net/filter/gzip_filter.h

   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 // GZipFilter applies gzip and deflate content encoding/decoding to a data
   6 // stream. As specified by HTTP 1.1, with gzip encoding the content is
   7 // wrapped with a gzip header, and with deflate encoding the content is in
   8 // a raw, headerless DEFLATE stream.
   9 //
  10 // Internally GZipFilter uses zlib inflate to do decoding.
  11 //
  12 // GZipFilter is a subclass of Filter. See the latter's header file filter.h
  13 // for sample usage.
  14
  15 #ifndef NET_FILTER_GZIP_FILTER_H_
  16 #define NET_FILTER_GZIP_FILTER_H_
  17
  18 #include "base/basictypes.h"
  19 #include "base/memory/scoped_ptr.h"
  20 #include "net/filter/filter.h"
  21
  22 typedef struct z_stream_s z_stream;
  23
  24 namespace net {
  25
  26 class GZipHeader;
  27
  28 class GZipFilter : public Filter {
  29  public:
  30   ~GZipFilter() override;
  31
  32   // Initializes filter decoding mode and internal control blocks.
  33   // Parameter filter_type specifies the type of filter, which corresponds to
  34   // either gzip or deflate decoding. The function returns true if success and
  35   // false otherwise.
  36   // The filter can only be initialized once.
  37   bool InitDecoding(Filter::FilterType filter_type);
  38
  39   // Decodes the pre-filter data and writes the output into the dest_buffer
  40   // passed in.
  41   // The function returns FilterStatus. See filter.h for its description.
  42   //
  43   // Upon entry, *dest_len is the total size (in number of chars) of the
  44   // destination buffer. Upon exit, *dest_len is the actual number of chars
  45   // written into the destination buffer.
  46   //
  47   // This function will fail if there is no pre-filter data in the
  48   // stream_buffer_. On the other hand, *dest_len can be 0 upon successful
  49   // return. For example, the internal zlib may process some pre-filter data
  50   // but not produce output yet.
  51   FilterStatus ReadFilteredData(char* dest_buffer, int* dest_len) override;
  52
  53  private:
  54   enum DecodingStatus {
  55     DECODING_UNINITIALIZED,
  56     DECODING_IN_PROGRESS,
  57     DECODING_DONE,
  58     DECODING_ERROR
  59   };
  60
  61   enum DecodingMode {
  62     DECODE_MODE_GZIP,
  63     DECODE_MODE_DEFLATE,
  64     DECODE_MODE_UNKNOWN
  65   };
  66
  67   enum GZipCheckHeaderState {
  68     GZIP_CHECK_HEADER_IN_PROGRESS,
  69     GZIP_GET_COMPLETE_HEADER,
  70     GZIP_GET_INVALID_HEADER
  71   };
  72
  73   static const int kGZipFooterSize = 8;
  74
  75   // Only to be instantiated by Filter::Factory.
  76   GZipFilter(FilterType type);
  77   friend class Filter;
  78
  79   // Parses and verifies the GZip header.
  80   // Upon exit, the function updates gzip_header_status_ accordingly.
  81   //
  82   // The function returns Filter::FILTER_OK if it gets a complete header and
  83   // there are more data in the pre-filter buffer.
  84   // The function returns Filter::FILTER_NEED_MORE_DATA if it parses all data
  85   // in the pre-filter buffer, either getting a complete header or a partial
  86   // header. The caller needs to check gzip_header_status_ and call this
  87   // function again for partial header.
  88   // The function returns Filter::FILTER_ERROR if error occurs.
  89   FilterStatus CheckGZipHeader();
  90
  91   // Internal function to decode the pre-filter data and writes the output into
  92   // the dest_buffer passed in.
  93   //
  94   // This is the internal version of ReadFilteredData. See the latter's
  95   // comments for the use of function.
  96   FilterStatus DoInflate(char* dest_buffer, int* dest_len);
  97
  98   // Inserts a zlib header to the data stream before calling zlib inflate.
  99   // This is used to work around server bugs. See more comments at the place
 100   // it is called in gzip_filter.cc.
 101   // The function returns true on success and false otherwise.
 102   bool InsertZlibHeader();
 103
 104   // Skip the 8 byte GZip footer after z_stream_end
 105   void SkipGZipFooter();
 106
 107   // Tracks the status of decoding.
 108   // This variable is initialized by InitDecoding and updated only by
 109   // ReadFilteredData.
 110   DecodingStatus decoding_status_;
 111
 112   // Indicates the type of content decoding the GZipFilter is performing.
 113   // This variable is set only once by InitDecoding.
 114   DecodingMode decoding_mode_;
 115
 116   // Used to parse the gzip header in gzip stream.
 117   // It is used when the decoding_mode_ is DECODE_MODE_GZIP.
 118   scoped_ptr<GZipHeader> gzip_header_;
 119
 120   // Tracks the progress of parsing gzip header.
 121   // This variable is maintained by gzip_header_.
 122   GZipCheckHeaderState gzip_header_status_;
 123
 124   // A flag used by InsertZlibHeader to record whether we've successfully added
 125   // a zlib header to this stream.
 126   bool zlib_header_added_;
 127
 128   // Tracks how many bytes of gzip footer have been received.
 129   int gzip_footer_bytes_;
 130
 131   // The control block of zlib which actually does the decoding.
 132   // This data structure is initialized by InitDecoding and updated only by
 133   // DoInflate, with InsertZlibHeader being the exception as a workaround.
 134   scoped_ptr<z_stream> zlib_stream_;
 135
 136   // For robustness, when we see the solo sdch filter, we chain in a gzip filter
 137   // in front of it, with this flag to indicate that the gzip decoding might not
 138   // be needed.  This handles a strange case where "Content-Encoding: sdch,gzip"
 139   // is reduced by an errant proxy to "Content-Encoding: sdch", while the
 140   // content is indeed really gzipped result of sdch :-/.
 141   // If this flag is set, then we will revert to being a pass through filter if
 142   // we don't get a valid gzip header.
 143   bool possible_sdch_pass_through_;
 144
 145   DISALLOW_COPY_AND_ASSIGN(GZipFilter);
 146 };
 147
 148 }  // namespace net
 149
 150 #endif  // NET_FILTER_GZIP_FILTER_H__