vproc_zlib_codec.py

import struct
import zlib
import os
import math

def readarc(f, verbose=False):

    
    
    fmt=">6sii"
    assert(14==struct.calcsize(fmt))

    contents=list()
    
#     try:
    if 1:
        i = 0
        while True:
            raw=f.read(struct.calcsize(fmt))
            lenraw = len(raw)
            if lenraw == 0:
                break
            
            hdr,unk,size=struct.unpack(fmt,raw)
            if verbose:
                print(i, hdr, unk, size)
            raw=f.read(size)
            decoded = zlib.decompress(raw)
#             print(decoded)
            contents.append(zlib.decompress(raw))
            i = i+1
#     except:
#         pass

    return contents


def save_text_vproc(filenamevproc, text,  fmt=">6sii", chunk=8192, verbose=True):

    # fmt=">6sii"
    assert(14==struct.calcsize(fmt))
    # chunk = 8192

    text = text

    N_chunks = math.ceil(len(text)/chunk)

    if verbose:
        print("N chunks:",N_chunks, chunk)


    if verbose:
        print("Saving to ", filenamevproc)

    with open(filenamevproc, "wb") as file:

        for i in range(N_chunks):
            text_part = text[i*chunk:(i+1)*chunk]
        #     print(text_part)

            text_size = len(text_part)

            #Compress the text_part
            compressed_text = zlib.compress(text_part.encode(encoding))
            chunk_size = len(compressed_text)


            # header writting
            if verbose:
                print(i, text_size, chunk_size)
            header = struct.pack(fmt, "!zlib!".encode(encoding), text_size , chunk_size)

            file.write(header)


            # write compressed
            file.write(compressed_text)





encoding = "iso-8859-1"
encoding = "utf-8"
encoding = "latin_1"

verbose = False


###########################################################
#### VPROC FILENAME HERE
filename_orig = "processes library.vproc"



print("Loading vproc file from ", filename_orig)

with open(filename_orig, "rb") as f:
    contents = readarc(f, verbose=verbose)
contents_str = [a.decode(encoding) for a in contents]
text = "".join(contents_str)


txtfilename = os.path.splitext(filename_orig)[0]
txtfilename = txtfilename + '.txt'
print("Saving decompressed text file to ",txtfilename)

with open(txtfilename, "w") as file:
    file.write(text)
    
    print("done")
    

    



############## TEXT FILENAME HERE
txtfilename = "processes library.txt"

#########################################################
print("Loading text file from ", txtfilename)


with open(txtfilename,"r") as file:
    text = file.read()
    
filename = txtfilename

filenamevproc = os.path.splitext(filename)[0]
filenamevproc = filenamevproc + '_compressed.vproc'

print("Saving compressed vproc file to ",filenamevproc)

save_text_vproc(filenamevproc, text, verbose=verbose)