microlink.io 0.0.5 → 0.2.0
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/bin/help.js +334 -0
- package/bin/index.js +277 -20
- package/bin/style.js +10 -0
- package/package.json +2 -3
- package/src/index.js +25 -11
- package/bin/help.txt +0 -47
package/bin/help.js
ADDED
|
@@ -0,0 +1,334 @@
|
|
|
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
|
+
['--api-key', 'Microlink API key (defaults to MICROLINK_API_KEY env)'],
|
|
16
|
+
['--endpoint', 'Microlink API endpoint'],
|
|
17
|
+
['--header, -H', "Extra request header as 'Name: value' (repeatable)"],
|
|
18
|
+
[
|
|
19
|
+
'--http.header.<name>',
|
|
20
|
+
'HTTP request header (e.g. --http.header.authorization)'
|
|
21
|
+
],
|
|
22
|
+
['--trace', 'Print request & response payload (API key masked)'],
|
|
23
|
+
['--trace-full', 'Same as --trace, including the full API key'],
|
|
24
|
+
['--help', 'Show this help']
|
|
25
|
+
]
|
|
26
|
+
|
|
27
|
+
const CLI_NO_TRACE = CLI.filter(
|
|
28
|
+
([name]) => name !== '--trace' && name !== '--trace-full'
|
|
29
|
+
)
|
|
30
|
+
|
|
31
|
+
const BROWSER = [
|
|
32
|
+
['--adblock', 'Block ads and trackers'],
|
|
33
|
+
['--animations', 'Enable CSS animations'],
|
|
34
|
+
['--cacheKey', 'Custom cache key'],
|
|
35
|
+
['--click', 'CSS selector(s) to click before capture'],
|
|
36
|
+
['--colorScheme', 'Color scheme: no-preference, light, dark'],
|
|
37
|
+
['--device', "Emulate a device (e.g. 'iPhone 11')"],
|
|
38
|
+
['--filename', 'Suggested download filename'],
|
|
39
|
+
['--filter', 'Pick response fields'],
|
|
40
|
+
['--force', 'Bypass the cache'],
|
|
41
|
+
['--javascript', 'Enable or disable JavaScript'],
|
|
42
|
+
['--mediaType', 'Emulate media: screen, print'],
|
|
43
|
+
['--modules', 'Inject ES module URLs'],
|
|
44
|
+
['--prerender', 'Prerender: auto, true, false'],
|
|
45
|
+
['--proxy', 'Proxy URL or country'],
|
|
46
|
+
['--retry', 'Retry count'],
|
|
47
|
+
['--scripts', 'Inject script URLs'],
|
|
48
|
+
['--scroll', 'CSS selector to scroll into view'],
|
|
49
|
+
['--staleTtl', 'Stale-while-revalidate TTL'],
|
|
50
|
+
['--styles', 'Inject stylesheet URLs'],
|
|
51
|
+
['--timeout', 'Request timeout'],
|
|
52
|
+
['--ttl', 'Cache TTL'],
|
|
53
|
+
['--viewport', 'Viewport as JSON (width, height, ...)'],
|
|
54
|
+
['--waitForSelector', 'Wait until a CSS selector matches'],
|
|
55
|
+
['--waitForTimeout', 'Wait for a duration before continuing'],
|
|
56
|
+
['--waitUntil', 'auto, load, domcontentloaded, networkidle0, networkidle2']
|
|
57
|
+
]
|
|
58
|
+
|
|
59
|
+
const CONTENT = [
|
|
60
|
+
['--selector', 'CSS selector to scope the extraction'],
|
|
61
|
+
['--selectorAll', 'CSS selector(s) matching many nodes'],
|
|
62
|
+
['--type', 'Cast the extracted value (url, image, ...)']
|
|
63
|
+
]
|
|
64
|
+
|
|
65
|
+
const COLLECTION = [
|
|
66
|
+
['--selector', 'CSS selector (single node)'],
|
|
67
|
+
['--selectorAll', 'CSS selector(s) matching many nodes'],
|
|
68
|
+
['--attr', 'Attribute to read (href, src, ...)'],
|
|
69
|
+
['--type', 'Cast each value (url, image, email, ...)']
|
|
70
|
+
]
|
|
71
|
+
|
|
72
|
+
const ALIAS = { run: 'function' }
|
|
73
|
+
|
|
74
|
+
const content = (name, desc) => ({
|
|
75
|
+
usage: `${name} <url> [options]`,
|
|
76
|
+
desc,
|
|
77
|
+
flags: CONTENT,
|
|
78
|
+
browser: true,
|
|
79
|
+
examples: [[`${name} https://example.com`, desc]]
|
|
80
|
+
})
|
|
81
|
+
|
|
82
|
+
const collection = (name, desc) => ({
|
|
83
|
+
usage: `${name} <url> [options]`,
|
|
84
|
+
desc,
|
|
85
|
+
flags: COLLECTION,
|
|
86
|
+
browser: true,
|
|
87
|
+
examples: [[`${name} https://example.com`, desc]]
|
|
88
|
+
})
|
|
89
|
+
|
|
90
|
+
const PRODUCTS = {
|
|
91
|
+
metadata: {
|
|
92
|
+
usage: ['metadata <url> [options]', '<url> [options]'],
|
|
93
|
+
desc: 'Unified metadata (title, description, image, ...); default',
|
|
94
|
+
flags: [
|
|
95
|
+
['--palette', 'Also extract dominant colors from images'],
|
|
96
|
+
['--meta', 'Include metadata fields, or false to skip']
|
|
97
|
+
],
|
|
98
|
+
browser: true,
|
|
99
|
+
examples: [
|
|
100
|
+
['https://example.com', 'unified metadata (default)'],
|
|
101
|
+
[
|
|
102
|
+
'https://example.com --trace',
|
|
103
|
+
'print request & response, API key masked'
|
|
104
|
+
]
|
|
105
|
+
]
|
|
106
|
+
},
|
|
107
|
+
logo: {
|
|
108
|
+
usage: 'logo <url> [options]',
|
|
109
|
+
