Lib/nntplib.py

   1 # An NNTP client class.  Based on RFC 977: Network News Transfer
   2 # Protocol, by Brian Kantor and Phil Lapsley.
   3
   4
   5 # Example:
   6 #
   7 # >>> from nntplib import NNTP
   8 # >>> s = NNTP('news')
   9 # >>> resp, count, first, last, name = s.group('comp.lang.python')
  10 # >>> print 'Group', name, 'has', count, 'articles, range', first, 'to', last
  11 # Group comp.lang.python has 51 articles, range 5770 to 5821
  12 # >>> resp, subs = s.xhdr('subject', first + '-' + last)
  13 # >>> resp = s.quit()
  14 # >>>
  15 #
  16 # Here 'resp' is the server response line.
  17 # Error responses are turned into exceptions.
  18 #
  19 # To post an article from a file:
  20 # >>> f = open(filename, 'r') # file containing article, including header
  21 # >>> resp = s.post(f)
  22 # >>>
  23 #
  24 # For descriptions of all methods, read the comments in the code below.
  25 # Note that all arguments and return values representing article numbers
  26 # are strings, not numbers, since they are rarely used for calculations.
  27
  28 # (xover, xgtitle, xpath, date methods by Kevan Heydon)
  29
  30
  31 # Imports
  32 import regex
  33 import socket
  34 import string
  35
  36
  37 # Exception raised when an error or invalid response is received
  38
  39 error_reply = 'nntplib.error_reply'     # unexpected [123]xx reply
  40 error_temp = 'nntplib.error_temp'       # 4xx errors
  41 error_perm = 'nntplib.error_perm'       # 5xx errors
  42 error_proto = 'nntplib.error_proto'     # response does not begin with [1-5]
  43 error_data = 'nntplib.error_data'       # error in response data
  44
  45
  46 # Standard port used by NNTP servers
  47 NNTP_PORT = 119
  48
  49
  50 # Response numbers that are followed by additional text (e.g. article)
  51 LONGRESP = ['100', '215', '220', '221', '222', '224', '230', '231', '282']
  52
  53
  54 # Line terminators (we always output CRLF, but accept any of CRLF, CR, LF)
  55 CRLF = '\r\n'
  56
  57
  58 # The class itself
  59
  60 class NNTP:
  61
  62         # Initialize an instance.  Arguments:
  63         # - host: hostname to connect to
  64         # - port: port to connect to (default the standard NNTP port)
  65
  66         def __init__(self, host, port = NNTP_PORT):
  67                 self.host = host
  68                 self.port = port
  69                 self.sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
  70                 self.sock.connect(self.host, self.port)
  71                 self.file = self.sock.makefile('rb')
  72                 self.debugging = 0
  73                 self.welcome = self.getresp()
  74
  75         # Get the welcome message from the server
  76         # (this is read and squirreled away by __init__()).
  77         # If the response code is 200, posting is allowed;
  78         # if it 201, posting is not allowed
  79
  80         def getwelcome(self):
  81                 if self.debugging: print '*welcome*', `self.welcome`
  82                 return self.welcome
  83
  84         # Set the debugging level.  Argument level means:
  85         # 0: no debugging output (default)
  86         # 1: print commands and responses but not body text etc.
  87         # 2: also print raw lines read and sent before stripping CR/LF
  88
  89         def set_debuglevel(self, level):
  90                 self.debugging = level
  91         debug = set_debuglevel
  92
  93         # Internal: send one line to the server, appending CRLF
  94         def putline(self, line):
  95                 line = line + CRLF
  96                 if self.debugging > 1: print '*put*', `line`
  97                 self.sock.send(line)
  98
  99         # Internal: send one command to the server (through putline())
 100         def putcmd(self, line):
 101                 if self.debugging: print '*cmd*', `line`
 102                 self.putline(line)
 103
 104         # Internal: return one line from the server, stripping CRLF.
