diff options
Diffstat (limited to 'common')
-rw-r--r-- | common/misc.js | 50 | ||||
-rw-r--r-- | common/observable.js | 28 | ||||
-rw-r--r-- | common/patterns.js | 167 | ||||
-rw-r--r-- | common/settings_query.js | 27 | ||||
-rw-r--r-- | common/signing.js | 73 | ||||
-rw-r--r-- | common/storage_light.js | 129 | ||||
-rw-r--r-- | common/storage_raw.js | 54 |
7 files changed, 369 insertions, 159 deletions
diff --git a/common/misc.js b/common/misc.js index 30a9e77..91d60d2 100644 --- a/common/misc.js +++ b/common/misc.js @@ -8,9 +8,7 @@ /* * IMPORTS_START - * IMPORT sha256 * IMPORT browser - * IMPORT is_chrome * IMPORT TYPE_NAME * IMPORT TYPE_PREFIX * IMPORTS_END @@ -45,32 +43,6 @@ function gen_nonce(length) // Default 16 return Uint8toHex(randomData); } -function get_secure_salt() -{ - if (is_chrome) - return browser.runtime.getManifest().key.substring(0, 50); - else - return browser.runtime.getURL("dummy"); -} - -function extract_signed(signature, data, times) -{ - const now = new Date(); - times = times || [[now], [now, -1]]; - - const reductor = - (ok, time) => ok || signature === sign_data(data, ...time); - if (!times.reduce(reductor, false)) - return undefined; - - try { - return JSON.parse(decodeURIComponent(data)); - } catch (e) { - /* This should not be reached - it's our self-produced valid JSON. */ - console.log("Unexpected internal error - invalid JSON smuggled!", e); - } -} - /* csp rule that blocks all scripts except for those injected by us */ function csp_rule(nonce) { @@ -89,7 +61,7 @@ const report_only_header_name = "content-security-policy-report-only"; function is_csp_header_name(string, include_report_only) { - string = string && string.toLowerCase() || ""; + string = string && string.toLowerCase().trim() || ""; return (include_report_only && string === report_only_header_name) || csp_header_names.has(string); @@ -112,17 +84,13 @@ function open_in_settings(prefix, name) window.open(url, "_blank"); } -/* Check if url corresponds to a browser's special page */ -function is_privileged_url(url) -{ - return !!/^(chrome(-extension)?|moz-extension):\/\/|^about:/i.exec(url); -} - -/* Sign a given string for a given time */ -function sign_data(data, now, hours_offset) { - let time = Math.floor(now / 3600000) + (hours_offset || 0); - return sha256(get_secure_salt() + time + data); -} +/* + * Check if url corresponds to a browser's special page (or a directory index in + * case of `file://' protocol). + */ +const privileged_reg = + /^(chrome(-extension)?|moz-extension):\/\/|^about:|^file:\/\/.*\/$/; +const is_privileged_url = url => privileged_reg.test(url); /* Parse a CSP header */ function parse_csp(csp) { @@ -193,8 +161,6 @@ const matchers = { /* * EXPORTS_START * EXPORT gen_nonce - * EXPORT extract_signed - * EXPORT sign_data * EXPORT csp_rule * EXPORT is_csp_header_name * EXPORT nice_name diff --git a/common/observable.js b/common/observable.js index 1fb0b0a..02f1c1b 100644 --- a/common/observable.js +++ b/common/observable.js @@ -6,28 +6,22 @@ * Redistribution terms are gathered in the `copyright' file. */ -function make() -{ - return new Set(); -} +const make = (value=undefined) => ({value, listeners: new Set()}); +const subscribe = (observable, cb) => observable.listeners.add(cb); +const unsubscribe = (observable, cb) => observable.listeners.delete(cb); -function subscribe(observable, cb) -{ - observable.add(cb); -} - -function unsubscribe(observable, cb) -{ - observable.delete(cb); -} +const silent_set = (observable, value) => observable.value = value; +const broadcast = (observable, ...values) => + observable.listeners.forEach(cb => cb(...values)); -function broadcast(observable, event) +function set(observable, value) { - for (const callback of observable) - callback(event); + const old_value = observable.value; + silent_set(observable, value); + broadcast(observable, value, old_value); } -const observables = {make, subscribe, unsubscribe, broadcast}; +const observables = {make, subscribe, unsubscribe, broadcast, silent_set, set}; /* * EXPORTS_START diff --git a/common/patterns.js b/common/patterns.js index be7c650..ebb55ab 100644 --- a/common/patterns.js +++ b/common/patterns.js @@ -5,35 +5,48 @@ * Redistribution terms are gathered in the `copyright' file. */ -const proto_re = "[a-zA-Z]*:\/\/"; +const proto_regex = /^(\w+):\/\/(.