Active_prgs/Relativity/HPPR_ReportParser.py

"""

HPPR_ReportParser.py

This program will take a history export from the HPPR and parse and hten make a spreadsheet
with the results.
"""

import os, shutil, time, ExcelLib

class DB_View:
    def __init__(self,begno,viewDate):
        self.begNo = begno
        self.viewCount = 1
        self.viewDateList = [viewDate,]
    def AddToViewDateList(self,date):
        """Adds value to view date list"""
        self.viewDateList.append(date)
    def IncrementViewCount (self,n=1):
        """Increments the viewcount. default if to increment by 1"""
        self.viewCount += n
    def ReturnInformation(self,info):
        data = getattr(self,info)
        return data
    def SaveInformation(self,field,info):
        data = setattr(self,field,info)

class DB_Print:
    def __init__(self,begno,printDate):
        self.begNo = begno
        self.printCount = 1
        self.printDateList = [printDate,]
    def AddToPrintDateList(self,date):
        """Adds value to print date list"""
        self.printDateList.append(date)
    def IncrementPrintCount (self,n=1):
        """Increments the viewcount. default if to increment by 1"""
        self.printCount += n
    def ReturnInformation(self,info):
        data = getattr(self,info)
        return data
    def SaveInformation(self,field,info):
        data = setattr(self,field,info)

class DB_Query:
    def __init__(self,begno):
        self.BegNo = begno
    def ReturnInformation(self,info):
        data = getattr(self,info)
        return data
    def SaveInformation(self,field,info):
        data = setattr(self,field,info)

class Reporting:
    def __init__(self,caseName):
        reportingFolder = r"\\BSTDD967DTW1\Users\eborges\Box Sync\Client\Honeywell\HPPR\Reporting"
        fileName = caseName+ time.strftime("_%Y%m%d") +"_HBPD_Report.xlsx"
        shutil.copy2(r"HBPD_Report_skel.xlsx",os.path.join(reportingFolder,fileName))

        self.spreadsheet = ExcelLib.ExcelConnection(os.path.join(reportingFolder,fileName))
        self.caseName = caseName

        self.activityByDocFile = r"C:\McDermott-Discovery\Client\activityByDocument.txt"

    def CreateDocsOpenedReport(self,viewedMatrix):
        """Creates the documents opened report"""
        myTab = "Docs Opened"

        self.spreadsheet.setCell(myTab,4,1,self.spreadsheet.getCell(myTab,4,1).replace("{DATE}",time.strftime("%m-%d-%Y")))
        self.spreadsheet.setCell(myTab,3,1,self.spreadsheet.getCell(myTab,3,1).replace("{Username /  Case}",self.caseName))

        docColumn = 2
        dateColumn = 4
        docNameColumn = 6
        rowCount = 7

        ## Need to make a new matrix with sorted date as key
        newMatrix = {}
        for i in viewedMatrix.keys():
            for dt in viewedMatrix[i].viewDateList:
                if dt in newMatrix.keys():
                    newMatrix[dt].append(i)
                else:
                    newMatrix[dt] = [i,]
        dateList = newMatrix.keys()
        dateList.sort()

        for dt in dateList:
            for document in newMatrix[dt]:
                self.spreadsheet.setCell(myTab,rowCount,docColumn,document)
                self.spreadsheet.setCell(myTab,rowCount,dateColumn,dt)
                rowCount = rowCount +1

    def CreateDocsPrintedReport(self,printedMatrix):
        """Creates the documents printed report"""
        myTab = "Docs Printed"

        self.spreadsheet.setCell(myTab,4,1,self.spreadsheet.getCell(myTab,4,1).replace("{DATE}",time.strftime("%m-%d-%Y")))
        self.spreadsheet.setCell(myTab,3,1,self.spreadsheet.getCell(myTab,3,1).replace("{Username /  Case}",self.caseName))

        docColumn = 2
        dateColumn = 4
        docNameColumn = 6
        rowCount = 7