desc: 'Brand logo of the site (--square prefers the square variant)',
|
|
110
|
+
flags: [
|
|
111
|
+
['--square', 'Prefer the square logo variant'],
|
|
112
|
+
['--palette', 'Also extract dominant colors']
|
|
113
|
+
],
|
|
114
|
+
browser: true,
|
|
115
|
+
examples: [['logo https://github.com --square', 'square brand logo']]
|
|
116
|
+
},
|
|
117
|
+
markdown: content('markdown', 'Page content as Markdown'),
|
|
118
|
+
html: content('html', 'Page content as HTML'),
|
|
119
|
+
text: content('text', 'Page content as plain text'),
|
|
120
|
+
video: {
|
|
121
|
+
usage: 'video <url> [options]',
|
|
122
|
+
desc: 'Primary video of the page (returns the asset object)',
|
|
123
|
+
flags: [['--meta', 'Include metadata fields, or false to skip']],
|
|
124
|
+
browser: true,
|
|
125
|
+
examples: [['video https://example.com', 'primary video asset']]
|
|
126
|
+
},
|
|
127
|
+
audio: {
|
|
128
|
+
usage: 'audio <url> [options]',
|
|
129
|
+
desc: 'Primary audio of the page (returns the asset object)',
|
|
130
|
+
flags: [['--meta', 'Include metadata fields, or false to skip']],
|
|
131
|
+
browser: true,
|
|
132
|
+
examples: [['audio https://example.com', 'primary audio asset']]
|
|
133
|
+
},
|
|
134
|
+
emails: collection('emails', 'Every email address present on the page'),
|
|
135
|
+
links: collection('links', 'Every absolute link URL on the page'),
|
|
136
|
+
images: collection('images', 'Every absolute image URL on the page'),
|
|
137
|
+
videos: collection('videos', 'Every absolute video URL on the page'),
|
|
138
|
+
audios: collection('audios', 'Every absolute audio URL on the page'),
|
|
139
|
+
extract: {
|
|
140
|
+
usage: 'extract <url> --data <json> [options]',
|
|
141
|
+
desc: 'Custom MQL data rules',
|
|
142
|
+
flags: [['--data', 'JSON data rules (e.g. --data \'{"field":{...}}\')']],
|
|
143
|
+
browser: true,
|
|
144
|
+
examples: [
|
|
145
|
+
[
|
|
146
|
+
'extract https://microlink.io --data \'{"image":{"selector":"meta[property=og:image]","attr":"content","type":"image"}}\'',
|
|
147
|
+
'extract og:image via MQL'
|
|
148
|
+
]
|
|
149
|
+
]
|
|
150
|
+
},
|
|
151
|
+
screenshot: {
|
|
152
|
+
usage: 'screenshot <url> [options]',
|
|
153
|
+
desc: 'Take a screenshot (returns the asset object)',
|
|
154
|
+
flags: [
|
|
155
|
+
['--fullPage', 'Capture the full scrollable page'],
|
|
156
|
+
['--type', 'Image format: png, jpeg'],
|
|
157
|
+
['--element', 'CSS selector of the element to capture'],
|
|
158
|
+
['--omitBackground', 'Transparent background (png)'],
|
|
159
|
+
['--optimizeForSpeed', 'Faster encode, larger file'],
|
|
160
|
+
['--overlay', 'Browser chrome overlay as JSON'],
|
|
161
|
+
['--codeScheme', 'Syntax theme for code pages'],
|
|
162
|
+
['--animated', 'Animated screenshot (GIF/MP4)'],
|
|
163
|
+
['--palette', 'Also extract dominant colors'],
|
|
164
|
+
['--quality', 'JPEG quality (0–100)']
|
|
165
|
+
],
|
|
166
|
+
browser: true,
|
|
167
|
+
examples: [
|
|
168
|
+
['screenshot https://example.com --fullPage', 'full-page screenshot']
|
|
169
|
+
]
|
|
170
|
+
},
|
|
171
|
+
pdf: {
|
|
172
|
+
usage: 'pdf <url> [options]',
|
|
173
|
+
desc: 'Generate a PDF (returns the asset object)',
|
|
174
|
+
flags: [
|
|
175
|
+
['--format', 'Page format: Letter, Legal, A4, ...'],
|
|
176
|
+
['--margin', "Margin (e.g. '0.5cm' or JSON)"],
|
|
177
|
+
['--scale', 'Scale (0.1–2)'],
|
|
178
|
+
['--landscape', 'Landscape orientation'],
|
|
179
|
+
['--pageRanges', "Pages to print (e.g. '1-3')"],
|
|
180
|
+
['--width', 'Page width'],
|
|
181
|
+
['--height', 'Page height'],
|
|
182
|
+
['--printBackground', 'Print background graphics']
|
|
183
|
+
],
|
|
184
|
+
browser: true,
|
|
185
|
+
examples: [['pdf https://example.com --format A4', 'A4 PDF']]
|
|
186
|
+
},
|
|
187
|
+
embed: {
|
|
188
|
+
usage: 'embed <url> [options]',
|
|
189
|
+
desc: 'oEmbed-style embeddable iframe ({ html, scripts })',
|
|
190
|
+
flags: [
|
|
191
|
+
['--maxWidth', 'Maximum iframe width'],
|
|
192
|
+
['--maxHeight', 'Maximum iframe height']
|
|
193
|
+
],
|
|
194
|
+
browser: true,
|
|
195
|
+
examples: [['embed https://example.com', 'embeddable iframe']]
|
|
196
|
+
},
|
|
197
|
+
technologies: {
|
|
198
|
+
usage: 'technologies <url> [options]',
|
|
199
|
+
desc: 'Detect the tech stack behind the site',
|
|
200
|
+
flags: [],
|
|
201
|
+
browser: true,
|
|
202
|
+
examples: [['technologies https://example.com', 'detect the tech stack']]
