Update README.md update paddle link. (#817)

* paddle_null_point_fix

* add core expception note

* delete yield

* modify test paddle for supporting enable_paddle()

Changelog

@@ -1,3 +1,15 @@
+2019-1-20: version 0.42.1
+1. 修复setup.py在python2.7版本无法工作的问题 (issue #809)
+
+2019-1-13: version 0.42
+1. 修复paddle模式空字符串coredump问题 @JesseyXujin
+2. 修复cut_all模式切分丢字问题 @fxsjy
+3. paddle安装检测优化 @vissssa
+
+2019-1-8: version 0.41
+1. 开启paddle模式更友好
+2. 修复cut_all模式不支持中英混合词的bug
+
 2019-12-25: version 0.40
 1. 支持基于paddle的深度学习分词模式(use_paddle=True); by @JesseyXujin, @xyzhou-puck
 2. 修复自定义Tokenizer实例的add_word方法指向全局的问题; by @linhx13 

README.md

@@ -13,7 +13,7 @@ jieba
     * 精确模式，试图将句子最精确地切开，适合文本分析；
     * 全模式，把句子中所有的可以成词的词语都扫描出来, 速度非常快，但是不能解决歧义；
     * 搜索引擎模式，在精确模式的基础上，对长词再次切分，提高召回率，适合用于搜索引擎分词。
-    * paddle模式，利用paddlepaddle深度学习框架，训练序列标注（双向GRU）网络模型实现分词。同时支持词性标注。如需使用，请先安装paddlepaddle-tiny，`pip install paddlepaddle-tiny==1.6.1`。 （www.paddlepaddle.org）
+    * paddle模式，利用PaddlePaddle深度学习框架，训练序列标注（双向GRU）网络模型实现分词。同时支持词性标注。paddle模式使用需安装paddlepaddle-tiny，`pip install paddlepaddle-tiny==1.6.1`。目前paddle模式支持jieba v0.40及以上版本。jieba v0.40以下版本，请升级jieba，`pip install jieba --upgrade` 。[PaddlePaddle官网](https://www.paddlepaddle.org.cn/)
 * 支持繁体分词
 * 支持自定义词典
 * MIT 授权协议
@@ -39,7 +39,7 @@ jieba
 =======
 1. 分词
 --------
-* `jieba.cut` 方法接受四个输入参数: 需要分词的字符串；cut_all 参数用来控制是否采用全模式；HMM 参数用来控制是否使用 HMM 模型；use_paddle 参数用来控制是否使用paddle模式下的分词模式（如需使用，安装paddlepaddle-tiny，`pip install paddlepaddle-tiny==1.6.1` ）；
+* `jieba.cut` 方法接受四个输入参数: 需要分词的字符串；cut_all 参数用来控制是否采用全模式；HMM 参数用来控制是否使用 HMM 模型；use_paddle 参数用来控制是否使用paddle模式下的分词模式，paddle模式采用延迟加载方式，通过enable_paddle接口安装paddlepaddle-tiny，并且import相关代码；
 * `jieba.cut_for_search` 方法接受两个参数：需要分词的字符串；是否使用 HMM 模型。该方法适合用于搜索引擎构建倒排索引的分词，粒度比较细
 * 待分词的字符串可以是 unicode 或 UTF-8 字符串、GBK 字符串。注意：不建议直接输入 GBK 字符串，可能无法预料地错误解码成 UTF-8
 * `jieba.cut` 以及 `jieba.cut_for_search` 返回的结构都是一个可迭代的 generator，可以使用 for 循环来获得分词后得到的每一个词语(unicode)，或者用
@@ -52,8 +52,11 @@ jieba
 # encoding=utf-8
 import jieba
 
-seg_list = jieba.cut("我来到北京清华大学", use_paddle=True)
-print("Paddle Mode: " + "/ ".join(seg_list))  # paddle模式
+jieba.enable_paddle()# 启动paddle模式。 0.40版之后开始支持，早期版本不支持
+strs=["我来到北京清华大学","乒乓球拍卖完了","中国科学技术大学"]
+for str in strs:
+    seg_list = jieba.cut(str,use_paddle=True) # 使用paddle模式
+    print("Paddle Mode: " + '/'.join(list(seg_list)))
 