        ## Need to make a new matrix with sorted date as key
        newMatrix = {}
        for i in printedMatrix.keys():
            for dt in printedMatrix[i].printDateList:
                if dt in newMatrix.keys():
                    newMatrix[dt].append(i)
                else:
                    newMatrix[dt] = [i,]
        dateList = newMatrix.keys()
        dateList.sort()

        for dt in dateList:
            for document in newMatrix[dt]:
                self.spreadsheet.setCell(myTab,rowCount,docColumn,document)
                self.spreadsheet.setCell(myTab,rowCount,dateColumn,dt)
                rowCount = rowCount +1

    def CreateDocsViewCountReport(self,viewedMatrix):
        """Creates the documents View Count report. Right now this is just docs viewed but might need to merge with printed."""
        myTab = "Docs by View Count"

        self.spreadsheet.setCell(myTab,4,1,self.spreadsheet.getCell(myTab,4,1).replace("{DATE}",time.strftime("%m-%d-%Y")))
        self.spreadsheet.setCell(myTab,3,1,self.spreadsheet.getCell(myTab,3,1).replace("{Username /  Case}",self.caseName))

        docColumn = 2
        viewsColumn = 3
        docNameColumn = 6
        rowCount = 6


        ## Need to make a new matrix with sorted date as key
        newMatrix = {}
        for i in viewedMatrix.keys():
            cnt=viewedMatrix[i].viewCount
            if cnt in newMatrix.keys():
                newMatrix[cnt].append(i)
            else:
                newMatrix[cnt] = [i,]
        cntList = newMatrix.keys()
        cntList.sort(reverse=True)

        for ct in cntList:
            for document in newMatrix[ct]:
                self.spreadsheet.setCell(myTab,rowCount,docColumn,document)
                self.spreadsheet.setCell(myTab,rowCount,viewsColumn,ct)
                rowCount = rowCount +1


    def CreateActivityByDocReport(self,viewedMatrix,printedMatrix):
        """ Creates the activity by document report"""
        outputFile = open(self.activityByDocFile,'w')
        viewedList = viewedMatrix.keys()
        printedList = printedMatrix.keys()
        combinedList = viewedList + list(set(printedList) - set(viewedList))

    def FinalizeReport(self):
        """Saves and closes the report. Without this, it wont save anything."""
        self.spreadsheet.save()
        self.spreadsheet.close()


if __name__ == '__main__':
    excelReport = Reporting("CA_Johnson")
    documentsQueried = []
    documentsPrintedMatrix = {}
    documentPrinted = []
    documentsViewedMatrix = {}
    documentsViewed = []
    #reportFile = r"\\BSTDD967DTW1\Users\eborges\Box Sync\Client\Honeywell\HPPR\Reporting\Don_report.dat"
    reportFile = r"\\BSTDD967DTW1\Users\eborges\Box Sync\Client\Honeywell\HPPR\Reporting\CA_Johnson_export.dat"
    
    contents = open(reportFile).readlines()
    contents = contents[1:]

    ##  First we pack the objs into the lists.

    for line in contents:
        line = line.replace("\n","")
        line = line.split("|")
        actionType = line[1]
        if actionType == "View":
            docID = line[0]
            date = line[4]
            if docID in documentsViewedMatrix.keys():
                documentsViewedMatrix[docID].IncrementViewCount()
                documentsViewedMatrix[docID].AddToViewDateList(date)
            else:
                docObj = DB_View(docID,date)
                documentsViewedMatrix[docID] = docObj
        elif actionType == "Document Query":
            pass
        elif actionType == "Print":
            docID = line[0]
            date = line[4]
            if docID in documentsPrintedMatrix.keys():
                documentsPrintedMatrix[docID].IncrementPrintCount()
                documentsPrintedMatrix[docID].AddToPrintDateList(date)
            else:
                docObj = DB_Print(docID,date)
                documentsPrintedMatrix[docID] = docObj
        else:
            print "ERROR: Unsupported action type - %s"% actionType
            
