add CLI interface made just for parsing

2025-06-05 22:09:25 +02:00 · 2020-07-14 22:00:59 +02:00
parent 094f0bf242
commit bf36383bff
3 changed files with 73 additions and 55 deletions
--- a/lib/cli.js
+++ b/lib/cli.js
@@ -0,0 +1,52 @@
+#!/usr/bin/env node
+
+const parseUsingLDJSONData = require('./services/ldjson-parser')
+const parseUsingDOM = require('./services/dom-parser')
+const { createParserError, createAppLogger } = require('./utils')
+
+const parseHTMLString = (html, { verbose }) => {
+  const dev = process.env.NODE_ENV === 'development'
+  const logger = createAppLogger({ dev })
+
+  if (!verbose) {
+    logger.transports.forEach((transport) => (transport.silent = true))
+  }
+
+  try {
+    const LDJSONEventData = parseUsingLDJSONData(html, { logger })
+    // TODO: adding empty string for URL argument since we don't know original
+    //       URL of document
+    const eventData = LDJSONEventData || parseUsingDOM(html, '', { logger })
+
+    if (!eventData) {
+      throw createParserError()
+      return
+    }
+
+    process.stdout.write(JSON.stringify(eventData))
+  } catch (err) {
+    console.error(err)
+    process.stderr.write(err.toString())
+    process.exit(1)
+  }
+}
+
+const argv = require('yargs')
+  .command('$0 <html> [options]', 'Parse Facebook event HTML page into JSON data.', (yargs) => {
+    yargs.positional('html', {
+      describe: 'valid HTML string',
+      type: 'string',
+    })
+  }, (argv) => {
+    const { html, verbose } = argv
+    parseHTMLString(html, { verbose })
+  })
+  .option('verbose', {
+    alias: 'v',
+    description: 'Add logging to output.',
+    type: 'boolean',
+  })
+  .alias('h', 'help')
+  .help('h')
+  .argv
+