support wildcard urls in settings

5 files changed, 142 insertions, 7 deletions

diff --git a/TODOS.org b/TODOS.org
index b389d06..bf449c9 100644
--- a/TODOS.org
+++ b/TODOS.org
@@ -4,8 +4,7 @@ TODO:
   or indirectly in a bag should be included multiple times or once
 - make it possible to provide backup urls for remote scripts
 - make it possible to cache remote scripts
-- make it possible to use wildcards or something similar to be able to assign a script set to -- CRUCIAL
-  a set of domains or to a set of possible queries at a url
+- optimize url querying
 - make it possible to automatically download page's served scripts and save them (of course, this by itself -- CRUCIAL
   would give little benefit, but it will make it easy to modify this set of scripts - useful, if some of
   those scripts are already free, as is often the case)
@@ -42,6 +41,7 @@ TODO:
 - perform never-ending refactoring of already-written code
 
 DONE:
+- make it possible to use wildcard urls in settings -- DONE 2021-05-14
 - port to gecko-based browsers -- DONE 2021-05-13
 - find a way to additionally block all other scripts using CSP -- DONE 2021-05-13
 - only allow a single injection payload for page -- DONE 2021-05-13
diff --git a/background/background.html b/background/background.html
index c6621e2..aac8ec8 100644
--- a/background/background.html
+++ b/background/background.html
@@ -12,6 +12,7 @@
     <script src="./storage_server.js"></script>
     <script src="/common/url_item.js"></script>
     <script src="/common/sha256.js"></script>
+    <script src="./settings_query.js"></script>
     <script src="./page_actions_server.js"></script>
     <script src="/common/gen_unique.js"></script>
     <script src="./policy_smuggler.js"></script>
diff --git a/background/page_actions_server.js b/background/page_actions_server.js
index 2dfcf9a..f96b659 100644
--- a/background/page_actions_server.js
+++ b/background/page_actions_server.js
@@ -16,15 +16,16 @@
     const CONNECTION_TYPE = window.CONNECTION_TYPE;
     const browser = window.browser;
     const listen_for_connection = window.listen_for_connection;
-    const url_item = window.url_item;
     const sha256 = window.sha256;
+    const get_query_best = window.get_query_best;
 
     var storage;
+    var query_best;
     var handler;
 
     function send_scripts(url, port)
     {
-	let settings = storage.get(TYPE_PREFIX.PAGE, url_item(url));
+	let [pattern, settings] = query_best(url);
 	if (settings === undefined)
 	    return;
 
@@ -142,6 +143,7 @@
     async function start()
     {
 	storage = await get_storage();
+	query_best = await get_query_best();
 
 	listen_for_connection(CONNECTION_TYPE.PAGE_ACTIONS, new_connection);
     }
diff --git a/background/policy_smuggler.js b/background/policy_smuggler.js
index 180dcb7..ad8d565 100644
--- a/background/policy_smuggler.js
+++ b/background/policy_smuggler.js
@@ -16,8 +16,10 @@
     const browser = window.browser;
     const url_item = window.url_item;
     const gen_unique = window.gen_unique;
+    const get_query_best = window.get_query_best;
 
     var storage;
+    var query_best;
 
     function redirect(request)
     {
@@ -35,12 +37,11 @@
 	    return {cancel : false};
 	}
 
-	let settings = storage.get(TYPE_PREFIX.PAGE, url);
-	console.log("got", storage.get(TYPE_PREFIX.PAGE, url), "for", url);
+	let [pattern, settings] = query_best(url);
 	if (settings === undefined || !settings.allow)
 	    return {cancel : false};
 
