microlink.io 0.1.1 → 0.5.0

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.
package/LICENSE.md CHANGED
File without changes
package/bin/config.js ADDED
@@ -0,0 +1,61 @@
1
+ 'use strict'
2
+
3
+ const { homedir } = require('os')
4
+ const { join } = require('path')
5
+ const {
6
+ mkdirSync,
7
+ readFileSync,
8
+ writeFileSync,
9
+ unlinkSync,
10
+ chmodSync
11
+ } = require('fs')
12
+
13
+ const configDir = () =>
14
+ join(process.env.XDG_CONFIG_HOME || join(homedir(), '.config'), 'microlink')
15
+
16
+ const configPath = () => join(configDir(), 'config.json')
17
+
18
+ const configPathDisplay = () => {
19
+ const file = configPath()
20
+ const home = homedir()
21
+ return file.startsWith(home + '/') ? `~${file.slice(home.length)}` : file
22
+ }
23
+
24
+ const readConfig = () => {
25
+ try {
26
+ return JSON.parse(readFileSync(configPath(), 'utf8'))
27
+ } catch {
28
+ return {}
29
+ }
30
+ }
31
+
32
+ const writeConfig = data => {
33
+ mkdirSync(configDir(), { recursive: true, mode: 0o700 })
34
+ writeFileSync(configPath(), JSON.stringify(data) + '\n', { mode: 0o600 })
35
+ chmodSync(configPath(), 0o600)
36
+ }
37
+
38
+ const clearConfig = () => {
39
+ try {
40
+ unlinkSync(configPath())
41
+ return true
42
+ } catch (error) {
43
+ if (error.code === 'ENOENT') return false
44
+ throw error
45
+ }
46
+ }
47
+
48
+ const readApiKey = () => {
49
+ const { apiKey } = readConfig()
50
+ return typeof apiKey === 'string' && apiKey ? apiKey : undefined
51
+ }
52
+
53
+ module.exports = {
54
+ configDir,
55
+ configPath,
56
+ configPathDisplay,
57
+ readConfig,
58
+ writeConfig,
59
+ clearConfig,
60
+ readApiKey
61
+ }
package/bin/help.js ADDED
@@ -0,0 +1,366 @@
1
+ 'use strict'
2
+
3
+ const { gray, white } = require('./style')
4
+
5
+ const col = (name, desc) => ` ${white(name.padEnd(21))} ${gray(desc)}`
6
+
7
+ const cmd = (rest, comment) =>
8
+ `${comment ? ` ${gray(`# ${comment}`)}\n` : ''} ${white(
9
+ 'microlink'
10
+ )} ${gray(rest)}`
11
+
12
+ const rows = items => items.map(([name, desc]) => col(name, desc)).join('\n')
13
+
14
+ const CLI = [
15
+ [
16
+ '--api-key',
17
+ 'Microlink API key (flag, MICROLINK_API_KEY, or `microlink login`)'
18
+ ],
19
+ ['--endpoint', 'Microlink API endpoint'],
20
+ ['--header, -H', "Extra request header as 'Name: value' (repeatable)"],
21
+ [
22
+ '--http.header.<name>',
23
+ 'HTTP request header (e.g. --http.header.authorization)'
24
+ ],
25
+ ['--trace', 'Print request & response payload (API key masked)'],
26
+ ['--trace-full', 'Same as --trace, including the full API key'],
27
+ ['--help', 'Show this help']
28
+ ]
29
+
30
+ const CLI_NO_TRACE = CLI.filter(
31
+ ([name]) => name !== '--trace' && name !== '--trace-full'
32
+ )
33
+
34
+ const BROWSER = [
35
+ ['--adblock', 'Block ads and trackers'],
36
+ ['--animations', 'Enable CSS animations'],
37
+ ['--cacheKey', 'Custom cache key'],
38
+ ['--click', 'CSS selector(s) to click before capture'],
39
+ ['--colorScheme', 'Color scheme: no-preference, light, dark'],
40
+ ['--device', "Emulate a device (e.g. 'iPhone 11')"],
41
+ ['--filename', 'Suggested download filename'],
42
+ ['--filter', 'Pick response fields'],
43
+ ['--force', 'Bypass the cache'],
44
+ ['--javascript', 'Enable or disable JavaScript'],
45
+ ['--mediaType', 'Emulate media: screen, print'],
46
+ ['--modules', 'Inject ES module URLs'],
47
+ ['--prerender', 'Prerender: auto, true, false'],
48
+ ['--proxy', 'Proxy URL or country'],
49
+ ['--retry', 'Retry count'],
50
+ ['--scripts', 'Inject script URLs'],
51
+ ['--scroll', 'CSS selector to scroll into view'],
52
+ ['--staleTtl', 'Stale-while-revalidate TTL'],
53
+ ['--styles', 'Inject stylesheet URLs'],
54
+ ['--timeout', 'Request timeout'],
55
+ ['--ttl', 'Cache TTL'],
56
+ ['--viewport', 'Viewport as JSON (width, height, ...)'],
57
+ ['--waitForSelector', 'Wait until a CSS selector matches'],
58
+ ['--waitForTimeout', 'Wait for a duration before continuing'],
59
+ ['--waitUntil', 'auto, load, domcontentloaded, networkidle0, networkidle2']
60
+ ]
61
+
62
+ const CONTENT = [
63
+ ['--selector', 'CSS selector to scope the extraction'],
64
+ ['--selectorAll', 'CSS selector(s) matching many nodes'],
65
+ ['--type', 'Cast the extracted value (url, image, ...)']
66
+ ]
67
+
68
+ const COLLECTION = [
69
+ ['--selector', 'CSS selector (single node)'],
70
+ ['--selectorAll', 'CSS selector(s) matching many nodes'],
71
+ ['--attr', 'Attribute to read (href, src, ...)'],
72
+ ['--type', 'Cast each value (url, image, email, ...)']
73
+ ]
74
+
75
+ const ALIAS = { run: 'function' }
76
+
77
+ const COMMANDS = {
78
+ login: {
79
+ usage: 'login',
80
+ desc: 'Save an API key from your Microlink account',
81
+ flags: [],
82
+ cli: [],
83
+ examples: [['login', 'open the dashboard and save an API key']]
84
+ },
85
+ logout: {
86
+ usage: 'logout',
87
+ desc: 'Remove the saved API key from this machine',
88
+ flags: [],
