ViewVC Help

View File | Revision Log | Show Annotations | View Changeset | Root Listing

root/ns_dev/Python/NinoCode/Active_prgs/Redgrave/NTRS-TopSenderAnalysis.py

Comparing Python/NinoCode/Active_prgs/Redgrave/NTRS-TopSenderAnalysis.py (file contents):
Revision 810 by nino.borges, Fri Jan 5 19:24:53 2024 UTC vs.
Revision 811 by nino.borges, Fri Jan 5 20:25:29 2024 UTC

#	Line 14 | Line 14 | To-Do: Method to QC for any parsing erro
14		import csv, os, re
15
16		class TopSendersAnalyzer(object):
17	<	version = "0.07"
17	>	version = "0.08"
18
19		def __init__(self):
20		self.startDir = r"C:\Users\eborges\Documents\Cases\Northern Trust\20231227 - FileNetTopSenderAnalysis-Req"
#	Line 41 | Line 41 | class TopSendersAnalyzer(object):
41		self.falsePositiveLanIdAddressesSet = set()
42		self.falsePositiveLanIdAddressesOutputFileName = r"C:\Users\eborges\Documents\Cases\Northern Trust\Extracted_FALSE-POSTIVE_LanAddresses.txt"
43
44	<	##  Sender values, deduplicated within each CSV only, where there is either no NTRS domain in the to or there is but it's not a lanID. Matrix of fileName:valuesSet.
45	<	self.senderEmailAddressesMatrix = {}
44	>	##  Sender values, deduplicated within each CSV only, where there is either no NTRS domain in the to or there is but it's not a lanID. Set of fileName|valuesSet.
45	>	##  Since there is only a single value in sender, I'm grabbing the entire value and not just the email address.
46	>	self.senderEmailAddressesAcrossCSVsSet = set()
47		self.senderEmailAddressesFileName = r"C:\Users\eborges\Documents\Cases\Northern Trust\SendersListSpecial.txt"
48
49		#self.lanIdRegExPattern = '[A-Za-z]{2}[0-9]{2,3}@NTRS.COM'
#	Line 76 | Line 77 | class TopSendersAnalyzer(object):
77		dateInPath = re.findall(self.dateInPathRegExPattern, root)
78		allEmailAddressesInCSVSet = set()
79
79	–	##  This is the list of unique Sender Addresses,per CSV, ONLY IF THE EMAIL IS NOT TO A LANID.  Diana asked for this list.
80	–	##  Since there is only a single value in sender, I'm grabbing the entire value and not just the email address.
81	–	senderEmailAddressesInCSVSet = set()
80
81		##  This is the full count of receipient addresses found in the CSV, not unique addresses
82		toFieldAddressesInCSVCount = 0
#	Line 121 | Line 119 | class TopSendersAnalyzer(object):
119		ntrsNoLanBucketCount +=1
120		## Also since no LAN ID was found in the TO field, add to the unique senders list.
121		senderValue = toValue = row['Sender']
122	<	senderEmailAddressesInCSVSet.add(senderValue.upper())
122	>	self.senderEmailAddressesAcrossCSVsSet.add(f"{fl}|{senderValue.upper()}")
123		else:
124		## No ntrs addresses found at all,
125		noNtrsDomainBucketCount +=1
126		## Also since no NTRS address was found in the TO field at all, add to the unique senders list.
127		senderValue = toValue = row['Sender']
128	<	senderEmailAddressesInCSVSet.add(senderValue.upper())
128	>	self.senderEmailAddressesAcrossCSVsSet.add(f"{fl}|{senderValue.upper()}")
129	>
130		scenario = self.CalculateScenario(ntrsAndLanBucketCount/fileRowCount, toFieldAddressesInCSVCount/fileRowCount)
131		print(f"{fl}|{dateInPath}|{fileRowCount}|{eightyCharRowCount}|{noAddressesInToFieldCount}|{nonEightyCharRowCount}|{charsOverEighty}|{len(allEmailAddressesInCSVSet)}|{toFieldAddressesInCSVCount/fileRowCount}|{noNtrsDomainBucketCount}|{noNtrsDomainBucketCount/fileRowCount}|{ntrsAndLanBucketCount}|{ntrsAndLanBucketCount/fileRowCount}|{ntrsNoLanBucketCount}|{ntrsNoLanBucketCount/fileRowCount}|{scenario}|{self.scenarioDescriptionMatrix[scenario]}")
132		csv_file.close()
133		##  Update the global all email addresses set, if they selected this option
134		if writeAllPossibleEmailAddressesLogFile:
135		self.allPossibleEmailAddressesSet.update(allEmailAddressesInCSVSet)
137	–
138	–	##  Update the global special senders matrix, if they selected this option.
139	–	if writeSpecialSendersListLogFile:
140	–	self.senderEmailAddressesMatrix[fl] = senderEmailAddressesInCSVSet
136
137		if writeTrueLanIDLogFile:
138		print("Writing the True LAN ID log file...")
#	Line 153 | Line 148 | class TopSendersAnalyzer(object):
148		print("Done.\n")
149		if writeSpecialSendersListLogFile:
150		print("Writing the deduplicated special senders data to log file....")
151	<
157	<
151	>	self.WriteLogFile(self.senderEmailAddressesAcrossCSVsSet, self.senderEmailAddressesFileName)
152		print("Done.\n")
153
154
#	Line 183 | Line 177 | class TopSendersAnalyzer(object):
177		tempList = list(setOfValues)
178		tempList.sort()
179		for i in tempList:
180	<	outFl.write(f"{i}\n")
180	>	try:
181	>	outFl.write(f"{i}\n")
182	>	except:
183	>	print(i)
184		outFl.close()
185
186		def CalculateScenario(self,rawNumber, averageNumbRecip):
#	Line 201 | Line 198 | class TopSendersAnalyzer(object):
198		if __name__ == '__main__':
199
200		tsa = TopSendersAnalyzer()
201	<	tsa.AnalyzeTopSenders(writeTrueLanIDLogFile = True, writeFalsePositiveLanIDLogFile = True, writeAllPossibleEmailAddressesLogFile = True)
201	>	tsa.AnalyzeTopSenders(writeTrueLanIDLogFile = True, writeFalsePositiveLanIDLogFile = True, writeAllPossibleEmailAddressesLogFile = True, writeSpecialSendersListLogFile = True)
202
203
204		#print(f"There are {len(allToLanAddressesSet)} unique LAN ID addresses.")

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines (old)
>	Changed lines (new)