Merge branch 'master' into jahoti-update

15 files changed, 114 insertions, 193 deletions

diff --git a/common/ajax.js b/common/ajax.js
index 8082bbe..7269a8a 100644
--- a/common/ajax.js
+++ b/common/ajax.js
@@ -1,6 +1,7 @@
 /**
- * part of Hachette
- * Wrapping XMLHttpRequest into a Promise.
+ * This file is part of Haketilo.
+ *
+ * Function: Wrapping XMLHttpRequest into a Promise.
  *
  * Copyright (C) 2021 Wojtek Kosior
  * Redistribution terms are gathered in the `copyright' file.
diff --git a/common/connection_types.js b/common/connection_types.js
index 88c6964..3e9df56 100644
--- a/common/connection_types.js
+++ b/common/connection_types.js
@@ -1,5 +1,7 @@
 /**
- * Hachette background scripts message connection types "enum"
+ * This file is part of Haketilo.
+ *
+ * Function: Define an "enum" of message connection types.
  *
  * Copyright (C) 2021 Wojtek Kosior
  * Redistribution terms are gathered in the `copyright' file.
diff --git a/common/lock.js b/common/lock.js
index 822ad1b..6cf0835 100644
--- a/common/lock.js
+++ b/common/lock.js
@@ -1,5 +1,7 @@
 /**
- * Hachette lock (aka binary semaphore aka mutex)
+ * This file is part of Haketilo.
+ *
+ * Function: Implement a lock (aka binary semaphore aka mutex).
  *
  * Copyright (C) 2021 Wojtek Kosior
  * Redistribution terms are gathered in the `copyright' file.
diff --git a/common/message_server.js b/common/message_server.js
index ea40487..c8c6696 100644
--- a/common/message_server.js
+++ b/common/message_server.js
@@ -1,5 +1,7 @@
 /**
- * Hachette message server
+ * This file is part of Haketilo.
+ *
+ * Function: Message server.
  *
  * Copyright (C) 2021 Wojtek Kosior
  * Redistribution terms are gathered in the `copyright' file.
diff --git a/common/misc.js b/common/misc.js
index 97fc2dc..9ffb7ff 100644
--- a/common/misc.js
+++ b/common/misc.js
@@ -1,5 +1,7 @@
 /**
- * Hachette miscellaneous operations refactored to a separate file
+ * This file is part of Haketilo.
+ *
+ * Function: Miscellaneous operations refactored to a separate file.
  *
  * Copyright (C) 2021 Wojtek Kosior
  * Copyright (C) 2021 jahoti
@@ -36,36 +38,26 @@ function Uint8toHex(data)
     return returnValue;
 }
 
-function gen_nonce(length) // Default 16
+function gen_nonce(length=16)
 {
-    let randomData = new Uint8Array(length || 16);
+    let randomData = new Uint8Array(length);
     crypto.getRandomValues(randomData);
     return Uint8toHex(randomData);
 }
 
-/* csp rule that blocks all scripts except for those injected by us */
-function csp_rule(nonce)
+/* CSP rule that blocks scripts according to policy's needs. */
+function make_csp_rule(policy)
 {
-    const rule = `'nonce-${nonce}'`;
-    return `script-src ${rule}; script-src-elem ${rule}; script-src-attr 'none'; prefetch-src 'none';`;
+    let rule = "prefetch-src 'none'; script-src-attr 'none';";
+    const script_src = policy.has_payload ?
+	  `'nonce-${policy.nonce}'` : "'none'";
+    rule += ` script-src ${script_src}; script-src-elem ${script_src};`;
+    return rule;
 }
 
 /* Check if some HTTP header might define CSP rules. */
-const csp_header_names = new Set([
-    "content-security-policy",
-    "x-webkit-csp",
-    "x-content-security-policy"
-]);
-
-const report_only_header_name = "content-security-policy-report-only";
-
-function is_csp_header_name(string, include_report_only)
-{
-    string = string && string.toLowerCase().trim() || "";
-
-    return (include_report_only && string === report_only_header_name) ||
-	csp_header_names.has(string);
-}
+const csp_header_regex =
+      /^\s*(content-security-policy|x-webkit-csp|x-content-security-policy)/i;
 