89
+ cli: [],
90
+ examples: [['logout', 'forget the saved API key']]
91
+ }
92
+ }
93
+
94
+ const content = (name, desc) => ({
95
+ usage: `${name} <url> [options]`,
96
+ desc,
97
+ flags: CONTENT,
98
+ browser: true,
99
+ examples: [[`${name} https://example.com`, desc]]
100
+ })
101
+
102
+ const collection = (name, desc) => ({
103
+ usage: `${name} <url> [options]`,
104
+ desc,
105
+ flags: COLLECTION,
106
+ browser: true,
107
+ examples: [[`${name} https://example.com`, desc]]
108
+ })
109
+
110
+ const PRODUCTS = {
111
+ metadata: {
112
+ usage: ['metadata <url> [options]', '<url> [options]'],
113
+ desc: 'Unified metadata (title, description, image, ...); default',
114
+ flags: [
115
+ ['--palette', 'Also extract dominant colors from images'],
116
+ ['--meta', 'Include metadata fields, or false to skip']
117
+ ],
118
+ browser: true,
119
+ examples: [
120
+ ['https://example.com', 'unified metadata (default)'],
121
+ [
122
+ 'https://example.com --trace',
123
+ 'print request & response, API key masked'
124
+ ]
125
+ ]
126
+ },
127
+ logo: {
128
+ usage: 'logo <url> [options]',
129
+ desc: 'Brand logo of the site (--square prefers the square variant)',
130
+ flags: [
131
+ ['--square', 'Prefer the square logo variant'],
132
+ ['--palette', 'Also extract dominant colors']
133
+ ],
134
+ browser: true,
135
+ examples: [['logo https://github.com --square', 'square brand logo']]
136
+ },
137
+ markdown: content('markdown', 'Page content as Markdown'),
138
+ html: content('html', 'Page content as HTML'),
139
+ text: content('text', 'Page content as plain text'),
140
+ video: {
141
+ usage: 'video <url> [options]',
142
+ desc: 'Primary video of the page (returns the asset object)',
143
+ flags: [['--meta', 'Include metadata fields, or false to skip']],
144
+ browser: true,
145
+ examples: [['video https://example.com', 'primary video asset']]
146
+ },
147
+ audio: {
148
+ usage: 'audio <url> [options]',
149
+ desc: 'Primary audio of the page (returns the asset object)',
150
+ flags: [['--meta', 'Include metadata fields, or false to skip']],
151
+ browser: true,
152
+ examples: [['audio https://example.com', 'primary audio asset']]
153
+ },
154
+ emails: collection('emails', 'Every email address present on the page'),
155
+ links: collection('links', 'Every absolute link URL on the page'),
156
+ images: collection('images', 'Every absolute image URL on the page'),
157
+ videos: collection('videos', 'Every absolute video URL on the page'),
158
+ audios: collection('audios', 'Every absolute audio URL on the page'),
159
+ extract: {
160
+ usage: 'extract <url> --data <json> [options]',
161
+ desc: 'Custom MQL data rules',
162
+ flags: [['--data', 'JSON data rules (e.g. --data \'{"field":{...}}\')']],
163
+ browser: true,
164
+ examples: [
165
+ [
166
+ 'extract https://microlink.io --data \'{"image":{"selector":"meta[property=og:image]","attr":"content","type":"image"}}\'',
167
+ 'extract og:image via MQL'
168
+ ]
169
+ ]
170
+ },
171
+ screenshot: {
172
+ usage: 'screenshot <url> [options]',
173
+ desc: 'Take a screenshot (returns the asset object)',
174
+ flags: [
175
+ ['--fullPage', 'Capture the full scrollable page'],
176
+ ['--type', 'Image format: png, jpeg'],
177
+ ['--element', 'CSS selector of the element to capture'],
178
+ ['--omitBackground', 'Transparent background (png)'],
179
+ ['--optimizeForSpeed', 'Faster encode, larger file'],
180
+ ['--overlay', 'Browser chrome overlay as JSON'],
181
+ ['--codeScheme', 'Syntax theme for code pages'],
182
+ ['--animated', 'Animated screenshot (GIF/MP4)'],
183
+ ['--palette', 'Also extract dominant colors'],
184
+ ['--quality', 'JPEG quality (0–100)']
185
+ ],
186
+ browser: true,
187
+ examples: [
188
+ ['screenshot https://example.com --fullPage', 'full-page screenshot']
189
+ ]
190
+ },
191
+ pdf: {
192
+ usage: 'pdf <url> [options]',
193
+ desc: 'Generate a PDF (returns the asset object)',
194
+ flags: [
195
+ ['--format', 'Page format: Letter, Legal, A4, ...'],
196
+ ['--margin', "Margin (e.g. '0.5cm' or JSON)"],
197
+ ['--scale', 'Scale (0.1–2)'],
198
+ ['--landscape', 'Landscape orientation'],
199
+ ['--pageRanges', "Pages to print (e.g. '1-3')"],
200
+ ['--width', 'Page width'],
201
+ ['--height', 'Page height'],
202
+ ['--printBackground', 'Print background graphics']
203
+ ],
204
+ browser: true,
205
+ examples: [['pdf https://example.com --format A4', 'A4 PDF']]
206
+ },
207
+ embed: {
208
+ usage: 'embed <url> [options]',
209
+ desc: 'oEmbed-style embeddable iframe ({ html, scripts })',
210
+ flags: [
211
+ ['--maxWidth', 'Maximum iframe width'],
212
+ ['--maxHeight', 'Maximum iframe height']
213
+ ],
214
+ browser: true,
215
+ examples: [['embed https://example.com', 'embeddable iframe']]
216
+ },
217
+ technologies: {
218
+ usage: 'technologies <url> [options]',
219
+ desc: 'Detect the tech stack behind the site',
220
+ flags: [],
221
+ browser: true,