 105         # Raise EOFError if the connection is closed
 106         def getline(self):
 107                 line = self.file.readline()
 108                 if self.debugging > 1:
 109                         print '*get*', `line`
 110                 if not line: raise EOFError
 111                 if line[-2:] == CRLF: line = line[:-2]
 112                 elif line[-1:] in CRLF: line = line[:-1]
 113                 return line
 114
 115         # Internal: get a response from the server.
 116         # Raise various errors if the response indicates an error
 117         def getresp(self):
 118                 resp = self.getline()
 119                 if self.debugging: print '*resp*', `resp`
 120                 c = resp[:1]
 121                 if c == '4':
 122                         raise error_temp, resp
 123                 if c == '5':
 124                         raise error_perm, resp
 125                 if c not in '123':
 126                         raise error_proto, resp
 127                 return resp
 128
 129         # Internal: get a response plus following text from the server.
 130         # Raise various errors if the response indicates an error
 131         def getlongresp(self):
 132                 resp = self.getresp()
 133                 if resp[:3] not in LONGRESP:
 134                         raise error_reply, resp
 135                 list = []
 136                 while 1:
 137                         line = self.getline()
 138                         if line == '.':
 139                                 break
 140                         list.append(line)
 141                 return resp, list
 142
 143         # Internal: send a command and get the response
 144         def shortcmd(self, line):
 145                 self.putcmd(line)
 146                 return self.getresp()
 147
 148         # Internal: send a command and get the response plus following text
 149         def longcmd(self, line):
 150                 self.putcmd(line)
 151                 return self.getlongresp()
 152
 153         # Process a NEWGROUPS command.  Arguments:
 154         # - date: string 'yymmdd' indicating the date
 155         # - time: string 'hhmmss' indicating the time
 156         # Return:
 157         # - resp: server response if succesful
 158         # - list: list of newsgroup names
 159
 160         def newgroups(self, date, time):
 161                 return self.longcmd('NEWGROUPS ' + date + ' ' + time)
 162
 163         # Process a NEWNEWS command.  Arguments:
 164         # - group: group name or '*'
 165         # - date: string 'yymmdd' indicating the date
 166         # - time: string 'hhmmss' indicating the time
 167         # Return:
 168         # - resp: server response if succesful
 169         # - list: list of article ids
 170
 171         def newnews(self, group, date, time):
 172                 cmd = 'NEWNEWS ' + group + ' ' + date + ' ' + time
 173                 return self.longcmd(cmd)
 174
 175         # Process a LIST command.  Return:
 176         # - resp: server response if succesful
 177         # - list: list of (group, last, first, flag) (strings)
 178
 179         def list(self):
 180                 resp, list = self.longcmd('LIST')
 181                 for i in range(len(list)):
 182                         # Parse lines into "group last first flag"
 183                         list[i] = string.split(list[i])
 184                 return resp, list
 185
 186         # Process a GROUP command.  Argument:
 187         # - group: the group name
 188         # Returns:
 189         # - resp: server response if succesful
 190         # - count: number of articles (string)
 191         # - first: first article number (string)
 192         # - last: last article number (string)
 193         # - name: the group name
 194
 195         def group(self, name):
 196                 resp = self.shortcmd('GROUP ' + name)
 197                 if resp[:3] <> '211':
 198                         raise error_reply, resp
 199                 words = string.split(resp)
 200                 count = first = last = 0
 201                 n = len(words)
 202                 if n > 1:
 203                         count = words[1]
 204                         if n > 2:
 205                                 first = words[2]
 206                                 if n > 3:
 207                                         last = words[3]
 208                                         if n > 4:
 209                                                 name = string.lower(words[4])
 210                 return resp, count, first, last, name
 211
 212         # Process a HELP command.  Returns:
 213         # - resp: server response if succesful
 214         # - list: list of strings
 215
 216         def help(self):
 217                 return self.longcmd('HELP')
