|
4 | 4 | import re |
5 | 5 | import tempfile |
6 | 6 |
|
| 7 | + |
7 | 8 | class _FileCacheError(Exception): |
8 | | - '''Base exception class for FileCache related errors''' |
| 9 | + """Base exception class for FileCache related errors""" |
| 10 | + |
9 | 11 |
|
10 | 12 | class _FileCache(object): |
| 13 | + DEPTH = 3 |
| 14 | + |
| 15 | + def __init__(self, root_directory=None): |
| 16 | + self._InitializeRootDirectory(root_directory) |
| 17 | + |
| 18 | + def Get(self, key): |
| 19 | + path = self._GetPath(key) |
| 20 | + if os.path.exists(path): |
| 21 | + return open(path).read() |
| 22 | + else: |
| 23 | + return None |
| 24 | + |
| 25 | + def Set(self, key, data): |
| 26 | + path = self._GetPath(key) |
| 27 | + directory = os.path.dirname(path) |
| 28 | + if not os.path.exists(directory): |
| 29 | + os.makedirs(directory) |
| 30 | + if not os.path.isdir(directory): |
| 31 | + raise _FileCacheError('%s exists but is not a directory' % directory) |
| 32 | + temp_fd, temp_path = tempfile.mkstemp() |
| 33 | + temp_fp = os.fdopen(temp_fd, 'w') |
| 34 | + temp_fp.write(data) |
| 35 | + temp_fp.close() |
| 36 | + if not path.startswith(self._root_directory): |
| 37 | + raise _FileCacheError('%s does not appear to live under %s' % |
| 38 | + (path, self._root_directory)) |
| 39 | + if os.path.exists(path): |
| 40 | + os.remove(path) |
| 41 | + os.rename(temp_path, path) |
| 42 | + |
| 43 | + def Remove(self, key): |
| 44 | + path = self._GetPath(key) |
| 45 | + if not path.startswith(self._root_directory): |
| 46 | + raise _FileCacheError('%s does not appear to live under %s' % |
| 47 | + (path, self._root_directory )) |
| 48 | + if os.path.exists(path): |
| 49 | + os.remove(path) |
| 50 | + |
| 51 | + def GetCachedTime(self, key): |
| 52 | + path = self._GetPath(key) |
| 53 | + if os.path.exists(path): |
| 54 | + return os.path.getmtime(path) |
| 55 | + else: |
| 56 | + return None |
| 57 | + |
| 58 | + def _GetUsername(self): |
| 59 | + """Attempt to find the username in a cross-platform fashion.""" |
| 60 | + try: |
| 61 | + return os.getenv('USER') or \ |
| 62 | + os.getenv('LOGNAME') or \ |
| 63 | + os.getenv('USERNAME') or \ |
| 64 | + os.getlogin() or \ |
| 65 | + 'nobody' |
| 66 | + except (AttributeError, IOError, OSError), e: |
| 67 | + return 'nobody' |
| 68 | + |
| 69 | + def _GetTmpCachePath(self): |
| 70 | + username = self._GetUsername() |
| 71 | + cache_directory = 'python.cache_' + username |
| 72 | + return os.path.join(tempfile.gettempdir(), cache_directory) |
| 73 | + |
| 74 | + def _InitializeRootDirectory(self, root_directory): |
| 75 | + if not root_directory: |
| 76 | + root_directory = self._GetTmpCachePath() |
| 77 | + root_directory = os.path.abspath(root_directory) |
| 78 | + if not os.path.exists(root_directory): |
| 79 | + os.mkdir(root_directory) |
| 80 | + if not os.path.isdir(root_directory): |
| 81 | + raise _FileCacheError('%s exists but is not a directory' % |
| 82 | + root_directory) |
| 83 | + self._root_directory = root_directory |
| 84 | + |
| 85 | + def _GetPath(self, key): |
| 86 | + try: |
| 87 | + hashed_key = md5(key).hexdigest() |
| 88 | + except TypeError: |
| 89 | + hashed_key = md5.new(key).hexdigest() |
| 90 | + |
| 91 | + return os.path.join(self._root_directory, |
| 92 | + self._GetPrefix(hashed_key), |
| 93 | + hashed_key) |
| 94 | + |
| 95 | + def _GetPrefix(self, hashed_key): |
| 96 | + return os.path.sep.join(hashed_key[0:_FileCache.DEPTH]) |
11 | 97 |
|
12 | | - DEPTH = 3 |
13 | | - |
14 | | - def __init__(self,root_directory=None): |
15 | | - self._InitializeRootDirectory(root_directory) |
16 | | - |
17 | | - def Get(self, key): |
18 | | - path = self._GetPath(key) |
19 | | - if os.path.exists(path): |
20 | | - return open(path).read() |
21 | | - else: |
22 | | - return None |
23 | | - |
24 | | - def Set(self, key, data): |
25 | | - path = self._GetPath(key) |
