Chore: simplify build infra

Build/build-anti-bogus-domain.ts

@@ -1,21 +1,20 @@
 // @ts-check
 import path from 'path';
-import { isIPv4, isIPv6 } from 'net';
 import { createRuleset } from './lib/create-file';
 import { fetchRemoteTextAndReadByLine, readFileByLine } from './lib/fetch-text-by-line';
 import { processLine } from './lib/process-line';
 import { task } from './lib/trace-runner';
 import { SHARED_DESCRIPTION } from './lib/constants';
+import { isProbablyIpv4, isProbablyIpv6 } from './lib/is-fast-ip';
 
 const getBogusNxDomainIPs = async () => {
-  /** @type {string[]} */
-  const result = [];
+  const result: string[] = [];
   for await (const line of await fetchRemoteTextAndReadByLine('https://raw.githubusercontent.com/felixonmars/dnsmasq-china-list/master/bogus-nxdomain.china.conf')) {
-    if (line.startsWith('bogus-nxdomain=')) {
+    if (line && line.startsWith('bogus-nxdomain=')) {
       const ip = line.slice(15).trim();
-      if (isIPv4(ip)) {
+      if (isProbablyIpv4(ip)) {
         result.push(`IP-CIDR,${ip}/32,no-resolve`);
-      } else if (isIPv6(ip)) {
+      } else if (isProbablyIpv6(ip)) {
         result.push(`IP-CIDR6,${ip}/128,no-resolve`);
       }
     }
@@ -26,21 +25,16 @@ const getBogusNxDomainIPs = async () => {
 export const buildAntiBogusDomain = task(import.meta.path, async () => {
   const bogusIpPromise = getBogusNxDomainIPs();
 
-  /** @type {string[]} */
-  const result = [];
+  const result: string[] = [];
   for await (const line of readFileByLine(path.resolve(import.meta.dir, '../Source/ip/reject.conf'))) {
-    if (line === '# --- [Anti Bogus Domain Replace Me] ---') {
-      // bogus ip is less than 200, no need to worry about "Maximum call stack size exceeded"
-      result.push(...(await bogusIpPromise));
-      continue;
-    } else {
-      const l = processLine(line);
-      if (l) {
-        result.push(l);
-      }
+    const l = processLine(line);
+    if (l) {
+      result.push(l);
     }
   }
 
+  result.push(...(await bogusIpPromise));
+
   const description = [
     ...SHARED_DESCRIPTION,
     '',

Build/build-cdn-conf.ts

@@ -63,16 +63,14 @@ const buildCdnConf = task(import.meta.path, async () => {
   const getS3OSSDomainsPromise: Promise<Set<string>> = getS3OSSDomains();
 
   for await (const l of readFileByLine(path.resolve(import.meta.dir, '../Source/non_ip/cdn.conf'))) {
-    if (l === '# --- [AWS S3 Replace Me] ---') {
-      (await getS3OSSDomainsPromise).forEach((domain: string) => { cdnDomainsList.push(`DOMAIN-SUFFIX,${domain}`); });
-      continue;
-    }
     const line = processLine(l);
     if (line) {
       cdnDomainsList.push(line);
     }
   }
 
+  (await getS3OSSDomainsPromise).forEach((domain: string) => { cdnDomainsList.push(`DOMAIN-SUFFIX,${domain}`); });
+
   const description: string[] = [
     ...SHARED_DESCRIPTION,
     '',

Build/build-common.ts

@@ -22,23 +22,25 @@ export const buildCommon = task(import.meta.path, async () => {
 
   const pw = new PathScurry(sourceDir);
   for await (const entry of pw) {
-    if (entry.isFile()) {
-      if (path.extname(entry.name) === '.js') {
-        continue;
-      }
+    if (!entry.isFile()) {
+      continue;
+    }
 
