@j0hanz/superfetch 1.2.2 → 1.2.3

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (53) hide show
  1. package/README.md +60 -45
  2. package/dist/config/formatting.d.ts +1 -1
  3. package/dist/config/types/content.d.ts +3 -3
  4. package/dist/config/types/runtime.d.ts +1 -1
  5. package/dist/config/types/tools.d.ts +12 -12
  6. package/dist/http/cors.js +23 -23
  7. package/dist/http/download-routes.js +9 -4
  8. package/dist/http/mcp-routes.js +2 -13
  9. package/dist/http/mcp-validation.js +1 -1
  10. package/dist/http/server-middleware.js +2 -1
  11. package/dist/http/server.js +2 -0
  12. package/dist/index.js +5 -0
  13. package/dist/middleware/error-handler.js +1 -1
  14. package/dist/resources/cached-content.js +8 -4
  15. package/dist/server.js +2 -0
  16. package/dist/services/cache.d.ts +1 -1
  17. package/dist/services/cache.js +20 -7
  18. package/dist/services/context.d.ts +2 -4
  19. package/dist/services/context.js +1 -1
  20. package/dist/services/extractor.js +26 -21
  21. package/dist/services/fetcher/interceptors.d.ts +22 -0
  22. package/dist/services/fetcher/interceptors.js +18 -8
  23. package/dist/services/fetcher/response.js +32 -24
  24. package/dist/services/fetcher.d.ts +0 -1
  25. package/dist/services/fetcher.js +5 -7
  26. package/dist/services/metadata-collector.d.ts +10 -0
  27. package/dist/services/metadata-collector.js +11 -0
  28. package/dist/services/parser.js +26 -25
  29. package/dist/services/transform-worker-pool.d.ts +14 -0
  30. package/dist/services/transform-worker-pool.js +167 -0
  31. package/dist/tools/handlers/fetch-markdown.tool.d.ts +9 -1
  32. package/dist/tools/handlers/fetch-markdown.tool.js +58 -30
  33. package/dist/tools/handlers/fetch-single.shared.d.ts +8 -3
  34. package/dist/tools/handlers/fetch-single.shared.js +42 -17
  35. package/dist/tools/handlers/fetch-url.tool.js +46 -16
  36. package/dist/tools/index.js +13 -0
  37. package/dist/tools/schemas.d.ts +33 -30
  38. package/dist/tools/schemas.js +4 -0
  39. package/dist/tools/utils/common.js +20 -16
  40. package/dist/tools/utils/content-transform-async.d.ts +6 -0
  41. package/dist/tools/utils/content-transform-async.js +33 -0
  42. package/dist/tools/utils/content-transform.d.ts +4 -1
  43. package/dist/tools/utils/content-transform.js +7 -2
  44. package/dist/tools/utils/fetch-pipeline.js +18 -10
  45. package/dist/utils/content-cleaner.d.ts +1 -1
  46. package/dist/utils/download-url.d.ts +9 -1
  47. package/dist/utils/download-url.js +9 -6
  48. package/dist/utils/tool-error-handler.d.ts +2 -2
  49. package/dist/utils/tool-error-handler.js +7 -7
  50. package/dist/utils/url-validator.js +38 -0
  51. package/dist/workers/transform-worker.d.ts +1 -0
  52. package/dist/workers/transform-worker.js +50 -0
  53. package/package.json +4 -6
@@ -1,22 +1,25 @@
1
1
  import { config } from '../config/index.js';
2
2
  import * as cache from '../services/cache.js';
3
3
  import { generateSafeFilename } from './filename-generator.js';
4
- export function buildFileDownloadInfo(options) {
5
- if (!config.runtime.httpMode) {
4
+ export function buildFileDownloadInfo(options, deps = {}) {
5
+ const resolvedConfig = deps.config ?? config;
6
+ const resolvedCache = deps.cache ?? cache;
7
+ const resolveFilename = deps.generateSafeFilename ?? generateSafeFilename;
8
+ if (!resolvedConfig.runtime.httpMode) {
6
9
  return null;
7
10
  }
8
- if (!config.cache.enabled || !options.cacheKey) {
11
+ if (!resolvedConfig.cache.enabled || !options.cacheKey) {
9
12
  return null;
10
13
  }
11
- const parts = cache.parseCacheKey(options.cacheKey);
14
+ const parts = resolvedCache.parseCacheKey(options.cacheKey);
12
15
  if (!parts)
13
16
  return null;
14
- const cacheEntry = cache.get(options.cacheKey);
17
+ const cacheEntry = resolvedCache.get(options.cacheKey);
15
18
  if (!cacheEntry)
16
19
  return null;
17
20
  const { expiresAt, title, url } = cacheEntry;
18
21
  const downloadUrl = buildDownloadUrl(parts.namespace, parts.urlHash);
19
