Guido van Rossum | 7c6ebb5 | 1994-03-22 12:05:32 +0000 | [diff] [blame^] | 1 | # Open an arbitrary URL |
| 2 | # |
| 3 | # See the following document for a tentative description of URLs: |
| 4 | # Uniform Resource Locators Tim Berners-Lee |
| 5 | # INTERNET DRAFT CERN |
| 6 | # IETF URL Working Group 14 July 1993 |
| 7 | # draft-ietf-uri-url-01.txt |
| 8 | # |
| 9 | # The object returned by URLopener().open(file) will differ per |
| 10 | # protocol. All you know is that is has methods read(), readline(), |
| 11 | # readlines(), fileno(), close() and info(). The read*(), fileno() |
| 12 | # and close() methods work like those of open files. |
| 13 | # The info() method returns an rfc822.Message object which can be |
| 14 | # used to query various info about the object, if available. |
| 15 | # (rfc822.Message objects are queried with the getheader() method.) |
| 16 | |
| 17 | import socket |
| 18 | import regex |
| 19 | |
| 20 | |
| 21 | # This really consists of two pieces: |
| 22 | # (1) a class which handles opening of all sorts of URLs |
| 23 | # (plus assorted utilities etc.) |
| 24 | # (2) a set of functions for parsing URLs |
| 25 | # XXX Should these be separated out into different modules? |
| 26 | |
| 27 | |
| 28 | # Shortcut for basic usage |
| 29 | _urlopener = None |
| 30 | def urlopen(url): |
| 31 | global _urlopener |
| 32 | if not _urlopener: |
| 33 | _urlopener = URLopener() |
| 34 | return _urlopener.open(url) |
| 35 | def urlretrieve(url): |
| 36 | global _urlopener |
| 37 | if not _urlopener: |
| 38 | _urlopener = URLopener() |
| 39 | return _urlopener.retrieve(url) |
| 40 | def urlcleanup(): |
| 41 | if _urlopener: |
| 42 | _urlopener.cleanup() |
| 43 | |
| 44 | |
| 45 | # Class to open URLs. |
| 46 | # This is a class rather than just a subroutine because we may need |
| 47 | # more than one set of global protocol-specific options. |
| 48 | ftpcache = {} |
| 49 | class URLopener: |
| 50 | |
| 51 | # Constructor |
| 52 | def __init__(self): |
| 53 | self.addheaders = [] |
| 54 | self.tempcache = {} |
| 55 | self.ftpcache = ftpcache |
| 56 | # Undocumented feature: you can use a different |
| 57 | # ftp cache by assigning to the .ftpcache member; |
| 58 | # in case you want logically independent URL openers |
| 59 | |
| 60 | def __del__(self): |
| 61 | self.close() |
| 62 | |
| 63 | def close(self): |
| 64 | self.cleanup() |
| 65 | |
| 66 | def cleanup(self): |
| 67 | import os |
| 68 | for url in self.tempcache.keys(): |
| 69 | try: |
| 70 | os.unlink(self.tempcache[url][0]) |
| 71 | except os.error: |
| 72 | pass |
| 73 | del self.tempcache[url] |
| 74 | |
| 75 | # Add a header to be used by the HTTP interface only |
| 76 | # e.g. u.addheader('Accept', 'sound/basic') |
| 77 | def addheader(self, *args): |
| 78 | self.addheaders.append(args) |
| 79 | |
| 80 | # External interface |
| 81 | # Use URLopener().open(file) instead of open(file, 'r') |
| 82 | def open(self, url): |
| 83 | type, url = splittype(unwrap(url)) |
| 84 | if not type: type = 'file' |
| 85 | name = 'open_' + type |
| 86 | if '-' in name: |
| 87 | import regsub |
