From b21d0424ad6bf3cec141abe7e1bb624744c12fc6 Mon Sep 17 00:00:00 2001
From: Louie S <louie@example.com>
Date: Wed, 18 Oct 2023 18:08:59 -0400
Subject: Rewrite to base off template; index more terms

---
 src/index.sh         | 95 ----------------------------------------------------
 src/index_pages.sh   | 42 +++++++++++++++++++++++
 src/index_terms.sh   | 39 +++++++++++++++++++++
 src/lib/create_table |  7 ++++
 src/lib/get_type     | 13 +++++++
 src/lib/insert       |  8 +++++
 6 files changed, 109 insertions(+), 95 deletions(-)
 delete mode 100755 src/index.sh
 create mode 100755 src/index_pages.sh
 create mode 100755 src/index_terms.sh
 create mode 100644 src/lib/create_table
 create mode 100644 src/lib/get_type
 create mode 100644 src/lib/insert

(limited to 'src')

diff --git a/src/index.sh b/src/index.sh
deleted file mode 100755
index a2a52d0..0000000
--- a/src/index.sh
+++ /dev/null
@@ -1,95 +0,0 @@
-#!/usr/bin/env sh
-
-DB_PATH="$1"
-shift
-
-get_title() {
-	FILE="$1"
-
-	PATTERN="<[tT][iI][tT][lL][eE]>.*</[tT][iI][tT][lL][eE]>"
-
-	#Find pattern in file
-	grep -Eo "$PATTERN" "$FILE" | 
-		#Remove tag
-		sed 's/<[^>]*>//g' | \
-		#Remove '(automake)'
-		sed 's/(Autoconf)//g' | \
-		#Remove trailing space
-		sed 's/[ ]*$//g' | \
-		#Remove trailing man categories
-		sed 's/ [0-9][mx]\?$//g' | \
-		#Replace '&amp' with '&'
-		sed 's/&amp/&/g' | \
-		# ReplACE '&ndash;' with '-'
-		sed 's/&ndash;/-/g' | \
-		#Replace '&lt;' with '<'
-		sed 's/&lt;/</g'
-}
-
-get_type() {
-	FILE="$1"
-	CATEGORY="$(echo "$FILE" | grep -Eo "\.[0-9].?\.html$")"
-
-	if [ -n "$CATEGORY" ]; then
-		case "$CATEGORY" in
-			.1*) echo "Command"  ;;
-			.2*) echo "Service"  ;;
-			.3*) echo "Function" ;;
-			*)  echo "Object"   ;;
-		esac
-	fi
-}
-
-insert() {
-	NAME="$1"
-	TYPE="$2"
-	PAGE_PATH="$3"
-
-	sqlite3 "$DB_PATH" "INSERT INTO searchIndex(name, type, path) VALUES (\"$NAME\",\"$TYPE\",\"$PAGE_PATH\");"
-}
-
-skip_check() {
-	NAME="$1"
-
-	case "$NAME" in
-		[A-Z])               return 1 ;;
-		"Source Browser")    return 1 ;;
-		terminal_interface*) return 1 ;;
-	esac
-}
-
-# 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 "$1")"
-			if skip_check "$PAGE_NAME"; then
-				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 "$1" | sed 's/^ncurses.docset\/Contents\/Resources\/Documents\///')"
-				fi
-			fi
-		fi
-
-		shift
-	done
-}
-
-# Create table
-sqlite3 "$DB_PATH" "CREATE TABLE searchIndex(id INTEGER PRIMARY KEY, name TEXT, type TEXT, path TEXT);"
-sqlite3 "$DB_PATH" "CREATE UNIQUE INDEX anchor ON searchIndex (name, type, path);"
-
-main "$@"
diff --git a/src/index_pages.sh b/src/index_pages.sh
new file mode 100755
index 0000000..030bda5
--- /dev/null
+++ b/src/index_pages.sh
@@ -0,0 +1,42 @@
+#!/usr/bin/env sh
+
+# shellcheck source=./lib/create_table
