def check_spdx_license(licenseText):
"""Check the license text against the spdx license list.
"""
licenseText = unicode(licenseText.decode('string_escape'), 'utf-8')
r = redis.StrictRedis(host=getRedisHost(), port=6379, db=0)
# if redis is empty build the spdx license list in the redis database
if r.keys('*') == []:
build_spdx_licenses()
spdxLicenseIds = r.keys()
spdxLicenseTexts = r.mget(spdxLicenseIds)
licenseData = dict(zip(spdxLicenseIds, spdxLicenseTexts))
matches = get_close_matches(licenseText, licenseData)
if not matches:
matchedLicenseIds = None
matchType = 'No match'
elif 1.0 in matches.values() or all(0.99 < score
for score in matches.values()):
matchedLicenseIds = matches.keys()
matchType = 'Perfect match'
else:
for licenseID in matches:
listedLicense = getListedLicense(licenseID)
isTextStandard = checkTextStandardLicense(listedLicense,
licenseText)
if not isTextStandard:
matchedLicenseIds = licenseID
matchType = 'Standard License match'
else:
matchedLicenseIds = max(matches, key=matches.get)
matchType = 'Close match'
return matchedLicenseIds, matchType
def matcher(text_file, threshold, build):
"""SPDX License matcher to match license text against the SPDX license list using an algorithm which finds close matches. """
try:
# For python 2
inputText = codecs.open(text_file, 'r',
encoding='string_escape').read()
inputText = unicode(inputText, 'utf-8')
except:
# For python 3
inputText = codecs.open(text_file, 'r',
encoding='unicode_escape').read()
if build or is_keys_empty():
click.echo('Building SPDX License List. This may take a while...')
build_spdx_licenses()
r = redis.StrictRedis(host=os.environ.get(key="SPDX_REDIS_HOST",
default="localhost"),
port=6379,
db=0)
keys = list(r.keys())
values = r.mget(keys)
licenseData = dict(list(zip(keys, values)))
matches = get_close_matches(inputText, licenseData, threshold)
matchingString = get_matching_string(matches, inputText)
if matchingString == '':
licenseID = max(matches, key=matches.get)
spdxLicenseText = get_spdx_license_text(licenseID)
similarityPercent = get_similarity_percent(spdxLicenseText, inputText)
click.echo(
colors(
'\nThe given license text matches {}% with that of {} based on Levenstein distance.'
.format(similarityPercent, licenseID), 94))
differences = generate_diff(spdxLicenseText, inputText)
for line in differences:
if line[0] == '+':
line = colors(line, 92)
if line[0] == '-':
line = colors(line, 91)
if line[0] == '@':
line = colors(line, 90)
click.echo(line)
else:
click.echo(colors(matchingString, 92))