remove(lambda doc: len(doc) == 0)
for i in range(printrandom):
printRandomDoc(corpus)
#save corpus
save_corpus(corpus=corpus, corpus_path=corpus_path, corpus_name=preCorpus_name)
#save corpus as labled, plain text
savelabledCorpiLines(corpus, de_plainpath)
return corpus
def main():
start = time.time()
THESAURUS = load_obj(path2thesaurus_dict)
WORDS = load_obj(path2wordsdict)
LEMMAS = load_obj(path2lemmadict)
DE_STOP_WORDS = load_obj(path2DEstopwordlist)
EN_STOP_WORDS = load_obj(path2ENstopwordlist)
NOUNS = load_obj(path2nouns_list)
VORNAMEN = load_obj(path2firstnameslist)
filter_tokens = [
keepNouns(NOUNS),
remove_words_containing_Numbers(),
removePOS(["PUNCT", "SPACE", "NUM"]),
removeWords(DE_STOP_WORDS + custom_words),
#removeWords(DE_STOP_WORDS),
remove_long_words(),
remove_short_words(),
remove_first_names()
]
clean_in_meta = {
"Solution": [removePOS(["SPACE"])],
"Subject": [removePOS(["SPACE", "PUNCT"])],
"categoryName": [removePOS(["SPACE", "PUNCT"])]
}
corpus = preprocessCorpus(corpus_de_path, filter_tokens, clean_in_meta, "de",printrandom=5)
end = time.time()
logprint("Time Elapsed Preprocessing:{0} min".format((end - start) / 60))
if __name__ == "__main__":
main()
"""
pipe=[
##String
fixUnicode(),
replaceHardS(),
resolveAbbrivations(),
remove_words_containing_topLVL(),
replaceSpecialChars(" "), (mit Leerzeichen erstzen, dadruch werden Terme wie 8203;verfügung getrennt
remove_words_containing_Numbers(),
##spacyParse
removeENT("PERSON"),
keepPOS(["NOUN"]),
#ODER
lemmatize(),
removeWords(de_stop_words + config.get("preprocessing","custom_words").split(",")),
# evtl.
spellCorrection(),
keepUniqeTokens(),
]
"""
"""
filter_tokens=[
#removeENT(["PERSON"]),
#idee addressen enfernen #bisher mit cut_after("gruss") --> postal.parser
#idee rechtschreibkorrektur --> PyEnchant
#idee thesaurus --> WordNet, eigener
remove_words_containing_Numbers(),
removePOS(["PUNCT","SPACE","NUM"]),
removeWords(de_stop_words+custom_words),
remove_long_words(),
remove_short_words(),
remove_first_names(),
keepPOS(["NOUN"]),
]
"""