|
|
203
|
+
},
|
|
204
|
+
lighthouse: {
|
|
205
|
+
usage: 'lighthouse <url> [options]',
|
|
206
|
+
desc: 'Run a Lighthouse report',
|
|
207
|
+
flags: [
|
|
208
|
+
['--onlyCategories', 'Limit to these categories'],
|
|
209
|
+
['--onlyAudits', 'Limit to these audits'],
|
|
210
|
+
['--skipAudits', 'Skip these audits'],
|
|
211
|
+
['--output', 'Report format: json, html, csv']
|
|
212
|
+
],
|
|
213
|
+
browser: true,
|
|
214
|
+
examples: [['lighthouse https://example.com', 'run a Lighthouse report']]
|
|
215
|
+
},
|
|
216
|
+
search: {
|
|
217
|
+
usage: 'search <query> [options]',
|
|
218
|
+
desc: 'Google as structured data (query instead of url)',
|
|
219
|
+
flags: [
|
|
220
|
+
[
|
|
221
|
+
'--type',
|
|
222
|
+
'news, images, videos, places, maps, shopping, scholar, patents, autocomplete'
|
|
223
|
+
],
|
|
224
|
+
['--limit', 'Maximum number of results'],
|
|
225
|
+
['--location', 'Country or locale (e.g. es)'],
|
|
226
|
+
['--period', 'Recency: hour, day, week, month, year'],
|
|
227
|
+
['--timeout', 'Request timeout']
|
|
228
|
+
],
|
|
229
|
+
cli: CLI_NO_TRACE,
|
|
230
|
+
examples: [
|
|
231
|
+
[
|
|
232
|
+
'search "best coffee" --limit 10 --location es',
|
|
233
|
+
'Google results in Spain, limit 10'
|
|
234
|
+
],
|
|
235
|
+
[
|
|
236
|
+
'search "open source llm" --type news --period week',
|
|
237
|
+
'news results from the past week'
|
|
238
|
+
]
|
|
239
|
+
]
|
|
240
|
+
},
|
|
241
|
+
function: {
|
|
242
|
+
usage: 'function <url> --file <path> [options]',
|
|
243
|
+
desc: 'Run code remotely with browser access',
|
|
244
|
+
flags: [['--file', 'Path to the code file']],
|
|
245
|
+
browser: true,
|
|
246
|
+
cli: CLI_NO_TRACE,
|
|
247
|
+
note: 'Extra flags are injected as variables in the function scope.',
|
|
248
|
+
examples: [
|
|
249
|
+
[
|
|
250
|
+
'function https://example.com --file ./fn.js',
|
|
251
|
+
'run ./fn.js with browser access'
|
|
252
|
+
]
|
|
253
|
+
]
|
|
254
|
+
}
|
|
255
|
+
}
|
|
256
|
+
|
|
257
|
+
const productList = Object.entries(PRODUCTS)
|
|
258
|
+
.map(([name, product]) => col(name, product.desc))
|
|
259
|
+
.join('\n')
|
|
260
|
+
|
|
261
|
+
const global = `Usage
|
|
262
|
+
${cmd('<url> [options]')}
|
|
263
|
+
${cmd('<product> <url|query> [options]')}
|
|
264
|
+
|
|
265
|
+
Products
|
|
266
|
+
${productList}
|
|
267
|
+
|
|
268
|
+
Options
|
|
269
|
+
${rows(CLI)}
|
|
270
|
+
|
|
271
|
+
Examples
|
|
272
|
+
${cmd('https://example.com', 'unified metadata (default)')}
|
|
273
|
+
${cmd(
|
|
274
|
+
'https://example.com --trace',
|
|
275
|
+
'print request & response, API key masked'
|
|
276
|
+
)}
|
|
277
|
+
${cmd(
|
|
278
|
+
'https://example.com --trace-full',
|
|
279
|
+
'same as --trace, including the full API key'
|
|
280
|
+
)}
|
|
281
|
+
${cmd('markdown https://example.com', 'page content as Markdown')}
|
|
282
|
+
${cmd('screenshot https://example.com --fullPage', 'full-page screenshot')}
|
|
283
|
+
${cmd('logo https://github.com --square', 'square brand logo')}
|
|
284
|
+
${cmd('links https://example.com', 'every absolute link on the page')}
|
|
285
|
+
${cmd(
|
|
286
|
+
'search "best coffee" --limit 10 --location es',
|
|
287
|
+
'Google results in Spain, limit 10'
|
|
288
|
+
)}
|
|
289
|
+
${cmd(
|
|
290
|
+
'search "open source llm" --type news --period week',
|
|
291
|
+
'news results from the past week'
|
|
292
|
+
)}
|
|
293
|
+
${cmd(
|
|
294
|
+
'extract https://microlink.io --data \'{"image":{"selector":"meta[property=og:image]","attr":"content","type":"image"}}\'',
|
|
295
|
+
'extract og:image via MQL'
|
|
296
|
+
)}
|
|
297
|
+
${cmd(
|
|
298
|
+
'function https://example.com --file ./fn.js',
|
|
299
|
+
'run ./fn.js with browser access'
|
|
300
|
+
)}
|
|
301
|
+
`
|
|
302
|
+
|
|
303
|
+
const render = (name, product) => {
|
|
304
|
+
const usage = []
|
|
305
|
+
.concat(product.usage)
|
|
306
|
+
.map(line => cmd(line))
|
|
307
|
+
.join('\n')
|
|
308
|
+
const cli = product.cli ?? CLI
|
|
309
|
+
const options = [...product.flags, ...cli]
|
|
310
|
+
const parts = [
|
|
311
|
+
'Usage',
|
|
312
|
+
usage,
|
|
313
|
+
'',
|
|
314
|
+
gray(product.desc),
|
|
315
|
+
'',
|
|
316
|
+
'Options',
|
|
317
|
+
rows(options)
|
|
318
|
+
]
|
|
319
|
+
if (product.browser) parts.push('', 'Browser', rows(BROWSER))
|
|
320
|
+
if (product.note) parts.push('', gray(product.note))
|
|
321
|
+
if (product.examples) {
