modules/luci-base/luasrc/http/protocol.lua

   1 -- Copyright 2008 Freifunk Leipzig / Jo-Philipp Wich <jow@openwrt.org>
   2 -- Licensed to the public under the Apache License 2.0.
   3
   4 -- This class contains several functions useful for http message- and content
   5 -- decoding and to retrive form data from raw http messages.
   6 module("luci.http.protocol", package.seeall)
   7
   8 local ltn12 = require("luci.ltn12")
   9
  10 HTTP_MAX_CONTENT      = 1024*8          -- 8 kB maximum content size
  11
  12 -- the "+" sign to " " - and return the decoded string.
  13 function urldecode( str, no_plus )
  14
  15         local function __chrdec( hex )
  16                 return string.char( tonumber( hex, 16 ) )
  17         end
  18
  19         if type(str) == "string" then
  20                 if not no_plus then
  21                         str = str:gsub( "+", " " )
  22                 end
  23
  24                 str = str:gsub( "%%([a-fA-F0-9][a-fA-F0-9])", __chrdec )
  25         end
  26
  27         return str
  28 end
  29
  30 -- from given url or string. Returns a table with urldecoded values.
  31 -- Simple parameters are stored as string values associated with the parameter
  32 -- name within the table. Parameters with multiple values are stored as array
  33 -- containing the corresponding values.
  34 function urldecode_params( url, tbl )
  35
  36         local params = tbl or { }
  37
  38         if url:find("?") then
  39                 url = url:gsub( "^.+%?([^?]+)", "%1" )
  40         end
  41
  42         for pair in url:gmatch( "[^&;]+" ) do
  43
  44                 -- find key and value
  45                 local key = urldecode( pair:match("^([^=]+)")     )
  46                 local val = urldecode( pair:match("^[^=]+=(.+)$") )
  47
  48                 -- store
  49                 if type(key) == "string" and key:len() > 0 then
  50                         if type(val) ~= "string" then val = "" end
  51
  52                         if not params[key] then
  53                                 params[key] = val
  54                         elseif type(params[key]) ~= "table" then
  55                                 params[key] = { params[key], val }
  56                         else
  57                                 table.insert( params[key], val )
  58                         end
  59                 end
  60         end
  61
  62         return params
  63 end
  64
  65 function urlencode( str )
  66
  67         local function __chrenc( chr )
  68                 return string.format(
  69                         "%%%02x", string.byte( chr )
  70                 )
  71         end
  72
  73         if type(str) == "string" then
  74                 str = str:gsub(
  75                         "([^a-zA-Z0-9$_%-%.!*'(),])",
  76                         __chrenc
  77                 )
  78         end
  79
  80         return str
  81 end
  82
  83 -- separated by "&". Tables are encoded as parameters with multiple values by
  84 -- repeating the parameter name with each value.
  85 function urlencode_params( tbl )
  86         local enc = ""
  87
  88         for k, v in pairs(tbl) do
  89                 if type(v) == "table" then
  90                         for i, v2 in ipairs(v) do
  91                                 enc = enc .. ( #enc > 0 and "&" or "" ) ..
  92                                         urlencode(k) .. "=" .. urlencode(v2)
  93                         end
  94                 else
  95                         enc = enc .. ( #enc > 0 and "&" or "" ) ..
  96                                 urlencode(k) .. "=" .. urlencode(v)
  97                 end
  98         end
  99
 100         return enc
 101 end
 102
 103 -- (Internal function)
 104 -- Initialize given parameter and coerce string into table when the parameter
 105 -- already exists.
 106 local function __initval( tbl, key )
 107         if tbl[key] == nil then
 108                 tbl[key] = ""
 109         elseif type(tbl[key]) == "string" then
 110                 tbl[key] = { tbl[key], "" }
 111         else
 112                 table.insert( tbl[key], "" )
 113         end
 114 end
 115
 116 -- (Internal function)
 117 -- Append given data to given parameter, either by extending the string value
 118 -- or by appending it to the last string in the parameter's value table.
 119 local function __appendval( tbl, key, chunk )
 120         if type(tbl[key]) == "table" then
 121                 tbl[key][#tbl[key]] = tbl[key][#tbl[key]] .. chunk
 122         else
 123                 tbl[key] = tbl[key] .. chunk
 124         end
 125 end
 126
 127 -- (Internal function)
 128 -- Finish the value of given parameter, either by transforming the string value
 129 -- or - in the case of multi value parameters - the last element in the
 130 -- associated values table.