*)$/; + +const user_re = "[^/?#@]+@" const domain_re = "[^/?#]+"; -const segments_re = "/[^?#]*"; -const query_re = "\\?[^#]*"; - -const url_regex = new RegExp(`\ -^\ -(${proto_re})\ -(${domain_re})\ -(${segments_re})?\ -(${query_re})?\ -#?.*\$\ -`); +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 ftp_regex = new RegExp(`^(${user_re})?(${domain_re})(${path_re}).*`); function deconstruct_url(url) { - const regex_match = url_regex.exec(url); - if (regex_match === null) + const proto_match = proto_regex.exec(url); + if (proto_match === null) return undefined; - let [_, proto, domain, path, query] = regex_match; + const deco = {proto: proto_match[1]}; + + if (deco.proto === "file") { + 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 { + 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); + } + + if (deco.domain) + deco.domain = deco.domain.split("."); - domain = domain.split("."); - let path_trailing_dash = - path && path[path.length - 1] === "/"; - path = (path || "").split("/").filter(s => s !== ""); - path.unshift(""); + 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 {proto, domain, path, query, path_trailing_dash}; + return deco; } /* Be sane: both arguments should be arrays of length >= 2 */ @@ -104,84 +117,70 @@ function url_matches(url, pattern) return false } - if (pattern_deco.proto !== url_deco.proto) - return false; - - return domain_matches(url_deco.domain, pattern_deco.domain) && - path_matches(url_deco.path, url_deco.path_trailing_dash, - pattern_deco.path, pattern_deco.path_trailing_dash); + 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); } -/* - * Call callback for every possible pattern that matches url. Return when there - * are no more patterns or callback returns false. - */ -function for_each_possible_pattern(url, callback) +function* each_domain_pattern(domain_segments) { - const deco = deconstruct_url(url); - - if (deco === undefined) { - console.log("bad url format", url); - return; + for (let slice = 0; slice < domain_segments.length; slice++) { + const domain_part = domain_segments.slice(slice).join("."); + const domain_wildcards = []; + if (slice === 0) + yield domain_part; + if (slice === 1) + yield "*." + domain_part; + if (slice > 1) + yield "**." + domain_part; + yield "***." + domain_part; } +} - for (let d_slice = 0; d_slice < deco.domain.length; d_slice++) { - const domain_part = deco.domain.slice(d_slice).join("."); - const 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 (const domain_wildcard of domain_wildcards) { - const domain_pattern = domain_wildcard + domain_part; - - for (let s_slice = deco.path.length; s_slice > 0; s_slice--) { - const path_part = deco.path.slice(0, s_slice).join("/"); - const path_wildcards = []; - if (s_slice === deco.path.length) { - if (deco.path_trailing_dash) - path_wildcards.push("/"); - path_wildcards.push(""); - } - if (s_slice === deco.path.length - 1 && - deco.path[s_slice] !== "*") - path_wildcards.push("/*"); - if (s_slice < deco.path.length && - (deco.path[s_slice] !== "**" || - s_slice < deco.path.length - 1)) - path_wildcards.push("/**"); - if (deco.path[s_slice] !== "***" || s_slice < deco.path.length) - path_wildcards.push("/***"); - - for (const path_wildcard of path_wildcards) { - const path_pattern = path_part + path_wildcard; - - const pattern = deco.proto + domain_pattern + path_pattern; - - if (callback(pattern) === false) - return; - } - } +function* each_path_pattern(path_segments, trailing_dash) +{ + for (let slice = path_segments.length; slice > 0; slice--) { + const path_part = path_segments.slice(0, slice).join("/"); + const path_wildcards = []; + if (slice === path_segments.length) { + if (trailing_dash) + yield path_part + "/"; + yield path_part; } + if (slice === path_segments.length - 1 && path_segments[slice] !== "*") + yield path_part + "/*"; + if (slice < path_segments.length - 1) + yield path_part + "/**"; + if (slice < path_segments.length - 1 || + path_segments[path_segments.length - 1] !