Home | History | Annotate | Download | only in filter 
      1 // Copyright 2014 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_FILTER_GZIP_HEADER_H_
     16 #define NET_FILTER_GZIP_HEADER_H_
     17 
     18 #include "base/basictypes.h"
     19 
     20 namespace net {
     21 
     22 class GZipHeader {
     23  public:
     24   enum Status {
     25     INCOMPLETE_HEADER,    // don't have all the bits yet...
     26     COMPLETE_HEADER,      // complete, valid header
     27     INVALID_HEADER,       // found something invalid in the header
     28   };
     29 
     30   GZipHeader();
     31   ~GZipHeader();
     32 
     33   // Wipe the slate clean and start from scratch.
     34   void Reset();
     35 
     36   // Attempt to parse the given buffer as the next installment of
     37   // bytes from a gzip header. If the bytes we've seen so far do not
     38   // yet constitute a complete gzip header, return
     39   // INCOMPLETE_HEADER. If these bytes do not constitute a *valid*
     40   // gzip header, return INVALID_HEADER. When we've seen a complete
     41   // gzip header, return COMPLETE_HEADER and set the pointer pointed
     42   // to by header_end to the first byte beyond the gzip header.
     43   Status ReadMore(const char* inbuf,
     44                   int inbuf_len,
     45                   const char** header_end);
     46  private:
     47   enum {                       // flags (see RFC)
     48     FLAG_FTEXT        = 0x01,  // bit 0 set: file probably ascii text
     49     FLAG_FHCRC        = 0x02,  // bit 1 set: header CRC present
     50     FLAG_FEXTRA       = 0x04,  // bit 2 set: extra field present
     51     FLAG_FNAME        = 0x08,  // bit 3 set: original file name present
     52     FLAG_FCOMMENT     = 0x10,  // bit 4 set: file comment present
     53     FLAG_RESERVED     = 0xE0,  // bits 5..7: reserved
     54   };
     55 
     56   enum State {
     57     // The first 10 bytes are the fixed-size header:
     58     IN_HEADER_ID1,
     59     IN_HEADER_ID2,
     60     IN_HEADER_CM,
     61     IN_HEADER_FLG,
     62     IN_HEADER_MTIME_BYTE_0,
     63     IN_HEADER_MTIME_BYTE_1,
     64     IN_HEADER_MTIME_BYTE_2,
     65     IN_HEADER_MTIME_BYTE_3,
     66     IN_HEADER_XFL,
     67     IN_HEADER_OS,
     68 
     69     IN_XLEN_BYTE_0,
     70     IN_XLEN_BYTE_1,
     71     IN_FEXTRA,
     72 
     73     IN_FNAME,
     74 
     75     IN_FCOMMENT,
     76 
     77     IN_FHCRC_BYTE_0,
     78     IN_FHCRC_BYTE_1,
     79 
     80     IN_DONE,
     81   };
     82 
     83   static const uint8 magic[];  // gzip magic header
     84 
     85   int    state_;  // our current State in the parsing FSM: an int so we can ++
     86   uint8  flags_;         // the flags byte of the header ("FLG" in the RFC)
     87   uint16 extra_length_;  // how much of the "extra field" we have yet to read
     88 
     89   DISALLOW_COPY_AND_ASSIGN(GZipHeader);
     90 };
     91 
     92 }  // namespace net
     93 
     94 #endif  // NET_FILTER_GZIP_HEADER_H_
     95