 218
 219         # Internal: parse the response of a STAT, NEXT or LAST command
 220         def statparse(self, resp):
 221                 if resp[:2] <> '22':
 222                         raise error_reply, resp
 223                 words = string.split(resp)
 224                 nr = 0
 225                 id = ''
 226                 n = len(words)
 227                 if n > 1:
 228                         nr = words[1]
 229                         if n > 2:
 230                                 id = string.lower(words[2])
 231                 return resp, nr, id
 232
 233         # Internal: process a STAT, NEXT or LAST command
 234         def statcmd(self, line):
 235                 resp = self.shortcmd(line)
 236                 return self.statparse(resp)
 237
 238         # Process a STAT command.  Argument:
 239         # - id: article number or message id
 240         # Returns:
 241         # - resp: server response if succesful
 242         # - nr:   the article number
 243         # - id:   the article id
 244
 245         def stat(self, id):
 246                 return self.statcmd('STAT ' + id)
 247
 248         # Process a NEXT command.  No arguments.  Return as for STAT
 249
 250         def next(self):
 251                 return self.statcmd('NEXT')
 252
 253         # Process a LAST command.  No arguments.  Return as for STAT
 254
 255         def last(self):
 256                 return self.statcmd('LAST')
 257
 258         # Internal: process a HEAD, BODY or ARTICLE command
 259         def artcmd(self, line):
 260                 resp, list = self.longcmd(line)
 261                 resp, nr, id = self.statparse(resp)
 262                 return resp, nr, id, list
 263
 264         # Process a HEAD command.  Argument:
 265         # - id: article number or message id
 266         # Returns:
 267         # - resp: server response if succesful
 268         # - list: the lines of the article's header
 269
 270         def head(self, id):
 271                 return self.artcmd('HEAD ' + id)
 272
 273         # Process a BODY command.  Argument:
 274         # - id: article number or message id
 275         # Returns:
 276         # - resp: server response if succesful
 277         # - list: the lines of the article's body
 278
 279         def body(self, id):
 280                 return self.artcmd('BODY ' + id)
 281
 282         # Process an ARTICLE command.  Argument:
 283         # - id: article number or message id
 284         # Returns:
 285         # - resp: server response if succesful
 286         # - list: the lines of the article
 287
 288         def article(self, id):
 289                 return self.artcmd('ARTICLE ' + id)
 290
 291         # Process a SLAVE command.  Returns:
 292         # - resp: server response if succesful
 293
 294         def slave(self):
 295                 return self.shortcmd('SLAVE')
 296
 297         # Process an XHDR command (optional server extension).  Arguments:
 298         # - hdr: the header type (e.g. 'subject')
 299         # - str: an article nr, a message id, or a range nr1-nr2
 300         # Returns:
 301         # - resp: server response if succesful
 302         # - list: list of (nr, value) strings
 303
 304         def xhdr(self, hdr, str):
 305                 resp, lines = self.longcmd('XHDR ' + hdr + ' ' + str)
 306                 for i in range(len(lines)):
 307                         line = lines[i]
 308                         n = regex.match('^[0-9]+', line)
 309                         nr = line[:n]
 310                         if n < len(line) and line[n] == ' ': n = n+1
 311                         lines[i] = (nr, line[n:])
 312                 return resp, lines
 313
 314         # Process an XOVER command (optional server extension) Arguments:
 315         # - start: start of range
 316         # - end: end of range
 317         # Returns:
 318         # - resp: server response if succesful
 319         # - list: list of (art-nr, subject, poster, date, id, refrences, size, lines)
 320
 321         def xover(self,start,end):
 322                 resp, lines = self.longcmd('XOVER ' + start + '-' + end)
 323                 xover_lines = []
 324                 for line in lines:
 325                         elem = string.splitfields(line,"\t")
 326                         try:
 327                                 xover_lines.append(elem[0],
 328                                                    elem[1],
 329                                                    elem[2],
 330                                                    elem[3],
 331                                                    elem[4],
 332                                                    elem[5:-2],
 333                                                    elem[-2],
 334                                                    elem[-1])
