From 80dbf5531816e8c56b41bcf895e47d655a87b75a Mon Sep 17 00:00:00 2001
From: Louie Shprung <lshprung@scu.edu>
Date: Tue, 14 Mar 2023 14:24:59 -0700
Subject: Bump default version; restructure source to use pup for html parsing

---
 .gitignore   |  4 ++--
 Makefile     |  2 +-
 README       |  1 +
 src/index.sh | 65 ++++++++++++++++++++++++++++--------------------------------
 4 files changed, 34 insertions(+), 38 deletions(-)

diff --git a/.gitignore b/.gitignore
index 9941ba2..9408f63 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,3 +1,3 @@
-/GNU_Make.docset
-/GNU_Make.tgz
+/GNU_Autoconf_Archive.docset
+/GNU_Autoconf_Archive.tgz
 /tmp
diff --git a/Makefile b/Makefile
index fd8a429..d5c95b3 100644
--- a/Makefile
+++ b/Makefile
@@ -10,7 +10,7 @@ INDEX_FILE      = $(RESOURCES_DIR)/docSet.dsidx
 ICON_FILE       = $(DOCSET_DIR)/icon.png
 ARCHIVE_FILE    = $(DOCSET_NAME).tgz
 
-MANUAL_VERSION = 2022.09.03
+MANUAL_VERSION = 2023.02.20
 MANUAL_URL  = https://ftp.gnu.org/gnu/autoconf-archive/autoconf-archive-$(MANUAL_VERSION).tar.xz
 MANUAL_SRC = tmp/autoconf-archive-$(MANUAL_VERSION)
 MANUAL_SRC_MAKEFILE = $(MANUAL_SRC)/Makefile
diff --git a/README b/README
index 0dbe50c..a345bcd 100644
--- a/README
+++ b/README
@@ -12,5 +12,6 @@ Requirements:
 - any POSIX-compliant shell
 - curl
 - make
+- pup
 - sqlite3
 - texinfo
diff --git a/src/index.sh b/src/index.sh
index 4d4b12b..05c066e 100755
--- a/src/index.sh
+++ b/src/index.sh
@@ -1,29 +1,20 @@
 #!/usr/bin/env sh
 
-DB_PATH="$1"
-shift
+create_table() {
+	sqlite3 "$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() {
 	FILE="$1"
 
-	PATTERN="<title>.*\(Autoconf Archive\).*</title>"
-
-	#Find pattern in file
-	grep -Eo "$PATTERN" "$FILE" | 
-		#Remove tag
-		sed 's/<[^>]*>//g' | \
-		#Remove '(automake)'
+	pup -p -f "$FILE" 'title text{}' | \
 		sed 's/(Autoconf Archive)//g' | \
-		#Remove trailing space
-		sed 's/[ ]*$//g' | \
-		#Replace '&amp' with '&'
-		sed 's/&amp/&/g' | \
-		#Replace '&lt;' with '<'
-		sed 's/&lt;/</g'
+		sed 's/\"/\"\"/g'
 }
 
 get_type() {
-	FILE="$(basename $1)"
+	FILE="$(basename "$1")"
 	MACRO_PATTERN="^ax_"
 
 	if echo "$FILE" | grep -q "$MACRO_PATTERN"; then
@@ -39,23 +30,27 @@ insert() {
 	sqlite3 "$DB_PATH" "INSERT INTO searchIndex(name, type, path) VALUES (\"$NAME\",\"$TYPE\",\"$PAGE_PATH\");"
 }
 
-# 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"
+insert_pages() {
+	# 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 "$1")"
 		fi
-		#echo "$PAGE_TYPE"
-		insert "$PAGE_NAME" "$PAGE_TYPE" "$(basename "$1")"
-	fi
-	shift
-done
+		shift
+	done
+}
+
+DB_PATH="$1"
+shift
+
+create_table
+insert_pages "$@"
-- 
cgit