 /*
  * Print item together with type, e.g.
@@ -111,52 +103,6 @@ function parse_csp(csp) {
     return directives;
 }
 
-/* Make CSP headers do our bidding, not interfere */
-function sanitize_csp_header(header, policy)
-{
-    const rule = `'nonce-${policy.nonce}'`;
-    const csp = parse_csp(header.value);
-
-    if (!policy.allow) {
-	/* No snitching */
-	delete csp['report-to'];
-	delete csp['report-uri'];
-
-	delete csp['script-src'];
-	delete csp['script-src-elem'];
-
-	csp['script-src-attr'] = ["'none'"];
-	csp['prefetch-src'] = ["'none'"];
-    }
-
-    if ('script-src' in csp)
-	csp['script-src'].push(rule);
-    else
-	csp['script-src'] = [rule];
-
-    if ('script-src-elem' in csp)
-	csp['script-src-elem'].push(rule);
-    else
-	csp['script-src-elem'] = [rule];
-
-    const new_csp = Object.entries(csp).map(
-	i => `${i[0]} ${i[1].join(' ')};`
-    );
-
-    return {name: header.name, value: new_csp.join('')};
-}
-
-/* csp rule that blocks all scripts except for those injected by us */
-function make_csp_rule(policy)
-{
-    let rule = "prefetch-src 'none'; ", nonce = `'nonce-${policy.nonce}'`;
-    if (!policy.allow) {
-	rule += `script-src ${nonce}; script-src-elem ${nonce}; ` +
-	    "script-src-attr 'none'; ";
-    }
-    return rule;
-}
-
 /* Regexes and objects to use as/in schemas for parse_json_with_schema(). */
 const nonempty_string_matcher = /.+/;
 
@@ -173,11 +119,10 @@ const matchers = {
  * EXPORTS_START
  * EXPORT gen_nonce
  * EXPORT make_csp_rule
- * EXPORT is_csp_header_name
+ * EXPORT csp_header_regex
  * EXPORT nice_name
  * EXPORT open_in_settings
  * EXPORT is_privileged_url
- * EXPORT sanitize_csp_header
  * EXPORT matchers
  * EXPORTS_END
  */
diff --git a/common/observable.js b/common/observable.js
index 02f1c1b..ab3b444 100644
--- a/common/observable.js
+++ b/common/observable.js
@@ -1,6 +1,7 @@
 /**
- * part of Hachette
- * Facilitate listening to events
+ * This file is part of Haketilo.
+ *
+ * Function: Facilitate listening to (internal, self-generated) events.
  *
  * Copyright (C) 2021 Wojtek Kosior
  * Redistribution terms are gathered in the `copyright' file.
diff --git a/common/once.js b/common/once.js
index 098b43f..93e842f 100644
--- a/common/once.js
+++ b/common/once.js
@@ -1,5 +1,8 @@
 /**
- * Hachette feature initialization promise
+ * This file is part of Haketilo.
+ *
+ * Function: Wrap APIs that depend on some asynchronous initialization into
+ *     promises.
  *
  * Copyright (C) 2021 Wojtek Kosior
  * Redistribution terms are gathered in the `copyright' file.
diff --git a/common/patterns.js b/common/patterns.js
index ebb55ab..625be05 100644
--- a/common/patterns.js
+++ b/common/patterns.js
@@ -1,10 +1,17 @@
 /**
- * Hachette operations on page url patterns
+ * This file is part of Haketilo.
+ *
+ * Function: Operations on page URL patterns.
  *
  * Copyright (C) 2021 Wojtek Kosior
  * 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 proto_regex = /^(\w+):\/\/(.*)$/;
 
 const user_re = "[^/?#@]+@"
@@ -37,103 +44,51 @@ function deconstruct_url(url)
 	[deco.domain, deco.path, deco.query] = http_match.slice(1, 4);
     }
 
