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 809 by nino.borges, Thu Jan 4 21:03:18 2024 UTC vs.
Revision 810 by nino.borges, Fri Jan 5 19:24:53 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 = "0.06"
17	>	version = "0.07"
18
19		def __init__(self):
20		self.startDir = r"C:\Users\eborges\Documents\Cases\Northern Trust\20231227 - FileNetTopSenderAnalysis-Req"
#	Line 40 | Line 40 | class TopSendersAnalyzer(object):
40		##  False positive NTRS LAN ID matches, per specification provided to me.  Close but just outside of specification. (for analysis)
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 = {}
46	+	self.senderEmailAddressesFileName = r"C:\Users\eborges\Documents\Cases\Northern Trust\SendersListSpecial.txt"
47
48		#self.lanIdRegExPattern = '[A-Za-z]{2}[0-9]{2,3}@NTRS.COM'
49		self.lanIdRegExPattern = '[A-Za-z]{1,15}[0-9]{1,3}@NTRS.COM'
#	Line 56 | Line 60 | class TopSendersAnalyzer(object):
60		#self.allPossibleEmailAddressesRegExPattern = r"[\w\.-]+@[\w\.-]+\.\w+"
61		self.allPossibleEmailAddressesRegExPattern = r"[\w.+-]+@[\w-]+\.[\w.-]+"
62
63	<	def AnalyzeTopSenders(self, writeTrueLanIDLogFile = False, writeFalsePositiveLanIDLogFile = False, writeAllPossibleEmailAddressesLogFile = False):
63	>	def AnalyzeTopSenders(self, writeTrueLanIDLogFile = False, writeFalsePositiveLanIDLogFile = False, writeAllPossibleEmailAddressesLogFile = False, writeSpecialSendersListLogFile = False):
64		"""Main Method in this program"""
65		print("FileName|CSV Date|Count no Header|80 Char TO Row Count|# Rows with NO EMAIL ADDRESSES|non-80 Char TO Row Count|TO over 80 Char?|Unique Email Addresses Count|Average # of Recipients|# Rows No NTRS Domains|% No NTRS Domains|# Rows NTRS and LAN IDs|% NTRS and LAN IDs|# Rows NTRS No LAN IDs|% NTRS No LAN IDs|Scenario|Scenario Short Description")
66		for (root,dirs,files) in os.walk(self.startDir):
#	Line 72 | Line 76 | class TopSendersAnalyzer(object):
76		dateInPath = re.findall(self.dateInPathRegExPattern, root)
77		allEmailAddressesInCSVSet = set()
78
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()
82	+
83		##  This is the full count of receipient addresses found in the CSV, not unique addresses
84		toFieldAddressesInCSVCount = 0
85
#	Line 111 | Line 119 | class TopSendersAnalyzer(object):
119		else:
120		## Not 1 true LAN ID was found, using the True Test
121		ntrsNoLanBucketCount +=1
122	+	## Also since no LAN ID was found in the TO field, add to the unique senders list.
123	+	senderValue = toValue = row['Sender']
124	+	senderEmailAddressesInCSVSet.add(senderValue.upper())
125		else:
126		## No ntrs addresses found at all,
127		noNtrsDomainBucketCount +=1
128	+	## Also since no NTRS address was found in the TO field at all, add to the unique senders list.
129	+	senderValue = toValue = row['Sender']
130	+	senderEmailAddressesInCSVSet.add(senderValue.upper())
131		scenario = self.CalculateScenario(ntrsAndLanBucketCount/fileRowCount, toFieldAddressesInCSVCount/fileRowCount)
132		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]}")
133		csv_file.close()
134		##  Update the global all email addresses set, if they selected this option
135		if writeAllPossibleEmailAddressesLogFile:
136		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
141	+
142		if writeTrueLanIDLogFile:
143		print("Writing the True LAN ID log file...")
144		self.WriteLogFile(self.trueLanIdAddressesSet, self.trueLanIdAddressesOutputFileName)
#	Line 132 | Line 151 | class TopSendersAnalyzer(object):
151		print("Writing the All Possible Email Addresses Across All CSV Files log file...")
152		self.WriteLogFile(self.allPossibleEmailAddressesSet, self.allPossibleEmailAddressesOutputFileName)
153		print("Done.\n")
154	+	if writeSpecialSendersListLogFile:
155	+	print("Writing the deduplicated special senders data to log file....")
156	+
157	+
158	+	print("Done.\n")
159
160
161		def LanIDTrueTest(self, listOfIds):

Diff Legend

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