    ##  Then we unpack into the report.
    printReport = open(r"\\BSTDD967DTW1\Users\eborges\Box Sync\Client\Honeywell\HPPR\Reporting\printReport.txt",'w')
    viewReport = open(r"\\BSTDD967DTW1\Users\eborges\Box Sync\Client\Honeywell\HPPR\Reporting\viewReport.txt","w")
    for i in documentsPrintedMatrix.keys():
        printReport.write("%s was printed %s times on the following dates:"% (i,documentsPrintedMatrix[i].printCount))
        for x in documentsPrintedMatrix[i].printDateList:
            printReport.write("%s,"%x)
        printReport.write("\n")
    for i in documentsViewedMatrix.keys():
        viewReport.write("%s was Viewed %s times on the following dates:"% (i,documentsViewedMatrix[i].viewCount))
        for x in documentsViewedMatrix[i].viewDateList:
            viewReport.write("%s,"%x)
        viewReport.write("\n")
    printReport.close()
    viewReport.close()

    ## Excel Reporting
    excelReport.CreateDocsOpenedReport(documentsViewedMatrix)
    excelReport.CreateDocsPrintedReport(documentsPrintedMatrix)
    excelReport.CreateDocsViewCountReport(documentsViewedMatrix)
    excelReport.FinalizeReport()
    
