Home | History | Annotate | Download | only in base 
      1 // Copyright (c) 2006-2008 The Chromium Authors. All rights reserved.
      2 // Use of this source code is governed by a BSD-style license that can be
      3 // found in the LICENSE file.
      4 
      5 // The GZipHeader class allows you to parse a gzip header, such as you
      6 // might find at the beginning of a file compressed by gzip (ie, a .gz
      7 // file), or at the beginning of an HTTP response that uses a gzip
      8 // Content-Encoding. See RFC 1952 for the specification for the gzip
      9 // header.
     10 //
     11 // The model is that you call ReadMore() for each chunk of bytes
     12 // you've read from a file or socket.
     13 //
     14 
     15 #ifndef NET_BASE_GZIP_HEADER_H_
     16 #define NET_BASE_GZIP_HEADER_H_
     17 
     18 #include "base/basictypes.h"
     19 
     20 class GZipHeader {
     21  public:
     22   GZipHeader() {
     23     Reset();
     24   }
     25   ~GZipHeader() {
     26   }
     27 
     28   // Wipe the slate clean and start from scratch.
     29   void Reset() {
     30     state_        = IN_HEADER_ID1;
     31     flags_        = 0;
     32     extra_length_ = 0;
     33   }
     34 
     35   enum Status {
     36     INCOMPLETE_HEADER,    // don't have all the bits yet...
     37     COMPLETE_HEADER,      // complete, valid header
     38     INVALID_HEADER,       // found something invalid in the header
     39   };
     40 
     41   // Attempt to parse the given buffer as the next installment of
     42   // bytes from a gzip header. If the bytes we've seen so far do not
     43   // yet constitute a complete gzip header, return
     44   // INCOMPLETE_HEADER. If these bytes do not constitute a *valid*
     45   // gzip header, return INVALID_HEADER. When we've seen a complete
     46   // gzip header, return COMPLETE_HEADER and set the pointer pointed
     47   // to by header_end to the first byte beyond the gzip header.
     48   Status ReadMore(const char* inbuf, int inbuf_len,
     49                   const char** header_end);
     50  private:
     51 
     52   static const uint8 magic[];  // gzip magic header
     53 
     54   enum {                       // flags (see RFC)
     55     FLAG_FTEXT        = 0x01,  // bit 0 set: file probably ascii text
     56     FLAG_FHCRC        = 0x02,  // bit 1 set: header CRC present
     57     FLAG_FEXTRA       = 0x04,  // bit 2 set: extra field present
     58     FLAG_FNAME        = 0x08,  // bit 3 set: original file name present
     59     FLAG_FCOMMENT     = 0x10,  // bit 4 set: file comment present
     60     FLAG_RESERVED     = 0xE0,  // bits 5..7: reserved
     61   };
     62 
     63   enum State {
     64     // The first 10 bytes are the fixed-size header:
     65     IN_HEADER_ID1,
     66     IN_HEADER_ID2,
     67     IN_HEADER_CM,
     68     IN_HEADER_FLG,
     69     IN_HEADER_MTIME_BYTE_0,
     70     IN_HEADER_MTIME_BYTE_1,
     71     IN_HEADER_MTIME_BYTE_2,
     72     IN_HEADER_MTIME_BYTE_3,
     73     IN_HEADER_XFL,
     74     IN_HEADER_OS,
     75 
     76     IN_XLEN_BYTE_0,
     77     IN_XLEN_BYTE_1,
     78     IN_FEXTRA,
     79 
     80     IN_FNAME,
     81 
     82     IN_FCOMMENT,
     83 
     84     IN_FHCRC_BYTE_0,
     85     IN_FHCRC_BYTE_1,
     86 
     87     IN_DONE,
     88   };
     89 
     90   int    state_;  // our current State in the parsing FSM: an int so we can ++
     91   uint8  flags_;         // the flags byte of the header ("FLG" in the RFC)
     92   uint16 extra_length_;  // how much of the "extra field" we have yet to read
     93 };
     94 
     95 #endif  // NET_BASE_GZIP_HEADER_H_
     96