Support embedding comments

enc/encode.cc

@@ -687,8 +687,16 @@ bool BrotliCompressor::WriteBrotliData(const bool is_last,
   assert(input_pos_ - last_flush_pos_ <= 1u << 24);
   const uint32_t metablock_size =
       static_cast<uint32_t>(input_pos_ - last_flush_pos_);
-  const size_t max_out_size = 2 * metablock_size + 500;
-  uint8_t* storage = GetBrotliStorage(max_out_size);
+  size_t metadata_size = params_.comment ? strlen(params_.comment) : 0;
+  size_t encoded_metadata_size = metadata_size ? 6 + metadata_size : 0;
+  const size_t max_out_size = encoded_metadata_size + 2 * metablock_size + 500;
+  uint8_t* output_storage = GetBrotliStorage(max_out_size);
+  if (encoded_metadata_size) {
+    if (!WriteMetadata(metadata_size, (const uint8_t*)params_.comment, false, &encoded_metadata_size, output_storage))
+      return false;
+    params_.comment = 0;
+  }
+  uint8_t* storage = output_storage + encoded_metadata_size;
   storage[0] = last_byte_;
   size_t storage_ix = last_byte_bits_;
   bool font_mode = params_.mode == BrotliParams::MODE_FONT;
@@ -711,8 +719,8 @@ bool BrotliCompressor::WriteBrotliData(const bool is_last,
   // Save the state of the distance cache in case we need to restore it for
   // emitting an uncompressed block.
   memcpy(saved_dist_cache_, dist_cache_, sizeof(dist_cache_));
-  *output = &storage[0];
-  *out_size = storage_ix >> 3;
+  *output = &output_storage[0];
+  *out_size = encoded_metadata_size + (storage_ix >> 3);
   return true;
 }
 

enc/encode.h

@@ -34,7 +34,8 @@ struct BrotliParams {
         enable_dictionary(true),
         enable_transforms(false),
         greedy_block_split(false),
-        enable_context_modeling(true) {}
+        enable_context_modeling(true),
+        comment(0) {}
 
   enum Mode {
     // Default compression mode. The compressor does not know anything in
@@ -62,6 +63,7 @@ struct BrotliParams {
   bool enable_transforms;
   bool greedy_block_split;
   bool enable_context_modeling;
+  const char* comment;
 };
 
 // An instance can not be reused for multiple brotli streams.

python/bro.py

@@ -81,6 +81,8 @@ def main(args=None):
                         'on the quality. Defaults to 0.')
     params.add_argument('--custom-dictionary', metavar="FILE", type=str, dest='dictfile',
                         help='Custom dictionary file.', default = None)
+    params.add_argument('--comment', type=str, dest='comment',
+                        help='Archive comment.', default = None)
     # set default values using global DEFAULT_PARAMS dictionary
     parser.set_defaults(**DEFAULT_PARAMS)
 
@@ -113,14 +115,18 @@ def main(args=None):
     else:
         custom_dictionary = ''
 
+    if options.comment:
+        archive_comment = options.comment
+    else:
+        archive_comment = ''
 
     try:
         if options.decompress:
             data = brotli.decompress(data, dictionary=custom_dictionary)
         else:
             data = brotli.compress(
                 data, mode=options.mode, quality=options.quality,
-                lgwin=options.lgwin, lgblock=options.lgblock, dictionary=custom_dictionary)
+                lgwin=options.lgwin, lgblock=options.lgblock, dictionary=custom_dictionary, comment=archive_comment)
     except brotli.error as e:
         parser.exit(1,'bro: error: %s: %s' % (e, options.infile or 'sys.stdin'))
 

python/brotlimodule.cc

@@ -107,6 +107,8 @@ PyDoc_STRVAR(compress__doc__,
 "    quality. Defaults to 0.\n"
 "  dictionary (bytes, optional): Custom dictionary. Only last sliding window\n"
 "     size bytes will be used.\n"
+"  comment (string, optional): A comment that will be added as the leading\n"
+"     metadata metablock.\n"
 "\n"
 "Returns:\n"
 "  The compressed byte string.\n"
@@ -118,26 +120,28 @@ static PyObject* brotli_compress(PyObject *self, PyObject *args, PyObject *keywd
   PyObject *ret = NULL;
   uint8_t *input, *output, *custom_dictionary;
   size_t length, output_length, custom_dictionary_length;
+  const char *comment = 0;
   BrotliParams::Mode mode = (BrotliParams::Mode) -1;
   int quality = -1;
   int lgwin = -1;
   int lgblock = -1;
   int ok;
 
   static const char *kwlist[] = {
-      "string", "mode", "quality", "lgwin", "lgblock", "dictionary", NULL};
+      "string", "mode", "quality", "lgwin", "lgblock", "dictionary", "comment", NULL};
 
   custom_dictionary = NULL;
   custom_dictionary_length = 0;
 
-  ok = PyArg_ParseTupleAndKeywords(args, keywds, "s#|O&O&O&O&s#:compress",
+  ok = PyArg_ParseTupleAndKeywords(args, keywds, "s#|O&O&O&O&s#s:compress",
                         const_cast<char **>(kwlist),
                         &input, &length,
                         &mode_convertor, &mode,
                         &quality_convertor, &quality,
                         &lgwin_convertor, &lgwin,
                         &lgblock_convertor, &lgblock,
-                        &custom_dictionary, &custom_dictionary_length);
+                        &custom_dictionary, &custom_dictionary_length,
+                        &comment);
   if (!ok)
     return NULL;
 
@@ -153,6 +157,7 @@ static PyObject* brotli_compress(PyObject *self, PyObject *args, PyObject *keywd
     params.lgwin = lgwin;
   if (lgblock != -1)
     params.lgblock = lgblock;
+  params.comment = comment;
 
   if (custom_dictionary_length == 0) {
     ok = BrotliCompressBuffer(params, length, input,