From 234d8b9d0fb568e2ad0f6ec9b88fbae87e7145ae Mon Sep 17 00:00:00 2001
From: Louie Shprung <lshprung@scu.edu>
Date: Fri, 30 Dec 2022 12:48:13 -0800
Subject: Rudimentary docset creation

---
 src/index.sh | 63 +++++++++++++++++++++++++++++++++++++-----------------------
 1 file changed, 39 insertions(+), 24 deletions(-)

(limited to 'src/index.sh')

diff --git a/src/index.sh b/src/index.sh
index e81a0f7..e506ae4 100755
--- a/src/index.sh
+++ b/src/index.sh
@@ -6,7 +6,7 @@ shift
 get_title() {
 	FILE="$1"
 
-	PATTERN="<title>.*\(Autoconf\).*</title>"
+	PATTERN="<[tT][iI][tT][lL][eE]>.*</[tT][iI][tT][lL][eE]>"
 
 	#Find pattern in file
 	grep -Eo "$PATTERN" "$FILE" | 
@@ -18,6 +18,8 @@ get_title() {
 		sed 's/[ ]*$//g' | \
 		#Replace '&amp' with '&'
 		sed 's/&amp/&/g' | \
+		# ReplACE '&ndash;' with '-'
+		sed 's/&ndash;/-/g' | \
 		#Replace '&lt;' with '<'
 		sed 's/&lt;/</g'
 }
@@ -27,13 +29,13 @@ get_type() {
 	PATTERN="The node you are looking for is at.*Limitations-of-.*\.html;Builtin
 	The node you are looking for is at;Macro"
 
-	echo "$PATTERN" | while read -r line; do
-		#echo "$line"
-		if grep -Eq "$(echo "$line" | cut -d ';' -f 1)" "$FILE"; then
-			echo "$line" | cut -d ';' -f 2
-			break
-		fi
-	done
+	#echo "$PATTERN" | while read -r line; do
+	#	#echo "$line"
+	#	if grep -Eq "$(echo "$line" | cut -d ';' -f 1)" "$FILE"; then
+	#		echo "$line" | cut -d ';' -f 2
+	#		break
+	#	fi
+	#done
 }
 
 insert() {
@@ -44,23 +46,36 @@ insert() {
 	sqlite3 "$DB_PATH" "INSERT INTO searchIndex(name, type, path) VALUES (\"$NAME\",\"$TYPE\",\"$PAGE_PATH\");"
 }
 
+# Get title and insert into table for each html file
+main() {
+	while [ -n "$1" ]; do
+		unset PAGE_NAME
+		unset PAGE_TYPE
+
+		echo "FILE: $1"
+		# Recurse into subdirectories
+		if [ -d "$1" ]; then
+			main "$1"/*
+		else
+			PAGE_NAME="$(get_title "$1")"
+			if [ -n "$PAGE_NAME" ]; then
+				PAGE_TYPE="$(get_type "$1")"
+				#get_type "$1"
+				if [ -z "$PAGE_TYPE" ]; then
+					PAGE_TYPE="Guide"
+				fi
+				echo "$PAGE_NAME"
+				echo "$PAGE_TYPE"
+				insert "$PAGE_NAME" "$PAGE_TYPE" "$(echo "$1" | sed 's/^ncurses.docset\/Contents\/Resources\/Documents\///')"
+			fi
+		fi
+
+		shift
+	done
+}
+
 # Create table
 sqlite3 "$DB_PATH" "CREATE TABLE searchIndex(id INTEGER PRIMARY KEY, name TEXT, type TEXT, path TEXT);"
 sqlite3 "$DB_PATH" "CREATE UNIQUE INDEX anchor ON searchIndex (name, type, path);"
 
-# Get title and insert into table for each html file
-while [ -n "$1" ]; do
-	unset PAGE_NAME
-	unset PAGE_TYPE
-	PAGE_NAME="$(get_title "$1")"
-	if [ -n "$PAGE_NAME" ]; then
-		PAGE_TYPE="$(get_type "$1")"
-		#get_type "$1"
-		if [ -z "$PAGE_TYPE" ]; then
-			PAGE_TYPE="Guide"
-		fi
-		#echo "$PAGE_TYPE"
-		insert "$PAGE_NAME" "$PAGE_TYPE" "$(basename "$1")"
-	fi
-	shift
-done
+main "$@"
-- 
cgit