 131 local function __finishval( tbl, key, handler )
 132         if handler then
 133                 if type(tbl[key]) == "table" then
 134                         tbl[key][#tbl[key]] = handler( tbl[key][#tbl[key]] )
 135                 else
 136                         tbl[key] = handler( tbl[key] )
 137                 end
 138         end
 139 end
 140
 141
 142 -- Table of our process states
 143 local process_states = { }
 144
 145 -- Extract "magic", the first line of a http message.
 146 -- Extracts the message type ("get", "post" or "response"), the requested uri
 147 -- or the status code if the line descripes a http response.
 148 process_states['magic'] = function( msg, chunk, err )
 149
 150         if chunk ~= nil then
 151                 -- ignore empty lines before request
 152                 if #chunk == 0 then
 153                         return true, nil
 154                 end
 155
 156                 -- Is it a request?
 157                 local method, uri, http_ver = chunk:match("^([A-Z]+) ([^ ]+) HTTP/([01]%.[019])$")
 158
 159                 -- Yup, it is
 160                 if method then
 161
 162                         msg.type           = "request"
 163                         msg.request_method = method:lower()
 164                         msg.request_uri    = uri
 165                         msg.http_version   = tonumber( http_ver )
 166                         msg.headers        = { }
 167
 168                         -- We're done, next state is header parsing
 169                         return true, function( chunk )
 170                                 return process_states['headers']( msg, chunk )
 171                         end
 172
 173                 -- Is it a response?
 174                 else
 175
 176                         local http_ver, code, message = chunk:match("^HTTP/([01]%.[019]) ([0-9]+) ([^\r\n]+)$")
 177
 178                         -- Is a response
 179                         if code then
 180
 181                                 msg.type           = "response"
 182                                 msg.status_code    = code
 183                                 msg.status_message = message
 184                                 msg.http_version   = tonumber( http_ver )
 185                                 msg.headers        = { }
 186
 187                                 -- We're done, next state is header parsing
 188                                 return true, function( chunk )
 189                                         return process_states['headers']( msg, chunk )
 190                                 end
 191                         end
 192                 end
 193         end
 194
 195         -- Can't handle it
 196         return nil, "Invalid HTTP message magic"
 197 end
 198
 199
 200 -- Extract headers from given string.
 201 process_states['headers'] = function( msg, chunk )
 202
 203         if chunk ~= nil then
 204
 205                 -- Look for a valid header format
 206                 local hdr, val = chunk:match( "^([A-Za-z][A-Za-z0-9%-_]+): +(.+)$" )
 207
 208                 if type(hdr) == "string" and hdr:len() > 0 and
 209                    type(val) == "string" and val:len() > 0
 210                 then
 211                         msg.headers[hdr] = val
 212
 213                         -- Valid header line, proceed
 214                         return true, nil
 215
 216                 elseif #chunk == 0 then
 217                         -- Empty line, we won't accept data anymore
 218                         return false, nil
 219                 else
 220                         -- Junk data
 221                         return nil, "Invalid HTTP header received"
 222                 end
 223         else
 224                 return nil, "Unexpected EOF"
 225         end
 226 end
 227
 228
 229 -- data line by line with the trailing \r\n stripped of.
 230 function header_source( sock )
 231         return ltn12.source.simplify( function()
 232
 233                 local chunk, err, part = sock:receive("*l")
 234
 235                 -- Line too long
 236                 if chunk == nil then
 237                         if err ~= "timeout" then
 238                                 return nil, part
 239                                         and "Line exceeds maximum allowed length"
 240                                         or  "Unexpected EOF"
 241                         else
 242                                 return nil, err
 243                         end
 244
 245                 -- Line ok
 246                 elseif chunk ~= nil then
 247
 248                         -- Strip trailing CR
 249                         chunk = chunk:gsub("\r$","")
 250
 251                         return chunk, nil
 252                 end
 253         end )
 254 end
 255
 256 -- Content-Type. Stores all extracted data associated with its parameter name
 257 -- in the params table withing the given message object. Multiple parameter
 258 -- values are stored as tables, ordinary ones as strings.
