update: stats calculation account for intra-ecc

Signed-off-by: Stephen L. <lrq3000@gmail.com>

Author: lrq3000

header_ecc.py

@@ -514,8 +514,8 @@ def main(argv=None):
             header_size_add = header_size
         else: # else for size smaller than the defined header size, it will just be the size of the file
             header_size_add = size
-        # Size of the ecc entry for this file will be: marker-bytes (\xFF bytes) + length-filepath-string + length-size-string + size of the ecc per block for all blocks in file header + size of the hash per block for all blocks in file header.
-        sizeheaders = sizeheaders + replication_rate * (len(entrymarker) + len(field_delim)*2 + len(relfilepath) + len(str(size)) + (int(math.ceil(float(header_size_add) / ecc_params["message_size"])) * (ecc_params["ecc_size"]+ecc_params["hash_size"])) ) # Compute the total number of bytes we will add with ecc + hash (accounting for the padding of the remaining characters at the end of the sequence in case it doesn't fit with the message_size, by using ceil() )
+        # Size of the ecc entry for this file will be: entrymarker-bytes + field_delim-bytes*occurrence + length-filepath-string + length-size-string + length-filepath-ecc + size of the ecc per block for all blocks in file header + size of the hash per block for all blocks in file header.
+        sizeheaders = sizeheaders + replication_rate * (len(entrymarker) + len(field_delim)*3 + len(relfilepath) + len(str(size)) + int(float(len(relfilepath))*resilience_rate_intra) + (int(math.ceil(float(header_size_add) / ecc_params["message_size"])) * (ecc_params["ecc_size"]+ecc_params["hash_size"])) ) # Compute the total number of bytes we will add with ecc + hash (accounting for the padding of the remaining characters at the end of the sequence in case it doesn't fit with the message_size, by using ceil() )
     ptee.write("Precomputing done.")
     # TODO: add the size of the ecc format header? (arguments string + PYHEADERECC identifier)
     total_pred_percentage = sizeheaders * 100 / sizetotal

structural_adaptive_ecc.py

@@ -557,8 +557,8 @@ def main(argv=None):
         else: # else for size smaller than the defined header size, it will just be the size of the file
             filesize_header = size
             filesize_content = 0
-        # Size of the ecc entry for this file will be: marker-bytes (\xFF bytes) + length-filepath-string + length-size-string + size of the ecc per block for all blocks in file header + size of the hash per block for all blocks in file header.
-        sizeecc += replication_rate * (len(entrymarker) + len(field_delim)*2 + len(relfilepath) + len(str(size)) + (int(math.ceil(float(filesize_header) / ecc_params_header["message_size"])) * (ecc_params_header["ecc_size"]+ecc_params_header["hash_size"])) + (int(math.ceil(float(filesize_content) / ecc_params_variable_average["message_size"])) * (ecc_params_variable_average["ecc_size"]+ecc_params_variable_average["hash_size"])) ) # Compute the total number of bytes we will add with ecc + hash (accounting for the padding of the remaining characters at the end of the sequence in case it doesn't fit with the message_size, by using ceil() )
+        # Size of the ecc entry for this file will be: entrymarker-bytes + field_delim-bytes*occurrence + length-filepath-string + length-size-string + length-filepath-ecc + size of the ecc per block for all blocks in file header + size of the hash per block for all blocks in file header.
+        sizeecc += replication_rate * (len(entrymarker) + len(field_delim)*3 + len(relfilepath) + len(str(size)) + int(float(len(relfilepath))*resilience_rate_intra) + (int(math.ceil(float(filesize_header) / ecc_params_header["message_size"])) * (ecc_params_header["ecc_size"]+ecc_params_header["hash_size"])) + (int(math.ceil(float(filesize_content) / ecc_params_variable_average["message_size"])) * (ecc_params_variable_average["ecc_size"]+ecc_params_variable_average["hash_size"])) ) # Compute the total number of bytes we will add with ecc + hash (accounting for the padding of the remaining characters at the end of the sequence in case it doesn't fit with the message_size, by using ceil() )
     ptee.write("Precomputing done.")
     if generate: # show statistics only if generating an ecc file
         # TODO: add the size of the ecc format header? (arguments string + PYHEADERECC identifier)