From d296cf0f97b59cfc45da21ec808981b41397b752 Mon Sep 17 00:00:00 2001
From: Frederik Baerentsen <drudoo@drudoo.com>
Date: Wed, 24 Jun 2020 08:41:55 +0200
Subject: [PATCH] Fixed errors

---
 lego_dl.sh | 78 +++++++++++++++++++++++++++++++++++++++++++++---------
 1 file changed, 65 insertions(+), 13 deletions(-)

diff --git a/lego_dl.sh b/lego_dl.sh
index a92ae7d..a680b76 100644
--- a/lego_dl.sh
+++ b/lego_dl.sh
@@ -1,17 +1,22 @@
 #!/bin/bash
 
+# Get themes.csv and sets.cvs from https://rebrickable.com/downloads
+
+
 echo "----> Starting Download of all LEGO instructions from https://brickset.com/exportscripts/instructions"
 firstline=0
 logfile="lego_errors.log"
-downloadFolder="Instructions"
+downloadFolder="../Instructions"
 #error_level: 
 #	0 no errors reported
 #	1 Download errors reported
 #	2 Existing files and download errors reported
-
-error_level=1
-
+error_level=0
 echo "" > $logfile
+
+THEMES="themes.csv"
+SETS="sets.csv"
+
 if [ -f "instructions" ]; then
 	echo "---> Using existing instructions file. Manually delete it to redownload..."
 else 
@@ -24,19 +29,63 @@ else
 		echo "Done!" 
 	fi
 fi
-while IFS=, read -r ID LINK NAME DESC ADDED RETRIVED
+
+echo "----> Converting csv with ',' to '|'..."
+head -1 instructions | sed 's/,/|/g' > instructions.csv
+sed 1,1d instructions | sed -r 's/("[^",]+),([^",]+")/\1###\2/g' | awk -F, '{print $1,$2,$3,$4,$5,$6}' | sed 's/###/,/g' | sed 's/" "/"|"/g' >> instructions.csv
+
+while IFS='|' read -r ID LINK NAME DESC ADDED RETRIVED
 do
+
+	#echo $ID $LINK $NAME $DESC $ADDED $RETRIVED
+
 	if [ "$firstline" = 0 ]; then
 		firstline=1
 	else
 		#echo "I got:$ID - $LINK - $NAME - $DESC - $ADDED - $RETRIVED"
-		
+			
 		tID=$(sed -e 's/^"//' -e 's/"$//' <<<"$ID")
 		tLINK=$(sed -e 's/^"//' -e 's/"$//' <<<"$LINK")
-		tNAME=$(sed -e 's/^"//' -e 's/"$//' <<<"$NAME")
+		#tNAME=$(sed -e 's/^"//' -e 's/"$//' <<<"$NAME")
+		ttNAME=$(cut -d, -f1-2 sets.csv | grep -w $tID | cut -d, -f2)
+		setYEAR=$(grep -w $tID sets.csv | cut -d, -f3)
+		setNAME=$(sed -e 's/[^A-Za-z0-9._-]/_/g' <<< $ttNAME)
+		themeID=$(grep -w $tID sets.csv | cut -d, -f4)
+		themeName=$(awk -F',' -v id="$themeID" '$1 == id' themes.csv)
+		IFS=',' read -r -a array <<< "$themeName"  
+		tempID=${array[2]}
+		if [[ $tempID != "" ]]; then
+			while [[ $tempID != "" ]]  
+			do 	 
+				#echo "tempID: $tempID"
+		 		tthemeName=$(awk -F',' -v id="$tempID" '$1 == id' themes.csv)
+		 		#echo $tthemeName 	 
+				IFS=',' read -r -a tArray <<< "$tthemeName"
+		 		#echo "<${tArray[2]}>"
+		 		tempID=${tArray[2]}
+		 		themeName=${tArray[1]}
+			done 
+		else  
+			themeName=${array[1]} 
+		fi
+
+		setTHEME=$(sed -e 's/[^A-Za-z0-9._-]/_/g' <<< $themeName)
+	
+		#echo $setYEAR	
+		
 		tADDED=$(sed -e 's/^"//' -e 's/"$//' <<<"$ADDED")
 		tDESC=$(sed -e 's/^"//' -e 's/"$//' <<<"$DESC")
-		filename=$tID"_"${tNAME// /_}"_("${tDESC// /_}")_"$tADDED".pdf"
+		ttDESC=$(echo $tDESC | grep -Eo '\s[0-9]{1}\s?\/\s?[0-9]{1,2}' | sed 's/ //g' | sed 's/\//_of_/g')
+		
+		if [ -z "$ttDESC" ]; then
+			setDESC="_"
+		else
+			setDESC="_("$ttDESC")_"
+		fi
+
+		tFilename=""$tID""$setDESC""${setNAME// /_}"_("$setTHEME"_"$setYEAR").pdf"
+		filename=$tFilename
+
 		if [ -f "$downloadFolder/$filename" ]; then
 			echo "-> $tID exists. Skipping..."
 			if [[ $error_level = 2 ]]; then
@@ -49,22 +98,25 @@ do
 					echo "$filename is not available." >> $logfile
 				fi
 			else  
+			 	#echo "___ URL $tLINK"
 				echo -ne "--> $tID downloading now..."
-				curl -H "Mozilla/5.0 (platform; rv:75.0) Gecko/20100101 Firefox/75.0" $tLINK --silent --output "Instructions/$filename"
+				curl -H "Mozilla/5.0 (platform; rv:75.0) Gecko/20100101 Firefox/75.0" -L $tLINK --silent --output "$downloadFolder/$filename"
+				#wget -T20 --tries=3 --retry-connrefused --continue -O "Instructions/$filename" $tLINK
 				if [ -f "$downloadFolder/$filename" ]; then
-					echo "Done!"
+					echo "Done! > $filename"
 				else
 					echo "ERROR!"
-					echo "--> Not downloaded. Try again manually..."
 					if [[ $error_level = 1 || $error_level = 2 ]]; then  
+					  echo 
+						echo "--> Not downloaded. Try again manually..."
 						echo "$filename was not downloaded. Check CURL" >> $logfile
 					fi
 				fi
 				#random sleep in order to not look like a script
-				sleep $(( ( RANDOM % 10 )  + 1 ))
+				sleep $(( ( RANDOM % 5 )  + 1 ))
 			fi
 		fi
 	fi
 	
-done < instructions
+done < instructions.csv