예제 #1
0
def loadTruths():
    loadDictionary()
    if cache.loadingDone:
        return
    print "Loading ground truths"
    fileThing = open(truthFile, mode="rt")
    reader = csv.DictReader(fileThing, fieldnames=["filename", "size", "type", "notes", "extensions", "accuracy", "metadata", "kind", "id", "Y", "digest"],
        restkey="rest", delimiter=";")
    for line in reader:

        filename = line["filename"]
        if filename == 'Filename+Ext':
            continue
        filename = filename.split(".")[0]
        truth = FileTruth()
        truth.fileName = filename
        for key,value in line.items():

            if truth.__dict__.has_key(key):
                truth.__dict__[key] = value
        truth.baseMime = getMime(int(truth.id),line["type"], cache.decode, cache.equivMimes)
        truth.version = getVersion(line["notes"], line["metadata"])
        if not truth.baseMime == []:
            cache.insert(filename,truth)
    cache.loadingDone = True
    fileThing.close()
    print "Ground truths loaded"
예제 #2
0
def loadTruths():
    loadDictionary()
    if cache.loadingDone:
        return
    print "Loading ground truths"
    fileThing = open(truthFile, mode="rt")
    # "Filename+Ext","Size","Description","Numbers Metadata Summary","Valid File Extensions","Accuracy","Text Metadata Summary","Content","Description Index","Extension Valid","SHA-1"
    # "000000.swf","80468","Shockwave Flash Object","Format v0.05, uncompressed","SWF","HIGH",,"Video, Graphic Image, Program Data","429","Y","931B45B6885310CB9E5F4C20321306CC24B919E6"
    reader = csv.DictReader(
        fileThing,
        fieldnames=[
            "filename",
            "size",
            "Description",
            "numbers",
            "extensions",
            "accuracy",
            "metadata",
            "kind",
            "id",
            "Y",
            "digest",
        ],
        restkey="rest",
        delimiter=",",
    )
    for line in reader:

        filename = line["filename"]
        if filename == "Filename+Ext":
            continue
        filename = filename.split(".")[0]
        truth = FileTruth()
        truth.fileName = filename
        for key, value in line.items():

            if truth.__dict__.has_key(key):
                truth.__dict__[key] = value
        # print truth.id
        truth.baseMime = getMime(int(truth.id), line["id"], cache.decode, cache.equivMimes)
        truth.version = getVersion(line["numbers"], line["metadata"])
        if not truth.baseMime == []:
            cache.insert(filename, truth)
    cache.loadingDone = True
    fileThing.close()
    print "Ground truths loaded"