 259 -- If an optional file callback function is given then it is feeded with the
 260 -- file contents chunk by chunk and only the extracted file name is stored
 261 -- within the params table. The callback function will be called subsequently
 262 -- with three arguments:
 263 --  o Table containing decoded (name, file) and raw (headers) mime header data
 264 --  o String value containing a chunk of the file data
 265 --  o Boolean which indicates wheather the current chunk is the last one (eof)
 266 function mimedecode_message_body( src, msg, filecb )
 267
 268         if msg and msg.env.CONTENT_TYPE then
 269                 msg.mime_boundary = msg.env.CONTENT_TYPE:match("^multipart/form%-data; boundary=(.+)$")
 270         end
 271
 272         if not msg.mime_boundary then
 273                 return nil, "Invalid Content-Type found"
 274         end
 275
 276
 277         local tlen   = 0
 278         local inhdr  = false
 279         local field  = nil
 280         local store  = nil
 281         local lchunk = nil
 282
 283         local function parse_headers( chunk, field )
 284
 285                 local stat
 286                 repeat
 287                         chunk, stat = chunk:gsub(
 288                                 "^([A-Z][A-Za-z0-9%-_]+): +([^\r\n]+)\r\n",
 289                                 function(k,v)
 290                                         field.headers[k] = v
 291                                         return ""
 292                                 end
 293                         )
 294                 until stat == 0
 295
 296                 chunk, stat = chunk:gsub("^\r\n","")
 297
 298                 -- End of headers
 299                 if stat > 0 then
 300                         if field.headers["Content-Disposition"] then
 301                                 if field.headers["Content-Disposition"]:match("^form%-data; ") then
 302                                         field.name = field.headers["Content-Disposition"]:match('name="(.-)"')
 303                                         field.file = field.headers["Content-Disposition"]:match('filename="(.+)"$')
 304                                 end
 305                         end
 306
 307                         if not field.headers["Content-Type"] then
 308                                 field.headers["Content-Type"] = "text/plain"
 309                         end
 310
 311                         if field.name and field.file and filecb then
 312                                 __initval( msg.params, field.name )
 313                                 __appendval( msg.params, field.name, field.file )
 314
 315                                 store = filecb
 316                         elseif field.name then
 317                                 __initval( msg.params, field.name )
 318
 319                                 store = function( hdr, buf, eof )
 320                                         __appendval( msg.params, field.name, buf )
 321                                 end
 322                         else
 323                                 store = nil
 324                         end
 325
 326                         return chunk, true
 327                 end
 328
 329                 return chunk, false
 330         end
 331
 332         local function snk( chunk )
 333
 334                 tlen = tlen + ( chunk and #chunk or 0 )
 335
 336                 if msg.env.CONTENT_LENGTH and tlen > tonumber(msg.env.CONTENT_LENGTH) + 2 then
 337                         return nil, "Message body size exceeds Content-Length"
 338                 end
 339
 340                 if chunk and not lchunk then
 341                         lchunk = "\r\n" .. chunk
 342
 343                 elseif lchunk then
 344                         local data = lchunk .. ( chunk or "" )
 345                         local spos, epos, found
 346
 347                         repeat
 348                                 spos, epos = data:find( "\r\n--" .. msg.mime_boundary .. "\r\n", 1, true )
 349
 350                                 if not spos then
 351                                         spos, epos = data:find( "\r\n--" .. msg.mime_boundary .. "--\r\n", 1, true )
 352                                 end
 353
 354
 355                                 if spos then
 356                                         local predata = data:sub( 1, spos - 1 )
 357
 358                                         if inhdr then
 359                                                 predata, eof = parse_headers( predata, field )
 360
 361                                                 if not eof then
 362                                                         return nil, "Invalid MIME section header"
 363                                                 elseif not field.name then
 364                                                         return nil, "Invalid Content-Disposition header"
 365                                                 end
 366                                         end
 367
 368                                         if store then
 369                                                 store( field, predata, true )
 370                                         end
 371
 372
 373                                         field = { headers = { } }
 374                                         found = found or true
 375
 376                                         data, eof = parse_headers( data:sub( epos + 1, #data ), field )
 377                                         inhdr = not eof
 378                                 end
 379                         until not spos
 380
 381                         if found then
 382                                 -- We found at least some boundary. Save
 383                                 -- the unparsed remaining data for the
 384                                 -- next chunk.
