Step 0 fixes

- Issue #1 created because of the workaround for Bookofrandomtables1.pdf

step-0-Cleaner.sh

@@ -1,8 +1,18 @@
 #!/bin/bash
+
+### Variables used in script
 infile="bookofrandomtables1.txt"
+##infile=$1
 cleanFile="clean-$infile"
 titleFile="titles-$infile"
 outpath='./Step-1'
+###
+
+### Debugger 
+source ~/bin/debugger
+debug=true
+###
+
 
 getTitles() {
 	cat $cleanFile | \
@@ -17,47 +27,62 @@ getTitles() {
 }
 
 fixScan(){
-	sed s/ffi/ffi/gm $infile > $cleanFile
+	cp $infile $cleanFile
+	sed -i 's/ffi/ffi/gm' $cleanFile
+	sed -i 's/Items in a Office/Items in an Office/gm' $cleanFile
 }
 
 GetListFromTitle() {
 	local titlecount=$(wc -l $titleFile | awk '{print $1}')
-	echo "TitleCount: $titlecount"
+	debugger "TitleCount: $titlecount"
 	local currentlinenum=1
 	while (( $currentlinenum < $titlecount )); do
-		echo ""
-		echo ""
-		echo ""
-		echo "$currentlinenum / $titlecount"
+		debugger ""
+		debugger ""
+		debugger ""
+		debugger "$currentlinenum / $titlecount"
 
 
 		lineTitle=$(sed -n "$currentlinenum p" $titleFile)
-		echo "Title from Title File= $lineTitle"
-	#	sed -n "$currentlinenum p" $titleFile
-		grepcommand="cat $cleanFile | grep -vE '\.' | $(badGrepper "$lineTitle")"
-		echo "Grep commands = $grepcommand"
-		local currentline="$(eval "$grepcommand")"
-		grepcommand=""
-		echo "Current Line = $currentline"
+		debugger "Title from Title File= $lineTitle"
+		
+		currentline=$(getTitleLine "$lineTitle")
+		debugger "Current Line $currentline"
 
 
-		echo "CurrentLineNum: $currentlinenum"
+#		currentline=""
+#		currentline="$(cat $cleanFile | grep "$lineTitle" | grep -v '\.' )"
+#		debugger "Title from $cleanFile: $currentline"
+
+
+#		if [ -z $currentline  ] ; then
+#			grepcommand="cat $cleanFile | grep -vE '\.' | $(badGrepper "$lineTitle")"
+#			debugger "Grep commands = $grepcommand"
+#			currentline="$(eval "$grepcommand")"
+#			grepcommand=""
+#		fi
+#		debugger "Current Line = $currentline"
+#
+#
+		debugger "CurrentLineNum: $currentlinenum"
 		local tableNum=$currentlinenum
 		currentlinenum=$(( $currentlinenum + 1 ))
-		echo "CurrentLineNum After+1: $currentlinenum"
+		debugger "CurrentLineNum After+1: $currentlinenum"
+
+		
 
 		nextLineTitle=$(sed -n "$currentlinenum p" $titleFile)
-		echo "Next Lines Title= $nextLineTitle"
-		grepcommand="cat $cleanFile | grep -vE '\.' | $(badGrepper "$nextLineTitle")"
-		echo "nexlines grepcommand: $grepcommand"
-		local nextline="$(eval "$grepcommand")"
-		echo "Next Line $nextline"
+		debugger "Next Lines Title= $nextLineTitle"
+
+
+		nextline=$(getTitleLine "$nextLineTitle")
+		debugger "Next Line $nextline"
 
 
 
 		local maxlines=$(wc -l $cleanFile | awk '{print $1}')
-		echo $currentline
-		echo $nextline
+		debugger $currentline
+		debugger $nextline
 #exit
 		cat $cleanFile | \
 			grep -A $maxlines "Good adventuring" | \
@@ -65,12 +90,33 @@ GetListFromTitle() {
 			grep -B $maxlines "$nextline" > "$outpath/$tableNum-$currentline.txt"
 	done
 	local currentline=$(sed -n "$currentlinenum p" $titleFile)
-	echo HEAERGAERGA
-	echo $currentline
-	echo EFAEFAWFE
+	debugger HEAERGAERGA
+	debugger $currentline
+	debugger EFAEFAWFE
 	cat $cleanFile | grep -A $maxlines "$currentline" > "$outpath/$titlecount-$currentline.txt"
 }
 
+function getTitleLine() {
+	
+	output_line=""
+	output_line="$(cat $cleanFile | grep "$1" | grep -v '\.' )"
+	echo "output before badgrepper: $output_line" > $(tty)
+ 	if [ -z "$output_line"  ] ; then
+		grepcommand="cat $cleanFile | grep -vE '\.' | $(badGrepper "$1")"
+		output_line="$(eval "$grepcommand")"
+		grepcommand=""
+	fi
+	echo "output after badgrepper: $output_line" > $(tty)
+ 	if [ -z "$output_line"  ] ; then
+		echo "Unable to match Table of Contents Title with title further down in document" >&2
+		exit 1
+	else
+		echo $output_line
+	fi
+
+}
+
+
 
 function dumpOfAllMyRandomCommandsAndStuff() {
 	# command for turning string variable into array variable
@@ -103,13 +149,13 @@ case $1 in
 		echo "The following is the command  '$grepresult'"
 		;;
 	*)
-		echo "Fixing spelling from certain pdftotext bugs"
+		debugger "Fixing spelling from certain pdftotext bugs"
 		fixScan
 		
-		echo "Getting Titles"
+		debugger "Getting Titles"
 		getTitles
 		mkdir -p $outpath
-		echo "Creating lists"
+		debugger "Creating lists"
 		GetListFromTitle
 		;;
 esac