222
+ examples: [['technologies https://example.com', 'detect the tech stack']]
223
+ },
224
+ lighthouse: {
225
+ usage: 'lighthouse <url> [options]',
226
+ desc: 'Run a Lighthouse report',
227
+ flags: [
228
+ ['--onlyCategories', 'Limit to these categories'],
229
+ ['--onlyAudits', 'Limit to these audits'],
230
+ ['--skipAudits', 'Skip these audits'],
231
+ ['--output', 'Report format: json, html, csv']
232
+ ],
233
+ browser: true,
234
+ examples: [['lighthouse https://example.com', 'run a Lighthouse report']]
235
+ },
236
+ search: {
237
+ usage: 'search <query> [options]',
238
+ desc: 'Google as structured data (query instead of url)',
239
+ flags: [
240
+ [
241
+ '--type',
242
+ 'news, images, videos, places, maps, shopping, scholar, patents, autocomplete'
243
+ ],
244
+ ['--limit', 'Maximum number of results'],
245
+ ['--page', 'Results page (1 default)'],
246
+ ['--html', 'Fetch HTML for the results page and each result'],
247
+ ['--markdown', 'Fetch Markdown for the results page and each result'],
248
+ ['--location', 'Country or locale (e.g. es)'],
249
+ ['--period', 'Recency: hour, day, week, month, year'],
250
+ ['--timeout', 'Request timeout']
251
+ ],
252
+ cli: CLI_NO_TRACE,
253
+ examples: [
254
+ [
255
+ 'search "best coffee" --limit 10 --location es',
256
+ 'Google results in Spain, limit 10'
257
+ ],
258
+ [
259
+ 'search "open source llm" --type news --period week',
260
+ 'news results from the past week'
261
+ ],
262
+ [
263
+ 'search "the matrix" --markdown',
264
+ 'include Markdown for the SERP and each result'
265
+ ],
266
+ ['search "the matrix" --page 2', 'second page of results']
267
+ ]
268
+ },
269
+ function: {
270
+ usage: 'function <url> --file <path> [options]',
271
+ desc: 'Run code remotely with browser access',
272
+ flags: [['--file', 'Path to the code file']],
273
+ browser: true,
274
+ cli: CLI_NO_TRACE,
275
+ note: 'Extra flags are injected as variables in the function scope.',
276
+ examples: [
277
+ [
278
+ 'function https://example.com --file ./fn.js',
279
+ 'run ./fn.js with browser access'
280
+ ]
281
+ ]
282
+ }
283
+ }
284
+
285
+ const productList = Object.entries(PRODUCTS)
286
+ .map(([name, product]) => col(name, product.desc))
287
+ .join('\n')
288
+
289
+ const commandList = Object.entries(COMMANDS)
290
+ .map(([name, command]) => col(name, command.desc))
291
+ .join('\n')
292
+
293
+ const global = `Usage
294
+ ${cmd('<url> [options]')}
295
+ ${cmd('<product> <url|query> [options]')}
296
+ ${cmd('login')}
297
+ ${cmd('logout')}
298
+
299
+ Commands
300
+ ${commandList}
301
+
302
+ Products
303
+ ${productList}
304
+
305
+ Options
306
+ ${rows(CLI)}
307
+
308
+ Examples
309
+ ${cmd('login', 'save an API key from your account')}
310
+ ${cmd('https://example.com', 'unified metadata (default)')}
311
+ ${cmd(
312
+ 'https://example.com --trace',
313
+ 'print request & response, API key masked'
314
+ )}
315
+ ${cmd(
316
+ 'https://example.com --trace-full',
317
+ 'same as --trace, including the full API key'
318
+ )}
319
+ ${cmd('markdown https://example.com', 'page content as Markdown')}
320
+ ${cmd('screenshot https://example.com --fullPage', 'full-page screenshot')}
321
+ ${cmd('logo https://github.com --square', 'square brand logo')}
322
+ ${cmd('links https://example.com', 'every absolute link on the page')}
323
+ ${cmd(
324
+ 'search "best coffee" --limit 10 --location es',
325
+ 'Google results in Spain, limit 10'
326
+ )}
327
+ ${cmd(
328
+ 'search "open source llm" --type news --period week',
329
+ 'news results from the past week'
330
+ )}
331
+ ${cmd(
332
+ 'extract https://microlink.io --data \'{"image":{"selector":"meta[property=og:image]","attr":"content","type":"image"}}\'',
333
+ 'extract og:image via MQL'
334
+ )}
335
+ ${cmd(
336
+ 'function https://example.com --file ./fn.js',
337
+ 'run ./fn.js with browser access'
338
+ )}
339
+ `
340
+
341
+ const render = (name, product) => {
342
+ const usage = []
343
+ .concat(product.usage)
344
+ .map(line => cmd(line))
345
+ .join('\n')
346
+ const cli = product.cli ?? CLI
347
+ const options = [...product.flags, ...cli]
348
+ const parts = ['Usage', usage, '', gray(product.desc)]
349
+ if (options.length > 0) parts.push('', 'Options', rows(options))
350
+ if (product.browser) parts.push('', 'Browser', rows(BROWSER))
351
+ if (product.note) parts.push('', gray(product.note))
352
+ if (product.examples) {
353
+ parts.push(
354
+ '',
355
+ 'Examples',
356
+ ...product.examples.map(([rest, comment]) => cmd(rest, comment))
357
+ )
358
+ }
359
+ return parts.join('\n') + '\n'
360
+ }
361
+
362
+ module.exports = command => {
363
+ const name = ALIAS[command] ?? command
364
+ if (COMMANDS[name]) return render(name, COMMANDS[name])
365
+ return PRODUCTS[name] ? render(name, PRODUCTS[name]) : global
366
+ }
package/bin/index.js CHANGED
@@ -1,10 +1,13 @@
1
1
  #!/usr/bin/env node
2
2
  'use strict'
3
3
 