 385                                 lchunk, data = data, nil
 386                         else
 387                                 -- There was a complete chunk without a boundary. Parse it as headers or
 388                                 -- append it as data, depending on our current state.
 389                                 if inhdr then
 390                                         lchunk, eof = parse_headers( data, field )
 391                                         inhdr = not eof
 392                                 else
 393                                         -- We're inside data, so append the data. Note that we only append
 394                                         -- lchunk, not all of data, since there is a chance that chunk
 395                                         -- contains half a boundary. Assuming that each chunk is at least the
 396                                         -- boundary in size, this should prevent problems
 397                                         store( field, lchunk, false )
 398                                         lchunk, chunk = chunk, nil
 399                                 end
 400                         end
 401                 end
 402
 403                 return true
 404         end
 405
 406         return ltn12.pump.all( src, snk )
 407 end
 408
 409 -- Content-Type. Stores all extracted data associated with its parameter name
 410 -- in the params table withing the given message object. Multiple parameter
 411 -- values are stored as tables, ordinary ones as strings.
 412 function urldecode_message_body( src, msg )
 413
 414         local tlen   = 0
 415         local lchunk = nil
 416
 417         local function snk( chunk )
 418
 419                 tlen = tlen + ( chunk and #chunk or 0 )
 420
 421                 if msg.env.CONTENT_LENGTH and tlen > tonumber(msg.env.CONTENT_LENGTH) + 2 then
 422                         return nil, "Message body size exceeds Content-Length"
 423                 elseif tlen > HTTP_MAX_CONTENT then
 424                         return nil, "Message body size exceeds maximum allowed length"
 425                 end
 426
 427                 if not lchunk and chunk then
 428                         lchunk = chunk
 429
 430                 elseif lchunk then
 431                         local data = lchunk .. ( chunk or "&" )
 432                         local spos, epos
 433
 434                         repeat
 435                                 spos, epos = data:find("^.-[;&]")
 436
 437                                 if spos then
 438                                         local pair = data:sub( spos, epos - 1 )
 439                                         local key  = pair:match("^(.-)=")
 440                                         local val  = pair:match("=([^%s]*)%s*$")
 441
 442                                         if key and #key > 0 then
 443                                                 __initval( msg.params, key )
 444                                                 __appendval( msg.params, key, val )
 445                                                 __finishval( msg.params, key, urldecode )
 446                                         end
 447
 448                                         data = data:sub( epos + 1, #data )
 449                                 end
 450                         until not spos
 451
 452                         lchunk = data
 453                 end
 454
 455                 return true
 456         end
 457
 458         return ltn12.pump.all( src, snk )
 459 end
 460
 461 -- version, message headers and resulting CGI environment variables from the
 462 -- given ltn12 source.
 463 function parse_message_header( src )
 464
 465         local ok   = true
 466         local msg  = { }
 467
 468         local sink = ltn12.sink.simplify(
 469                 function( chunk )
 470                         return process_states['magic']( msg, chunk )
 471                 end
 472         )
 473
 474         -- Pump input data...
 475         while ok do
 476
 477                 -- get data
 478                 ok, err = ltn12.pump.step( src, sink )
 479
 480                 -- error
 481                 if not ok and err then
 482                         return nil, err
 483
 484                 -- eof
 485                 elseif not ok then
 486
 487                         -- Process get parameters
 488                         if ( msg.request_method == "get" or msg.request_method == "post" ) and
 489                            msg.request_uri:match("?")
 490                         then
 491                                 msg.params = urldecode_params( msg.request_uri )