-	second_target = (first_target || "") + (second_target || "")
+	second_target = (first_target || "") + (second_target || "");
 
 	console.log(["redirecting", request.url,
 		     (base_url + unique + second_target)]);
@@ -52,6 +53,7 @@
 
     async function start() {
 	storage = await get_storage();
+	query_best = await get_query_best();
 
 	chrome.webRequest.onBeforeRequest.addListener(
 	    redirect,
diff --git a/background/settings_query.js b/background/settings_query.js
new file mode 100644
index 0000000..9101913
--- /dev/null
+++ b/background/settings_query.js
@@ -0,0 +1,130 @@
+/**
+* Myext querying page settings with regard to wildcard records
+*
+* Copyright (C) 2021 Wojtek Kosior
+*
+* Dual-licensed under:
+*   - 0BSD license
+*   - GPLv3 or (at your option) any later version
+*/
+
+"use strict";
+
+(() => {
+    const make_once = window.make_once;
+    const get_storage = window.get_storage;
+
+    var storage;
+
+    var exports = {};
+
+    async function init(fun)
+    {
+	storage = await get_storage();
+
+	return fun;
+    }
+
+    // TODO: also support urls with specified ports as well as `data:' urls
+    function query(url, multiple)
+    {
+	let proto_re = "[a-zA-Z]*:\/\/";
+	let domain_re = "[^/?#]+";
+	let segments_re = "/[^?#]*";
+	let query_re = "\\?[^#]*";
+
+	let url_regex = new RegExp(`\
+^\
+(${proto_re})\
+(${domain_re})\
+(${segments_re})?\
+(${query_re})?\
+#?.*\$\
+`);
+
+	let regex_match = url_regex.exec(url);
+	if (regex_match === null) {
+	    console.log("bad url format", url);
+	    return multiple ? [] : [undefined, undefined];
+	}
+
+	let [_, proto, domain, segments, query] = regex_match;
+
+	domain = domain.split(".");
+	let segments_trailing_dash =
+	    segments && segments[segments.length - 1] === "/";
+	segments = (segments || "").split("/").filter(s => s !== "");
+	segments.unshift("");
+
+	let matched = [];
+
+	for (let d_slice = 0; d_slice < domain.length; d_slice++) {
+	    let domain_part = domain.slice(d_slice).join(".");
+	    let domain_wildcards = [];
+	    if (d_slice === 0)
+		domain_wildcards.push("");
+	    if (d_slice === 1)
+		domain_wildcards.push("*.");
+	    if (d_slice > 0)
+		domain_wildcards.push("**.");
+	    domain_wildcards.push("***.");
+
+	    for (let domain_wildcard of domain_wildcards) {
+		let domain_pattern = domain_wildcard + domain_part;
+
+		for (let s_slice = segments.length; s_slice > 0; s_slice--) {
+		    let segments_part = segments.slice(0, s_slice).join("/");
+		    let segments_wildcards = [];
+		    if (s_slice === segments.length) {
+			if (segments_trailing_dash)
+			    segments_wildcards.push("/");
+			segments_wildcards.push("");
+		    }
+		    if (s_slice === segments.length - 1) {
+			if (segments[s_slice] !== "*")
+			    segments_wildcards.push("/*");
+		    }
+		    if (s_slice < segments.length &&
+			(segments[s_slice] !== "**" ||
+			 s_slice < segments.length - 1))
+			segments_wildcards.push("/**");
+		    if (segments[s_slice] !== "***" ||
+			s_slice < segments.length)
+			segments_wildcards.push("/***");
+
+		    for (let segments_wildcard of segments_wildcards) {
+			let segments_pattern =
+			    segments_part + segments_wildcard;
+
+			let pattern = proto + domain_pattern + segments_pattern;
+			console.log("trying", pattern);
+			let settings = storage.get(TYPE_PREFIX.PAGE, pattern);
+
+			if (settings === undefined)
+			    continue;
+
+			if (!multiple)
+			    return [pattern, settings];
+
+			matched.push([pattern, settings]);
+		    }
+		}
+	    }
+	}
+
+	return multiple ?  matched : [undefined, undefined];
+    }
+
+    function query_best(url)
+    {
+	return query(url, false);
+    }
+
+    function query_all(url)
+    {
+	return query(url, true);
+    }
+
+    window.get_query_best = make_once(() => init(query_best));
+    window.get_query_all = make_once(() => init(query_all));
+})();