-      const relativePath = entry.relative();
-      if (relativePath.startsWith('domainset/')) {
-        promises.push(transformDomainset(entry.fullpath(), relativePath));
-        continue;
-      }
-      if (
-        relativePath.startsWith('ip/')
+    if (path.extname(entry.name) === '.js') {
+      continue;
+    }
+
+    const relativePath = entry.relative();
+    if (relativePath.startsWith('domainset/')) {
+      promises.push(transformDomainset(entry.fullpath(), relativePath));
+      continue;
+    }
+    if (
+      relativePath.startsWith('ip/')
         || relativePath.startsWith('non_ip/')
-      ) {
-        promises.push(transformRuleset(entry.fullpath(), relativePath));
-        continue;
-      }
+    ) {
+      promises.push(transformRuleset(entry.fullpath(), relativePath));
+      continue;
     }
   }
 

Build/build-internal-cdn-rules.ts

@@ -15,9 +15,6 @@ export const buildInternalCDNDomains = task(import.meta.path, async () => {
   const set = new Set<string>();
   const keywords = new Set<string>();
 
-  const gorhill = await getGorhillPublicSuffixPromise();
-  const domainSorter = createDomainSorter(gorhill);
-
   const addApexDomain = (input: string) => {
     // We are including the private domains themselves
     const d = tldts.getDomain(input, { allowPrivateDomains: false });
@@ -61,7 +58,8 @@ export const buildInternalCDNDomains = task(import.meta.path, async () => {
     }
   };
 
-  await Promise.all([
+  const [domainSorter] = await Promise.all([
+    getGorhillPublicSuffixPromise().then(createDomainSorter),
     processLocalRuleSet(path.resolve(import.meta.dir, '../List/non_ip/cdn.conf')),
     processLocalRuleSet(path.resolve(import.meta.dir, '../List/non_ip/global.conf')),
     processLocalRuleSet(path.resolve(import.meta.dir, '../List/non_ip/global_plus.conf')),

Build/build-reject-domainset.ts

@@ -18,15 +18,12 @@ import * as tldts from 'tldts';
 import { SHARED_DESCRIPTION } from './lib/constants';
 import { getPhishingDomains } from './lib/get-phishing-domains';
 
-/** Whitelists */
-const filterRuleWhitelistDomainSets = new Set(PREDEFINED_WHITELIST);
-/** @type {Set<string>} Dedupe domains inclued by DOMAIN-KEYWORD */
-const domainKeywordsSet = new Set<string>();
-/** @type {Set<string>} Dedupe domains included by DOMAIN-SUFFIX */
-const domainSuffixSet = new Set<string>();
-
 export const buildRejectDomainSet = task(import.meta.path, async () => {
-  /** @type Set<string> */
+  /** Whitelists */
+  const filterRuleWhitelistDomainSets = new Set(PREDEFINED_WHITELIST);
+  const domainKeywordsSet = new Set<string>();
+  const domainSuffixSet = new Set<string>();
+
   const domainSets = new Set<string>();
 
   // Parse from AdGuard Filters
@@ -38,9 +35,7 @@ export const buildRejectDomainSet = task(import.meta.path, async () => {
       // Parse from remote hosts & domain lists
       ...HOSTS.map(entry => processHosts(entry[0], entry[1]).then(hosts => {
         hosts.forEach(host => {
-          if (host) {
-            domainSets.add(host);
-          }
+          domainSets.add(host);
         });
       })),
       ...ADGUARD_FILTERS.map(input => {
@@ -61,12 +56,8 @@ export const buildRejectDomainSet = task(import.meta.path, async () => {
         'https://raw.githubusercontent.com/AdguardTeam/AdGuardSDNSFilter/master/Filters/exceptions.txt',
         'https://raw.githubusercontent.com/AdguardTeam/AdGuardSDNSFilter/master/Filters/exclusions.txt'
       ].map(input => processFilterRules(input).then(({ white, black }) => {
-        white.forEach(i => {
-          filterRuleWhitelistDomainSets.add(i);
-        });
-        black.forEach(i => {
-          filterRuleWhitelistDomainSets.add(i);
-        });
+        white.forEach(i => filterRuleWhitelistDomainSets.add(i));
+        black.forEach(i => filterRuleWhitelistDomainSets.add(i));
       }))),
       getPhishingDomains().then(([purePhishingDomains, fullDomainSet]) => {
         fullDomainSet.forEach(host => {
@@ -74,10 +65,16 @@ export const buildRejectDomainSet = task(import.meta.path, async () => {
             domainSets.add(host);
           }
         });
-        purePhishingDomains.forEach(suffix => {
-          domainSets.add(`.${suffix}`);
-        });
-      })
+        purePhishingDomains.forEach(suffix => domainSets.add(`.${suffix}`));
+      }),
+      (async () => {
+        for await (const line of readFileByLine(path.resolve(import.meta.dir, '../Source/domainset/reject_sukka.conf'))) {
+          const l = processLine(line);
+          if (l) {
+            domainSets.add(l);
+          }
+        }
+      })()
     ]);
 