- const fileName = generateSafeFilename(url, title ?? options.title, parts.urlHash, resolveExtension(parts.namespace));
22
+ const fileName = resolveFilename(url, title ?? options.title, parts.urlHash, resolveExtension(parts.namespace));
20
23
  return { downloadUrl, fileName, expiresAt };
21
24
  }
22
25
  function buildDownloadUrl(namespace, hash) {
@@ -1,3 +1,3 @@
1
1
  import type { ToolErrorResponse } from '../config/types/tools.js';
2
- export declare function createToolErrorResponse(message: string, url: string, code: string): ToolErrorResponse;
3
- export declare function handleToolError(error: unknown, url: string, fallbackMessage?: string): ToolErrorResponse;
2
+ export declare function createToolErrorResponse(message: string, url: string, code: string, details?: Record<string, unknown>): ToolErrorResponse;
3
+ export declare function handleToolError(error: unknown, url: string, fallbackMessage?: string, details?: Record<string, unknown>): ToolErrorResponse;
@@ -22,12 +22,12 @@ function normalizeToolErrorCode(code) {
22
22
  return String(ErrorCode.InternalError);
23
23
  return MCP_ERROR_CODE_MAP[code] ?? code;
24
24
  }
25
- export function createToolErrorResponse(message, url, code) {
25
+ export function createToolErrorResponse(message, url, code, details = {}) {
26
26
  const structuredContent = {
27
+ ...details,
27
28
  error: message,
28
29
  url,
29
30
  errorCode: normalizeToolErrorCode(code),
30
- errorType: code,
31
31
  };
32
32
  return {
33
33
  content: [{ type: 'text', text: JSON.stringify(structuredContent) }],
@@ -42,19 +42,19 @@ function formatErrorMessage(baseMessage, error, fallback) {
42
42
  }
43
43
  return message;
44
44
  }
45
- export function handleToolError(error, url, fallbackMessage = 'Operation failed') {
45
+ export function handleToolError(error, url, fallbackMessage = 'Operation failed', details = {}) {
46
46
  if (isValidationError(error)) {
47
- return createToolErrorResponse(error.message, url, 'VALIDATION_ERROR');
47
+ return createToolErrorResponse(error.message, url, 'VALIDATION_ERROR', details);
48
48
  }
49
49
  if (error instanceof FetchError) {
50
50
  const message = formatErrorMessage(error.message, error);
51
- return createToolErrorResponse(message, url, error.code);
51
+ return createToolErrorResponse(message, url, error.code, details);
52
52
  }
53
53
  if (error instanceof Error) {
54
54
  const message = formatErrorMessage(error.message, error, fallbackMessage);
55
- return createToolErrorResponse(message, url, 'UNKNOWN_ERROR');
55
+ return createToolErrorResponse(message, url, 'UNKNOWN_ERROR', details);
56
56
  }
57
- return createToolErrorResponse(`${fallbackMessage}: Unknown error`, url, 'UNKNOWN_ERROR');
57
+ return createToolErrorResponse(`${fallbackMessage}: Unknown error`, url, 'UNKNOWN_ERROR', details);
58
58
  }
59
59
  function isValidationError(error) {
60
60
  return (error instanceof Error &&
@@ -32,6 +32,35 @@ for (const entry of BLOCKED_IPV6_SUBNETS) {
32
32
  BLOCK_LIST.addSubnet(entry.subnet, entry.prefix, 'ipv6');
33
33
  }
34
34
  const DNS_LOOKUP_TIMEOUT_MS = 5000;
35
+ const DNS_DECISION_TTL_MS = 60000;
36
+ const DNS_DECISION_MAX = 1000;
37
+ const dnsDecisionCache = new Map();
38
+ function getCachedDnsDecision(hostname) {
39
+ const cached = dnsDecisionCache.get(hostname);
40
+ if (!cached)
41
+ return null;
42
+ if (cached.expiresAt <= Date.now()) {
43
+ dnsDecisionCache.delete(hostname);
44
+ return null;
45
+ }
46
+ return cached;
47
+ }
48
+ function setCachedDnsDecision(hostname, ok) {
49
+ dnsDecisionCache.set(hostname, {
50
+ ok,
51
+ expiresAt: Date.now() + DNS_DECISION_TTL_MS,
52
+ });
53
+ if (dnsDecisionCache.size <= DNS_DECISION_MAX)
54
+ return;
55
+ const evictCount = Math.ceil(DNS_DECISION_MAX * 0.05);
56
+ const iterator = dnsDecisionCache.keys();
57
+ for (let i = 0; i < evictCount; i++) {
58
+ const { value, done } = iterator.next();
59
+ if (done)
60
+ break;
61
+ dnsDecisionCache.delete(value);
62
+ }
63
+ }
35
64
  function matchesBlockedIpPatterns(resolvedIp) {
36
65
  for (const pattern of config.security.blockedIpPatterns) {
