publicHealthProject.ahk

;========
; SELECT
;========

		SetWorkingDir %A_ScriptDir%
		FileDelete results.txt
		
		top = CALL#`tLC CLASS`tTITLE`tAUTHOR`tPUBLISHER`tPUBLICATION YEAR`tSUBJECT`tBARCODE`tISBN
		FileAppend, %top%`n, results.txt
		
;=======
; PARSE
;=======
		
		Loop, read, _list.txt
		{
			line := A_LoopReadLine
			
			StringReplace, line, line, ",,All
			;"
			StringSplit, l, line, %A_Tab%
		
			LocationCode := l1
			LocationName := l2
			callNum := l3
			DisplayCallNumberNormalized := l4
			Title := l5
			Enumeration := l6	
			CopyNumber := l7
			PossibleDuplicate := l8
			Author := l9
			Publisher := l10
			PublicationYear := l11
			Edition := l12
			PrimaryLanguage := l13
			Series := l14
			RecordedUsesItem := l15
			RecordedUsesTitle := l16
			LastChargeDateItem := l17
			LastChargeDateTitle := l18
			LastAddDateTitle := l19
			GlobalHoldingsSameEdition := l20
			USAHoldingsSamEdition := l21
			USAHoldingsAnyEdition := l22
			CaliforniaHoldingsSameEdition := l23	
			CaliforniaHoldingsAnyEdition := l24
			AllComparatorLibraryHoldingsSameEdition := l25
			AllComparatorLibraryHoldingsAnyEdition := l26
			OtherCSUsSameEdition := l27
			OtherCSUsAnyEdition := l28
			UCsSameEdition := l29
			UCsAnyEdition := l30
			SelectedLinlPlusSameEdition := l31
			SelectedLinkPlusAnyEdition := l32
			RegionalComparatorsSameEdition := l33	
			RegionalComparatorsAnyEdition := l34
			LocalComparatorsSameEdition := l35
			LocalComparatorsAnyEdition := l36
			Protected := l37
			HathiTrustPublicDomain := l38
			HathiTrustInCopyright := l39
			HathiTrustURL := l40
			OPACURL := l41
			WorldCatURL := l42
			OCLCWorkID := l43
			WorldCatOCLCNumber := l44	
			BibRecordNumber := l45
			BibControlNumber := l46
			ItemControlNumber := l47
			ItemTypeCode := l48
			ItemStatusCode := l49
			ISBN := l50
			Barcode := l51
			
;=========================
; NORMALIZE - CALL NUMBER
;=========================
			
			callNum := RegExReplace(callNum, " v.*", "")
			callNum := RegExReplace(callNum, " pt.*", "")
			callNum := RegExReplace(callNum, "REF ", "")
			callNum := RegExReplace(callNum, "INT GTR", "")
			
;===================
; NORMALIZE - TITLE
;===================

			title := RegExReplace(title, "/.*", "")
			title := RegExReplace(title, "/.*", "")
			title := RegExReplace(title, "#.*", "")
			title := RegExReplace(title, "/[/]", "")

;====================
; NORMALIZE - AUTHOR
;====================

			author := RegExReplace(author, "\s\w\.", "")
			author := RegExReplace(author, "\..*", "")
			author := RegExReplace(author, "\d\d\d\d.*", "@")
			author := RegExReplace(author, ",\s@", "")
			author := RegExReplace(author, ",\s\s@", "")
			author := RegExReplace(author, ",\s\r", "")
			author := RegExReplace(author, "\(.*", "")
			author := RegExReplace(author, ", author", "")
			author := RegExReplace(author, ", Ph", "")
			author := RegExReplace(author, ", editor", "")
			author := RegExReplace(author, "#", " - ")
			
;=========================
; NORMALIZE - ENUMERATION
;=========================
			
			enumeration := RegExReplace(enumeration, "\*", "")
			
;=======================
; NORMALIZE - PUBLISHER
;=======================
			
			publisher := RegExReplace(publisher, ".*: ", "")
			publisher := RegExReplace(publisher, "\d\d\d\d", "")
			publisher := RegExReplace(publisher, ",.*", "")
			
;======================
; NORMALIZE - PUB DATE
;======================

			RegExMatch(publicationYear, "\d\d\d\d", p)
			publicationYear := p
			
;=====================
; NORMALIZE - EDITION
;=====================

			edition := RegExReplace(edition, "\d(.*)", "")

;==========
; PULL OUT
;==========

		subject = 

		if callNum not contains .
		{
			callNum := RegExReplace(callNum, "\s", ".")
		}
		
		StringSplit, c, callNum, .
		