|
|
322
|
+
parts.push(
|
|
323
|
+
'',
|
|
324
|
+
'Examples',
|
|
325
|
+
...product.examples.map(([rest, comment]) => cmd(rest, comment))
|
|
326
|
+
)
|
|
327
|
+
}
|
|
328
|
+
return parts.join('\n') + '\n'
|
|
329
|
+
}
|
|
330
|
+
|
|
331
|
+
module.exports = command => {
|
|
332
|
+
const name = ALIAS[command] ?? command
|
|
333
|
+
return PRODUCTS[name] ? render(name, PRODUCTS[name]) : global
|
|
334
|
+
}
|
package/bin/index.js
CHANGED
|
@@ -3,16 +3,223 @@
|
|
|
3
3
|
|
|
4
4
|
const { readFileSync } = require('fs')
|
|
5
5
|
const path = require('path')
|
|
6
|
-
const jsome = require('jsome')
|
|
7
6
|
const mri = require('mri')
|
|
7
|
+
const helpText = require('./help')
|
|
8
|
+
const { gray, white, green, red, styleText } = require('./style')
|
|
8
9
|
|
|
9
10
|
const create = require('../src')
|
|
10
11
|
|
|
11
|
-
const
|
|
12
|
-
|
|
12
|
+
const SHOW_CURSOR = '\u001b[?25h'
|
|
13
|
+
const HIDE_CURSOR = '\u001b[?25l'
|
|
14
|
+
const CLEAR_LINE = '\r\u001b[K'
|
|
15
|
+
const FRAMES = ['⠋', '⠙', '⠹', '⠸', '⠼', '⠴', '⠦', '⠧', '⠇', '⠏']
|
|
16
|
+
|
|
17
|
+
const label = (text, color) =>
|
|
18
|
+
styleText(['inverse', 'bold', color], ` ${text.toUpperCase()} `)
|
|
19
|
+
const keyValue = (key, value) => key + ' ' + gray(value)
|
|
20
|
+
|
|
21
|
+
const prettyMs = ms => {
|
|
22
|
+
if (!Number.isFinite(ms)) return 'unknown'
|
|
23
|
+
const sign = ms < 0 ? '-' : ''
|
|
24
|
+
let n = Math.abs(ms)
|
|
25
|
+
if (n < 1000) return `${sign}${Math.round(n)}ms`
|
|
26
|
+
n /= 1000
|
|
27
|
+
if (n < 60) return `${sign}${n.toFixed(1).replace(/\.0$/, '')}s`
|
|
28
|
+
const hours = Math.floor(n / 3600)
|
|
29
|
+
n %= 3600
|
|
30
|
+
const mins = Math.floor(n / 60)
|
|
31
|
+
const secs = (n % 60).toFixed(1).replace(/\.0$/, '')
|
|
32
|
+
if (hours) return `${sign}${hours}h ${mins}m ${secs}s`
|
|
33
|
+
return secs === '0' ? `${sign}${mins}m` : `${sign}${mins}m ${secs}s`
|
|
34
|
+
}
|
|
35
|
+
|
|
36
|
+
const prettyBytes = n => {
|
|
37
|
+
if (!Number.isFinite(n) || n < 1000) return `${Math.round(n || 0)} B`
|
|
38
|
+
if (n < 1e6) {
|
|
39
|
+
const val = n / 1000
|
|
40
|
+
return `${
|
|
41
|
+
val >= 100 ? Math.round(val) : val.toFixed(1).replace(/\.0$/, '')
|
|
42
|
+
} kB`
|
|
43
|
+
}
|
|
44
|
+
return `${(n / 1e6).toFixed(1).replace(/\.0$/, '')} MB`
|
|
45
|
+
}
|
|
46
|
+
|
|
47
|
+
const toPlainHeaders = headers => {
|
|
48
|
+
if (!headers) return {}
|
|
49
|
+
if (typeof headers.entries === 'function') {
|
|
50
|
+
return Object.fromEntries(headers.entries())
|
|
51
|
+
}
|
|
52
|
+
return headers
|
|
53
|
+
}
|
|
54
|
+
|
|
55
|
+
const humanizeApiKey = apiKey => `${String(apiKey).slice(0, 5)}…`
|
|
56
|
+
|
|
57
|
+
const quote = str =>
|
|
58
|
+
gray('"') + white(JSON.stringify(str).slice(1, -1)) + gray('"')
|
|
59
|
+
|
|
60
|
+
const printPretty = (value, indent = 0) => {
|
|
61
|
+
if (value === null) return white('null')
|
|
62
|
+
if (typeof value === 'string') return quote(value)
|
|
63
|
+
if (typeof value !== 'object') return white(String(value))
|
|
64
|
+
|
|
65
|
+
const isArray = Array.isArray(value)
|
|
66
|
+
const keys = isArray ? value : Object.keys(value)
|
|
67
|
+
if (keys.length === 0) return gray(isArray ? '[]' : '{}')
|
|
68
|
+
|
|
69
|
+
const pad = ' '.repeat(indent)
|
|
70
|
+
const inner = ' '.repeat(indent + 1)
|
|
71
|
+
const open = gray(isArray ? '[' : '{')
|
|
72
|
+
const close = gray(isArray ? ']' : '}')
|
|
73
|
+
const lines = keys.map(key => {
|
|
74
|
+
if (isArray) return inner + printPretty(key, indent + 1)
|
|
75
|
+
const name = /^[A-Za-z_$][\w$]*$/.test(key) ? white(key) : quote(key)
|
|
76
|
+
return inner + name + gray(':') + ' ' + printPretty(value[key], indent + 1)
|
|
77
|
+
})
|
|
78
|
+
return open + '\n' + lines.join(gray(',') + '\n') + '\n' + pad + close
|
|
79
|
+
}
|
|
80
|
+
|
|
81
|
+
const printJson = payload => {
|
|
82
|
+
console.log(
|
|
83
|
+
process.stdout.hasColors?.()
|
|
84
|
+
? printPretty(payload)
|
|
85
|
+
: JSON.stringify(payload, null, 2)
|
|
86
|
+
)
|
|
87
|
+
}
|
|
88
|
+
|
|
89
|
+
const tracePayload = ({
|
|
90
|
+
requestUrl,
|
|
91
|
+
requestOptions = {},
|
|
92
|
+
response,
|
|
93
|
+
full = false
|
|
94
|
+
}) => {
|
|
95
|
+
const rest = { ...requestOptions }
|
|
96
|
+
delete rest.responseType
|
|
97
|
+
const headers = { ...rest.headers }
|
|
98
|
+
if (!full) {
|
|
99
|
+