 seg_list = jieba.cut("我来到北京清华大学", cut_all=True)
 print("Full Mode: " + "/ ".join(seg_list))  # 全模式
@@ -190,12 +193,14 @@ https://github.com/fxsjy/jieba/blob/master/test/extract_tags.py
 -----------
 * `jieba.posseg.POSTokenizer(tokenizer=None)` 新建自定义分词器，`tokenizer` 参数可指定内部使用的 `jieba.Tokenizer` 分词器。`jieba.posseg.dt` 为默认词性标注分词器。
 * 标注句子分词后每个词的词性，采用和 ictclas 兼容的标记法。
-* 除了jieba默认分词模式，提供paddle模式下的词性标注功能。如需使用，请先安装paddlepaddle-tiny，`pip install paddlepaddle-tiny==1.6.1`。
+* 除了jieba默认分词模式，提供paddle模式下的词性标注功能。paddle模式采用延迟加载方式，通过enable_paddle()安装paddlepaddle-tiny，并且import相关代码；
 * 用法示例
 
 ```pycon
+>>> import jieba
 >>> import jieba.posseg as pseg
 >>> words = pseg.cut("我爱北京天安门") #jieba默认模式
+>>> jieba.enable_paddle() #启动paddle模式。 0.40版之后开始支持，早期版本不支持
 >>> words = pseg.cut("我爱北京天安门",use_paddle=True) #paddle模式
 >>> for word, flag in words:
 ...    print('%s %s' % (word, flag))

jieba/__init__.py

@@ -1,26 +1,24 @@
 from __future__ import absolute_import, unicode_literals
-__version__ = '0.40'
+
+__version__ = '0.42.1'
 __license__ = 'MIT'
 
-import re
-import os
-import sys
-import time
-import logging
 import marshal
+import re
 import tempfile
 import threading
-from math import log
+import time
 from hashlib import md5
-from ._compat import *
+from math import log
+
 from . import finalseg
+from ._compat import *
 
 if os.name == 'nt':
     from shutil import move as _replace_file
 else:
     _replace_file = os.rename
 
-
 _get_abs_path = lambda path: os.path.normpath(os.path.join(os.getcwd(), path))
 
 DEFAULT_DICT = None
@@ -46,13 +44,12 @@ re_eng = re.compile('[a-zA-Z0-9]', re.U)
 re_han_default = re.compile("([\u4E00-\u9FD5a-zA-Z0-9+#&\._%\-]+)", re.U)
 
 re_skip_default = re.compile("(\r\n|\s)", re.U)
-re_han_cut_all = re.compile("([\u4E00-\u9FD5]+)", re.U)
-re_skip_cut_all = re.compile("[^a-zA-Z0-9+#\n]", re.U)
+
 
 def setLogLevel(log_level):
-    global logger
     default_logger.setLevel(log_level)
 
+
 class Tokenizer(object):
 
     def __init__(self, dictionary=DEFAULT_DICT):
@@ -71,7 +68,8 @@ class Tokenizer(object):
     def __repr__(self):
         return '<Tokenizer dictionary=%r>' % self.dictionary
 
-    def gen_pfdict(self, f):
+    @staticmethod
+    def gen_pfdict(f):
         lfreq = {}
         ltotal = 0
         f_name = resolve_filename(f)
@@ -130,7 +128,7 @@ class Tokenizer(object):
 
             load_from_cache_fail = True
             if os.path.isfile(cache_file) and (abs_path == DEFAULT_DICT or
-                os.path.getmtime(cache_file) > os.path.getmtime(abs_path)):
+                                               os.path.getmtime(cache_file) > os.path.getmtime(abs_path)):
                 default_logger.debug(
                     "Loading model from cache %s" % cache_file)
                 try:
@@ -200,15 +198,30 @@ class Tokenizer(object):
     def __cut_all(self, sentence):
         dag = self.get_DAG(sentence)
         old_j = -1
+        eng_scan = 0
+        eng_buf = u''
         for k, L in iteritems(dag):
+            if eng_scan == 1 and not re_eng.match(sentence[k]):
+                eng_scan = 0
+                yield eng_buf
             if len(L) == 1 and k > old_j:
-                yield sentence[k:L[0] + 1]
+                word = sentence[k:L[0] + 1]
+                if re_eng.match(word):
+                    if eng_scan == 0:
+                        eng_scan = 1
+                        eng_buf = word
+                    else:
+                        eng_buf += word
+                if eng_scan == 0:
+                    yield word
                 old_j = L[0]
             else:
                 for j in L:
                     if j > k:
                         yield sentence[k:j + 1]
                         old_j = j
+        if eng_scan == 1:
+            yield eng_buf
 
     def __cut_DAG_NO_HMM(self, sentence):
         DAG = self.get_DAG(sentence)
@@ -273,8 +286,8 @@ class Tokenizer(object):
                 for elem in buf:
                     yield elem
 
-    def cut(self, sentence, cut_all = False, HMM = True,use_paddle = False):
-        '''
+    def cut(self, sentence, cut_all=False, HMM=True, use_paddle=False):
+        """
         The main function that segments an entire sentence that contains
         Chinese characters into separated words.
 
@@ -282,15 +295,12 @@ class Tokenizer(object):
             - sentence: The str(unicode) to be segmented.
             - cut_all: Model type. True for full pattern, False for accurate pattern.
             - HMM: Whether to use the Hidden Markov Model.