     // remove pre-defined enforced blacklist from whitelist
@@ -94,17 +91,7 @@ export const buildRejectDomainSet = task(import.meta.path, async () => {
   }
 
   let previousSize = domainSets.size;
-  console.log(`Import ${previousSize} rules from Hosts / AdBlock Filter Rules!`);
-
-  for await (const line of readFileByLine(path.resolve(import.meta.dir, '../Source/domainset/reject_sukka.conf'))) {
-    const l = processLine(line);
-    if (l) {
-      domainSets.add(l);
-    }
-  }
-
-  previousSize = domainSets.size - previousSize;
-  console.log(`Import ${previousSize} rules from reject_sukka.conf!`);
+  console.log(`Import ${previousSize} rules from Hosts / AdBlock Filter Rules & reject_sukka.conf!`);
 
   for await (const line of readFileByLine(path.resolve(import.meta.dir, '../Source/non_ip/reject.conf'))) {
     const [type, keyword] = line.split(',');
@@ -150,11 +137,11 @@ export const buildRejectDomainSet = task(import.meta.path, async () => {
         domainSets.delete(domain);
       }
     }
-  });
-
-  console.log(`Deduped ${previousSize} - ${domainSets.size} = ${previousSize - domainSets.size} from black keywords and suffixes!`);
 
+    console.log(`Deduped ${previousSize} - ${domainSets.size} = ${previousSize - domainSets.size} from black keywords and suffixes!`);
+  });
   previousSize = domainSets.size;
+
   // Dedupe domainSets
   const dudupedDominArray = traceSync('* Dedupe from covered subdomain', () => domainDeduper(Array.from(domainSets)));
   console.log(`Deduped ${previousSize - dudupedDominArray.length} rules!`);
@@ -180,9 +167,6 @@ export const buildRejectDomainSet = task(import.meta.path, async () => {
     })
   );
 
-  const domainSorter = createDomainSorter(gorhill);
-  const domainset = traceSync('* Sort reject domainset', () => dudupedDominArray.sort(domainSorter));
-
   const description = [
     ...SHARED_DESCRIPTION,
     '',
@@ -198,7 +182,7 @@ export const buildRejectDomainSet = task(import.meta.path, async () => {
       'Sukka\'s Ruleset - Reject Base',
       description,
       new Date(),
-      domainset,
+      traceSync('* Sort reject domainset', () => dudupedDominArray.sort(createDomainSorter(gorhill))),
       'domainset',
       path.resolve(import.meta.dir, '../List/domainset/reject.conf'),
       path.resolve(import.meta.dir, '../Clash/domainset/reject.txt')

Build/build-telegram-cidr.ts

@@ -2,14 +2,14 @@
 import { defaultRequestInit, fetchWithRetry } from './lib/fetch-retry';
 import { createReadlineInterfaceFromResponse } from './lib/fetch-text-by-line';
 import path from 'path';
-import { isIPv4, isIPv6 } from 'net';
+import { isProbablyIpv4, isProbablyIpv6 } from './lib/is-fast-ip';
 import { processLine } from './lib/process-line';
 import { createRuleset } from './lib/create-file';
 import { task } from './lib/trace-runner';
 import { SHARED_DESCRIPTION } from './lib/constants';
 
 export const buildTelegramCIDR = task(import.meta.path, async () => {
-  const resp = await fetchWithRetry('https://core.telegram.org/resources/cidr.txt', defaultRequestInit);
+  const resp = await fetchWithRetry('https://core.telegram.org/resources/cidr.txt', defaultRequestInit) as Response;
   const lastModified = resp.headers.get('last-modified');
   const date = lastModified ? new Date(lastModified) : new Date();
 