26 | | - directory = os.path.dirname(path) |
27 | | - if not os.path.exists(directory): |
28 | | - os.makedirs(directory) |
29 | | - if not os.path.isdir(directory): |
30 | | - raise _FileCacheError('%s exists but is not a directory' % directory) |
31 | | - temp_fd, temp_path = tempfile.mkstemp() |
32 | | - temp_fp = os.fdopen(temp_fd, 'w') |
33 | | - temp_fp.write(data) |
34 | | - temp_fp.close() |
35 | | - if not path.startswith(self._root_directory): |
36 | | - raise _FileCacheError('%s does not appear to live under %s' % |
37 | | - (path, self._root_directory)) |
38 | | - if os.path.exists(path): |
39 | | - os.remove(path) |
40 | | - os.rename(temp_path, path) |
41 | | - |
42 | | - def Remove(self, key): |
43 | | - path = self._GetPath(key) |
44 | | - if not path.startswith(self._root_directory): |
45 | | - raise _FileCacheError('%s does not appear to live under %s' % |
46 | | - (path, self._root_directory )) |
47 | | - if os.path.exists(path): |
48 | | - os.remove(path) |
49 | | - |
50 | | - def GetCachedTime(self, key): |
51 | | - path = self._GetPath(key) |
52 | | - if os.path.exists(path): |
53 | | - return os.path.getmtime(path) |
54 | | - else: |
55 | | - return None |
56 | | - |
57 | | - def _GetUsername(self): |
58 | | - '''Attempt to find the username in a cross-platform fashion.''' |
59 | | - try: |
60 | | - return os.getenv('USER') or \ |
61 | | - os.getenv('LOGNAME') or \ |
62 | | - os.getenv('USERNAME') or \ |
63 | | - os.getlogin() or \ |
64 | | - 'nobody' |
65 | | - except (AttributeError, IOError, OSError), e: |
66 | | - return 'nobody' |
67 | | - |
68 | | - def _GetTmpCachePath(self): |
69 | | - username = self._GetUsername() |
70 | | - cache_directory = 'python.cache_' + username |
71 | | - return os.path.join(tempfile.gettempdir(), cache_directory) |
72 | | - |
73 | | - def _InitializeRootDirectory(self, root_directory): |
74 | | - if not root_directory: |
75 | | - root_directory = self._GetTmpCachePath() |
76 | | - root_directory = os.path.abspath(root_directory) |
77 | | - if not os.path.exists(root_directory): |
78 | | - os.mkdir(root_directory) |
79 | | - if not os.path.isdir(root_directory): |
80 | | - raise _FileCacheError('%s exists but is not a directory' % |
81 | | - root_directory) |
82 | | - self._root_directory = root_directory |
83 | | - |
84 | | - def _GetPath(self, key): |
85 | | - try: |
86 | | - hashed_key = md5(key).hexdigest() |
87 | | - except TypeError: |
88 | | - hashed_key = md5.new(key).hexdigest() |
89 | | - |
90 | | - return os.path.join(self._root_directory, |
91 | | - self._GetPrefix(hashed_key), |
92 | | - hashed_key) |
93 | | - |
94 | | - def _GetPrefix(self, hashed_key): |
95 | | - return os.path.sep.join(hashed_key[0:_FileCache.DEPTH]) |
96 | 98 |
|
97 | 99 | class ParseTweet: |
98 | | - # compile once on import |
99 | | - regexp = { "RT": "^RT", "MT":r"^MT", "ALNUM": r"(@[a-zA-Z0-9_]+)", |
100 | | - "HASHTAG": r"(#[\w\d]+)", "URL": r"([http://]?[a-zA-Z\d\/]+[\.]+[a-zA-Z\d\/\.]+)" } |
101 | | - regexp = dict((key,re.compile(value)) for key,value in regexp.items()) |
102 | | - |
103 | | - def __init__(self,timeline_owner,tweet): |
104 | | - ''' timeline_owner : twitter handle of user account. tweet - 140 chars from feed; object does all computation on construction |
105 | | - properties: |
106 | | - RT, MT - boolean |
107 | | - URLs - list of URL |
108 | | - Hashtags - list of tags |
109 | | - ''' |
110 | | - self.Owner = timeline_owner |
111 | | - self.tweet = tweet |
112 | | - self.UserHandles = ParseTweet.getUserHandles(tweet) |
113 | | - self.Hashtags = ParseTweet.getHashtags(tweet) |
114 | | - self.URLs = ParseTweet.getURLs(tweet) |
115 | | - self.RT = ParseTweet.getAttributeRT(tweet) |
116 | | - self.MT = ParseTweet.getAttributeMT(tweet) |
117 | | - |
118 | | - # additional intelligence |
119 | | - if ( self.RT and len(self.UserHandles) > 0 ): #change the owner of tweet? |
