import?os
import?re
from?mrjob.job?import?MRJob
from?mrjob.step?import?MRStep

word_search_re?=?re.compile（r“[\w‘]+“）


class?ExtractPosts（MRJob）:

????post_start?=?False
????post?=?[]

????def?mapper（self?key?line）:
????????filename?=?os.environ[“map_input_file“]
????????gender?=?filename.split（“.“）[1]
????????try:
????????????docnum?=?int（filename[0]）
????????except:
????????????docnum?=?8
????????if?filename.startswith（“51“）:
????????????#?remove?leading?and?trailing?whitespace
????????????line?=?line.strip（）
????????????if?line?==?““:
????????????????self.post_start?=?True
????????????elif?line?==?““:
????????????????self.post_start?=?False
????????????????yield?gender?repr（“\n“.join（self.post））
????????????????self.post?=?[]
????????????elif?self.post_start:
????????????????self.post.append（line）



if?__name__?==?‘__main__‘:
????ExtractPosts.run（）

?屬性????????????大小?????日期????時間???名稱
-----------?---------??----------?-----??----
?????目錄???????????0??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?1\
?????文件????????1000??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?1\affinity_dataset.txt
?????文件???????16777??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?1\ch1_affinity.ipynb
?????文件????????3670??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?1\ch1_affinity_create.ipynb
?????文件???????13847??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?1\ch1_oner_application.ipynb
?????目錄???????????0??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?10\
?????文件???????78925??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?10\Chapter?10?Clusterer.ipynb
?????目錄???????????0??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?11\
?????文件???????59206??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?11\Chapter?11?（CIFAR）.ipynb
?????文件???????62409??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?11\Chapter?11?（Theano?and?Lasagne）.ipynb
?????目錄???????????0??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?12\
?????文件???????38759??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?12\CH12?MapReduce?Basics.ipynb
?????文件???????10578??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?12\Chapter?12?（NB?Predict）.ipynb
?????文件????????1730??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?12\Chapter?12?（Test?load）.ipynb
?????文件?????????882??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?12\extract_posts.py
?????文件????????1986??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?12\nb_predict.py
?????文件????????2021??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?12\nb_train.py
?????目錄???????????0??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?2\
?????文件??????143291??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?2\Ionosphere?Nearest?Neighbour.ipynb
?????目錄???????????0??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?3\
?????文件???????45385??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?3\Basketball?Results.ipynb
?????目錄???????????0??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?4\
?????文件???????46259??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?4\ch4?Affinity?Analysis.ipynb
?????目錄???????????0??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?5\
?????文件????????1034??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?5\adult_tests.py
?????文件???????13293??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?5\ch5_adult.ipynb
?????文件???????29985??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?5\ch5_advertisements.ipynb
?????目錄???????????0??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?6\
?????文件????????4977??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?6\ch6_classify_twitter.ipynb
?????文件???????13853??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?6\ch6_get_twitter.ipynb
?????文件????????6717??2018-03-20?01:15??python數(shù)據(jù)挖掘入門與實踐\Chapter?6\ch6_label_twitter.ipynb
............此處省略12個文件信息