4
- const { styleText } = require('node:util')
5
4
  const { readFileSync } = require('fs')
6
5
  const path = require('path')
7
6
  const mri = require('mri')
7
+ const helpText = require('./help')
8
+ const { readApiKey, clearConfig } = require('./config')
9
+ const login = require('./login')
10
+ const { gray, white, green, red, orange, link, styleText } = require('./style')
8
11
 
9
12
  const create = require('../src')
10
13
 
@@ -13,12 +16,8 @@ const HIDE_CURSOR = '\u001b[?25l'
13
16
  const CLEAR_LINE = '\r\u001b[K'
14
17
  const FRAMES = ['⠋', '⠙', '⠹', '⠸', '⠼', '⠴', '⠦', '⠧', '⠇', '⠏']
15
18
 
16
- const gray = str => styleText('gray', str)
17
- const white = str => styleText('white', str)
18
- const green = str => styleText('green', str)
19
- const red = str => styleText('red', str)
20
19
  const label = (text, color) =>
21
- styleText(['inverse', 'bold', color], ` ${text.toUpperCase()} `)
20
+ styleText(['inverse', 'bold'], color(` ${text.toUpperCase()} `))
22
21
  const keyValue = (key, value) => key + ' ' + gray(value)
23
22
 
24
23
  const prettyMs = ms => {
@@ -66,7 +65,9 @@ const printPretty = (value, indent = 0) => {
66
65
  if (typeof value !== 'object') return white(String(value))
67
66
 
68
67
  const isArray = Array.isArray(value)
69
- const keys = isArray ? value : Object.keys(value)
68
+ const keys = isArray
69
+ ? value.filter(item => typeof item !== 'function')
70
+ : Object.keys(value).filter(key => typeof value[key] !== 'function')
70
71
  if (keys.length === 0) return gray(isArray ? '[]' : '{}')
71
72
 
72
73
  const pad = ' '.repeat(indent)
@@ -98,8 +99,10 @@ const tracePayload = ({
98
99
  const rest = { ...requestOptions }
99
100
  delete rest.responseType
100
101
  const headers = { ...rest.headers }
101
- if (!full && headers['x-api-key']) {
102
- headers['x-api-key'] = humanizeApiKey(headers['x-api-key'])
102
+ if (!full) {
103
+ for (const key of ['x-api-key', 'authorization', 'cookie']) {
104
+ if (headers[key]) headers[key] = humanizeApiKey(headers[key])
105
+ }
103
106
  }
104
107
  return {
105
108
  request: { url: requestUrl, ...rest, headers },
@@ -168,7 +171,7 @@ const printFooter = ({ duration, response }) => {
168
171
 
169
172
  if (process.stdout.isTTY) console.error()
170
173
  console.error(
171
- label('success', 'green'),
174
+ label('success', green),
172
175
  gray(`${prettyBytes(size)} in ${time}`)
173
176
  )
174
177
  console.error()
@@ -188,37 +191,62 @@ const printFooter = ({ duration, response }) => {
188
191
  keyValue(green('mode'), `${fetchMode} ${gray(fetchTime)}`.trim())
189
192
  )
190
193
  }
191
- if (uri) console.error(' ', keyValue(green('uri'), uri))
194
+ if (uri) console.error(' ', keyValue(green('uri'), link(uri)))
192
195
  if (id) console.error(' ', keyValue(green('id'), id))
193
196
  }
194
197
 
198
+ const isClientError = statusCode => statusCode >= 400 && statusCode < 500
199
+
200
+ /**
201
+ * The API reports the actionable reason per field under `data`, keeping
202
+ * `message` as a generic pointer to it.
203
+ */
204
+ const reasons = error => {
205
+ const values = Object.values(error.data ?? {}).filter(
206
+ value => typeof value === 'string'
207
+ )
208
+ return values.length > 0
209
+ ? values
210
+ : [String(error.message).replace(`${error.code}, `, '')]
211
+ }
212
+
195
213
  const printFail = error => {
214
+ const color = isClientError(error.statusCode) ? orange : red
215
+ const status = error.status || 'fail'
216
+ const [reason, ...rest] = reasons(error)
217
+ const indent = ' '.repeat(status.length + 2)
196
218
  if (process.stdout.isTTY) console.error()
197
- console.error(
198
- label(error.status || 'fail', 'red'),
199
- gray(String(error.message).replace(`${error.code}, `, ''))
200
- )
219
+ console.error(label(status, color), gray(reason))
220
+ for (const extra of rest) console.error(indent, gray(extra))
201
221
  console.error()
202
222
  const id = error.headers?.['x-request-id']
203
- if (id) console.error(' ', keyValue(red('id'), id))
204
- if (error.url) console.error(' ', keyValue(red('uri'), error.url))
223
+ if (id) console.error(' ', keyValue(color('id'), id))
224
+ if (error.url) console.error(' ', keyValue(color('uri'), link(error.url)))
205
225
  if (error.code) {
206
226
  console.error(
207
227
  ' ',
208
228
  keyValue(
209
- red('code'),
229
+ color('code'),
210
230
  `${error.code}${error.statusCode ? ` (${error.statusCode})` : ''}`
211
231
  )
212
232
  )
213
233
  }
214
- if (error.more) console.error(' ', keyValue(red('more'), error.more))
234
+ if (error.more) console.error(' ', keyValue(color('more'), link(error.more)))
235
+ if (error.statusCode === 429) {
236
+ console.error(
237
+ ' ',
238
+ keyValue(color('hint'), 'run `microlink login` to use an API key')
239
+ )
240
+ }
215
241
  }
216
242
 
