From 5c583de820c0d5f666a830ca1e8205fe7d55e61e Mon Sep 17 00:00:00 2001
From: Wojtek Kosior <koszko@koszko.org>
Date: Wed, 1 Dec 2021 21:08:03 +0100
Subject: start implementing more efficient querying of URL patterns

---
 common/patterns.js | 29 ++++++++++++++++++-----------
 1 file changed, 18 insertions(+), 11 deletions(-)

(limited to 'common/patterns.js')

diff --git a/common/patterns.js b/common/patterns.js
index 635b128..054e610 100644
--- a/common/patterns.js
+++ b/common/patterns.js
@@ -17,16 +17,25 @@ const MAX = {
 const proto_regex = /^(\w+):\/\/(.*)$/;
 
 const user_re = "[^/?#@]+@"
-const domain_re = "[.a-zA-Z0-9-]+";
+const domain_re = "[.*a-zA-Z0-9-]+";
 const path_re = "[^?#]*";
 const query_re = "\\??[^#]*";
 
 const http_regex = new RegExp(`^(${domain_re})(${path_re})(${query_re}).*`);
 
-const file_regex = new RegExp(`^(${path_re}).*`);
+const file_regex = new RegExp(`^(/${path_re}).*`);
 
 const ftp_regex = new RegExp(`^(${user_re})?(${domain_re})(${path_re}).*`);
 
+function match_or_throw(regex, string, error_msg)
+{
+    const match = regex.exec(string);
+    if (match === null)
+	throw error_msg;
+
+    return match;
+}
+
 function deconstruct_url(url, use_limits=true)
 {
     const max = MAX;
@@ -35,21 +44,19 @@ function deconstruct_url(url, use_limits=true)
 	    max[key] = Infinity;
     }
 
-    const proto_match = proto_regex.exec(url);
-    if (proto_match === null)
-	throw `bad url '${url}'`;
+    const matcher = (re, str) => match_or_throw(re, str, `bad url '${url}'`)
 
+    const proto_match = matcher(proto_regex, url);
     const deco = {proto: proto_match[1]};
 
     if (deco.proto === "file") {
-	deco.path = file_regex.exec(proto_match[2])[1];
+	deco.path = matcher(file_regex, proto_match[2])[1];
     } else if (deco.proto === "ftp") {
-	[deco.domain, deco.path] = ftp_regex.exec(proto_match[2]).slice(2, 4);
+	[deco.domain, deco.path] =
+	    matcher(ftp_regex, proto_match[2]).slice(2, 4);
     } else if (deco.proto === "http" || deco.proto === "https") {
-	const http_match = http_regex.exec(proto_match[2]);
-	if (!http_match)
-	    return undefined;
-	[deco.domain, deco.path, deco.query] = http_match.slice(1, 4);
+	[deco.domain, deco.path, deco.query] =
+	    matcher(http_regex, proto_match[2]).slice(1, 4);
 	deco.domain = deco.domain.toLowerCase();
     } else {
 	throw `unsupported protocol in url '${url}'`;
-- 
cgit v1.2.3