for (const key of ['x-api-key', 'authorization', 'cookie']) {
|
|
100
|
+
if (headers[key]) headers[key] = humanizeApiKey(headers[key])
|
|
101
|
+
}
|
|
102
|
+
}
|
|
103
|
+
return {
|
|
104
|
+
request: { url: requestUrl, ...rest, headers },
|
|
105
|
+
response: {
|
|
106
|
+
...response,
|
|
107
|
+
headers: toPlainHeaders(response?.headers)
|
|
108
|
+
}
|
|
109
|
+
}
|
|
110
|
+
}
|
|
111
|
+
|
|
112
|
+
const shouldSpin = () =>
|
|
113
|
+
!process.env.NO_COLOR &&
|
|
114
|
+
process.env.FORCE_COLOR !== '0' &&
|
|
115
|
+
Boolean(process.stdout?.hasColors?.())
|
|
116
|
+
|
|
117
|
+
const spinner = () => {
|
|
118
|
+
const now = Date.now()
|
|
119
|
+
let i = 0
|
|
120
|
+
let timer
|
|
121
|
+
const draw = () => {
|
|
122
|
+
process.stderr.write(
|
|
123
|
+
`${CLEAR_LINE}${FRAMES[i++ % FRAMES.length]} ${prettyMs(
|
|
124
|
+
Date.now() - now
|
|
125
|
+
)}`
|
|
126
|
+
)
|
|
127
|
+
}
|
|
128
|
+
return {
|
|
129
|
+
start () {
|
|
130
|
+
process.stderr.write(HIDE_CURSOR)
|
|
131
|
+
draw()
|
|
132
|
+
process.on('SIGINT', () => {
|
|
133
|
+
process.stderr.write(CLEAR_LINE + SHOW_CURSOR)
|
|
134
|
+
process.exit(130)
|
|
135
|
+
})
|
|
136
|
+
timer = setInterval(draw, 50)
|
|
137
|
+
},
|
|
138
|
+
stop () {
|
|
139
|
+
clearInterval(timer)
|
|
140
|
+
process.stderr.write(CLEAR_LINE + SHOW_CURSOR)
|
|
141
|
+
}
|
|
142
|
+
}
|
|
143
|
+
}
|
|
144
|
+
|
|
145
|
+
const printFooter = ({ duration, response }) => {
|
|
146
|
+
const headers = toPlainHeaders(response?.headers)
|
|
147
|
+
const time = prettyMs(duration)
|
|
148
|
+
const size = Number(headers['content-length']) || 0
|
|
149
|
+
const serverTiming = headers['server-timing']
|
|
150
|
+
const id = headers['x-request-id']
|
|
151
|
+
const edgeCacheStatus = headers['cf-cache-status']
|
|
152
|
+
const unifiedCacheStatus = headers['x-cache-status']
|
|
153
|
+
const cacheStatus =
|
|
154
|
+
unifiedCacheStatus === 'MISS' && edgeCacheStatus === 'HIT'
|
|
155
|
+
? edgeCacheStatus
|
|
156
|
+
: unifiedCacheStatus
|
|
157
|
+
const timestamp = Number(headers['x-timestamp'])
|
|
158
|
+
const ttl = Number(headers['x-cache-ttl'])
|
|
159
|
+
const expires = timestamp + ttl - Date.now()
|
|
160
|
+
const expiredAt =
|
|
161
|
+
cacheStatus === 'HIT' && Number.isFinite(expires)
|
|
162
|
+
? `(${prettyMs(expires)})`
|
|
163
|
+
: ''
|
|
164
|
+
const fetchMode = headers['x-fetch-mode']
|
|
165
|
+
const fetchTime = fetchMode && `(${headers['x-fetch-time']})`
|
|
166
|
+
const uri = response?.url
|
|
167
|
+
|
|
168
|
+
if (process.stdout.isTTY) console.error()
|
|
169
|
+
console.error(
|
|
170
|
+
label('success', 'green'),
|
|
171
|
+
gray(`${prettyBytes(size)} in ${time}`)
|
|
172
|
+
)
|
|
173
|
+
console.error()
|
|
174
|
+
|
|
175
|
+
if (serverTiming) {
|
|
176
|
+
console.error(' ', keyValue(green('timing'), serverTiming))
|
|
177
|
+
}
|
|
178
|
+
if (cacheStatus) {
|
|
179
|
+
console.error(
|
|
180
|
+
' ',
|
|
181
|
+
keyValue(green('cache'), `${cacheStatus} ${gray(expiredAt)}`.trim())
|
|
182
|
+
)
|
|
183
|
+
}
|
|
184
|
+
if (fetchMode) {
|
|
185
|
+
console.error(
|
|
186
|
+
' ',
|
|
187
|
+
keyValue(green('mode'), `${fetchMode} ${gray(fetchTime)}`.trim())
|
|
188
|
+
)
|
|
189
|
+
}
|
|
190
|
+
if (uri) console.error(' ', keyValue(green('uri'), uri))
|
|
191
|
+
if (id) console.error(' ', keyValue(green('id'), id))
|
|
192
|
+
}
|
|
193
|
+
|
|
194
|
+
const printFail = error => {
|
|
195
|
+
if (process.stdout.isTTY) console.error()
|
|
196
|
+
console.error(
|
|
197
|
+
label(error.status || 'fail', 'red'),
|
|
198
|
+
gray(String(error.message).replace(`${error.code}, `, ''))
|
|
199
|
+
)
|
|
200
|
+
console.error()
|
|
201
|
+
const id = error.headers?.['x-request-id']
|
|
202
|
+
if (id) console.error(' ', keyValue(red('id'), id))
|
|
203
|
+
if (error.url) console.error(' ', keyValue(red('uri'), error.url))
|
|
204
|
+
if (error.code) {
|
|
205
|
+
console.error(
|
|
206
|
+
' ',
|
|
207
|
+
keyValue(
|
|
208
|
+
red('code'),
|
|
209
|
+
`${error.code}${error.statusCode ? ` (${error.statusCode})` : ''}`
|
|
210
|
+
)
|
|
211
|
+
)
|
|
212
|
+
}
|
|
213
|
+
if (error.more) console.error(' ', keyValue(red('more'), error.more))
|
|
214
|
+
}
|
|
215
|
+
|
|
216
|
+
const showHelp = command => {
|
|
217
|
+
console.log(helpText(command).trimEnd())
|
|
13
218
|
process.exit(0)
|
|
14
219
|
}
|
|
15
220
|
|
|
221
|
+
const HTTP_HEADER = 'http.header.'