217
- const showHelp = () => {
218
- console.log(readFileSync(path.join(__dirname, 'help.txt'), 'utf8'))
243
+ const showHelp = command => {
244
+ console.log(helpText(command).trimEnd())
219
245
  process.exit(0)
220
246
  }
221
247
 
248
+ const HTTP_HEADER = 'http.header.'
249
+
222
250
  const parseHeaders = input => {
223
251
  const headers = {}
224
252
  for (const item of [].concat(input ?? [])) {
@@ -231,10 +259,24 @@ const parseHeaders = input => {
231
259
  return headers
232
260
  }
233
261
 
262
+ const takeHttpHeaders = flags => {
263
+ const headers = {}
264
+ for (const key of Object.keys(flags)) {
265
+ if (!key.startsWith(HTTP_HEADER)) continue
266
+ let value = flags[key]
267
+ delete flags[key]
268
+ if (Array.isArray(value)) value = value.at(-1)
269
+ if (typeof value !== 'string' && typeof value !== 'number') continue
270
+ const name = key.slice(HTTP_HEADER.length).toLowerCase()
271
+ if (name) headers[name] = String(value)
272
+ }
273
+ return headers
274
+ }
275
+
234
276
  const argv = mri(process.argv.slice(2), {
235
277
  alias: { H: 'header' },
236
- boolean: ['trace', 'trace-full'],
237
- string: ['header', 'api-key', 'data', 'file']
278
+ boolean: ['trace', 'trace-full', 'help', 'html', 'markdown'],
279
+ string: ['header', 'api-key', 'data', 'file', 'endpoint']
238
280
  })
239
281
 
240
282
  let {
@@ -245,70 +287,114 @@ let {
245
287
  file,
246
288
  'api-key': apiKeyFlag,
247
289
  apiKey: apiKeyCamel,
290
+ endpoint: endpointFlag,
248
291
  trace,
249
292
  'trace-full': traceFull,
293
+ html: htmlFlag,
294
+ markdown: markdownFlag,
250
295
  ...flags
251
296
  } = argv
252
297
 
253
298
  const isTrace = trace || traceFull
254
299
 
255
- if (help || !command) showHelp()
300
+ if (!command) showHelp()
301
+
302
+ if (command === 'login' || command === 'logout') {
303
+ if (help) showHelp(command)
304
+ if (command === 'logout') {
305
+ console.error(clearConfig() ? 'Logged out.' : 'Already logged out.')
306
+ process.exit(0)
307
+ }
308
+ login().then(
309
+ () => process.exit(0),
310
+ error => {
311
+ console.error(error.message)
312
+ process.exit(error.code === 'ABORT' ? 130 : 1)
313
+ }
314
+ )
315
+ } else {
316
+ const apiKey =
317
+ apiKeyFlag || apiKeyCamel || process.env.MICROLINK_API_KEY || readApiKey()
318
+ const endpoint = endpointFlag
319
+ const client = create({
320
+ ...(apiKey && { apiKey }),
321
+ ...(endpoint && { endpoint })
322
+ })
323
+
324
+ if (typeof client[command] !== 'function') {
325
+ if (!target && URL.canParse(command)) {
326
+ target = command
327
+ command = 'metadata'
328
+ } else if (help) {
329
+ showHelp()
330
+ } else {
331
+ console.error(
332
+ `Unknown command \`${command}\`. Run \`microlink --help\` to see the available commands.`
333
+ )
334
+ process.exit(1)
335
+ }
336
+ }
256
337
 
257
- const apiKey = apiKeyFlag || apiKeyCamel || process.env.MICROLINK_API_KEY
258
- const client = create(apiKey ? { apiKey } : {})
338
+ if (help || !target) showHelp(command)
259
339
 
260
- if (typeof client[command] !== 'function') {
261
- if (!target && URL.canParse(command)) {
262
- target = command
263
- command = 'metadata'
264
- } else {
265
- console.error(
266
- `Unknown command \`${command}\`. Run \`microlink --help\` to see the available commands.`
267
- )
340
+ if (
341
+ isTrace &&
342
+ (command === 'search' || command === 'function' || command === 'run')
343
+ ) {
344
+ console.error(`\`--trace\` is not supported for \`${command}\`.`)
268
345
  process.exit(1)
269
346
  }
270
- }
271
347
 
272
- if (
273
- isTrace &&
274
- (command === 'search' || command === 'function' || command === 'run')
275
- ) {
276
- console.error(`\`--trace\` is not supported for \`${command}\`.`)
277
- process.exit(1)
278
- }
348
+ const options = { ...flags }
349
+ const headers = { ...takeHttpHeaders(options), ...parseHeaders(header) }
350
+ if (Object.keys(headers).length > 0) options.headers = headers
279
351
 