| 88 | name = regsub.gsub('-', '_', name) |
| 89 | if not hasattr(self, name): |
| 90 | raise IOError, ('url error', 'unknown url type', type) |
| 91 | try: |
| 92 | return getattr(self, name)(url) |
| 93 | except socket.error, msg: |
| 94 | raise IOError, ('socket error', msg) |
| 95 | |
| 96 | # External interface |
| 97 | # retrieve(url) returns (filename, None) for a local object |
| 98 | # or (tempfilename, headers) for a remote object |
| 99 | def retrieve(self, url): |
| 100 | if self.tempcache.has_key(url): |
| 101 | return self.tempcache[url] |
| 102 | url1 = unwrap(url) |
| 103 | if self.tempcache.has_key(url1): |
| 104 | self.tempcache[url] = self.tempcache[url1] |
| 105 | return self.tempcache[url1] |
| 106 | type, url1 = splittype(url1) |
| 107 | if not type or type == 'file': |
| 108 | try: |
| 109 | fp = self.open_local_file(url1) |
| 110 | del fp |
| 111 | return splithost(url1)[1], None |
| 112 | except IOError, msg: |
| 113 | pass |
| 114 | fp = self.open(url) |
| 115 | headers = fp.info() |
| 116 | import tempfile |
| 117 | tfn = tempfile.mktemp() |
| 118 | self.tempcache[url] = result = tfn, headers |
| 119 | tfp = open(tfn, 'w') |
| 120 | bs = 1024*8 |
| 121 | block = fp.read(bs) |
| 122 | while block: |
| 123 | tfp.write(block) |
| 124 | block = fp.read(bs) |
| 125 | del fp |
| 126 | del tfp |
| 127 | return result |
| 128 | |
| 129 | # Each method named open_<type> knows how to open that type of URL |
| 130 | |
| 131 | # Use HTTP protocol |
| 132 | def open_http(self, url): |
| 133 | import httplib |
| 134 | host, selector = splithost(url) |
| 135 | h = httplib.HTTP(host) |
| 136 | h.putrequest('GET', selector) |
| 137 | for args in self.addheaders: apply(h.putheader, args) |
| 138 | errcode, errmsg, headers = h.getreply() |
| 139 | if errcode == 200: return addinfo(h.getfile(), headers) |
| 140 | else: raise IOError, ('http error', errcode, errmsg, headers) |
| 141 | |
| 142 | # Use Gopher protocol |
| 143 | def open_gopher(self, url): |
| 144 | import gopherlib |
| 145 | host, selector = splithost(url) |
| 146 | type, selector = splitgophertype(selector) |
| 147 | selector, query = splitquery(selector) |
| 148 | if query: fp = gopherlib.send_query(selector, query, host) |
| 149 | else: fp = gopherlib.send_selector(selector, host) |
| 150 | return addinfo(fp, noheaders()) |
| 151 | |
| 152 | # Use local file or FTP depending on form of URL |
| 153 | def open_file(self, url): |
| 154 | try: |
| 155 | return self.open_local_file(url) |
| 156 | except IOError: |
| 157 | return self.open_ftp(url) |
| 158 | |
| 159 | # Use local file |
| 160 | def open_local_file(self, url): |
| 161 | host, file = splithost(url) |
| 162 | if not host: return addinfo(open(file, 'r'), noheaders()) |
| 163 | host, port = splitport(host) |
| 164 | if not port and socket.gethostbyname(host) in ( |
| 165 | localhost(), thishost()): |
| 166 | return addinfo(open(file, 'r'), noheaders()) |
| 167 | raise IOError, ('local file error', 'not on local host') |
| 168 | |
| 169 | # Use FTP protocol |
| 170 | def open_ftp(self, url): |
| 171 | host, file = splithost(url) |
