Rewrite in POSIX shell; index entries from the index pages

4 files changed, 128 insertions, 33 deletions

diff --git a/Makefile b/Makefile
index 836fe3c..096ceee 100644
--- a/Makefile
+++ b/Makefile
@@ -47,9 +47,11 @@ $(DOCUMENTS_DIR): $(RESOURCES_DIR) $(MANUAL_FILE)
 $(INFO_PLIST_FILE): src/Info.plist $(CONTENTS_DIR)
 	cp src/Info.plist $@
 
-$(INDEX_FILE): src/index.rb $(DOCUMENTS_DIR)
+$(INDEX_FILE): src/index.sh $(DOCUMENTS_DIR)
 	rm -f $@
-	ruby src/index.rb $(DOCUMENTS_DIR)/*.html | sqlite3 $@
+	src/index.sh $@ $(DOCUMENTS_DIR)/*.html
+	src/index.sh -i $@ $(DOCUMENTS_DIR)/Concept-Index.html
+	src/index.sh -i -c "Function Variable Directive" $@ $(DOCUMENTS_DIR)/Name-Index.html
 
 $(ICON_FILE): src/icon.png $(DOCSET_DIR)
 	cp src/icon.png $@
diff --git a/README b/README
index 541e5e4..5ddd230 100644
--- a/README
+++ b/README
@@ -9,7 +9,8 @@ will be downloaded from www.gnu.org and packaged appropriately.
 
 Requirements:
 
+- any POSIX-compliant shell
 - curl
 - make
-- ruby
+- pup
 - sqlite3
diff --git a/src/index.rb b/src/index.rb
deleted file mode 100755
index da58535..0000000
--- a/src/index.rb
+++ /dev/null
@@ -1,30 +0,0 @@
-require 'pathname'
-
-puts %Q[
-  CREATE TABLE searchIndex(id INTEGER PRIMARY KEY, name TEXT, type TEXT, path TEXT);
-  CREATE UNIQUE INDEX anchor ON searchIndex (name, type, path);
-]
-
-INSERT_SQL = %Q[
-  INSERT INTO searchIndex(name, type, path) VALUES ('%s','%s','%s');
-]
-
-PATTERN = %r[<title>(.+)</title>]
-
-def quote(s)
-  s.gsub(/&amp;/, '&').gsub(/'/, "\\'")
-end
-
-ARGV.each do |arg|
-  Pathname.glob(arg) do |path|
-    match = path.each_line.lazy.map { |line| PATTERN.match(line) }.find { |m| m }
-    if match
-      title = match[1]
-      title.delete_prefix!('GNU make: ') # older docs
-      title.delete_suffix!(' (GNU make)') # newer docs
-      printf INSERT_SQL, quote(title), 'Guide', path.basename
-    else
-      $stderr.puts "#{path}: no title found"
-    end
-  end
-end
diff --git a/src/index.sh b/src/index.sh
new file mode 100755
index 0000000..3b4cacd
--- /dev/null
+++ b/src/index.sh
@@ -0,0 +1,122 @@
+#!/usr/bin/env sh
+
+create_table() {
+	sqlite3 "$DB_PATH" "CREATE TABLE IF NOT EXISTS searchIndex(id INTEGER PRIMARY KEY, name TEXT, type TEXT, path TEXT);"
+	sqlite3 "$DB_PATH" "CREATE UNIQUE INDEX IF NOT EXISTS anchor ON searchIndex (name, type, path);"
+}
+
+get_title() {
+	FILE="$1"
+
+	pup -p -f "$FILE" 'title text{}' | \
+		tr -d \\n | \
+		sed 's/(GNU make)//g' | \
+		sed 's/\"/\"\"/g'
+}
+
+get_type() {
+	LINK="$(echo "$1" | sed 's/#[^#]*$//')"
+	LINK_TITLE="$(get_title "$2/$LINK")"
+
+	set -- $POSSIBLE_TYPES
+
+	while [ -n "$1" ]; do
+		if echo "$LINK_TITLE" | grep -iq "$1"; then
+			echo "$1"
+			return
+		fi
+		shift
+	done
+
+	echo "Entry"
+}
+
+insert() {
+	NAME="$1"
+	TYPE="$2"
+	PAGE_PATH="$3"
+
+	sqlite3 "$DB_PATH" "INSERT INTO searchIndex(name, type, path) VALUES (\"$NAME\",\"$TYPE\",\"$PAGE_PATH\");"
+}
+
+insert_index_terms() {
+	# Get each term from an index page and insert
+	while [ -n "$1" ]; do
+		grep -Eo "<a href.*></a>:" "$1" | while read -r line; do
+			insert_term "$line" "$(dirname "$1")"
+		done
+
+		shift
+	done
+}
+
+
+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="Guide"
+			insert "$PAGE_NAME" "$PAGE_TYPE" "$(basename "$1")"
+		fi
+		shift
+	done
+}
+
+insert_term() {
+	LINK="$1"
+	PAGE_DIR="$2"
+
+	NAME="$(echo "$LINK" | pup -p 'a text{}' | tr -d \\n | sed 's/"/\"\"/g')"
+	TYPE="$INDEX_TYPE"
+	PAGE_PATH="$(echo "$LINK" | pup -p 'a attr{href}')"
+	if [ -n "$POSSIBLE_TYPES" ]; then
+		TYPE="$(get_type "$PAGE_PATH" "$PAGE_DIR")"
+	elif [ -z "$TYPE" ]; then
+		TYPE="Entry"
+	fi
+
+	insert "$NAME" "$TYPE" "$PAGE_PATH"
+}
+
+TYPE="PAGES"
+
+# Check flags
+while true; do
+	case "$1" in
+		-c|--check)
+			# List of space-separated possible index entry types (overwrites -t)
+			shift
+			POSSIBLE_TYPES="$1"
+			shift
+			;;
+		-i|--index)
+			# Set the script to handle index pages
+			TYPE="INDEX"
+			shift
+			;;
+		-t|--type)
+			# Set a type for index entries
+			shift
+			INDEX_TYPE="$1"
+			shift
+			;;
+		*)
+			break
+	esac
+done
+
+DB_PATH="$1"
+shift
+
+create_table
+case "$TYPE" in
+	PAGES)
+		insert_pages "$@"
+		;;
+	INDEX)
+		insert_index_terms "$@"
+		;;
+esac