From 69e53743393664ed2db59bbe9bbeaf6f124754f3 Mon Sep 17 00:00:00 2001
From: Wojtek Kosior <koszko@koszko.org>
Date: Wed, 24 Nov 2021 15:53:00 +0100
Subject: implement more efficient querying of URL patterns

---
 common/patterns.js | 47 ++++++++++++++++++++++++++++-------------------
 1 file changed, 28 insertions(+), 19 deletions(-)

(limited to 'common/patterns.js')

diff --git a/common/patterns.js b/common/patterns.js
index 625be05..790ce1b 100644
--- a/common/patterns.js
+++ b/common/patterns.js
@@ -7,10 +7,12 @@
  * Redistribution terms are gathered in the `copyright' file.
  */
 
-const MAX_URL_PATH_LEN = 12;
-const MAX_URL_PATH_CHARS = 255;
-const MAX_DOMAIN_LEN = 7;
-const MAX_DOMAIN_CHARS = 100;
+const MAX = {
+    URL_PATH_LEN:   12,
+    URL_PATH_CHARS: 255,
+    DOMAIN_LEN:     7,
+    DOMAIN_CHARS:   100
+};
 
 const proto_regex = /^(\w+):\/\/(.*)$/;
 
@@ -25,11 +27,17 @@ const file_regex = new RegExp(`^(${path_re}).*`);
 
 const ftp_regex = new RegExp(`^(${user_re})?(${domain_re})(${path_re}).*`);
 
-function deconstruct_url(url)
+function deconstruct_url(url, use_limits=true)
 {
+    const max = MAX;
+    if (!use_limits) {
+	for (key in MAX)
+	    max[key] = Infinity;
+    }
+
     const proto_match = proto_regex.exec(url);
     if (proto_match === null)
-	return undefined;
+	throw `bad url '${url}'`;
 
     const deco = {proto: proto_match[1]};
 
@@ -37,20 +45,21 @@ function deconstruct_url(url)
 	deco.path = file_regex.exec(proto_match[2])[1];
     } else if (deco.proto === "ftp") {
 	[deco.domain, deco.path] = ftp_regex.exec(proto_match[2]).slice(2, 4);
-    } else {
+    } 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);
+    } else {
+	throw `unsupported protocol in url '${url}'`;
     }
 
-    const leading_dash = deco.path[0] === "/";
     deco.trailing_dash = deco.path[deco.path.length - 1] === "/";
 
     if (deco.domain) {
-	if (deco.domain.length > MAX_DOMAIN_CHARS) {
+	if (deco.domain.length > max.DOMAIN_CHARS) {
 	    const idx = deco.domain.indexOf(".", deco.domain.length -
-					    MAX_DOMAIN_CHARS);
+					    max.DOMAIN_CHARS);
 	    if (idx === -1)
 		deco.domain = [];
 	    else
@@ -59,7 +68,7 @@ function deconstruct_url(url)
 	    deco.domain_truncated = true;
 	}
 
-	if (deco.path.length > MAX_URL_PATH_CHARS) {
+	if (deco.path.length > max.URL_PATH_CHARS) {
 	    deco.path = deco.path.substring(0, deco.path.lastIndexOf("/"));
 	    deco.path_truncated = true;
 	}
@@ -67,16 +76,14 @@ function deconstruct_url(url)
 
     if (typeof deco.domain === "string") {
 	deco.domain = deco.domain.split(".");
-	if (deco.domain.splice(0, deco.domain.length - MAX_DOMAIN_LEN).length
+	if (deco.domain.splice(0, deco.domain.length - max.DOMAIN_LEN).length
 	    > 0)
 	    deco.domain_truncated = true;
     }
 
     deco.path = deco.path.split("/").filter(s => s !== "");
-    if (deco.domain && deco.path.splice(MAX_URL_PATH_LEN).length > 0)
+    if (deco.domain && deco.path.splice(max.URL_PATH_LEN).length > 0)
 	deco.path_truncated = true;
-    if (leading_dash || deco.path.length === 0)
-	deco.path.unshift("");
 
     return deco;
 }
@@ -98,13 +105,14 @@ function* each_domain_pattern(deco)
 
 function* each_path_pattern(deco)
 {
-    for (let slice = deco.path.length; slice > 0; slice--) {
-	const path_part = deco.path.slice(0, slice).join("/");
+    for (let slice = deco.path.length; slice >= 0; slice--) {
+	const path_part = ["", ...deco.path.slice(0, slice)].join("/");
 	const path_wildcards = [];
 	if (slice === deco.path.length && !deco.path_truncated) {
-	    if (deco.trailing_dash)
+	    if (deco.trailing_dash && path_part !== )
 		yield path_part + "/";
-	    yield path_part;
+	    if (part_part !== "" || deco.proto !== "file")
+		yield path_part;
 	}
 	if (slice === deco.path.length - 1 && !deco.path_truncated &&
 	    deco.path[slice] !== "*")
@@ -137,5 +145,6 @@ function* each_url_pattern(url)
 /*
  * EXPORTS_START
  * EXPORT each_url_pattern
+ * EXPORT deconstruct_url
  * EXPORTS_END
  */
-- 
cgit v1.2.3