From 3c530aafe415db3515465e8b0662373065afd925 Mon Sep 17 00:00:00 2001
From: Louie Shprung <lshprung@scu.edu>
Date: Sun, 26 Mar 2023 13:58:50 -0700
Subject: Index terms

---
 src/index-terms.sh | 38 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 38 insertions(+)
 create mode 100755 src/index-terms.sh

(limited to 'src/index-terms.sh')

diff --git a/src/index-terms.sh b/src/index-terms.sh
new file mode 100755
index 0000000..8ac3163
--- /dev/null
+++ b/src/index-terms.sh
@@ -0,0 +1,38 @@
+#!/usr/bin/env sh
+
+# shellcheck source=./lib/create_table
+. "$(dirname "$0")"/lib/create_table
+# shellcheck source=./lib/insert
+. "$(dirname "$0")"/lib/insert
+
+TYPE="$1"
+shift
+DB_PATH="$1"
+shift
+
+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"
+		done
+
+		shift
+	done
+}
+
+insert_term() {
+	LINK="$1"
+	NAME="$(echo "$LINK" | pup -p 'a text{}' | sed 's/\"\"//g' | tr -d \\n)"
+	PAGE_PATH="$(echo "$LINK" | pup -p 'a attr{href}')"
+
+	# DEBUG
+	#echo "$NAME"
+	#echo "$PAGE_PATH"
+	#echo 
+
+	insert "$DB_PATH" "$NAME" "$TYPE" "$PAGE_PATH"
+}
+
+create_table "$DB_PATH"
+insert_index_terms "$@"
-- 
cgit