| 172 | if not host: raise IOError, ('ftp error', 'no host given') |
| 173 | host, port = splitport(host) |
| 174 | host = socket.gethostbyname(host) |
| 175 | if not port: |
| 176 | import ftplib |
| 177 | port = ftplib.FTP_PORT |
| 178 | key = (host, port) |
| 179 | try: |
| 180 | if not self.ftpcache.has_key(key): |
| 181 | self.ftpcache[key] = ftpwrapper(host, port) |
| 182 | return addinfo(self.ftpcache[key].retrfile(file), |
| 183 | noheaders()) |
| 184 | except ftperrors(), msg: |
| 185 | raise IOError, ('ftp error', msg) |
| 186 | |
| 187 | |
| 188 | # Utility functions |
| 189 | |
| 190 | # Return the IP address of the magic hostname 'localhost' |
| 191 | _localhost = None |
| 192 | def localhost(): |
| 193 | global _localhost |
| 194 | if not _localhost: |
| 195 | _localhost = socket.gethostbyname('localhost') |
| 196 | return _localhost |
| 197 | |
| 198 | # Return the IP address of the current host |
| 199 | _thishost = None |
| 200 | def thishost(): |
| 201 | global _thishost |
| 202 | if not _thishost: |
| 203 | _thishost = socket.gethostbyname(socket.gethostname()) |
| 204 | return _thishost |
| 205 | |
| 206 | # Return the set of errors raised by the FTP class |
| 207 | _ftperrors = None |
| 208 | def ftperrors(): |
| 209 | global _ftperrors |
| 210 | if not _ftperrors: |
| 211 | import ftplib |
| 212 | _ftperrors = (ftplib.error_reply, |
| 213 | ftplib.error_temp, |
| 214 | ftplib.error_perm, |
| 215 | ftplib.error_proto) |
| 216 | return _ftperrors |
| 217 | |
| 218 | # Return an empty rfc822.Message object |
| 219 | _noheaders = None |
| 220 | def noheaders(): |
| 221 | global _noheaders |
| 222 | if not _noheaders: |
| 223 | import rfc822 |
| 224 | _noheaders = rfc822.Message(open('/dev/null', 'r')) |
| 225 | _noheaders.fp.close() # Recycle file descriptor |
| 226 | return _noheaders |
| 227 | |
| 228 | |
| 229 | # Utility classes |
| 230 | |
| 231 | # Class used by open_ftp() for cache of open FTP connections |
| 232 | class ftpwrapper: |
| 233 | def __init__(self, host, port): |
| 234 | self.host = host |
| 235 | self.port = port |
| 236 | self.init() |
| 237 | def init(self): |
| 238 | import ftplib |
| 239 | self.ftp = ftplib.FTP() |
| 240 | self.ftp.connect(self.host, self.port) |
| 241 | self.ftp.login() |
| 242 | def retrfile(self, file): |
| 243 | import ftplib |
| 244 | try: |
| 245 | self.ftp.voidcmd('TYPE I') |
| 246 | except ftplib.all_errors: |
| 247 | self.init() |
| 248 | self.ftp.voidcmd('TYPE I') |
| 249 | conn = None |
| 250 | if file: |
| 251 | try: |
| 252 | cmd = 'RETR ' + file |
| 253 | conn = self.ftp.transfercmd(cmd) |
| 254 | except ftplib.error_perm, reason: |
| 255 | if reason[:3] != '550': |
| 256 | raise IOError, ('ftp error', reason) |
| 257 | if not conn: |
| 258 | # Try a directory listing |
| 259 | if file: cmd = 'LIST ' + file |
| 260 | else: cmd = 'LIST' |
| 261 | conn = self.ftp.transfercmd(cmd) |
| 262 | return addclosehook(conn.makefile('r'), self.ftp.voidresp) |
| 263 | |
| 264 | # Base class for addinfo and addclosehook |
| 265 | class addbase: |
| 266 | def __init__(self, fp): |
| 267 | self.fp = fp |