37
66
  if (pattern.test(resolvedIp)) {
@@ -79,6 +108,13 @@ function lookupWithTimeout(hostname) {
79
108
  });
80
109
  }
81
110
  export async function assertResolvedAddressesAllowed(hostname) {
111
+ const cached = getCachedDnsDecision(hostname);
112
+ if (cached) {
113
+ if (!cached.ok) {
114
+ throw createValidationError(`Blocked IP range resolved from hostname: ${hostname}`);
115
+ }
116
+ return;
117
+ }
82
118
  try {
83
119
  const result = await lookupWithTimeout(hostname);
84
120
  const addresses = Array.isArray(result) ? result : [result];
@@ -87,9 +123,11 @@ export async function assertResolvedAddressesAllowed(hostname) {
87
123
  }
88
124
  for (const { address } of addresses) {
89
125
  if (isBlockedIp(address.toLowerCase())) {
126
+ setCachedDnsDecision(hostname, false);
90
127
  throw createValidationError(`Blocked IP range resolved from hostname: ${hostname}`);
91
128
  }
92
129
  }
130
+ setCachedDnsDecision(hostname, true);
93
131
  }
94
132
  catch (error) {
95
133
  const code = error?.code;
@@ -0,0 +1 @@
1
+ export {};
@@ -0,0 +1,50 @@
1
+ import { parentPort } from 'node:worker_threads';
2
+ import { transformHtmlToJsonl, transformHtmlToMarkdown, transformHtmlToMarkdownWithBlocks, } from '../tools/utils/content-transform.js';
3
+ function isTransformJob(value) {
4
+ if (!value || typeof value !== 'object')
5
+ return false;
6
+ const record = value;
7
+ return (typeof record.id === 'number' &&
8
+ typeof record.mode === 'string' &&
9
+ typeof record.html === 'string' &&
10
+ typeof record.url === 'string');
11
+ }
12
+ function resolveTransform(job) {
13
+ if (job.mode === 'markdown') {
14
+ return transformHtmlToMarkdown(job.html, job.url, job.options);
15
+ }
16
+ if (job.mode === 'markdown-blocks') {
17
+ return transformHtmlToMarkdownWithBlocks(job.html, job.url, {
18
+ ...job.options,
19
+ includeContentBlocks: job.options.includeContentBlocks ?? true,
20
+ });
21
+ }
22
+ return transformHtmlToJsonl(job.html, job.url, job.options);
23
+ }
24
+ function sendResponse(response) {
25
+ if (!parentPort)
26
+ return;
27
+ parentPort.postMessage(response);
28
+ }
29
+ function handleMessage(message) {
30
+ if (!isTransformJob(message)) {
31
+ sendResponse({
32
+ id: -1,
33
+ ok: false,
34
+ error: 'Invalid transform job payload',
35
+ });
36
+ return;
37
+ }
38
+ try {
39
+ const result = resolveTransform(message);
40
+ sendResponse({ id: message.id, ok: true, result });
41
+ }
42
+ catch (error) {
43
+ sendResponse({
44
+ id: message.id,
45
+ ok: false,
46
+ error: error instanceof Error ? error.message : String(error),
47
+ });
48
+ }
49
+ }
50
+ parentPort?.on('message', handleMessage);
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@j0hanz/superfetch",
3
- "version": "1.2.2",
3
+ "version": "1.2.3",
4
4
  "mcpName": "io.github.j0hanz/superfetch",
5
5
  "description": "Intelligent web content fetcher MCP server that converts HTML to clean, AI-readable JSONL format",
6
6
  "type": "module",
@@ -44,8 +44,8 @@
44
44
  "type-check": "tsc --noEmit",
45
45
  "lint": "eslint .",
46
46
  "lint:fix": "eslint . --fix",
47
- "test": "vitest run",
48
- "test:coverage": "vitest run --coverage",
47
+ "test": "npm run build --silent && node --test --experimental-transform-types",
48
+ "test:coverage": "npm run build --silent && node --test --experimental-transform-types --experimental-test-coverage",
49
49
  "bench": "npm run build && node scripts/bench.mjs",
50
50
  "knip": "knip",
51
51
  "knip:fix": "knip --fix"
@@ -67,7 +67,6 @@
67
67
  "@types/express": "^5.0.6",
68
68
  "@types/node": "^22.19.3",
69
69
  "@types/turndown": "^5.0.6",
70
- "@vitest/coverage-v8": "^2.1.9",
71
70
  "eslint": "^9.23.2",
72
71
  "eslint-config-prettier": "^10.1.8",
73
72
  "eslint-plugin-unused-imports": "^4.3.0",
@@ -76,8 +75,7 @@
76
75
  "shx": "^0.4.0",
77
76
  "tsx": "^4.21.0",
78
77
  "typescript": "^5.9.3",
79
- "typescript-eslint": "^8.51.0",
80
- "vitest": "^2.1.9"
78
+ "typescript-eslint": "^8.51.0"
81
79
  },
82
80
  "engines": {
83
81
  "node": ">=20.12.0"