280
- const options = { ...flags }
281
- const headers = parseHeaders(header)
282
- if (Object.keys(headers).length > 0) options.headers = headers
283
-
284
- const invoke = () => {
352
+ let rules
285
353
  if (command === 'extract') {
286
- return client.extract(target, JSON.parse(data), options)
287
- }
288
- if (command === 'function' || command === 'run') {
289
- const code = readFileSync(path.resolve(file), 'utf8')
290
- return client.function(target, code, options)
354
+ try {
355
+ rules = JSON.parse(data)
356
+ } catch {
357
+ printFail({ message: 'Invalid --data JSON' })
358
+ process.exit(1)
359
+ }
291
360
  }
292
- return client[command](target, options)
293
- }
294
361
 
295
- const spin = !isTrace && shouldSpin() ? spinner() : null
296
-
297
- ;(async () => {
298
- spin?.start()
299
- const started = Date.now()
300
- try {
301
- const result = await invoke()
302
- const duration = Date.now() - started
303
- spin?.stop()
304
- if (isTrace) printJson(tracePayload({ ...client.last, full: traceFull }))
305
- else if (typeof result === 'string') console.log(result)
306
- else printJson({ status: 'success', data: result })
307
- if (!isTrace) printFooter({ duration, response: client.last.response })
308
- process.exit(0)
309
- } catch (error) {
310
- spin?.stop()
311
- printFail(error)
312
- process.exit(1)
362
+ const invoke = () => {
363
+ if (command === 'extract') {
364
+ return client.extract(target, rules, options)
365
+ }
366
+ if (command === 'function' || command === 'run') {
367
+ const code = readFileSync(path.resolve(file), 'utf8')
368
+ return client.function(target, code, options)
369
+ }
370
+ if (command === 'search') {
371
+ return client.search(target, {
372
+ ...options,
373
+ ...(htmlFlag && { html: true }),
374
+ ...(markdownFlag && { markdown: true })
375
+ })
376
+ }
377
+ return client[command](target, options)
313
378
  }
314
- })()
379
+
380
+ const spin = !isTrace && shouldSpin() ? spinner() : null
381
+
382
+ ;(async () => {
383
+ spin?.start()
384
+ const started = Date.now()
385
+ try {
386
+ const result = await invoke()
387
+ const duration = Date.now() - started
388
+ spin?.stop()
389
+ if (isTrace) printJson(tracePayload({ ...client.last, full: traceFull }))
390
+ else if (typeof result === 'string') console.log(result)
391
+ else printJson({ status: 'success', data: result })
392
+ if (!isTrace) printFooter({ duration, response: client.last.response })
393
+ process.exit(0)
394
+ } catch (error) {
395
+ spin?.stop()
396
+ printFail(error)
397
+ process.exit(1)
398
+ }
399
+ })()
400
+ }
package/bin/login.js ADDED
@@ -0,0 +1,148 @@
1
+ 'use strict'
2
+
3
+ const { randomBytes } = require('crypto')
4
+ const { spawn } = require('child_process')
5
+ const http = require('http')
6
+ const { writeConfig, readApiKey, configPathDisplay } = require('./config')
7
+ const select = require('./select')
8
+ const { gray } = require('./style')
9
+
10
+ const TIMEOUT_MS = 5 * 60 * 1000
11
+
12
+ const dashboardUrl = () =>
13
+ process.env.MICROLINK_DASHBOARD_URL || 'https://dashboard.microlink.io'
14
+
15
+ const openUrl = url => {
16
+ const { platform } = process
17
+ const child =
18
+ platform === 'win32'
19
+ ? spawn('cmd', ['/c', 'start', '""', `"${url}"`], {
20
+ detached: true,
21
+ stdio: 'ignore',
22
+ windowsVerbatimArguments: true
23
+ })
24
+ : spawn(platform === 'darwin' ? 'open' : 'xdg-open', [url], {
25
+ detached: true,
26
+ stdio: 'ignore'
27
+ })
28
+ child.on('error', () => {})
29
+ child.unref()
30
+ }
31
+
32
+ const listen = state =>
33
+ new Promise((resolve, reject) => {
34
+ let settle
35
+ const token = new Promise((resolve, reject) => {
36
+ settle = { resolve, reject }
37
+ })
38
+
39
+ const cors = res => {
40
+ res.setHeader('Access-Control-Allow-Origin', '*')
41
+ res.setHeader('Access-Control-Allow-Methods', 'POST, OPTIONS')
42
+ res.setHeader('Access-Control-Allow-Headers', 'content-type')
43
+ res.setHeader('Access-Control-Allow-Private-Network', 'true')
44
+ }
45
+
46
+ const server = http.createServer((req, res) => {
47
+ cors(res)
48
+ if (req.method === 'OPTIONS') {
49
+ res.writeHead(204)
50
+ res.end()
51
+ return
52
+ }
53
+ if (req.method !== 'POST') {
54
+ res.writeHead(405)
55
+ res.end()
56
+ return
57
+ }
58
+ const chunks = []
59
+ req.on('data', chunk => chunks.push(chunk))
60
+ req.on('end', () => {
61
+ try {
62
+ const body = JSON.parse(Buffer.concat(chunks).toString())
63
+ if (body.state !== state || typeof body.token !== 'string') {
64
+ res.writeHead(400)
65
+ res.end()
66
+ return
67
+ }
68
+ res.writeHead(204)
69
+ res.end()
70
+ clearTimeout(timer)
71
+ settle.resolve(body.token)
72
+ } catch {
73
+ res.writeHead(400)
74
+ res.end()
75
+ }
76
+ })
77
+ })
78
+
79
+ const close = () => {
80
+ clearTimeout(timer)
81
+ server.close()