@@ -20,10 +20,10 @@ export const buildTelegramCIDR = task(import.meta.path, async () => {
     if (!cidr) continue;
 
     const [subnet] = cidr.split('/');
-    if (isIPv4(subnet)) {
+    if (isProbablyIpv4(subnet)) {
       results.push(`IP-CIDR,${cidr},no-resolve`);
     }
-    if (isIPv6(subnet)) {
+    if (isProbablyIpv6(subnet)) {
       results.push(`IP-CIDR6,${cidr},no-resolve`);
     }
   }

Build/lib/create-file.ts

@@ -2,6 +2,7 @@
 import { readFileByLine } from './fetch-text-by-line';
 import { surgeDomainsetToClashDomainset, surgeRulesetToClashClassicalTextRuleset } from './clash';
 import { traceAsync } from './trace-runner';
+import picocolors from 'picocolors';
 
 export async function compareAndWriteFile(linesA: string[], filePath: string) {
   let isEqual = true;
@@ -45,11 +46,11 @@ export async function compareAndWriteFile(linesA: string[], filePath: string) {
   }
 
   if (isEqual) {
-    console.log(`Same Content, bail out writing: ${filePath}`);
+    console.log(picocolors.gray(`Same Content, bail out writing: ${filePath}`));
     return;
   }
 
-  await traceAsync(`Writing ${filePath}`, async () => {
+  await traceAsync(picocolors.gray(`Writing ${filePath}`), async () => {
     if (linesALen < 10000) {
       return Bun.write(file, `${linesA.join('\n')}\n`);
     }
@@ -63,7 +64,7 @@ export async function compareAndWriteFile(linesA: string[], filePath: string) {
 
     await writer.flush();
     return writer.end();
-  });
+  }, picocolors.gray);
 }
 
 export const withBannerArray = (title: string, description: string[], date: Date, content: string[]) => {

Build/lib/fetch-retry.ts

@@ -1,4 +1,5 @@
 import retry from 'async-retry';
+import picocolors from 'picocolors';
 
 // retry settings
 const MIN_TIMEOUT = 10;
@@ -86,7 +87,7 @@ function createFetchRetry($fetch: typeof fetch): typeof fetch {
               err.name === 'AbortError'
               || ('digest' in err && err.digest === 'AbortError')
             ) {
-              console.log('[fetch abort]', url.toString());
+              console.log(picocolors.gray('[fetch abort]'), picocolors.gray(url.toString()));
               return bail(err);
             }
           }

Build/lib/is-fast-ip.ts

@@ -31,3 +31,47 @@ export function isProbablyIpv4(hostname: string): boolean {
     && /* '.' */ hostname.charCodeAt(hostname.length - 1) !== 46 /* '.' */
   );
 }
+
+export function isProbablyIpv6(hostname: string): boolean {
+  if (hostname.length < 3) {
+    return false;
+  }
+
+  let start = hostname[0] === '[' ? 1 : 0;
+  let end = hostname.length;
+
+  if (hostname[end - 1] === ']') {
+    end -= 1;
+  }
+
+  // We only consider the maximum size of a normal IPV6. Note that this will
+  // fail on so-called "IPv4 mapped IPv6 addresses" but this is a corner-case
+  // and a proper validation library should be used for these.
+  if (end - start > 39) {
+    return false;
+  }
+
+  /* eslint-disable sukka/no-single-return -- here it goes */
+  let hasColon = false;
+
+  for (; start < end; start += 1) {
+    const code = hostname.charCodeAt(start);
+
+    if (code === 58 /* ':' */) {
+      hasColon = true;
+    } else if (
+      !(
+        (
+          (code >= 48 && code <= 57) // 0-9
+          || (code >= 97 && code <= 102) // a-f
+          || (code >= 65 && code <= 90) // A-F
+        )
+      )
+    ) {
+      return false;
+    }
+  }
+
+  return hasColon;
+  /* eslint-enable sukka/no-single-return -- here it goes */
+}

Build/lib/parse-filter.ts

@@ -51,11 +51,7 @@ export async function processDomainLists(domainListsUrl: string, includeAllSubDo
       foundDebugDomain = true;
     }
 