-    if (deco.domain)
-	deco.domain = deco.domain.split(".");
-
     const leading_dash = deco.path[0] === "/";
     deco.trailing_dash = deco.path[deco.path.length - 1] === "/";
-    deco.path = deco.path.split("/").filter(s => s !== "");
-    if (leading_dash || deco.path.length === 0)
-	deco.path.unshift("");
 
-    return deco;
-}
+    if (deco.domain) {
+	if (deco.domain.length > MAX_DOMAIN_CHARS) {
+	    const idx = deco.domain.indexOf(".", deco.domain.length -
+					    MAX_DOMAIN_CHARS);
+	    if (idx === -1)
+		deco.domain = [];
+	    else
+		deco.domain = deco.domain.substring(idx + 1);
 
-/* Be sane: both arguments should be arrays of length >= 2 */
-function domain_matches(url_domain, pattern_domain)
-{
-    const length_difference = url_domain.length - pattern_domain.length;
-
-    for (let i = 1; i <= url_domain.length; i++) {
-	const url_part = url_domain[url_domain.length - i];
-	const pattern_part = pattern_domain[pattern_domain.length - i];
-
-	if (pattern_domain.length === i) {
-	    if (pattern_part === "*")
-		return length_difference === 0;
-	    if (pattern_part === "**")
-		return length_difference > 0;
-	    if (pattern_part === "***")
-		return true;
-	    return length_difference === 0 && pattern_part === url_part;
+	    deco.domain_truncated = true;
 	}
 
-	if (pattern_part !== url_part)
-	    return false;
-    }
-
-    return pattern_domain.length === url_domain.length + 1 &&
-	pattern_domain[0] === "***";
-}
-
-function path_matches(url_path, url_trailing_dash,
-		      pattern_path, pattern_trailing_dash)
-{
-    const dashes_ok = !(pattern_trailing_dash && !url_trailing_dash);
-
-    if (pattern_path.length === 0)
-	return url_path.length === 0 && dashes_ok;
-
-    const length_difference = url_path.length - pattern_path.length;
-
-    for (let i = 0; i < url_path.length; i++) {
-	if (pattern_path.length === i + 1) {
-	    if (pattern_path[i] === "*")
-		return length_difference === 0;
-	    if (pattern_path[i] === "**") {
-		return length_difference > 0 ||
-		    (url_path[i] === "**" && dashes_ok);
-	    }
-	    if (pattern_path[i] === "***")
-		return length_difference >= 0;
-	    return length_difference === 0 &&
-		pattern_path[i] === url_path[i] && dashes_ok;
+	if (deco.path.length > MAX_URL_PATH_CHARS) {
+	    deco.path = deco.path.substring(0, deco.path.lastIndexOf("/"));
+	    deco.path_truncated = true;
 	}
-
-	if (pattern_path[i] !== url_path[i])
-	    return false;
     }
 
-    return false;
-}
-
-function url_matches(url, pattern)
-{
-    const url_deco = deconstruct_url(url);
-    const pattern_deco = deconstruct_url(pattern);
-
-    if (url_deco === undefined || pattern_deco === undefined) {
-	console.log(`bad comparison: ${url} and ${pattern}`);
-	return false
+    if (typeof deco.domain === "string") {
+	deco.domain = deco.domain.split(".");
+	if (deco.domain.splice(0, deco.domain.length - MAX_DOMAIN_LEN).length
+	    > 0)
+	    deco.domain_truncated = true;
     }
 
-    return pattern_deco.proto === url_deco.proto &&
-	!(pattern_deco.proto === "file" && pattern_deco.trailing_dash) &&
-	!!url_deco.domain === !!pattern_deco.domain &&
-	(!url_deco.domain ||
-	 domain_matches(url_deco.domain, pattern_deco.domain)) &&
-	path_matches(url_deco.path, url_deco.trailing_dash,
-		     pattern_deco.path, pattern_deco.trailing_dash);
+    deco.path = deco.path.split("/").filter(s => s !== "");
+    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;
 }
 