    
Revision:	569
Committed:	Wed Jan 21 22:48:01 2015 UTC (11 years, 2 months ago) by nino.borges
Content type:	text/x-python
File size:	9083 byte(s)
Log Message:	Added support for creating reports in Excel. Viewed and printed tabs are done. doc count viewed is almost done.
#	User	Rev	Content
1	nino.borges	561	"""
2
3			HPPR_ReportParser.py
4
5			This program will take a history export from the HPPR and parse and hten make a spreadsheet
6			with the results.
7			"""
8
9	nino.borges	569	import os, shutil, time, ExcelLib
10	nino.borges	561
11			class DB_View:
12	nino.borges	564	def __init__(self,begno,viewDate):
13			self.begNo = begno
14			self.viewCount = 1
15			self.viewDateList = [viewDate,]
16			def AddToViewDateList(self,date):
17			"""Adds value to view date list"""
18			self.viewDateList.append(date)
19			def IncrementViewCount (self,n=1):
20			"""Increments the viewcount. default if to increment by 1"""
21			self.viewCount += n
22	nino.borges	561	def ReturnInformation(self,info):
23			data = getattr(self,info)
24			return data
25			def SaveInformation(self,field,info):
26			data = setattr(self,field,info)
27
28			class DB_Print:
29	nino.borges	564	def __init__(self,begno,printDate):
30			self.begNo = begno
31			self.printCount = 1
32			self.printDateList = [printDate,]
33			def AddToPrintDateList(self,date):
34			"""Adds value to print date list"""
35			self.printDateList.append(date)
36			def IncrementPrintCount (self,n=1):
37			"""Increments the viewcount. default if to increment by 1"""
38			self.printCount += n
39	nino.borges	561	def ReturnInformation(self,info):
40			data = getattr(self,info)
41			return data
42			def SaveInformation(self,field,info):
43			data = setattr(self,field,info)
44
45			class DB_Query:
46			def __init__(self,begno):
47			self.BegNo = begno
48			def ReturnInformation(self,info):
49			data = getattr(self,info)
50			return data
51			def SaveInformation(self,field,info):
52			data = setattr(self,field,info)
53
54	nino.borges	564	class Reporting:
55	nino.borges	569	def __init__(self,caseName):
56			reportingFolder = r"\\BSTDD967DTW1\Users\eborges\Box Sync\Client\Honeywell\HPPR\Reporting"
57			fileName = caseName+ time.strftime("_%Y%m%d") +"_HBPD_Report.xlsx"
58			shutil.copy2(r"HBPD_Report_skel.xlsx",os.path.join(reportingFolder,fileName))
59
60			self.spreadsheet = ExcelLib.ExcelConnection(os.path.join(reportingFolder,fileName))
61			self.caseName = caseName
62
63	nino.borges	564	self.activityByDocFile = r"C:\McDermott-Discovery\Client\activityByDocument.txt"
64	nino.borges	569
65			def CreateDocsOpenedReport(self,viewedMatrix):
66			"""Creates the documents opened report"""
67			myTab = "Docs Opened"
68
69			self.spreadsheet.setCell(myTab,4,1,self.spreadsheet.getCell(myTab,4,1).replace("{DATE}",time.strftime("%m-%d-%Y")))
70			self.spreadsheet.setCell(myTab,3,1,self.spreadsheet.getCell(myTab,3,1).replace("{Username / Case}",self.caseName))
71
72			docColumn = 2
73			dateColumn = 4
74			docNameColumn = 6
75			rowCount = 7
76
77			## Need to make a new matrix with sorted date as key
78			newMatrix = {}
79			for i in viewedMatrix.keys():
80			for dt in viewedMatrix[i].viewDateList:
81			if dt in newMatrix.keys():
82			newMatrix[dt].append(i)
83			else:
84			newMatrix[dt] = [i,]
85			dateList = newMatrix.keys()
86			dateList.sort()
87
88			for dt in dateList:
89			for document in newMatrix[dt]:
90			self.spreadsheet.setCell(myTab,rowCount,docColumn,document)
91			self.spreadsheet.setCell(myTab,rowCount,dateColumn,dt)
92			rowCount = rowCount +1
93
94			def CreateDocsPrintedReport(self,printedMatrix):
95			"""Creates the documents printed report"""
96			myTab = "Docs Printed"
97
98			self.spreadsheet.setCell(myTab,4,1,self.spreadsheet.getCell(myTab,4,1).replace("{DATE}",time.strftime("%m-%d-%Y")))
99			self.spreadsheet.setCell(myTab,3,1,self.spreadsheet.getCell(myTab,3,1).replace("{Username / Case}",self.caseName))
100
101			docColumn = 2
102			dateColumn = 4
103			docNameColumn = 6
104			rowCount = 7
105
106			## Need to make a new matrix with sorted date as key
107			newMatrix = {}
108			for i in printedMatrix.keys():
109			for dt in printedMatrix[i].printDateList:
110			if dt in newMatrix.keys():
111			newMatrix[dt].append(i)
112			else:
113			newMatrix[dt] = [i,]
114			dateList = newMatrix.keys()
115			dateList.sort()
116
117			for dt in dateList:
118			for document in newMatrix[dt]:
119			self.spreadsheet.setCell(myTab,rowCount,docColumn,document)
120			self.spreadsheet.setCell(myTab,rowCount,dateColumn,dt)
121			rowCount = rowCount +1
