licenses changes

ecosystemDataManager/ecosystemDataManager.py

@@ -411,9 +411,11 @@ def groupsEvolution(self):
 			history = package.getHistory()
 			for i in range(len(history) - 1):
 				versionFrom = history[i]
+				versionTo = history[i + 1]
 				if not versionFrom.getDatetime():
 					continue
-				versionTo = history[i + 1]
+				if not versionTo.getDatetime():
+					continue
 				licensesFrom = versionFrom.getLicenses()
 				licensesTo = versionTo.getLicenses()
 				if not licensesFrom:

generateCSVLicensesChanges.py

@@ -0,0 +1,68 @@
+from ecosystemDataManager.ecosystemDataManager import EcosystemDataManager
+from ecosystemDataManager.group import Group
+import csv
+import sys
+
+if __name__ == '__main__':
+	if len(sys.argv) < 2:
+		print("Usage:", sys.argv[0], "<ecosystem> [<file>]")
+		sys.exit(1)
+	if len(sys.argv) > 2:
+		file = sys.argv[2]
+	else:
+		print("file not provided. using default.")
+		file = "licensesChanges.csv"
+	try:
+		open(file)
+		print("file loaded")
+		file = open(file, "a")
+		file = csv.writer(file, delimiter = ';')
+	except Exception as e:
+		file = open(file, "w")
+		file = csv.writer(file, delimiter = ';')
+		file.writerow(["ecosystem","package","versions","unknownTOknown", "knownTOunknown", "latestGroup"])
+		print("file not loaded. initialized.")
+	ecosystem = sys.argv[1]
+	ecosystemDataManager = EcosystemDataManager(ecosystem)
+	for package in ecosystemDataManager.getPackages():
+		unknownTOknown = 0
+		knownTOunknown = 0
+		history = package.getHistory()
+		for i in range(len(history) - 1):
+			versionFrom = history[i]
+			versionTo = history[i + 1]
+			if not versionFrom.getDatetime():
+				continue
+			if not versionTo.getDatetime():
+				continue
+			licensesFrom = versionFrom.getLicenses()
+			licensesTo = versionTo.getLicenses()
+			if not licensesFrom:
+				if licensesTo:
+					for licenseTo in licensesTo:
+						groupTo = licenseTo.getGroup()
+						if groupTo == Group.KNOWN:
+							unknownTOknown += 1
+			elif not licensesTo:
+				for licenseFrom in licensesFrom:
+					groupFrom = licenseFrom.getGroup()
+					if groupFrom == Group.KNOWN:
+						knownTOunknown += 1
+			else:
+				for licenseFrom in licensesFrom:
+					for licenseTo in licensesTo:
+						groupFrom = licenseFrom.getGroup()
+						groupTo = licenseTo.getGroup()
+						if groupFrom == Group.KNOWN:
+							if groupTo == Group.NONE or groupTo == Group.UNDEFINED or groupTo == Group.UNKNOWN:
+								knownTOunknown += 1
+						if groupFrom == Group.NONE or groupFrom == Group.UNDEFINED or groupFrom == Group.UNKNOWN:
+							if groupTo == Group.KNOWN:
+								unknownTOknown += 1
+		try:
+			if unknownTOknown > 0 or knownTOunknown > 0:
+				versions = len(package)
+				latestGroup = ",".join([license.getGroup().name for license in package.getLatestVersion().getLicenses()])
+				file.writerow([ecosystem, package, versions, unknownTOknown, knownTOunknown, latestGroup])
+		except Exception as e:
+			print(package, len(package), "discarted")

generateCSVUnknown.py

@@ -32,9 +32,21 @@
 				if license.getGroup() == Group.NONE or license.getGroup() == Group.UNKNOWN or license.getGroup() == Group.UNDEFINED:
 					group = license.getGroup().name
 					author = version.getAuthor()
+					if author:
+						author = author.replace("\"", "").replace("'", "").replace(";", "")
+						author = "\"" + author + "\""
 					email = version.getEmail()
+					if email:
+						email = email.replace("\"", "").replace("'", "").replace(";", "")
+						email = "\"" + email + "\""
 					repository = package.getRepository()
+					if repository:
+						repository = repository.replace("\"", "").replace("'", "").replace(";", "")
+						repository = "\"" + repository + "\""
 					licenses = ", ".join(version.getOriginalLicenses())
+					if licenses:
+						licenses = licenses.replace("\"", "").replace("'", "").replace(";", "")
+						licenses = "\"" + licenses + "\""
 					normalized = ", ".join([str(license) for license in version.getLicenses()])
 					parents = version.getContextSize()
 					file.writerow([ecosystem, package, version, group, author, email, repository, licenses, normalized, parents])