为这个问题痛苦了很久！所以就决定把本该放在 Wiki 上的东西放这了！起因是明明设置了环境变量，却不能在 python 文件中获取。

问题描述

非常诡异的一件事
Jupter 和 python shell 都能运行的代码：

>>> from nltk.parse.stanford import StanfordParser
>>> parser=StanfordParser(model_path="edu/stanford/nlp/models/lexparser/englishPCFG.ser.gz")
>>> print list(parser.raw_parse("the quick brown fox jumps over the lazy dog"))
[Tree('ROOT', [Tree('NP', [Tree('NP', [Tree('DT', ['the']), Tree('JJ', ['quick']), Tree('JJ', ['brown']), Tree('NN', ['fox'])]), Tree('NP', [Tree('NP', [Tree('NNS', ['jumps'])]), Tree('PP', [Tree('IN', ['over']), Tree('NP', [Tree('DT', ['the']), Tree('JJ', ['lazy']), Tree('NN', ['dog'])])])])])])]

运行 py 文件就不行

from nltk.parse.stanford import StanfordParser
def getParserTree(line):
    parser=StanfordParser(model_path="edu/stanford/nlp/models/lexparser/englishPCFG.ser.gz")
    print list(parser.raw_parse(line))
getParserTree("the quick brown fox jumps over the lazy dog")

Error:

Traceback (most recent call last):
  File "Helper.py", line 70, in <module>
    getParserTree(doc1)
  File "Helper.py", line 55, in getParserTree
    parser=StanfordParser(model_path="edu/stanford/nlp/models/lexparser/englishPCFG.ser.gz")
  File "/Library/Python/2.7/site-packages/nltk/parse/stanford.py", line 51, in __init__
    key=lambda model_name: re.match(self._JAR, model_name)
  File "/Library/Python/2.7/site-packages/nltk/internals.py", line 714, in find_jar_iter
    raise LookupError('\n\n%s\n%s\n%s' % (div, msg, div))
LookupError:
===========================================================================
  NLTK was unable to find stanford-parser\.jar! Set the CLASSPATH
  environment variable.
  For more information, on stanford-parser\.jar, see:
    <http://nlp.stanford.edu/software/lex-parser.shtml>
===========================================================================

问题探索

来看一下环境变量，同样诡异的事就发生了，用 jupter 或 python shell 运行下面的代码，可以得到环境变量路径

1
2
3

>>> import os
>>> os.environ.get('CLASSPATH')
/Users/sure/stanford-tools//stanford-postagger-full-2015-04-20/stanford-postagger.jar:/Users/sure/stanford-tools//stanford-ner-2015-04-20/stanford-ner.jar:/Users/sure/stanford-tools//stanford-parser-full-2015-04-20/stanford-parser.jar:/Users/sure/stanford-tools//stanford-parser-full-2015-04-20/stanford-parser-3.5.2-models.jar

然而放到 py 文件里，就只能 output 出一个 None

为啥！明明设置好了环境变量！看 ~/.bash_profile

export STANFORDTOOLSDIR=/Users/sure/stanford-tools/
export CLASSPATH=$STANFORDTOOLSDIR/stanford-postagger-full-2015-04-20/stanford-postagger.jar:$STANFORDTOOLSDIR/stanford-ner-2015-04-20/stanford-ner.jar:$STANFORDTOOLSDIR/stanford-parser-full-2015-04-20/stanford-parser.jar:$STANFORDTOOLSDIR/stanford-parser-full-2015-04-20/stanford-parser-3.5.2-models.jar
export STANFORD_MODELS=$STANFORDTOOLSDIR/stanford-postagger-full-2015-04-20/models:$STANFORDTOOLSDIR/stanford-ner-2015-04-20/classifiers

索性一个简单粗暴的方法是直接在 py 文件里，StanfordParser 传进绝对地址参数，第一个参数是 path/to/jar，第二个参数是 path/to/module，然后就可以运行文件了。

1
2
3

parser = StanfordParser('/Users/sure/stanford-tools//stanford-parser-full-2015-04-20/stanford-parser.jar', '/Users/sure/stanford-tools//stanford-parser-full-2015-04-20/stanford-parser-3.5.2-models.jar',model_path="edu/stanford/nlp/models/lexparser/englishPCFG.ser.gz")
print list(parser.raw_parse("the quick brown fox jumps over the lazy dog"))

然而想一想，我们要用 Stanford PoS tagging，要用 NER，要用 parser，还要用 dependency parser，那得多麻烦！而且作为一个 team，share code，每个人的路径都不一样。。。

有两个解决方案，一是把对应的 jar 包，model 之类的所有东西放进当前目录下，然后在 py 文件里 get 当前目录，再进行拼接，传入参数；二是单独设置一个配置文件，每个人都配好各自的环境，在 py 文件里调用配置文件，这两种方法都可以解决个人路径不一样的问题。

然而，为什么还是这么麻烦。为什么 get 不到环境变量！

问题解决

来看一下 os.environ 这个函数，才发现 os.environ 保存的是 python 运行的当前 shell 中的环境变量。于是就想到了：我们根本没有在当前 shell 中 mark & export varible 好不好！所以：

$ STANFORDTOOLSDIR=/Users/sure/stanford-tools/
$ CLASSPATH=$STANFORDTOOLSDIR/stanford-postagger-full-2015-04-20/stanford-postagger.jar:$STANFORDTOOLSDIR/stanford-ner-2015-04-20/stanford-ner.jar:$STANFORDTOOLSDIR/stanford-parser-full-2015-04-20/stanford-parser.jar:$STANFORDTOOLSDIR/stanford-parser-full-2015-04-20/stanford-parser-3.5.2-models.jar
$ STANFORD_MODELS=$STANFORDTOOLSDIR/stanford-postagger-full-2015-04-20/models:$STANFORDTOOLSDIR/stanford-ner-2015-04-20/classifiers
$ export STANFORDTOOLSDIR
$ export CLASSPATH
$ export STANFORD_MODELS
$ python2.7 -c 'import os;print os.environ.get("CLASSPATH")'
/Users/sure/stanford-tools//stanford-postagger-full-2015-04-20/stanford-postagger.jar:/Users/sure/stanford-tools//stanford-ner-2015-04-20/stanford-ner.jar:/Users/sure/stanford-tools//stanford-parser-full-2015-04-20/stanford-parser.jar:/Users/sure/stanford-tools//stanford-parser-full-2015-04-20/stanford-parser-3.5.2-models.jar

Done!

1
2
3

$ python2.7 Helper.py
/Users/sure/stanford-tools//stanford-postagger-full-2015-04-20/stanford-postagger.jar:/Users/sure/stanford-tools//stanford-ner-2015-04-20/stanford-ner.jar:/Users/sure/stanford-tools//stanford-parser-full-2015-04-20/stanford-parser.jar:/Users/sure/stanford-tools//stanford-parser-full-2015-04-20/stanford-parser-3.5.2-models.jar
[Tree('ROOT', [Tree('NP', [Tree('NP', [Tree('DT', ['the']), Tree('JJ', ['quick']), Tree('JJ', ['brown']), Tree('NN', ['fox'])]), Tree('NP', [Tree('NP', [Tree('NNS', ['jumps'])]), Tree('PP', [Tree('IN', ['over']), Tree('NP', [Tree('DT', ['the']), Tree('JJ', ['lazy']), Tree('NN', ['dog'])])])])])])]