122
123			def CreateDocsViewCountReport(self,viewedMatrix):
124			"""Creates the documents View Count report. Right now this is just docs viewed but might need to merge with printed."""
125			myTab = "Docs by View Count"
126
127			self.spreadsheet.setCell(myTab,4,1,self.spreadsheet.getCell(myTab,4,1).replace("{DATE}",time.strftime("%m-%d-%Y")))
128			self.spreadsheet.setCell(myTab,3,1,self.spreadsheet.getCell(myTab,3,1).replace("{Username / Case}",self.caseName))
129
130			docColumn = 2
131			viewsColumn = 3
132			docNameColumn = 6
133			rowCount = 6
134
135
136			## Need to make a new matrix with sorted date as key
137			newMatrix = {}
138			for i in viewedMatrix.keys():
139			cnt=viewedMatrix[i].viewCount
140			if cnt in newMatrix.keys():
141			newMatrix[cnt].append(i)
142			else:
143			newMatrix[cnt] = [i,]
144			cntList = newMatrix.keys()
145			cntList.sort(reverse=True)
146
147			for ct in cntList:
148			for document in newMatrix[ct]:
149			self.spreadsheet.setCell(myTab,rowCount,docColumn,document)
150			self.spreadsheet.setCell(myTab,rowCount,viewsColumn,ct)
151			rowCount = rowCount +1
152
153
154			def CreateActivityByDocReport(self,viewedMatrix,printedMatrix):
155	nino.borges	564	""" Creates the activity by document report"""
156			outputFile = open(self.activityByDocFile,'w')
157			viewedList = viewedMatrix.keys()
158			printedList = printedMatrix.keys()
159			combinedList = viewedList + list(set(printedList) - set(viewedList))
160
161	nino.borges	569	def FinalizeReport(self):
162			"""Saves and closes the report. Without this, it wont save anything."""
163			self.spreadsheet.save()
164			self.spreadsheet.close()
165
166
167	nino.borges	561	if __name__ == '__main__':
168	nino.borges	569	excelReport = Reporting("CA_Johnson")
169	nino.borges	564	documentsQueried = []
170			documentsPrintedMatrix = {}
171			documentPrinted = []
172			documentsViewedMatrix = {}
173			documentsViewed = []
174	nino.borges	569	#reportFile = r"\\BSTDD967DTW1\Users\eborges\Box Sync\Client\Honeywell\HPPR\Reporting\Don_report.dat"
175			reportFile = r"\\BSTDD967DTW1\Users\eborges\Box Sync\Client\Honeywell\HPPR\Reporting\CA_Johnson_export.dat"
176	nino.borges	564
177	nino.borges	561	contents = open(reportFile).readlines()
178	nino.borges	564	contents = contents[1:]
179
180			## First we pack the objs into the lists.
181
182	nino.borges	561	for line in contents:
183	nino.borges	564	line = line.replace("\n","")
184	nino.borges	561	line = line.split("\|")
185	nino.borges	564	actionType = line[1]
186			if actionType == "View":
187			docID = line[0]
188			date = line[4]
189			if docID in documentsViewedMatrix.keys():
190			documentsViewedMatrix[docID].IncrementViewCount()
191			documentsViewedMatrix[docID].AddToViewDateList(date)
192			else:
193			docObj = DB_View(docID,date)
194			documentsViewedMatrix[docID] = docObj
195			elif actionType == "Document Query":
196			pass
197			elif actionType == "Print":
198			docID = line[0]
199			date = line[4]
200			if docID in documentsPrintedMatrix.keys():
201			documentsPrintedMatrix[docID].IncrementPrintCount()
202			documentsPrintedMatrix[docID].AddToPrintDateList(date)
203			else:
204			docObj = DB_Print(docID,date)
205			documentsPrintedMatrix[docID] = docObj
206			else:
207			print "ERROR: Unsupported action type - %s"% actionType
208
209			## Then we unpack into the report.
210	nino.borges	569	printReport = open(r"\\BSTDD967DTW1\Users\eborges\Box Sync\Client\Honeywell\HPPR\Reporting\printReport.txt",'w')
211			viewReport = open(r"\\BSTDD967DTW1\Users\eborges\Box Sync\Client\Honeywell\HPPR\Reporting\viewReport.txt","w")
212	nino.borges	564	for i in documentsPrintedMatrix.keys():
213			printReport.write("%s was printed %s times on the following dates:"% (i,documentsPrintedMatrix[i].printCount))
214			for x in documentsPrintedMatrix[i].printDateList:
215			printReport.write("%s,"%x)
216			printReport.write("\n")
217			for i in documentsViewedMatrix.keys():
218			viewReport.write("%s was Viewed %s times on the following dates:"% (i,documentsViewedMatrix[i].viewCount))
219			for x in documentsViewedMatrix[i].viewDateList:
220			viewReport.write("%s,"%x)
221			viewReport.write("\n")
222			printReport.close()
223			viewReport.close()
224	nino.borges	569
225			## Excel Reporting
226			excelReport.CreateDocsOpenedReport(documentsViewedMatrix)
227			excelReport.CreateDocsPrintedReport(documentsPrintedMatrix)
228			excelReport.CreateDocsViewCountReport(documentsViewedMatrix)
229			excelReport.FinalizeReport()
230	nino.borges	564
231