-    if (includeAllSubDomain) {
-      domainSets.add(`.${domainToAdd}`);
-    } else {
-      domainSets.add(domainToAdd);
-    }
+    domainSets.add(includeAllSubDomain ? `.${domainToAdd}` : domainToAdd);
   }
 
   return domainSets;
@@ -90,6 +86,8 @@ export async function processHosts(hostsUrl: string, includeAllSubDomain = false
       }
     }
 
+    console.log(picocolors.gray('[process hosts]'), picocolors.gray(hostsUrl), picocolors.gray(domainSets.size));
+
     return domainSets;
   });
 }
@@ -159,7 +157,7 @@ export async function processFilterRules(
           warningMessages.push(hostname);
           break;
         default:
-          throw new Error(`Unknown flag: ${flag as any}`);
+          break;
       }
     };
 
@@ -187,6 +185,13 @@ export async function processFilterRules(
     );
   });
 
+  console.log(
+    picocolors.gray('[process filter]'),
+    picocolors.gray(filterRulesUrl),
+    picocolors.gray(`white: ${whitelistDomainSets.size}`),
+    picocolors.gray(`black: ${blacklistDomainSets.size}`)
+  );
+
   return {
     white: whitelistDomainSets,
     black: blacklistDomainSets,
@@ -569,25 +574,23 @@ class CustomAbortError extends Error {
   public readonly digest = 'AbortError';
 }
 
-function sleepWithAbort(ms: number, signal: AbortSignal) {
-  return new Promise<void>((resolve, reject) => {
-    signal.throwIfAborted();
-    signal.addEventListener('abort', stop);
-    Bun.sleep(ms).then(done).catch(doReject);
+const sleepWithAbort = (ms: number, signal: AbortSignal) => new Promise<void>((resolve, reject) => {
+  signal.throwIfAborted();
+  signal.addEventListener('abort', stop);
+  Bun.sleep(ms).then(done).catch(doReject);
 
-    function done() {
-      signal.removeEventListener('abort', stop);
-      resolve();
-    }
-    function stop(this: AbortSignal) {
-      reject(this.reason);
-    }
-    function doReject(reason: unknown) {
-      signal.removeEventListener('abort', stop);
-      reject(reason);
-    }
-  });
-}
+  function done() {
+    signal.removeEventListener('abort', stop);
+    resolve();
+  }
+  function stop(this: AbortSignal) {
+    reject(this.reason);
+  }
+  function doReject(reason: unknown) {
+    signal.removeEventListener('abort', stop);
+    reject(reason);
+  }
+});
 
 async function fetchAssets(url: string, fallbackUrls: string[] | readonly string[]) {
   const controller = new AbortController();
@@ -602,7 +605,7 @@ async function fetchAssets(url: string, fallbackUrls: string[] | readonly string
   const createFetchFallbackPromise = async (url: string, index: number) => {
     // Most assets can be downloaded within 250ms. To avoid wasting bandwidth, we will wait for 350ms before downloading from the fallback URL.
     try {
-      await sleepWithAbort(200 + (index + 1) * 10, controller.signal);
+      await sleepWithAbort(300 + (index + 1) * 20, controller.signal);
     } catch {
       console.log(picocolors.gray('[fetch cancelled early]'), picocolors.gray(url));
       throw new CustomAbortError();

Source/ip/reject.conf

@@ -58,4 +58,3 @@ IP-CIDR,222.73.156.235/32,no-resolve
 
 # --- Anti-Bogus Domain ---
 # https://github.com/felixonmars/dnsmasq-china-list/blob/master/bogus-nxdomain.china.conf
-# --- [Anti Bogus Domain Replace Me] ---

Source/non_ip/cdn.conf

@@ -108,4 +108,3 @@ DOMAIN-SUFFIX,cos.eu-frankfurt.myqcloud.com
 DOMAIN-SUFFIX,ks3-cn-hk-1.ksyuncs.com
 DOMAIN-SUFFIX,ks3-rus.ksyuncs.com
 DOMAIN-SUFFIX,ks3-sgp.ksyuncs.com
-# --- [AWS S3 Replace Me] ---