-        '''
-        is_paddle_installed = False
-        if use_paddle == True:
-            import_paddle_check = import_paddle()
-            is_paddle_installed = check_paddle_install()
+        """
+        is_paddle_installed = check_paddle_install['is_paddle_installed']
         sentence = strdecode(sentence)
-        if use_paddle == True and is_paddle_installed == True and import_paddle_check == True:
-            if sentence is None or sentence == "" or sentence == u"":
-                yield sentence
+        if use_paddle and is_paddle_installed:
+            # if sentence is null, it will raise core exception in paddle.
+            if sentence is None or len(sentence) == 0:
                 return
             import jieba.lac_small.predict as predict
             results = predict.get_sent(sentence)
@@ -299,12 +309,8 @@ class Tokenizer(object):
                     continue
                 yield sent
             return
-        if cut_all:
-            re_han = re_han_cut_all
-            re_skip = re_skip_cut_all
-        else:
-            re_han = re_han_default
-            re_skip = re_skip_default
+        re_han = re_han_default
+        re_skip = re_skip_default
         if cut_all:
             cut_block = self.__cut_all
         elif HMM:

jieba/_compat.py

@@ -1,38 +1,54 @@
 # -*- coding: utf-8 -*-
+import logging
 import os
 import sys
-import imp
-import logging
 
 log_console = logging.StreamHandler(sys.stderr)
 default_logger = logging.getLogger(__name__)
 default_logger.setLevel(logging.DEBUG)
 
+
 def setLogLevel(log_level):
-    global logger
     default_logger.setLevel(log_level)
 
+
+check_paddle_install = {'is_paddle_installed': False}
+
 try:
     import pkg_resources
+
     get_module_res = lambda *res: pkg_resources.resource_stream(__name__,
                                                                 os.path.join(*res))
 except ImportError:
     get_module_res = lambda *res: open(os.path.normpath(os.path.join(
-                            os.getcwd(), os.path.dirname(__file__), *res)), 'rb')
+        os.getcwd(), os.path.dirname(__file__), *res)), 'rb')
 
 
-def import_paddle():
-    import_paddle_check = False
+def enable_paddle():
     try:
         import paddle
-        if paddle.__version__ >= '1.6.1' or paddle.__version__ >= u'1.6.1':
-            import paddle.fluid as fluid
+    except ImportError:
+        default_logger.debug("Installing paddle-tiny, please wait a minute......")
+        os.system("pip install paddlepaddle-tiny")
+        try:
+            import paddle
+        except ImportError:
+            default_logger.debug(
+                "Import paddle error, please use command to install: pip install paddlepaddle-tiny==1.6.1."
+                "Now, back to jieba basic cut......")
+    if paddle.__version__ < '1.6.1':
+        default_logger.debug("Find your own paddle version doesn't satisfy the minimum requirement (1.6.1), "
+                             "please install paddle tiny by 'pip install --upgrade paddlepaddle-tiny', "
+                             "or upgrade paddle full version by "
+                             "'pip install --upgrade paddlepaddle (-gpu for GPU version)' ")
+    else:
+        try:
             import jieba.lac_small.predict as predict
-        import_paddle_check = True
-    except ImportError:   
-        default_logger.debug("Import paddle error, please use command to install: pip install paddlepaddle-tiny==1.6.1. Now, back to jieba basic cut......")
-        return False
-    return import_paddle_check
+            default_logger.debug("Paddle enabled successfully......")
+            check_paddle_install['is_paddle_installed'] = True
+        except ImportError:
+            default_logger.debug("Import error, cannot find paddle.fluid and jieba.lac_small.predict module. "
+                                 "Now, back to jieba basic cut......")
 
 
 PY2 = sys.version_info[0] == 2
@@ -56,6 +72,7 @@ else:
     itervalues = lambda d: iter(d.values())
     iteritems = lambda d: iter(d.items())
 
+
 def strdecode(sentence):
     if not isinstance(sentence, text_type):
         try:
@@ -64,25 +81,9 @@ def strdecode(sentence):
             sentence = sentence.decode('gbk', 'ignore')
     return sentence
 
+
 def resolve_filename(f):
     try:
         return f.name
     except AttributeError:
         return repr(f)
-
-
-def check_paddle_install():
-    is_paddle_installed =  False
-    try:
