def validate_file(f, kind):
kml = "kml" in f.name
linenumber = 0
print("Validate UTF-8 encoding in ", f.name)
for lineb in f:
linenumber += 1
lineb = lineb[:-1] # remove the \n
try:
line = lineb.decode('utf-8')
except UnicodeDecodeError as e:
print("{}:{} {} {}".format(f.name, linenumber, str(e),
asbinary(lineb)))
continue
if bulk_extractor_reader.is_comment_line(line):
continue # don't test comments
if bulk_extractor_reader.is_histogram_line(line):
continue # don't test
if kind == FEATURE_FILE:
fields = line.split("\t")
r = invalid_feature_file_line(line, fields)
if r:
print("{}: {:8} {} Invalid feature file line: {}".format(
f.name, linenumber, r, line))
if kml and fields[1].count("kml") != 2:
print("{}: {:8} Invalid KML line: {}".format(
f.name, linenumber, line))
def validate_file(f,kind):
kml = "kml" in f.name
linenumber = 0
print("Validate UTF-8 encoding in ",f.name)
for lineb in f:
linenumber += 1
lineb = lineb[:-1] # remove the \n
try:
line = lineb.decode('utf-8')
except UnicodeDecodeError as e:
print("{}:{} {} {}".format(f.name,linenumber,str(e),asbinary(lineb)))
continue
if bulk_extractor_reader.is_comment_line(line):
continue # don't test comments
if bulk_extractor_reader.is_histogram_line(line):
continue # don't test
if kind==FEATURE_FILE:
fields = line.split("\t")
r = invalid_feature_file_line(line,fields)
if r:
print("{}: {:8} {} Invalid feature file line: {}".format(f.name,linenumber,r,line))
if kml and fields[1].count("kml")!=2:
print("{}: {:8} Invalid KML line: {}".format(f.name,linenumber,line))
def validate_openfile(f):
fn = f.name
if fn.endswith('.xml') or fn.endswith('.dmp') or fn.endswith("_tags.txt") or "wordlist" in fn:
is_feature_file = False
return
else:
is_feature_file = True
# now read
linenumber = 0
print("Validate ",fn)
for lineb in f:
linenumber += 1
lineb = lineb[:-1]
try:
line = lineb.decode('utf-8')
except UnicodeDecodeError as e:
print("{}:{} {} {}".format(fn,linenumber,str(e),asbinary(lineb)))
if bulk_extractor_reader.is_comment_line(line):
continue # don't test
if bulk_extractor_reader.is_histogram_line(line):
continue # don't test
if is_feature_file and not valid_feature_file_line(line):
print("{}: {:8} Invalid feature file line: {}".format(fn,linenumber,line))
