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:
??????????

?屬性????????????大小?????日期????時(shí)間???名稱
-----------?---------??----------?-----??----
?????目錄???????????0??2015-07-08?11:25??Code_REWRITE\Chapter?1\
?????目錄???????????0??2015-07-08?11:25??Code_REWRITE\Chapter?1\.ipynb_checkpoints\
?????文件???????14242??2015-05-28?09:33??Code_REWRITE\Chapter?1\.ipynb_checkpoints\ch1_affinity-checkpoint.ipynb
?????文件???????13797??2015-05-28?09:33??Code_REWRITE\Chapter?1\.ipynb_checkpoints\ch1_oner_application-checkpoint.ipynb
?????文件????????1000??2014-10-10?03:13??Code_REWRITE\Chapter?1\affinity_dataset.txt
?????文件???????14242??2015-05-28?09:33??Code_REWRITE\Chapter?1\ch1_affinity.ipynb
?????文件????????3544??2014-10-10?03:13??Code_REWRITE\Chapter?1\ch1_affinity_create.ipynb
?????文件???????13797??2015-05-28?09:33??Code_REWRITE\Chapter?1\ch1_oner_application.ipynb
?????目錄???????????0??2015-07-08?11:25??Code_REWRITE\Chapter?10\
?????目錄???????????0??2015-07-08?11:25??Code_REWRITE\Chapter?10\.ipynb_checkpoints\
?????文件??????434745??2015-06-28?15:31??Code_REWRITE\Chapter?10\.ipynb_checkpoints\Chapter?10?（Cluster?Types）-checkpoint.ipynb
?????文件???????64766??2015-06-28?15:36??Code_REWRITE\Chapter?10\.ipynb_checkpoints\Chapter?10?Clusterer-checkpoint.ipynb
?????文件?????1100672??2015-06-28?15:34??Code_REWRITE\Chapter?10\.ipynb_checkpoints\Chapter?10?Image?creation?10_02.png-checkpoint.ipynb
?????文件??????????72??2015-06-28?15:29??Code_REWRITE\Chapter?10\.ipynb_checkpoints\Chapter?10-checkpoint.ipynb
?????文件???????78925??2015-06-28?15:40??Code_REWRITE\Chapter?10\Chapter?10?Clusterer.ipynb
?????目錄???????????0??2015-07-08?11:25??Code_REWRITE\Chapter?11\
?????文件???????58837??2015-06-28?15:17??Code_REWRITE\Chapter?11\Chapter?11?（CIFAR）.ipynb
?????文件???????62409??2015-06-28?15:02??Code_REWRITE\Chapter?11\Chapter?11?（Theano?and?Lasagne）.ipynb
?????目錄???????????0??2015-07-08?11:25??Code_REWRITE\Chapter?12\
?????文件???????38759??2015-06-28?15:22??Code_REWRITE\Chapter?12\CH12?MapReduce?Basics.ipynb
?????文件???????10578??2015-05-08?19:56??Code_REWRITE\Chapter?12\Chapter?12?（NB?Predict）.ipynb
?????文件????????1730??2015-05-08?06:35??Code_REWRITE\Chapter?12\Chapter?12?（Test?load）.ipynb
?????文件?????????882??2015-05-08?16:49??Code_REWRITE\Chapter?12\extract_posts.py
?????文件????????1986??2015-05-08?09:37??Code_REWRITE\Chapter?12\nb_predict.py
?????文件????????2021??2015-05-08?08:44??Code_REWRITE\Chapter?12\nb_train.py
?????目錄???????????0??2015-07-08?11:25??Code_REWRITE\Chapter?2\
?????目錄???????????0??2015-07-08?11:25??Code_REWRITE\Chapter?2\.ipynb_checkpoints\
?????文件??????128669??2015-05-28?10:08??Code_REWRITE\Chapter?2\.ipynb_checkpoints\Ionosphere?Nearest?Neighbour-checkpoint.ipynb
?????文件??????128669??2015-05-28?10:08??Code_REWRITE\Chapter?2\Ionosphere?Nearest?Neighbour.ipynb
?????目錄???????????0??2015-07-08?11:25??Code_REWRITE\Chapter?3\
?????目錄???????????0??2015-07-08?11:25??Code_REWRITE\Chapter?3\.ipynb_checkpoints\
............此處省略33個(gè)文件信息