|
|
222
|
+
|
|
16
223
|
const parseHeaders = input => {
|
|
17
224
|
const headers = {}
|
|
18
225
|
for (const item of [].concat(input ?? [])) {
|
|
@@ -25,12 +232,27 @@ const parseHeaders = input => {
|
|
|
25
232
|
return headers
|
|
26
233
|
}
|
|
27
234
|
|
|
235
|
+
const takeHttpHeaders = flags => {
|
|
236
|
+
const headers = {}
|
|
237
|
+
for (const key of Object.keys(flags)) {
|
|
238
|
+
if (!key.startsWith(HTTP_HEADER)) continue
|
|
239
|
+
let value = flags[key]
|
|
240
|
+
delete flags[key]
|
|
241
|
+
if (Array.isArray(value)) value = value.at(-1)
|
|
242
|
+
if (typeof value !== 'string' && typeof value !== 'number') continue
|
|
243
|
+
const name = key.slice(HTTP_HEADER.length).toLowerCase()
|
|
244
|
+
if (name) headers[name] = String(value)
|
|
245
|
+
}
|
|
246
|
+
return headers
|
|
247
|
+
}
|
|
248
|
+
|
|
28
249
|
const argv = mri(process.argv.slice(2), {
|
|
29
250
|
alias: { H: 'header' },
|
|
30
|
-
|
|
251
|
+
boolean: ['trace', 'trace-full', 'help'],
|
|
252
|
+
string: ['header', 'api-key', 'data', 'file', 'endpoint']
|
|
31
253
|
})
|
|
32
254
|
|
|
33
|
-
|
|
255
|
+
let {
|
|
34
256
|
_: [command, target],
|
|
35
257
|
header,
|
|
36
258
|
help,
|
|
@@ -38,23 +260,49 @@ const {
|
|
|
38
260
|
file,
|
|
39
261
|
'api-key': apiKeyFlag,
|
|
40
262
|
apiKey: apiKeyCamel,
|
|
263
|
+
endpoint: endpointFlag,
|
|
264
|
+
trace,
|
|
265
|
+
'trace-full': traceFull,
|
|
41
266
|
...flags
|
|
42
267
|
} = argv
|
|
43
268
|
|
|
44
|
-
|
|
269
|
+
const isTrace = trace || traceFull
|
|
270
|
+
|
|
271
|
+
if (!command) showHelp()
|
|
45
272
|
|
|
46
273
|
const apiKey = apiKeyFlag || apiKeyCamel || process.env.MICROLINK_API_KEY
|
|
47
|
-
const
|
|
274
|
+
const endpoint = endpointFlag
|
|
275
|
+
const client = create({
|
|
276
|
+
...(apiKey && { apiKey }),
|
|
277
|
+
...(endpoint && { endpoint })
|
|
278
|
+
})
|
|
48
279
|
|
|
49
280
|
if (typeof client[command] !== 'function') {
|
|
50
|
-
|
|
51
|
-
|
|
52
|
-
|
|
281
|
+
if (!target && URL.canParse(command)) {
|
|
282
|
+
target = command
|
|
283
|
+
command = 'metadata'
|
|
284
|
+
} else if (help) {
|
|
285
|
+
showHelp()
|
|
286
|
+
} else {
|
|
287
|
+
console.error(
|
|
288
|
+
`Unknown command \`${command}\`. Run \`microlink --help\` to see the available commands.`
|
|
289
|
+
)
|
|
290
|
+
process.exit(1)
|
|
291
|
+
}
|
|
292
|
+
}
|
|
293
|
+
|
|
294
|
+
if (help || !target) showHelp(command)
|
|
295
|
+
|
|
296
|
+
if (
|
|
297
|
+
isTrace &&
|
|
298
|
+
(command === 'search' || command === 'function' || command === 'run')
|
|
299
|
+
) {
|
|
300
|
+
console.error(`\`--trace\` is not supported for \`${command}\`.`)
|
|
53
301
|
process.exit(1)
|
|
54
302
|
}
|
|
55
303
|
|
|
56
304
|
const options = { ...flags }
|
|
57
|
-
const headers = parseHeaders(header)
|
|
305
|
+
const headers = { ...takeHttpHeaders(options), ...parseHeaders(header) }
|
|
58
306
|
if (Object.keys(headers).length > 0) options.headers = headers
|
|
59
307
|
|
|
60
308
|
const invoke = () => {
|
|
@@ -68,14 +316,23 @@ const invoke = () => {
|
|
|
68
316
|
return client[command](target, options)
|
|
69
317
|
}
|
|
70
318
|
|
|
71
|
-
|
|
72
|
-
|
|
73
|
-
|
|
74
|
-
|
|
75
|
-
|
|
319
|
+
const spin = !isTrace && shouldSpin() ? spinner() : null
|
|
320
|
+
|
|
321
|
+
;(async () => {
|
|
322
|
+
spin?.start()
|
|
323
|
+
const started = Date.now()
|
|
324
|
+
try {
|
|
325
|
+
const result = await invoke()
|
|
326
|
+
const duration = Date.now() - started
|
|
327
|
+
spin?.stop()
|
|
328
|
+
if (isTrace) printJson(tracePayload({ ...client.last, full: traceFull }))
|
|
329
|
+
else if (typeof result === 'string') console.log(result)
|
|
330
|
+
else printJson({ status: 'success', data: result })
|
|
331
|
+
if (!isTrace) printFooter({ duration, response: client.last.response })
|
|
76
332
|
process.exit(0)
|
|
77
|
-
})
|
|
78
|
-
|
|
79
|
-
|
|
333
|
+
} catch (error) {
|
|
334
|
+
spin?.stop()
|
|
335
|
+
printFail(error)
|
|
80
336
|
process.exit(1)
|
|
81
|
-
}
|
|
337
|
+
}
|
|
338
|
+
})()
|
package/bin/style.js
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