82
+ }
83
+
84
+ const timer = setTimeout(() => {
85
+ close()
86
+ settle.reject(new Error('Timed out waiting for dashboard authorization'))
87
+ }, TIMEOUT_MS)
88
+
89
+ server.listen(0, '127.0.0.1', () => {
90
+ resolve({ port: server.address().port, token, close })
91
+ })
92
+ server.on('error', reject)
93
+ })
94
+
95
+ const fetchKeys = async token => {
96
+ const res = await fetch(new URL('/api/v1/connect/keys', dashboardUrl()), {
97
+ headers: { authorization: `Bearer ${token}` }
98
+ })
99
+ if (!res.ok) {
100
+ throw new Error(`Could not load API keys (${res.status})`)
101
+ }
102
+ const body = await res.json()
103
+ return Array.isArray(body) ? body : body.keys
104
+ }
105
+
106
+ const asChoice = key => ({
107
+ name: key.label,
108
+ hint: key.maskedKey,
109
+ value: key.apiKey
110
+ })
111
+
112
+ const login = async () => {
113
+ const state = randomBytes(16).toString('hex')
114
+ const { port, token: tokenP, close } = await listen(state)
115
+ const url = new URL('/connect', dashboardUrl())
116
+ url.searchParams.set('port', String(port))
117
+ url.searchParams.set('state', state)
118
+ process.stderr.write(`Opening ${url}\n`)
119
+ openUrl(url.toString())
120
+
121
+ try {
122
+ const keys = await fetchKeys(await tokenP)
123
+ if (!keys?.length) {
124
+ throw new Error(
125
+ `No API keys on this account. Create a plan at ${dashboardUrl()}/plans`
126
+ )
127
+ }
128
+
129
+ const choices = keys.map(asChoice)
130
+ const picked =
131
+ choices.length === 1
132
+ ? choices[0]
133
+ : await select({
134
+ message: 'Which API key?',
135
+ choices,
136
+ current: readApiKey()
137
+ })
138
+
139
+ writeConfig({ apiKey: picked.value })
140
+ process.stderr.write(
141
+ `\n${gray('Saved')} ${picked.name} ${gray(`to ${configPathDisplay()}`)}\n`
142
+ )
143
+ } finally {
144
+ close()
145
+ }
146
+ }
147
+
148
+ module.exports = login
package/bin/select.js ADDED
@@ -0,0 +1,98 @@
1
+ 'use strict'
2
+
3
+ const readline = require('readline')
4
+ const { gray, white, styleText } = require('./style')
5
+
6
+ const HIDE = '\u001b[?25l'
7
+ const SHOW = '\u001b[?25h'
8
+ const UP_N = n => `\u001b[${n}A`
9
+
10
+ const label = (choice, current) => {
11
+ const mark = current && choice.value === current ? gray(' (current)') : ''
12
+ return `${choice.name} ${gray(`(${choice.hint})`)}${mark}`
13
+ }
14
+
15
+ const numbered = ({ message, choices, current }) =>
16
+ new Promise((resolve, reject) => {
17
+ for (const [i, choice] of choices.entries()) {
18
+ process.stderr.write(` ${i + 1}) ${label(choice, current)}\n`)
19
+ }
20
+ const rl = readline.createInterface({
21
+ input: process.stdin,
22
+ output: process.stderr
23
+ })
24
+ rl.question(`${message} `, answer => {
25
+ rl.close()
26
+ const choice = choices[Number.parseInt(answer, 10) - 1]
27
+ if (!choice) reject(new Error('Invalid selection'))
28
+ else resolve(choice)
29
+ })
30
+ })
31
+
32
+ const arrows = ({ message, choices, current }) => {
33
+ let index = Math.max(
34
+ 0,
35
+ choices.findIndex(choice => choice.value === current)
36
+ )
37
+ const lines = choices.length + 1
38
+
39
+ const draw = first => {
40
+ if (!first) process.stderr.write(UP_N(lines))
41
+ process.stderr.write(`${styleText('cyan', '?')} ${white(message)}\n`)
42
+ for (const [i, choice] of choices.entries()) {
43
+ const active = i === index
44
+ const prefix = active ? styleText('cyan', '❯') : ' '
45
+ const text = active
46
+ ? styleText('cyan', label(choice, current))
47
+ : label(choice, current)
48
+ process.stderr.write(`\u001b[2K${prefix} ${text}\n`)
49
+ }
50
+ }
51
+
52
+ return new Promise((resolve, reject) => {
53
+ readline.emitKeypressEvents(process.stdin)
54
+ process.stdin.setRawMode(true)
55
+ process.stderr.write(HIDE)
56
+ draw(true)
57
+
58
+ const cleanup = () => {
59
+ process.stdin.setRawMode(false)
60
+ process.stdin.off('keypress', onKey)
61
+ process.stderr.write(SHOW)
62
+ }
63
+
64
+ const onKey = (_str, key) => {
65
+ if (key.name === 'up') {
66
+ index = (index - 1 + choices.length) % choices.length
67
+ draw()
68
+ return
69
+ }
70
+ if (key.name === 'down') {
71
+ index = (index + 1) % choices.length
72
+ draw()
73
+ return
74
+ }
75
+ if (key.name === 'return') {
76
+ cleanup()
77
+ resolve(choices[index])
78
+ return
79
+ }
80
+ if (key.ctrl && key.name === 'c') {
81
+ cleanup()
82
+ reject(Object.assign(new Error('Aborted'), { code: 'ABORT' }))
83
+ }
84
+ }
85
+
86
+ process.stdin.on('keypress', onKey)
87
+ })
88
+ }
89
+
90
+ const select = ({ message, choices, current }) => {
91
+ if (choices.length === 1) return Promise.resolve(choices[0])
92
+ if (!process.stdin.isTTY || !process.stdout.isTTY) {
93
+ return numbered({ message, choices, current })
94
+ }
95
+ return arrows({ message, choices, current })
96
+ }
97
+
98
+ module.exports = select