| 268 | self.read = self.fp.read |
| 269 | self.readline = self.fp.readline |
| 270 | self.readlines = self.fp.readlines |
| 271 | self.fileno = self.fp.fileno |
| 272 | def __repr__(self): |
| 273 | return '<%s at %s whose fp = %s>' % ( |
| 274 | self.__class__.__name__, `id(self)`, `self.fp`) |
| 275 | def __del__(self): |
| 276 | self.close() |
| 277 | def close(self): |
| 278 | self.read = None |
| 279 | self.readline = None |
| 280 | self.readlines = None |
| 281 | self.fileno = None |
| 282 | self.fp = None |
| 283 | |
| 284 | # Class to add a close hook to an open file |
| 285 | class addclosehook(addbase): |
| 286 | def __init__(self, fp, closehook, *hookargs): |
| 287 | addbase.__init__(self, fp) |
| 288 | self.closehook = closehook |
| 289 | self.hookargs = hookargs |
| 290 | def close(self): |
| 291 | if self.closehook: |
| 292 | apply(self.closehook, self.hookargs) |
| 293 | self.closehook = None |
| 294 | self.hookargs = None |
| 295 | addbase.close(self) |
| 296 | |
| 297 | # class to add an info() method to an open file |
| 298 | class addinfo(addbase): |
| 299 | def __init__(self, fp, headers): |
| 300 | addbase.__init__(self, fp) |
| 301 | self.headers = headers |
| 302 | def info(self): |
| 303 | return self.headers |
| 304 | |
| 305 | |
| 306 | # Utility to combine a URL with a base URL to form a new URL |
| 307 | |
| 308 | def basejoin(base, url): |
| 309 | type, path = splittype(url) |
| 310 | if type: return url |
| 311 | host, path = splithost(path) |
| 312 | basetype, basepath = splittype(base) |
| 313 | basehost, basepath = splithost(basepath) |
| 314 | basepath, basetag = splittag(basepath) |
| 315 | basepath, basequery = splitquery(basepath) |
| 316 | type = basetype or 'file' |
| 317 | if path[:1] != '/': |
| 318 | import string |
| 319 | i = string.rfind(basepath, '/') |
| 320 | if i < 0: basepath = '/' |
| 321 | else: basepath = basepath[:i+1] |
| 322 | path = basepath + path |
| 323 | if not host: host = basehost |
| 324 | if host: return type + '://' + host + path |
| 325 | else: return type + ':' + path |
| 326 | |
| 327 | |
| 328 | # Utilities to parse URLs: |
| 329 | # unwrap('<URL:type//host/path>') --> 'type//host/path' |
| 330 | # splittype('type:opaquestring') --> 'type', 'opaquestring' |
| 331 | # splithost('//host[:port]/path') --> 'host[:port]', '/path' |
| 332 | # splitport('host:port') --> 'host', 'port' |
| 333 | # splitquery('/path?query') --> '/path', 'query' |
| 334 | # splittag('/path#tag') --> '/path', 'tag' |
| 335 | # splitgophertype('/Xselector') --> 'X', 'selector' |
| 336 | # unquote('abc%20def') -> 'abc def' |
| 337 | # quote('abc def') -> 'abc%20def') |
| 338 | |
| 339 | def unwrap(url): |
| 340 | import string |
| 341 | url = string.strip(url) |
| 342 | if url[:1] == '<' and url[-1:] == '>': |
| 343 | url = string.strip(url[1:-1]) |
| 344 | if url[:4] == 'URL:': url = string.strip(url[4:]) |
| 345 | return url |
| 346 | |
| 347 | _typeprog = regex.compile('^\([^/:]+\):\(.*\)$') |
| 348 | def splittype(url): |
| 349 | if _typeprog.match(url) >= 0: return _typeprog.group(1, 2) |
| 350 | return None, url |