		call := c1
		
		if c2 is integer
		{
			call = %c1%`.%c2%
		}
		
		RegExMatch(call, "[A-Za-z]*", let)
		RegExMatch(call, "\d.*", num)
		
		num := RegExReplace(num, "\s.*", "")
		
		callNormalized := let
		
		If let = GE
		{
			subject = Environmental Sciences
		}
		
		If let = GF
		{
			if num between 51 and 74.999
			{
				subject = Environmental influences on man
			}
			
			if num between 75 and 79.999
			{
				subject = Man's influences on the environment
			}
		}
		
		if let = GV
		{
			if num between 443 and 443.999
			{
				subject = Physical education for children
			}
			
			if num between 446 and 446.999
			{
				subject = Special physical education
			}
			
			if num between 447 and 449.999
			{
				subject = Physical education for the aged
			}
			
			if num between 452 and 545.4999
			{
				subject = Excerise, gymnastics
			}
		}
		
		if let = HM
		{
			if num between 856 and 865.999
			{
				subject = Environment. Social ecology
			}
		}
		
		if let = HN
		{
			if num between 1 and 24.999
			{
				subject = Social history and conditions. Social problems
			}
			
			if num between 25 and 28.999
			{
				subject = Social history and confitions. General
			}
		}
		
		if let = HV
		{
			if num between 680 and 696.999
			{
				subject = Medical and other professional services
			}
			
			if num between 697 and 4996.999
			{
				subject = Social pathology. Special cases
			}
			
			if num between 4997 and 6000.999
			{
				subject = Substance abuse
			}
		}
		
		if let = KF
		{
			if num between 3300 and 3700.999
			{
				subject = Social legislation
			}
			
			if num between 3771 and 3774.999
			{
				subject = Birth control. Family planning
			}
			
			if num between 3775 and 3834.999
			{
				subject = Medical legislation
			}
			
			if num between 3861 and 3940.999
			{
				subject = Food. Drugs. Alcohol
			}
		}
		
		if let = LB
		{
			if num between 3401 and 3640.999
			{
				subject = School hygiene. School life
			}
		}
		
		if let = QH
		{
			if num between 323 and 323.999
			{
				subject = School hygiene. School life
			}
		}
		
		if let = QP
		{
			if num between 31 and 33.999
			{
				subject = Physiology. General
			}
			
			if num between 82 and 82.999
			{
				subject = Environmental influences, e.g. altitude, cold, pesticides, radiation, stress, etc.
			}
			
			if num between 83 and 87.999
			{
				subject = Developmental physiology, e.g. growth. aging, death
			}
			
			if num between 136 and 144.999
			{
				subject = Nutrition
			}
			
			if num between 301 and 350.999
			{
				subject = Movements. Kinesiology. Voice and speech
			}
		}
			
		if let = QR
		{
			if num between 48 and 50.999
			{
				subject = Sanitary microbiology
			}
			
			if num between 180 and 200.999
			{
				subject = Immunology
			}
			
			if num between 201 and 300.999
			{
				subject = Pathogenic micro-organisms
			}
			
			if num between 352 and 354.999
			{
				subject = Mycoplasmas. Rickettsias
			}
			
			if num between 355 and 502.999
			{
				subject = Virology
			}
		}
		
		if let = RA
		{
			subject = Public Aspects of Medicine
		}
		
		if let = RJ
		{
			if num between 59 and 60.999
			{
				subject = Infant and neonatal morbidity and mortality
			}
			
			if num between 101 and 124.999
			{
				subject = Child health and wellness
			}
			
			if num between 240 and 244.999
			{
				subject = Immunization. Hospital care
			}
			
			if num between 370 and 599.999
			{
				subject = Diseases
			}
		}
		
		if let = RK
		{
			if num between 52 and 57.999
			{
				subject = Public health
			}
		}
		
		if let = TD
		{
			if num between 1 and 157.999
			{
				subject = Environmental Technology. Sanitary Engineering. General. History
			}
			
			if num between 172 and 192.999
			{
				subject = Environmental pollution. Remediation
			}
			
			if num between 194 and 195.399
			{
				subject = Effects of industries and plants
			}
			
			if num between 201 and 510.999
			{
				subject = Water supply
			}
		}
		
		if let = TX
		{
			If num between 341 and 644.999
			{
				subject = Nutrition. Food supply
			}
		}
			
;=======
; WRITE
;=======

		if subject !=
		{
			write = %callNum%`t%callNormalized%`t%Title%`t%Author%`t%publisher%`t%PublicationYear%`t%subject%`t%barcode%`t%isbn%
			
			FileAppend, %write%`n, results.txt
		}
	}
		
;========
; FINISH
;========
		
		msgbox Done.