1
|
+
'use strict'
|
|
2
|
+
|
|
3
|
+
const { styleText } = require('node:util')
|
|
4
|
+
|
|
5
|
+
const gray = str => styleText('gray', str)
|
|
6
|
+
const white = str => styleText('white', str)
|
|
7
|
+
const green = str => styleText('green', str)
|
|
8
|
+
const red = str => styleText('red', str)
|
|
9
|
+
|
|
10
|
+
module.exports = { gray, white, green, red, 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": "0.0
|
|
5
|
+
"version": "0.2.0",
|
|
6
6
|
"types": "./src/index.d.ts",
|
|
7
7
|
"main": "./src/index.js",
|
|
8
8
|
"exports": {
|
|
@@ -46,7 +46,6 @@
|
|
|
46
46
|
"@microlink/function": "0.3.0",
|
|
47
47
|
"@microlink/google": "1.0.6",
|
|
48
48
|
"@microlink/mql": "0.18.0",
|
|
49
|
-
"jsome": "~2.5.0",
|
|
50
49
|
"mri": "~1.2.0"
|
|
51
50
|
},
|
|
52
51
|
"devDependencies": {
|
|
@@ -77,5 +76,5 @@
|
|
|
77
76
|
},
|
|
78
77
|
"directory": "test"
|
|
79
78
|
},
|
|
80
|
-
"gitHead": "
|
|
79
|
+
"gitHead": "3766602b11599a8284e4aa31815d409e0ffa0788"
|
|
81
80
|
}
|
package/src/index.js
CHANGED
|
@@ -41,6 +41,17 @@ const LIGHTHOUSE_KEYS = ['onlyCategories', 'onlyAudits', 'skipAudits', 'output']
|
|
|
41
41
|
const isEmpty = obj => Object.keys(obj).length === 0
|
|
42
42
|
|
|
43
43
|
const create = (ctx = {}) => {
|
|
44
|
+
const last = {}
|
|
45
|
+
const request = (...args) => {
|
|
46
|
+
const [requestUrl, requestOptions] = mql.getApiUrl(...args)
|
|
47
|
+
last.requestUrl = requestUrl
|
|
48
|
+
last.requestOptions = requestOptions
|
|
49
|
+
return mql(...args).then(result => {
|
|
50
|
+
last.response = result.response
|
|
51
|
+
return result
|
|
52
|
+
})
|
|
53
|
+
}
|
|
54
|
+
|
|
44
55
|
/**
|
|
45
56
|
* Split the single options bag into the three destinations mql has:
|
|
46
57
|
* `got.headers` (HTTP layer, 3rd arg), `sub` (capability nested keys)
|
|
@@ -61,7 +72,7 @@ const create = (ctx = {}) => {
|
|
|
61
72
|
|
|
62
73
|
const content = field => (url, options) => {
|
|
63
74
|
const { top, sub, got } = route(options, CONTENT_KEYS)
|
|
64
|
-
return
|
|
75
|
+
return request(
|
|
65
76
|
url,
|
|
66
77
|
{ ...top, meta: false, data: { [field]: { attr: field, ...sub } } },
|
|
67
78
|
got
|
|
@@ -70,7 +81,7 @@ const create = (ctx = {}) => {
|
|
|
70
81
|
|
|
71
82
|
const collection = (field, rule) => (url, options) => {
|
|
72
83
|
const { top, sub, got } = route(options, COLLECTION_KEYS)
|
|
73
|
-
return
|
|
84
|
+
return request(
|
|
74
85
|
url,
|
|
75
86
|
{ ...top, meta: false, data: { [field]: { ...rule, ...sub } } },
|
|
76
87
|
got
|
|
@@ -79,7 +90,7 @@ const create = (ctx = {}) => {
|
|
|
79
90
|
|
|
80
91
|
const capability = (field, nested) => (url, options) => {
|
|
81
92
|
const { top, sub, got } = route(options, nested)
|
|
82
|
-
return
|
|
93
|
+
return request(
|
|
83
94
|
url,
|
|
84
95
|
{ ...top, meta: false, [field]: isEmpty(sub) ? true : sub },
|
|
85
96
|
got
|
|
@@ -89,7 +100,7 @@ const create = (ctx = {}) => {
|
|
|
89
100
|
/* Primary media detection (`data.video` / `data.audio`). */
|
|
90
101
|
const media = field => (url, options) => {
|
|
91
102
|
const { top, got } = route(options)
|
|
92
|
-
return
|
|
103
|
+
return request(url, { ...top, meta: false, [field]: true }, got).then(
|
|
93
104
|
({ data }) => data[field]
|
|
94
105
|
)
|
|
95
106
|
}
|
|
@@ -101,14 +112,14 @@ const create = (ctx = {}) => {
|
|
|
101
112
|
return fn(code, top, got)(url)
|
|
102
113
|
}
|
|
103
114
|
|
|
104
|
-
|
|
115
|
+
const client = {
|
|
105
116
|
metadata: (url, options) => {
|
|
106
117
|
const { top, got } = route(options)
|
|
107
|
-
return
|
|
118
|
+
return request(url, top, got).then(({ data }) => data)
|
|
108
119
|
},
|
|
109
120
|
logo: (url, options) => {
|
|
110
121
|
const { top, sub, got } = route(options, LOGO_KEYS)
|
|
111
|
-
return
|
|
122
|
+
return request(
|
|
112
123
|
url,
|
|
113
124
|
{ ...top, meta: isEmpty(sub) ? true : { logo: sub } },
|
|
114
125
|
got
|
|
@@ -142,7 +153,7 @@ const create = (ctx = {}) => {
|
|
|
142
153
|
}),
|
|
143
154
|
extract: (url, rules, options) => {
|
|
144
155
|