| 351 | |
| 352 | _hostprog = regex.compile('^//\([^/]+\)\(.*\)$') |
| 353 | def splithost(url): |
| 354 | if _hostprog.match(url) >= 0: return _hostprog.group(1, 2) |
| 355 | return None, url |
| 356 | |
| 357 | _portprog = regex.compile('^\(.*\):\([0-9]+\)$') |
| 358 | def splitport(host): |
| 359 | if _portprog.match(host) >= 0: return _portprog.group(1, 2) |
| 360 | return host, None |
| 361 | |
| 362 | _queryprog = regex.compile('^\(.*\)\?\([^?]*\)$') |
| 363 | def splitquery(url): |
| 364 | if _queryprog.match(url) >= 0: return _queryprog.group(1, 2) |
| 365 | return url, None |
| 366 | |
| 367 | _tagprog = regex.compile('^\(.*\)#\([^#]*\)$') |
| 368 | def splittag(url): |
| 369 | if _tagprog.match(url) >= 0: return _tagprog.group(1, 2) |
| 370 | return url, None |
| 371 | |
| 372 | def splitgophertype(selector): |
| 373 | if selector[:1] == '/' and selector[1:2]: |
| 374 | return selector[1], selector[2:] |
| 375 | return None, selector |
| 376 | |
| 377 | _quoteprog = regex.compile('%[0-9a-fA-F][0-9a-fA-F]') |
| 378 | def unquote(s): |
| 379 | import string |
| 380 | i = 0 |
| 381 | n = len(s) |
| 382 | res = '' |
| 383 | while 0 <= i < n: |
| 384 | j = _quoteprog.search(s, i) |
| 385 | if j < 0: |
| 386 | res = res + s[i:] |
| 387 | break |
| 388 | res = res + (s[i:j] + chr(eval('0x' + s[j+1:j+3]))) |
| 389 | i = j+3 |
| 390 | return res |
| 391 | |
| 392 | _acceptable = \ |
| 393 | 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789-._@' |
| 394 | def quote(s): |
| 395 | res = '' |
| 396 | for c in s: |
| 397 | if c in _acceptable: res = res + c |
| 398 | else: res = res + '%%%02x' % ord(c) |
| 399 | return res |
| 400 | |
| 401 | # Test and time quote() and unquote() |
| 402 | def test1(): |
| 403 | import time |
| 404 | s = '' |
| 405 | for i in range(256): s = s + chr(i) |
| 406 | s = s*4 |
| 407 | t0 = time.time() |
| 408 | qs = quote(s) |
| 409 | uqs = unquote(qs) |
| 410 | t1 = time.time() |
| 411 | if uqs != s: |
| 412 | print 'Wrong!' |
| 413 | print `s` |
| 414 | print `qs` |
| 415 | print `uqs` |
| 416 | print round(t1 - t0, 3), 'sec' |
| 417 | |
| 418 | |
| 419 | # Test program |
| 420 | def test(): |
| 421 | import sys |
| 422 | import regsub |
| 423 | args = sys.argv[1:] |
| 424 | if not args: |
| 425 | args = [ |
| 426 | '/etc/passwd', |
| 427 | 'file:/etc/passwd', |
| 428 | 'file://localhost/etc/passwd', |
| 429 | 'ftp://ftp.cwi.nl/etc/passwd', |
| 430 | 'gopher://gopher.cwi.nl/11/', |
| 431 | 'http://www.cwi.nl/index.html', |
| 432 | ] |
| 433 | try: |
| 434 | for url in args: |
| 435 | print '-'*10, url, '-'*10 |
| 436 | fn, h = urlretrieve(url) |
| 437 | print fn, h |
| 438 | if h: |
| 439 | print '======' |
| 440 | for k in h.keys(): print k + ':', h[k] |
| 441 | print '======' |
| 442 | fp = open(fn, 'r') |
| 443 | data = fp.read() |
| 444 | del fp |
| 445 | print regsub.gsub('\r', '', data) |
| 446 | fn, h = None, None |
| 447 | print '-'*40 |
| 448 | finally: |
| 449 | urlcleanup() |
| 450 | |
| 451 | # Run test program when run as a script |
| 452 | if __name__ == '__main__': |
| 453 | test1() |
| 454 | test() |