-function* each_domain_pattern(domain_segments)
+function* each_domain_pattern(deco)
 {
-    for (let slice = 0; slice < domain_segments.length; slice++) {
-	const domain_part = domain_segments.slice(slice).join(".");
+    for (let slice = 0; slice < deco.domain.length - 1; slice++) {
+	const domain_part = deco.domain.slice(slice).join(".");
 	const domain_wildcards = [];
-	if (slice === 0)
+	if (slice === 0 && !deco.domain_truncated)
 	    yield domain_part;
-	if (slice === 1)
+	if (slice === 1 && !deco.domain_truncated)
 	    yield "*." + domain_part;
 	if (slice > 1)
 	    yield "**." + domain_part;
@@ -141,22 +96,23 @@ function* each_domain_pattern(domain_segments)
     }
 }
 
-function* each_path_pattern(path_segments, trailing_dash)
+function* each_path_pattern(deco)
 {
-    for (let slice = path_segments.length; slice > 0; slice--) {
-	const path_part = path_segments.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 === path_segments.length) {
-	    if (trailing_dash)
+	if (slice === deco.path.length && !deco.path_truncated) {
+	    if (deco.trailing_dash)
 		yield path_part + "/";
 	    yield path_part;
 	}
-	if (slice === path_segments.length - 1 && path_segments[slice] !== "*")
+	if (slice === deco.path.length - 1 && !deco.path_truncated &&
+	    deco.path[slice] !== "*")
 	    yield path_part + "/*";
-	if (slice < path_segments.length - 1)
+	if (slice < deco.path.length - 1)
 	    yield path_part + "/**";
-	if (slice < path_segments.length - 1 ||
-	    path_segments[path_segments.length - 1] !== "***")
+	if (slice !== deco.path.length - 1 || deco.path_truncated ||
+	    deco.path[slice] !== "***")
 	    yield path_part + "/***";
     }
 }
@@ -167,20 +123,19 @@ function* each_url_pattern(url)
     const deco = deconstruct_url(url);
 
     if (deco === undefined) {
-	console.log("bad url format", url);
+	console.error("bad url format", url);
 	return false;
     }
 
-    const all_domains = deco.domain ? each_domain_pattern(deco.domain) : [""];
+    const all_domains = deco.domain ? each_domain_pattern(deco) : [""];
     for (const domain of all_domains) {
-	for (const path of each_path_pattern(deco.path, deco.trailing_dash))
+	for (const path of each_path_pattern(deco))
 	    yield `${deco.proto}://${domain}${path}`;
     }
 }
 
 /*
  * EXPORTS_START
- * EXPORT url_matches
  * EXPORT each_url_pattern
  * EXPORTS_END
  */
diff --git a/common/sanitize_JSON.js b/common/sanitize_JSON.js
index 8b86d2d..4cf1ef4 100644
--- a/common/sanitize_JSON.js
+++ b/common/sanitize_JSON.js
@@ -1,6 +1,7 @@
 /**
- * part of Hachette
- * Powerful, full-blown format enforcer for externally-obtained JSON
+ * This file is part of Haketilo.
+ *
+ * Function: Powerful, full-blown format enforcer for externally-obtained JSON.
  *
  * Copyright (C) 2021 Wojtek Kosior
  * Redistribution terms are gathered in the `copyright' file.
diff --git a/common/settings_query.js b/common/settings_query.js
index b54e580..7e1315e 100644
--- a/common/settings_query.js
+++ b/common/settings_query.js
@@ -1,5 +1,7 @@
 /**
- * Hachette querying page settings with regard to wildcard records
+ * This file is part of Haketilo.
+ *
+ * Function: Querying page settings.
  *
  * Copyright (C) 2021 Wojtek Kosior
  * Redistribution terms are gathered in the `copyright' file.
diff --git a/common/signing.js b/common/signing.js
index 2171714..11cd442 100644
--- a/common/signing.js
+++ b/common/signing.js
@@ -1,6 +1,7 @@
 /**
- * part of Hachette
- * Functions related to "signing" of data, refactored to a separate file.
+ * This file is part of Haketilo.
+ *
+ * Functions: Operations related to "signing" of data.
  *
  * Copyright (C) 2021 Wojtek Kosior
  * Redistribution terms are gathered in the `copyright' file.
@@ -10,13 +11,13 @@
  * IMPORTS_START
  * IMPORT sha256
  * IMPORT browser
- * IMPORT is_chrome
+ * IMPORT is_mozilla
  * IMPORTS_END
  */
 