const { top, got } = route(options)
|
|
145
|
-
return
|
|
156
|
+
return request(url, { ...top, meta: false, data: rules }, got).then(
|
|
146
157
|
({ data }) => data
|
|
147
158
|
)
|
|
148
159
|
},
|
|
@@ -150,7 +161,7 @@ const create = (ctx = {}) => {
|
|
|
150
161
|
pdf: capability('pdf', PDF_KEYS),
|
|
151
162
|
embed: (url, options) => {
|
|
152
163
|
const { top, sub, got } = route(options, EMBED_KEYS)
|
|
153
|
-
return
|
|
164
|
+
return request(
|
|
154
165
|
url,
|
|
155
166
|
{ ...top, meta: false, iframe: isEmpty(sub) ? true : sub },
|
|
156
167
|
got
|
|
@@ -158,7 +169,7 @@ const create = (ctx = {}) => {
|
|
|
158
169
|
},
|
|
159
170
|
technologies: (url, options) => {
|
|
160
171
|
const { top, got } = route(options)
|
|
161
|
-
return
|
|
172
|
+
return request(
|
|
162
173
|
url,
|
|
163
174
|
{
|
|
164
175
|
...top,
|
|
@@ -170,7 +181,7 @@ const create = (ctx = {}) => {
|
|
|
170
181
|
},
|
|
171
182
|
lighthouse: (url, options) => {
|
|
172
183
|
const { top, sub, got } = route(options, LIGHTHOUSE_KEYS)
|
|
173
|
-
return
|
|
184
|
+
return request(
|
|
174
185
|
url,
|
|
175
186
|
{
|
|
176
187
|
...top,
|
|
@@ -190,6 +201,9 @@ const create = (ctx = {}) => {
|
|
|
190
201
|
function: run,
|
|
191
202
|
run
|
|
192
203
|
}
|
|
204
|
+
|
|
205
|
+
Object.defineProperty(client, 'last', { value: last })
|
|
206
|
+
return client
|
|
193
207
|
}
|
|
194
208
|
|
|
195
209
|
module.exports = create
|
package/bin/help.txt
DELETED
|
@@ -1,47 +0,0 @@
|
|
|
1
|
-
Usage
|
|
2
|
-
$ microlink <product> <url|query> [options]
|
|
3
|
-
|
|
4
|
-
Products
|
|
5
|
-
metadata Unified metadata (title, description, image, ...)
|
|
6
|
-
logo Brand logo of the site (--square prefers the square variant)
|
|
7
|
-
markdown Page content as Markdown
|
|
8
|
-
html Page content as HTML
|
|
9
|
-
text Page content as plain text
|
|
10
|
-
video Primary video of the page (returns the asset object)
|
|
11
|
-
audio Primary audio of the page (returns the asset object)
|
|
12
|
-
emails Every email address present on the page
|
|
13
|
-
links Every absolute link URL on the page
|
|
14
|
-
images Every absolute image URL on the page
|
|
15
|
-
videos Every absolute video URL on the page
|
|
16
|
-
audios Every absolute audio URL on the page
|
|
17
|
-
extract Custom MQL data rules (--data '{"field":{...}}')
|
|
18
|
-
screenshot Take a screenshot (returns the asset object)
|
|
19
|
-
pdf Generate a PDF (returns the asset object)
|
|
20
|
-
embed oEmbed-style embeddable iframe ({ html, scripts })
|
|
21
|
-
technologies Detect the tech stack behind the site
|
|
22
|
-
lighthouse Run a Lighthouse report
|
|
23
|
-
search Google as structured data (query instead of url); --type
|
|
24
|
-
routes to news, images, videos, places, maps, shopping,
|
|
25
|
-
scholar, patents or autocomplete
|
|
26
|
-
function Run code remotely with browser access (--file ./fn.js);
|
|
27
|
-
extra flags are injected as variables in the function scope
|
|
28
|
-
|
|
29
|
-
Options
|
|
30
|
-
--api-key Microlink API key (defaults to MICROLINK_API_KEY env)
|
|
31
|
-
--header, -H Extra request header as 'Name: value' (repeatable)
|
|
32
|
-
--data JSON data rules for the extract command
|
|
33
|
-
--file Path to the code file for the function command
|
|
34
|
-
--help Show this help
|
|
35
|
-
|
|
36
|
-
Any other flag is passed as an option to the product, e.g. --fullPage,
|
|
37
|
-
--device 'iPhone 11', --waitUntil networkidle0, --selector article.
|
|
38
|
-
|
|
39
|
-
Examples
|
|
40
|
-
$ microlink markdown https://example.com
|
|
41
|
-
$ microlink screenshot https://example.com --fullPage
|
|
42
|
-
$ microlink logo https://github.com --square
|
|
43
|
-
$ microlink links https://example.com
|
|
44
|
-
$ microlink search "best coffee" --limit 10 --location es
|
|
45
|
-
$ microlink search "open source llm" --type news --period week
|
|
46
|
-
$ microlink extract https://microlink.io --data '{"image":{"selector":"meta[property=og:image]","attr":"content","type":"image"}}'
|
|
47
|
-
$ microlink function https://example.com --file ./fn.js
|