Shell script dir

shell-scripts.d/canticle-to-xml.sh

@@ -0,0 +1,18 @@
+#!/bin/bash
+
+# A script which converts plaintext cantos from one canticle to XML formatting.
+# Must be run inside canticle directory: [inferno,purgatorio,paradiso]-xml.d
+# Tag scripts must be in the above directory.
+
+# 1) word-tags.sh --- add <word></word> tags to each word.
+# 2) terzina-tags.sh --- add <terzina num="..."> tags to each terzina
+# 3) top-level-tags.sh --- add XML header; add <canto> and <terzine> tags
+
+# Check parent directory
+if [[ "$(pwd | awk -F'/' '{print $NF}')" =~ inferno-xml.d|purgatorio-xml.d|paradiso-xml.d ]]; then
+	clear
+	echo "Starting..."
+	cp ../shell-scripts.d/word-tags.sh . && ./word-tags.sh && rm word-tags.sh
+	cp ../shell-scripts.d/terzina-tags.sh . && ./terzina-tags.sh && rm terzina-tags.sh
+	cp ../shell-scripts.d/top-level-tags.sh . && ./top-level-tags.sh && rm top-level-tags.sh
+fi

shell-scripts.d/terzina-tags.sh

@@ -0,0 +1,33 @@
+#!/bin/bash
+
+# Add terzina tags to each canto.xml file in a directory
+# Also add canto number attribute.
+
+echo -e "\nAdding terzina tags and numbers."
+
+for file in *;
+do
+	if [[ "$file" != 'terzina-tags.sh' ]]; then		# exclude this script
+		if [[ "$file" != sed* ]]; then			# exclude sed file made later
+			echo -e "Modifying $file"
+			# Every third line (mod 3) add a closing tag, newline, and opening tag.
+			awk -i inplace '{print}; NR % 3 == 0 {print "\t\t</terzina>\n\t\t<terzina num=~NUM~>"}' "$file"
+			# Add an opening tag at the beginning and a closing tag at the end because the above awk doesn't.
+			sed -i '1s/^/\t\t<terzina num="1">\n/' "$file"
+			echo -e "\t\t</terzina>" >> "$file"
+			# Add the canto number attribute (at most 160/3 ~ 54 terzine)
+			for i in $(seq 1 54);
+			do
+				# Substitute in the actual number with sed. I couldn't figure out
+				# how to include the $i variable with the surrounding attribute
+				# quotes in the awk statement, so I just used sed here separately.
+
+				# We start at i=2 because the awk command above
+				# doesn't add a terzina tag to the first terzina.
+				TERZ_NUM=$((i + 1))
+				# Replace only first occurence of ~NUM~.
+				sed -i -e "0,/~NUM~/ s/~NUM~/\"$TERZ_NUM\"/" "$file"
+			done
+		fi
+	fi
+done

shell-scripts.d/top-level-tags.sh

@@ -0,0 +1,31 @@
+#!/bin/bash
+
+# This script adds top-level tags to your XML files.
+# - XML header
+# - <canto> and <terzine> tags with attributes
+
+echo -e "\nAdding top-level tags."
+
+for file in *;
+do
+	if [[ "$file" != 'top-level-tags.sh' ]]; then
+		CANTO_NUM="$(echo "'$file'" | tr -d -c 0-9)"	# get canto number
+		if [[ "$file" == inferno* ]]; then
+			COMM_NUM="$CANTO_NUM"			# same as inferno canto number
+		elif [[ "$file" == purgatorio* ]]; then
+			COMM_NUM="$((CANTO_NUM + 34))"		# add 34 if in purgatory
+		else
+			COMM_NUM="$((CANTO_NUM + 67))"		# add 68 if in paradise
+		fi
+		# Add tags to first line of file
+		sed -i '1s/^/<?xml version="1.0" encoding="UTF-8"?>\n\n<canto num=~CANTO_NUM~ num_comm=~COMM_NUM~>\n\t<terzine>\n/' "$file"
+		# Replace canto number and comedy number
+		sed -i "s/~CANTO_NUM~/\"$CANTO_NUM\"/" "$file"
+		sed -i "s/~COMM_NUM~/\"$COMM_NUM\"/" "$file"
+		# Add tags to end of file
+		echo -e "\t</terzine>" >> "$file"
+		echo -e "</canto>" >> "$file"
+		# Add a `.xml` extension
+		mv "$file" "$file.xml"
+	fi
+done

shell-scripts.d/word-tags.sh

@@ -0,0 +1,19 @@
+#!/bin/bash
+
+# For all files in directory:
+# 	- replace all spaces with `</word> <word>`
+# 	- add `<word>` to the start of each line
+# 	- add `</word>` to the end of each line
+# Also add proper indentation.
+# This script treats punctuation as words.
+
+echo -e "\nAdding word tags."
+
+for file in *;
+do
+	if [[ "$file" != 'word-tags.sh' ]]; then
+		sed -i 's/\ /<\/word>\ <word>/g' "$file"
+		sed -i 's/^/\t\t\t<word>/' "$file"
+		sed -i 's/$/<\/word>/' "$file"
+	fi
+done

shell-scripts.d/xpath.sh

@@ -0,0 +1,60 @@
+#!/bin/bash
+
+# An interactive xpath query tool
+
+touch .prev-xpath.tmp
+
+GREEN='\033[0;32m'
+NC='\033[0m' # no color
+
+clear
+read -p "Choose a canto: [i,r,p][1-34]" CANTO
+CANTICLE="$(echo $CANTO | tr -d -c a-z)"
+CANTO_NUM="$(echo $CANTO | tr -d -c 0-9)"
+
+file="-${CANTO_NUM}.xml"
+
+if [[ "$CANTICLE" == 'i' ]]; then
+	CANTICLE='Inferno'
+	file="inferno${file}"
+elif [[ "$CANTICLE" == 'r' ]]; then
+	CANTICLE='Purgatorio'
+	file="purgatorio${file}"
+elif [[ "$CANTICLE" == 'p' ]]; then
+	CANTICLE='Paradiso'
+	file="paradiso${file}"
+else
+	echo "Not a valid canto file: $file"
+fi
+
+if [[ -e "$file" ]]; then
+	clear && echo -e "You are searching ${CANTICLE^} ${CANTO_NUM}.\nAdd a trailing slash ${GREEN}\`/\`${NC} to your search before submitting it. The default first two search choices are ${GREEN}\`canto\`${NC} and then ${GREEN}\`terzine\`${NC}. Press ${GREEN}<enter>${NC} for those defaults and type something else to search that.\n"
+	i='0'
+	read -p "Load previous: (Y)/n " PREV
+	case "$PREV" in
+		N|n)
+			search=''
+			;;
+		Y|y|*)
+			search="$(cat .prev-xpath.tmp)"
+			;;
+	esac
+	while true; do
+		read -e -i "$search" -p "$(echo -e "\nSearch: \n  "${GREEN}"===> "${NC})" nsearch
+		if [[ "${nsearch: -1}" == '/' ]]; then			# if last char trailing slash then search
+			path="${nsearch::-1}"
+			echo "$path" > .prev-xpath.tmp			# remove trailing slash
+			xmllint --xpath "$path" $file | batcat -l 'xml'
+			exit
+		elif [[ "$i" == '0' ]] && [[ "$nsearch" == '' ]]; then
+			search="/canto"
+		elif [[ "$nsearch" == '/canto' ]]; then
+			search="/canto/terzine"
+		else
+			search="${nsearch}"
+		fi
+		i=$[$i+1]
+	done	
+else
+	echo "Canto file not found. What directory are you in?"
+fi