 /*
  * In order to make certain data synchronously accessible in certain contexts,
- * hachette smuggles it in string form in places like cookies, URLs and headers.
+ * Haketilo smuggles it in string form in places like cookies, URLs and headers.
  * When using the smuggled data, we first need to make sure it isn't spoofed.
  * For that, we use this pseudo-signing mechanism.
  *
@@ -30,18 +31,18 @@
  *
  * The secret shared between execution contexts has to be available
  * synchronously. Under Mozilla, this is the extension's per-session id. Under
- * Chromium, this is the key that resides in the manifest.
- *
- * An idea to (under Chromium) instead store the secret in a file fetched
- * synchronously using XMLHttpRequest is being considered.
+ * Chromium, this is a dummy web-accessible-resource name that resides in the
+ * manifest and is supposed to be constructed by each user using a unique value
+ * (this is done automatically by `build.sh').
  */
 
 function get_secret()
 {
-    if (is_chrome)
-	return browser.runtime.getManifest().key.substring(0, 50);
-    else
+    if (is_mozilla)
 	return browser.runtime.getURL("dummy");
+
+    return chrome.runtime.getManifest().web_accessible_resources
+	.map(r => /^chromium-key-dummy-file-(.*)/.exec(r)).filter(r => r)[0][1];
 }
 
 function extract_signed(signature, signed_data)
diff --git a/common/storage_client.js b/common/storage_client.js
index 2b2f495..ef4a0b8 100644
--- a/common/storage_client.js
+++ b/common/storage_client.js
@@ -1,5 +1,7 @@
 /**
- * Hachette storage through connection (client side)
+ * This file is part of Haketilo.
+ *
+ * Function: Storage through messages (client side).
  *
  * Copyright (C) 2021 Wojtek Kosior
  * Redistribution terms are gathered in the `copyright' file.
diff --git a/common/storage_light.js b/common/storage_light.js
index 067bf0c..32e3b1f 100644
--- a/common/storage_light.js
+++ b/common/storage_light.js
@@ -1,6 +1,7 @@
 /**
- * part of Hachette
- * Storage manager, lighter than the previous one.
+ * This file is part of Haketilo.
+ *
+ * Function: Storage manager, lighter than the previous one.
  *
  * Copyright (C) 2021 Wojtek Kosior
  * Redistribution terms are gathered in the `copyright' file.
diff --git a/common/storage_raw.js b/common/storage_raw.js
index 4c02ee4..e354b6b 100644
--- a/common/storage_raw.js
+++ b/common/storage_raw.js
@@ -1,6 +1,7 @@
 /**
- * part of Hachette
- * Basic wrappers for storage API functions.
+ * This file is part of Haketilo.
+ *
+ * Function: Basic wrappers for storage API functions.
  *
  * Copyright (C) 2021 Wojtek Kosior
  * Redistribution terms are gathered in the `copyright' file.
diff --git a/common/stored_types.js b/common/stored_types.js
index bfceba6..a693b1c 100644
--- a/common/stored_types.js
+++ b/common/stored_types.js
@@ -1,5 +1,7 @@
 /**
- * Hachette stored item types "enum"
+ * This file is part of Haketilo.
+ *
+ * Function: Define an "enum" of stored item types.
  *
  * Copyright (C) 2021 Wojtek Kosior
  * Redistribution terms are gathered in the `copyright' file.