 492                         else
 493                                 msg.params = { }
 494                         end
 495
 496                         -- Populate common environment variables
 497                         msg.env = {
 498                                 CONTENT_LENGTH    = msg.headers['Content-Length'];
 499                                 CONTENT_TYPE      = msg.headers['Content-Type'] or msg.headers['Content-type'];
 500                                 REQUEST_METHOD    = msg.request_method:upper();
 501                                 REQUEST_URI       = msg.request_uri;
 502                                 SCRIPT_NAME       = msg.request_uri:gsub("?.+$","");
 503                                 SCRIPT_FILENAME   = "";         -- XXX implement me
 504                                 SERVER_PROTOCOL   = "HTTP/" .. string.format("%.1f", msg.http_version);
 505                                 QUERY_STRING      = msg.request_uri:match("?")
 506                                         and msg.request_uri:gsub("^.+?","") or ""
 507                         }
 508
 509                         -- Populate HTTP_* environment variables
 510                         for i, hdr in ipairs( {
 511                                 'Accept',
 512                                 'Accept-Charset',
 513                                 'Accept-Encoding',
 514                                 'Accept-Language',
 515                                 'Connection',
 516                                 'Cookie',
 517                                 'Host',
 518                                 'Referer',
 519                                 'User-Agent',
 520                         } ) do
 521                                 local var = 'HTTP_' .. hdr:upper():gsub("%-","_")
 522                                 local val = msg.headers[hdr]
 523
 524                                 msg.env[var] = val
 525                         end
 526                 end
 527         end
 528
 529         return msg
 530 end
 531
 532 -- This function will examine the Content-Type within the given message object
 533 -- to select the appropriate content decoder.
 534 -- Currently the application/x-www-urlencoded and application/form-data
 535 -- mime types are supported. If the encountered content encoding can't be
 536 -- handled then the whole message body will be stored unaltered as "content"
 537 -- property within the given message object.
 538 function parse_message_body( src, msg, filecb )
 539         -- Is it multipart/mime ?
 540         if msg.env.REQUEST_METHOD == "POST" and msg.env.CONTENT_TYPE and
 541            msg.env.CONTENT_TYPE:match("^multipart/form%-data")
 542         then
 543
 544                 return mimedecode_message_body( src, msg, filecb )
 545
 546         -- Is it application/x-www-form-urlencoded ?
 547         elseif msg.env.REQUEST_METHOD == "POST" and msg.env.CONTENT_TYPE and
 548                msg.env.CONTENT_TYPE:match("^application/x%-www%-form%-urlencoded")
 549         then
 550                 return urldecode_message_body( src, msg, filecb )
 551
 552
 553         -- Unhandled encoding
 554         -- If a file callback is given then feed it chunk by chunk, else
 555         -- store whole buffer in message.content
 556         else
 557
 558                 local sink
 559
 560                 -- If we have a file callback then feed it
 561                 if type(filecb) == "function" then
 562                         local meta = {
 563                                 name = "raw",
 564                                 encoding = msg.env.CONTENT_TYPE
 565                         }
 566                         sink = function( chunk )
 567                                 if chunk then
 568                                         return filecb(meta, chunk, false)
 569                                 else
 570                                         return filecb(meta, nil, true)
 571                                 end
 572                         end
 573                 -- ... else append to .content
 574                 else
 575                         msg.content = ""
 576                         msg.content_length = 0
 577
 578                         sink = function( chunk )
 579                                 if chunk then
 580                                         if ( msg.content_length + #chunk ) <= HTTP_MAX_CONTENT then
 581                                                 msg.content        = msg.content        .. chunk
 582                                                 msg.content_length = msg.content_length + #chunk
 583                                                 return true
 584                                         else
 585                                                 return nil, "POST data exceeds maximum allowed length"
 586                                         end
 587                                 end
 588                                 return true
 589                         end
 590                 end
 591
 592                 -- Pump data...
 593                 while true do
 594                         local ok, err = ltn12.pump.step( src, sink )
 595
 596                         if not ok and err then
 597                                 return nil, err
 598                         elseif not ok then -- eof
 599                                 return true
 600                         end
 601                 end
 602
 603                 return true
 604         end
 605 end
 606
 607 statusmsg = {
 608         [200] = "OK",
 609         [206] = "Partial Content",
 610         [301] = "Moved Permanently",
 611         [302] = "Found",
 612         [304] = "Not Modified",
 613         [400] = "Bad Request",
 614         [403] = "Forbidden",
 615         [404] = "Not Found",
 616         [405] = "Method Not Allowed",
 617         [408] = "Request Time-out",
 618         [411] = "Length Required",
 619         [412] = "Precondition Failed",
 620         [416] = "Requested range not satisfiable",
 621         [500] = "Internal Server Error",
 622         [503] = "Server Unavailable",
 623 }