-        import paddle
-        if imp.find_module('paddle') and (paddle.__version__ >= '1.6.1' or paddle.__version__ >= u'1.6.1'):
-            is_paddle_installed = True
-        elif paddle.__version__ < '1.6.1':
-            is_paddle_installed = False
-            default_logger.debug("Check the paddle version is not correct, please\
-            use command to install paddle: pip uninstall paddlepaddle(-gpu), \
-            pip install paddlepaddle-tiny==1.6.1. Now, back to jieba basic cut......")
-    except ImportError:
-        default_logger.debug("import paddle error, back to jieba basic cut......")
-        is_paddle_installed = False
-    return is_paddle_installed

jieba/lac_small/reader_small.py

@@ -64,7 +64,7 @@ class Dataset(object):
     
     @property
     def vocab_size(self):
-        """vocabuary size"""
+        """vocabulary size"""
         return max(self.word2id_dict.values()) + 1
     
     @property
@@ -97,4 +97,4 @@ class Dataset(object):
         word_ids = self.word_to_ids(words)
         return word_ids
     
-   
+   

jieba/posseg/__init__.py

@@ -1,11 +1,11 @@
 from __future__ import absolute_import, unicode_literals
-import os
-import re
-import sys
-import jieba
+
 import pickle
-from .._compat import *
+import re
+
+import jieba
 from .viterbi import viterbi
+from .._compat import *
 
 PROB_START_P = "prob_start.p"
 PROB_TRANS_P = "prob_trans.p"
@@ -252,6 +252,7 @@ class POSTokenizer(object):
     def lcut(self, *args, **kwargs):
         return list(self.cut(*args, **kwargs))
 
+
 # default Tokenizer instance
 
 dt = POSTokenizer(jieba.dt)
@@ -276,20 +277,17 @@ def cut(sentence, HMM=True, use_paddle=False):
     Note that this only works using dt, custom POSTokenizer
     instances are not supported.
     """
-    is_paddle_installed = False
-    if use_paddle == True:
-        import_paddle_check = import_paddle()
-        is_paddle_installed = check_paddle_install()
-    if use_paddle==True and is_paddle_installed == True and import_paddle_check == True:
+    is_paddle_installed = check_paddle_install['is_paddle_installed']
+    if use_paddle and is_paddle_installed:
+        # if sentence is null, it will raise core exception in paddle.
         if sentence is None or sentence == "" or sentence == u"":
-            yield pair(None, None)
             return
         import jieba.lac_small.predict as predict
-        sents,tags = predict.get_result(strdecode(sentence))
-        for i,sent in enumerate(sents):
+        sents, tags = predict.get_result(strdecode(sentence))
+        for i, sent in enumerate(sents):
             if sent is None or tags[i] is None:
                 continue
-            yield pair(sent,tags[i])
+            yield pair(sent, tags[i])
         return
     global dt
     if jieba.pool is None:
@@ -306,5 +304,7 @@ def cut(sentence, HMM=True, use_paddle=False):
                 yield w
 
 
-def lcut(sentence, HMM=True):
+def lcut(sentence, HMM=True, use_paddle=False):
+    if use_paddle:
+        return list(cut(sentence, use_paddle=True))
     return list(cut(sentence, HMM))

setup.py

@@ -43,7 +43,7 @@ GitHub: https://github.com/fxsjy/jieba
 """
 
 setup(name='jieba',
-      version='0.40',
+      version='0.42.1',
       description='Chinese Words Segmentation Utilities',
       long_description=LONGDOC,
       author='Sun, Junyi',
@@ -71,5 +71,5 @@ setup(name='jieba',
       keywords='NLP,tokenizing,Chinese word segementation',
       packages=['jieba'],
       package_dir={'jieba':'jieba'},
-      package_data={'jieba':['*.*','finalseg/*','analyse/*','posseg/*', 'lac_small/*','lac_small/model_baseline/*']}
+      package_data={'jieba':['*.*','finalseg/*','analyse/*','posseg/*', 'lac_small/*.py','lac_small/*.dic', 'lac_small/model_baseline/*']}
 )

test/test_cutall.py

@@ -96,3 +96,6 @@ if __name__ == "__main__":
     cuttest('AT&T是一件不错的公司，给你发offer了吗？')
     cuttest('C++和c#是什么关系？11+122=133，是吗？PI=3.14159')
     cuttest('你认识那个和主席握手的的哥吗？他开一辆黑色的士。')
+    jieba.add_word('超敏C反应蛋白')
+    cuttest('超敏C反应蛋白是什么, java好学吗?,小潘老板都学Python')
+    cuttest('steel健身爆发力运动兴奋补充剂')

test/test_paddle.py

@@ -2,7 +2,7 @@
 import sys
 sys.path.append("../")
 import jieba
-
+jieba.enable_paddle()
 
 def cuttest(test_sent):
     result = jieba.cut(test_sent, use_paddle=True)

test/test_paddle_postag.py

@@ -2,7 +2,8 @@
 import sys
 sys.path.append("../")
 import jieba.posseg as pseg
-
+import jieba
+jieba.enable_paddle()
 
 def cuttest(test_sent):
     result = pseg.cut(test_sent, use_paddle=True)