== "***") + yield path_part + "/***"; } } -function possible_patterns(url) +/* Generate every possible pattern that matches url. */ +function* each_url_pattern(url) { - const patterns = []; - for_each_possible_pattern(url, patterns.push); + const deco = deconstruct_url(url); - return patterns; + if (deco === undefined) { + console.log("bad url format", url); + return false; + } + + const all_domains = deco.domain ? each_domain_pattern(deco.domain) : [""]; + for (const domain of all_domains) { + for (const path of each_path_pattern(deco.path, deco.trailing_dash)) + yield `${deco.proto}://${domain}${path}`; + } } /* * EXPORTS_START * EXPORT url_matches - * EXPORT for_each_possible_pattern - * EXPORT possible_patterns + * EXPORT each_url_pattern * EXPORTS_END */ diff --git a/common/settings_query.js b/common/settings_query.js index e85ae63..b54e580 100644 --- a/common/settings_query.js +++ b/common/settings_query.js @@ -8,30 +8,25 @@ /* * IMPORTS_START * IMPORT TYPE_PREFIX - * IMPORT for_each_possible_pattern + * IMPORT each_url_pattern * IMPORTS_END */ -function check_pattern(storage, pattern, multiple, matched) -{ - const settings = storage.get(TYPE_PREFIX.PAGE, pattern); - - if (settings === undefined) - return; - - matched.push([pattern, settings]); - - if (!multiple) - return false; -} - function query(storage, url, multiple) { const matched = []; const cb = p => check_pattern(storage, p, multiple, matched); - for_each_possible_pattern(url, cb); + for (const pattern of each_url_pattern(url)) { + const result = [pattern, storage.get(TYPE_PREFIX.PAGE, pattern)]; + if (result[1] === undefined) + continue; + + if (!multiple) + return result; + matched.push(result); + } - return multiple ? matched : (matched[0] || [undefined, undefined]); + return multiple ? matched : [undefined, undefined]; } function query_best(storage, url) diff --git a/common/signing.js b/common/signing.js new file mode 100644 index 0000000..2171714 --- /dev/null +++ b/common/signing.js @@ -0,0 +1,73 @@ +/** + * part of Hachette + * Functions related to "signing" of data, refactored to a separate file. + * + * Copyright (C) 2021 Wojtek Kosior + * Redistribution terms are gathered in the `copyright' file. + */ + +/* + * IMPORTS_START + * IMPORT sha256 + * IMPORT browser + * IMPORT is_chrome + * 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. + * When using the smuggled data, we first need to make sure it isn't spoofed. + * For that, we use this pseudo-signing mechanism. + * + * Despite what name suggests, no assymetric cryptography is involved, as it + * would bring no additional benefits and would incur bigger performance + * overhead. Instead, we hash the string data together with some secret value + * that is supposed to be known only by this browser instance. Resulting hash + * sum plays the role of the signature. In the hash we also include current + * time. This way, even if signed data leaks (which shouldn't happen in the + * first place), an attacker won't be able to re-use it indefinitely. + * + * 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. + */ + +function get_secret() +{ + if (is_chrome) + return browser.runtime.getManifest().key.substring(0, 50); + else + return browser.runtime.getURL("dummy"); +} + +function extract_signed(signature, signed_data) +{ + const match = /^([1-9][0-9]{12}|0)_(.*)$/.exec(signed_data); + if (!match) + return {fail: "bad format"}; + + const result = {time: parseInt(match[1]), data: match[2]}; + if (sign_data(result.data, result.time)[0] !== signature) + result.fail = "bad signature"; + + return result; +} + +/* + * Sign a given string for a given time. Time should be either 0 or in the range + * 10^12 <= time < 10^13. + */ +function sign_data(data, time) { + return [sha256(get_secret() + time + data), `${time}_${data}`]; +} + +/* + * EXPORTS_START + * EXPORT extract_signed + * EXPORT sign_data + * EXPORTS_END + */ diff --git a/common/storage_light.js b/common/storage_light.js new file mode 100644 index 0000000..067bf0c --- /dev/null +++ b/common/storage_light.js @@ -0,0 +1,129 @@ +/** + * part of Hachette + * Storage manager, lighter than the previous one. + * + * Copyright (C) 2021 Wojtek