+. "$(dirname "$0")"/lib/create_table
+# shellcheck source=./lib/insert
+. "$(dirname "$0")"/lib/insert
+# shellcheck source=./lib/get_type
+. "$(dirname "$0")"/lib/get_type
+
+DB_PATH="$1"
+shift
+
+get_title() {
+	FILE="$1"
+
+	pup -p -f "$FILE" 'title text{}' | \
+		tr -d \\n | \
+		#Remove trailing man categories
+		sed 's/ [0-9][mx]\?$//g' | \
+		sed 's/\"/\"\"/g'
+}
+
+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 "$1")"
+			if [ -z "$PAGE_TYPE" ]; then
+				PAGE_TYPE="Guide"
+			fi
+			insert "$DB_PATH" "$PAGE_NAME" "$PAGE_TYPE" "$(echo "$1" | sed 's/^ncurses.docset\/Contents\/Resources\/Documents\///')"
+		fi
+		shift
+	done
+}
+
+create_table "$DB_PATH"
+insert_pages "$@"
diff --git a/src/index_terms.sh b/src/index_terms.sh
new file mode 100755
index 0000000..a2b39fb
--- /dev/null
+++ b/src/index_terms.sh
@@ -0,0 +1,39 @@
+#!/usr/bin/env sh
+
+# shellcheck source=./lib/create_table
+. "$(dirname "$0")"/lib/create_table
+# shellcheck source=./lib/insert
+. "$(dirname "$0")"/lib/insert
+# shellcheck source=./lib/get_type
+. "$(dirname "$0")"/lib/get_type
+
+DB_PATH="$1"
+shift
+
+insert_index_terms() {
+	# Get each term from an index page and insert
+	while [ -n "$1" ]; do
+		grep -Eoi "^[a-zA-Z_ ]{45}<STRONG><A HREF=\"[^\"]*\">[^<]*</A></STRONG>" "$1" | while read -r line; do
+			insert_term "$line"
+		done
+
+		shift
+	done
+}
+
+insert_term() {
+	LINK="$1"
+	#NAME="$(echo "$LINK" | pup -p 'a text{}' | sed 's/\"\"//g' | tr -d \\n)"
+	NAME="$(echo "$LINK" | cut -d ' ' -f 1)"
+	PAGE_PATH="$(echo "$LINK" | pup -p 'a attr{href}')"
+
+	TYPE=$(get_type "$PAGE_PATH")
+	if [ -z "$TYPE" ]; then
+		TYPE="Guide"
+	fi
+
+	insert "$DB_PATH" "$NAME" "$TYPE" "man/$PAGE_PATH"
+}
+
+create_table "$DB_PATH"
+insert_index_terms "$@"
diff --git a/src/lib/create_table b/src/lib/create_table
new file mode 100644
index 0000000..a783c50
--- /dev/null
+++ b/src/lib/create_table
@@ -0,0 +1,7 @@
+create_table() {
+	DB_PATH="$1"
+
+	sqlite3 "$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);"
+}
+
diff --git a/src/lib/get_type b/src/lib/get_type
new file mode 100644
index 0000000..9f04b53
--- /dev/null
+++ b/src/lib/get_type
@@ -0,0 +1,13 @@
+get_type() {
+	FILE="$1"
+	CATEGORY="$(echo "$FILE" | grep -Eo "\.[0-9].?\.html$")"
+
+	if [ -n "$CATEGORY" ]; then
+		case "$CATEGORY" in
+			.1*) echo "Command"  ;;
+			.2*) echo "Service"  ;;
+			.3*) echo "Function" ;;
+			*)  echo "Object"   ;;
+		esac
+	fi
+}
diff --git a/src/lib/insert b/src/lib/insert
new file mode 100644
index 0000000..31c1b4c
--- /dev/null
+++ b/src/lib/insert
@@ -0,0 +1,8 @@
+insert() {
+	DB_PATH="$1"
+	NAME="$2"
+	TYPE="$3"
+	PAGE_PATH="$4"
+
+	sqlite3 "$DB_PATH" "INSERT INTO searchIndex(name, type, path) VALUES (\"$NAME\",\"$TYPE\",\"$PAGE_PATH\");"
+}
-- 
cgit