From 20aa42ad2c7740d64247e900e02638317b650209 Mon Sep 17 00:00:00 2001
From: Louie Shprung <lshprung@scu.edu>
Date: Tue, 22 Nov 2022 21:45:29 -0800
Subject: Rewrite in POSIX shell

---
 src/index.rb | 40 ------------------------------------
 src/index.sh | 66 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 66 insertions(+), 40 deletions(-)
 delete mode 100755 src/index.rb
 create mode 100755 src/index.sh

(limited to 'src')

diff --git a/src/index.rb b/src/index.rb
deleted file mode 100755
index 81b042a..0000000
--- a/src/index.rb
+++ /dev/null
@@ -1,40 +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>(.*)\(Autoconf\)(.*)</title>]
-
-BUILTIN_PATTERN = /The node you are looking for is at.*Limitations-of-.*\.html/
-MACRO_PATTERN = /The node you are looking for is at/
-
-def quote(s)
-  s.gsub(/&amp;/, '&').gsub(/'/, "\\'").gsub(/&lt;/, '<')
-end
-
-ARGV.each do |arg|
-  Pathname.glob(arg) do |path|
-    macro_match = path.each_line.lazy.map { |line| MACRO_PATTERN.match(line) }.find { |m| m }
-    builtin_match = path.each_line.lazy.map { |line| BUILTIN_PATTERN.match(line) }.find { |m| m }
-    if builtin_match
-      type = "Builtin"
-    elsif macro_match
-      type = "Macro"
-    else
-      type = "Guide"
-    end
-
-    match = path.each_line.lazy.map { |line| PATTERN.match(line) }.find { |m| m }
-    if match
-      printf INSERT_SQL, quote(match[1]), type, path.basename
-    else
-      $stderr.puts "%{path.basename}: no title found"
-    end
-  end
-end
diff --git a/src/index.sh b/src/index.sh
new file mode 100755
index 0000000..e81a0f7
--- /dev/null
+++ b/src/index.sh
@@ -0,0 +1,66 @@
+#!/usr/bin/env sh
+
+DB_PATH="$1"
+shift
+
+get_title() {
+	FILE="$1"
+
+	PATTERN="<title>.*\(Autoconf\).*</title>"
+
+	#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' | \
+		#Replace '&amp' with '&'
+		sed 's/&amp/&/g' | \
+		#Replace '&lt;' with '<'
+		sed 's/&lt;/</g'
+}
+
+get_type() {
+	FILE="$1"
+	PATTERN="The node you are looking for is at.*Limitations-of-.*\.html;Builtin
+	The node you are looking for is at;Macro"
+
+	echo "$PATTERN" | while read -r line; do
+		#echo "$line"
+		if grep -Eq "$(echo "$line" | cut -d ';' -f 1)" "$FILE"; then
+			echo "$line" | cut -d ';' -f 2
+			break
+		fi
+	done
+}
+
+insert() {
+	NAME="$1"
+	TYPE="$2"
+	PAGE_PATH="$3"
+
+	sqlite3 "$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"
+		fi
+		#echo "$PAGE_TYPE"
+		insert "$PAGE_NAME" "$PAGE_TYPE" "$(basename "$1")"
+	fi
+	shift
+done
-- 
cgit