package/bin/style.js ADDED
@@ -0,0 +1,21 @@
1
+ 'use strict'
2
+
3
+ const { styleText } = require('node:util')
4
+ const { default: terminalLink } = require('terminal-link')
5
+
6
+ const gray = str => styleText('gray', str)
7
+ const white = str => styleText('white', str)
8
+ const green = str => styleText('green', str)
9
+ const red = str => styleText('red', str)
10
+
11
+ const ORANGE_256 = '\u001b[38;5;208m'
12
+ const DEFAULT_FOREGROUND = '\u001b[39m'
13
+
14
+ const orange = str =>
15
+ process.stdout.hasColors?.()
16
+ ? `${ORANGE_256}${str}${DEFAULT_FOREGROUND}`
17
+ : String(str)
18
+
19
+ const link = url => terminalLink.stderr(url, url, { fallback: false })
20
+
21
+ module.exports = { gray, white, green, red, orange, link, styleText }
package/package.json CHANGED
@@ -2,7 +2,7 @@
2
2
  "name": "microlink.io",
3
3
  "description": "The Microlink API organized into products, each returning a direct result",
4
4
  "homepage": "https://github.com/microlinkhq/microlink",
5
- "version": "0.1.1",
5
+ "version": "0.5.0",
6
6
  "types": "./src/index.d.ts",
7
7
  "main": "./src/index.js",
8
8
  "exports": {
@@ -14,7 +14,8 @@
14
14
  }
15
15
  },
16
16
  "bin": {
17
- "microlink": "bin/index.js"
17
+ "microlink": "bin/index.js",
18
+ "microlink.io": "bin/index.js"
18
19
  },
19
20
  "author": {
20
21
  "email": "hello@microlink.io",
@@ -43,10 +44,11 @@
43
44
  "seo"
44
45
  ],
45
46
  "dependencies": {
46
- "@microlink/function": "0.3.0",
47
- "@microlink/google": "1.0.6",
48
- "@microlink/mql": "0.18.0",
49
- "mri": "~1.2.0"
47
+ "mri": "~1.2.0",
48
+ "terminal-link": "~5.0.0",
49
+ "@microlink/google": "1.2.0",
50
+ "@microlink/function": "0.3.2",
51
+ "@microlink/mql": "0.18.1"
50
52
  },
51
53
  "devDependencies": {
52
54
  "ava": "latest",
@@ -61,9 +63,6 @@
61
63
  "bin",
62
64
  "src"
63
65
  ],
64
- "scripts": {
65
- "test": "ava && tsd"
66
- },
67
66
  "preferGlobal": true,
68
67
  "license": "MIT",
69
68
  "ava": {
@@ -76,5 +75,7 @@
76
75
  },
77
76
  "directory": "test"
78
77
  },
79
- "gitHead": "a12fcd2a7bc918e741a71b50fd2cd93876f4ca8f"
80
- }
78
+ "scripts": {
79
+ "test": "ava && tsd"
80
+ }
81
+ }
package/bin/help.txt DELETED
@@ -1,53 +0,0 @@
1
- Usage
2
- $ microlink <url> [options]
3
- $ microlink <product> <url|query> [options]
4
-
5
- Products
6
- metadata Unified metadata (title, description, image, ...); default
7
- logo Brand logo of the site (--square prefers the square variant)
8
- markdown Page content as Markdown
9
- html Page content as HTML
10
- text Page content as plain text
11
- video Primary video of the page (returns the asset object)
12
- audio Primary audio of the page (returns the asset object)
13
- emails Every email address present on the page
14
- links Every absolute link URL on the page
15
- images Every absolute image URL on the page
16
- videos Every absolute video URL on the page
17
- audios Every absolute audio URL on the page
18
- extract Custom MQL data rules (--data '{"field":{...}}')
19
- screenshot Take a screenshot (returns the asset object)
20
- pdf Generate a PDF (returns the asset object)
21
- embed oEmbed-style embeddable iframe ({ html, scripts })
22
- technologies Detect the tech stack behind the site
23
- lighthouse Run a Lighthouse report
24
- search Google as structured data (query instead of url); --type
25
- routes to news, images, videos, places, maps, shopping,
26
- scholar, patents or autocomplete
27
- function Run code remotely with browser access (--file ./fn.js);
28
- extra flags are injected as variables in the function scope
29
-
30
- Options
31
- --api-key Microlink API key (defaults to MICROLINK_API_KEY env)
32
- --header, -H Extra request header as 'Name: value' (repeatable)
33
- --data JSON data rules for the extract command
34
- --file Path to the code file for the function command
35
- --trace Print request & response payload (API key masked)
36
- --trace-full Same as --trace, including the full API key
37
- --help Show this help
38
-
39
- Any other flag is passed as an option to the product, e.g. --fullPage,
40
- --device 'iPhone 11', --waitUntil networkidle0, --selector article.
41
-
42
- Examples
43
- $ microlink https://example.com
44
- $ microlink https://example.com --trace
45
- $ microlink https://example.com --trace-full
46
- $ microlink markdown https://example.com
47
- $ microlink screenshot https://example.com --fullPage
48
- $ microlink logo https://github.com --square
49
- $ microlink links https://example.com
50
- $ microlink search "best coffee" --limit 10 --location es
51
- $ microlink search "open source llm" --type news --period week
52
- $ microlink extract https://microlink.io --data '{"image":{"selector":"meta[property=og:image]","attr":"content","type":"image"}}'
53
- $ microlink function https://example.com --file ./fn.js