Kosior + * Redistribution terms are gathered in the `copyright' file. + */ + +/* + * IMPORTS_START + * IMPORT TYPE_PREFIX + * IMPORT raw_storage + * IMPORT is_mozilla + * IMPORT observables + */ + +const reg_spec = new Set(["\\", "[", "]", "(", ")", "{", "}", ".", "*", "+"]); +const escape_reg_special = c => reg_spec.has(c) ? "\\" + c : c; + +function make_regex(name) +{ + return new RegExp(`^${name.split("").map(escape_reg_special).join("")}\$`); +} + +const listeners_by_callback = new Map(); + +function listen(callback, prefix, name) +{ + let by_prefix = listeners_by_callback.get(callback); + if (!by_prefix) { + by_prefix = new Map(); + listeners_by_callback.set(callback, by_prefix); + } + + let by_name = by_prefix.get(prefix); + if (!by_name) { + by_name = new Map(); + by_prefix.set(prefix, by_name); + } + + let name_reg = by_name.get(name); + if (!name_reg) { + name_reg = name.test ? name : make_regex(name); + by_name.set(name, name_reg); + } +} + +function no_listen(callback, prefix, name) +{ + const by_prefix = listeners_by_callback.get(callback); + if (!by_prefix) + return; + + const by_name = by_prefix.get(prefix); + if (!by_name) + return; + + const name_reg = by_name.get(name); + if (!name_reg) + return; + + by_name.delete(name); + + if (by_name.size === 0) + by_prefix.delete(prefix); + + if (by_prefix.size === 0) + listeners_by_callback.delete(callback); +} + +function storage_change_callback(changes, area) +{ + if (is_mozilla && area !== "local") + {console.log("area", area);return;} + + for (const item of Object.keys(changes)) { + for (const [callback, by_prefix] of listeners_by_callback.entries()) { + const by_name = by_prefix.get(item[0]); + if (!by_name) + continue; + + for (const reg of by_name.values()) { + if (!reg.test(item.substring(1))) + continue; + + try { + callback(item, changes[item]); + } catch(e) { + console.error(e); + } + } + } + } +} + +raw_storage.listen(storage_change_callback); + + +const created_observables = new Map(); + +async function observe(prefix, name) +{ + const observable = observables.make(); + const callback = (it, ch) => observables.set(observable, ch.newValue); + listen(callback, prefix, name); + created_observables.set(observable, [callback, prefix, name]); + observables.silent_set(observable, await raw_storage.get(prefix + name)); + + return observable; +} + +const observe_var = name => observe(TYPE_PREFIX.VAR, name); + +function no_observe(observable) +{ + no_listen(...created_observables.get(observable) || []); + created_observables.delete(observable); +} + +const light_storage = {}; +Object.assign(light_storage, raw_storage); +Object.assign(light_storage, + {listen, no_listen, observe, observe_var, no_observe}); + +/* + * EXPORTS_START + * EXPORT light_storage + * EXPORTS_END + */ diff --git a/common/storage_raw.js b/common/storage_raw.js new file mode 100644 index 0000000..4c02ee4 --- /dev/null +++ b/common/storage_raw.js @@ -0,0 +1,54 @@ +/** + * part of Hachette + * Basic wrappers for storage API functions. + * + * Copyright (C) 2021 Wojtek Kosior + * Redistribution terms are gathered in the `copyright' file. + */ + +/* + * IMPORTS_START + * IMPORT TYPE_PREFIX + * IMPORT browser + * IMPORT is_chrome + * IMPORTS_END + */ + +async function get(key) +{ + /* Fix for fact that Chrome does not use promises here */ + const promise = is_chrome ? + new Promise(resolve => chrome.storage.local.get(key, resolve)) : + browser.storage.local.get(key); + + return (await promise)[key]; +} + +async function set(key_or_object, value) +{ + const arg = typeof key_or_object === "object" ? + key_or_object : {[key_or_object]: value}; + return browser.storage.local.set(arg); +} + +async function set_var(name, value) +{ + return set(TYPE_PREFIX.VAR + name, value); +} + +async function get_var(name) +{ + return get(TYPE_PREFIX.VAR + name); +} + +const on_changed = browser.storage.onChanged || browser.storage.local.onChanged; +const listen = cb => on_changed.addListener(cb); +const no_listen = cb => on_changed.removeListener(cb); + +const raw_storage = {get, set, get_var, set_var, listen, no_listen}; + +/* + * EXPORTS_START + * EXPORT raw_storage + * EXPORTS_END + */ |