120 | | - self.Owner = self.UserHandles[0] |
121 | | - return |
122 | | - |
123 | | - def __str__(self): |
124 | | - ''' for display method ''' |
125 | | - return "owner %s, urls: %d, hashtags %d, user_handles %d, len_tweet %d, RT = %s, MT = %s"%(self.Owner,len(self.URLs),len(self.Hashtags),len(self.UserHandles), len(self.tweet), self.RT,self.MT) |
126 | | - |
127 | | - @staticmethod |
128 | | - def getAttributeRT( tweet ): |
129 | | - """ see if tweet is a RT """ |
130 | | - return re.search(ParseTweet.regexp["RT"],tweet.strip()) != None |
131 | | - |
132 | | - @staticmethod |
133 | | - def getAttributeMT( tweet ): |
134 | | - """ see if tweet is a MT """ |
135 | | - return re.search(ParseTweet.regexp["MT"],tweet.strip()) != None |
136 | | - |
137 | | - @staticmethod |
138 | | - def getUserHandles( tweet ): |
139 | | - """ given a tweet we try and extract all user handles in order of occurrence""" |
140 | | - return re.findall(ParseTweet.regexp["ALNUM"],tweet) |
141 | | - |
142 | | - @staticmethod |
143 | | - def getHashtags( tweet ): |
144 | | - """ return all hashtags""" |
145 | | - return re.findall(ParseTweet.regexp["HASHTAG"],tweet) |
146 | | - |
147 | | - @staticmethod |
148 | | - def getURLs( tweet ): |
149 | | - """ URL : [http://]?[\w\.?/]+""" |
150 | | - return re.findall(ParseTweet.regexp["URL"],tweet) |
| 100 | + # compile once on import |
| 101 | + regexp = {"RT": "^RT", "MT": r"^MT", "ALNUM": r"(@[a-zA-Z0-9_]+)", |
| 102 | + "HASHTAG": r"(#[\w\d]+)", "URL": r"([http://]?[a-zA-Z\d\/]+[\.]+[a-zA-Z\d\/\.]+)"} |
| 103 | + regexp = dict((key, re.compile(value)) for key, value in regexp.items()) |
| 104 | + |
| 105 | + def __init__(self, timeline_owner, tweet): |
| 106 | + """ timeline_owner : twitter handle of user account. tweet - 140 chars from feed; object does all computation on construction |
| 107 | + properties: |
| 108 | + RT, MT - boolean |
| 109 | + URLs - list of URL |
| 110 | + Hashtags - list of tags |
| 111 | + """ |
| 112 | + self.Owner = timeline_owner |
| 113 | + self.tweet = tweet |
| 114 | + self.UserHandles = ParseTweet.getUserHandles(tweet) |
| 115 | + self.Hashtags = ParseTweet.getHashtags(tweet) |
| 116 | + self.URLs = ParseTweet.getURLs(tweet) |
| 117 | + self.RT = ParseTweet.getAttributeRT(tweet) |
| 118 | + self.MT = ParseTweet.getAttributeMT(tweet) |
| 119 | + |
| 120 | + # additional intelligence |
| 121 | + if ( self.RT and len(self.UserHandles) > 0 ): # change the owner of tweet? |
| 122 | + self.Owner = self.UserHandles[0] |
| 123 | + return |
| 124 | + |
| 125 | + def __str__(self): |
| 126 | + """ for display method """ |
| 127 | + return "owner %s, urls: %d, hashtags %d, user_handles %d, len_tweet %d, RT = %s, MT = %s" % ( |
| 128 | + self.Owner, len(self.URLs), len(self.Hashtags), len(self.UserHandles), len(self.tweet), self.RT, self.MT) |
| 129 | + |
| 130 | + @staticmethod |
| 131 | + def getAttributeRT(tweet): |
| 132 | + """ see if tweet is a RT """ |
| 133 | + return re.search(ParseTweet.regexp["RT"], tweet.strip()) != None |
| 134 | + |
| 135 | + @staticmethod |
| 136 | + def getAttributeMT(tweet): |
| 137 | + """ see if tweet is a MT """ |
| 138 | + return re.search(ParseTweet.regexp["MT"], tweet.strip()) != None |
| 139 | + |
| 140 | + @staticmethod |
| 141 | + def getUserHandles(tweet): |
| 142 | + """ given a tweet we try and extract all user handles in order of occurrence""" |
| 143 | + return re.findall(ParseTweet.regexp["ALNUM"], tweet) |
| 144 | + |
| 145 | + @staticmethod |
| 146 | + def getHashtags(tweet): |
| 147 | + """ return all hashtags""" |
| 148 | + return re.findall(ParseTweet.regexp["HASHTAG"], tweet) |
| 149 | + |
| 150 | + @staticmethod |
| 151 | + def getURLs(tweet): |
| 152 | + """ URL : [http://]?[\w\.?/]+""" |
| 153 | + return re.findall(ParseTweet.regexp["URL"], tweet) |
0 commit comments