From 03699605e4ba3443ff3f1e34d2be9e70e152d42a Mon Sep 17 00:00:00 2001
From: Louie Shprung <lshprung@tutanota.com>
Date: Thu, 20 Jun 2024 17:26:12 -0400
Subject: Move index-terms.sh to shared scripts for GNU docsets

---
 src/scripts/gnu/index-terms.sh | 33 +++++++++++++++++++++++++++++++++
 1 file changed, 33 insertions(+)
 create mode 100755 src/scripts/gnu/index-terms.sh

(limited to 'src/scripts')

diff --git a/src/scripts/gnu/index-terms.sh b/src/scripts/gnu/index-terms.sh
new file mode 100755
index 0000000..0eaffa7
--- /dev/null
+++ b/src/scripts/gnu/index-terms.sh
@@ -0,0 +1,33 @@
+#!/usr/bin/env sh
+
+# shellcheck source=../create_table.sh
+. "$(dirname "$0")"/../create_table.sh
+# shellcheck source=../insert.sh
+. "$(dirname "$0")"/../insert.sh
+
+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 "$LINK" | pup -p 'a attr{href}')"
+
+	insert "$DB_PATH" "$NAME" "$TYPE" "$PAGE_PATH"
+}
+
+create_table "$DB_PATH"
+insert_index_terms "$@"
-- 
cgit