 335                         except IndexError:
 336                                 raise error_data,line
 337                 return resp,xover_lines
 338
 339         # Process an XGTITLE command (optional server extension) Arguments:
 340         # - group: group name wildcard (i.e. news.*)
 341         # Returns:
 342         # - resp: server response if succesful
 343         # - list: list of (name,title) strings
 344
 345         def xgtitle(self, group):
 346                 line_pat = regex.compile("^\([^ \t]+\)[ \t]+\(.*\)$")
 347                 resp, raw_lines = self.longcmd('XGTITLE ' + group)
 348                 lines = []
 349                 for raw_line in raw_lines:
 350                         if line_pat.search(string.strip(raw_line)) == 0:
 351                                 lines.append(line_pat.group(1),
 352                                              line_pat.group(2))
 353
 354                 return resp, lines
 355
 356         # Process an XPATH command (optional server extension) Arguments:
 357         # - id: Message id of article
 358         # Returns:
 359         # resp: server response if succesful
 360         # path: directory path to article
 361
 362         def xpath(self,id):
 363                 resp = self.shortcmd("XPATH " + id)
 364                 if resp[:3] <> '223':
 365                         raise error_reply, resp
 366                 try:
 367                         [resp_num, path] = string.split(resp)
 368                 except ValueError:
 369                         raise error_reply, resp
 370                 else:
 371                         return resp, path
 372
 373         # Process the DATE command. Arguments:
 374         # None
 375         # Returns:
 376         # resp: server response if succesful
 377         # date: Date suitable for newnews/newgroups commands etc.
 378         # time: Time suitable for newnews/newgroups commands etc.
 379
 380         def date (self):
 381                 resp = self.shortcmd("DATE")
 382                 if resp[:3] <> '111':
 383                         raise error_reply, resp
 384                 elem = string.split(resp)
 385                 if len(elem) != 2:
 386                         raise error_data, resp
 387                 date = elem[1][2:8]
 388                 time = elem[1][-6:]
 389                 if len(date) != 6 or len(time) != 6:
 390                         raise error_data, resp
 391                 return resp, date, time
 392
 393
 394         # Process a POST command.  Arguments:
 395         # - f: file containing the article
 396         # Returns:
 397         # - resp: server response if succesful
 398
 399         def post(self, f):
 400                 resp = self.shortcmd('POST')
 401                 # Raises error_??? if posting is not allowed
 402                 if resp[0] <> '3':
 403                         raise error_reply, resp
 404                 while 1:
 405                         line = f.readline()
 406                         if not line:
 407                                 break
 408                         if line[-1] == '\n':
 409                                 line = line[:-1]
 410                         if line == '.':
 411                                 line = '..'
 412                         self.putline(line)
 413                 self.putline('.')
 414                 return self.getresp()
 415
 416         # Process an IHAVE command.  Arguments:
 417         # - id: message-id of the article
 418         # - f:  file containing the article
 419         # Returns:
 420         # - resp: server response if succesful
 421         # Note that if the server refuses the article an exception is raised
 422
 423         def ihave(self, id, f):
 424                 resp = self.shortcmd('IHAVE ' + id)
 425                 # Raises error_??? if the server already has it
 426                 if resp[0] <> '3':
 427                         raise error_reply, resp
 428                 while 1:
 429                         line = f.readline()
 430                         if not line:
 431                                 break
 432                         if line[-1] == '\n':
 433                                 line = line[:-1]
 434                         if line == '.':
 435                                 line = '..'
 436                         self.putline(line)
 437                 self.putline('.')
 438                 return self.getresp()
 439
 440          # Process a QUIT command and close the socket.  Returns:
 441          # - resp: server response if succesful
 442
 443         def quit(self):
 444                 resp = self.shortcmd('QUIT')
 445                 self.file.close()
 446                 self.sock.close()
 447                 del self.file, self.sock
 448                 return resp