mirror of
https://github.com/postlight/mercury-parser
synced 2024-11-11 01:10:35 +00:00
64 lines
2.0 KiB
JavaScript
64 lines
2.0 KiB
JavaScript
/* eslint-disable */
|
|
const bot = require('@jesses/circle-github-bot').default.create();
|
|
const Mercury = require('../dist/mercury.js');
|
|
const fs = require("fs");
|
|
|
|
const run = () => {
|
|
const screenshotPath = process.argv[2];
|
|
|
|
const fixture = screenshotPath.split('tmp/artifacts/')[1].slice(0, -4);
|
|
|
|
const html = fs.readFileSync(`${fixture}`);
|
|
|
|
// first parse is just to get the url
|
|
Mercury.parse('http://example.com', html, { fallback: false }).then(({ url, domain, excerpt, word_count, direction }) => {
|
|
// with the url, second pass will test the correct parser
|
|
Mercury.parse(url, html, { fallback: false }).then(json => {
|
|
// removing excerpt b/c this comes from content, not necessary
|
|
delete json.excerpt
|
|
|
|
// adding items that aren't pulled in custom parser w/out fallback
|
|
Object.assign(json, { url, domain, word_count, direction });
|
|
|
|
// a quick preview of the parsed content in an html file
|
|
const previewHtml = `<h1>${json.title}</h1><img src=${json.lead_image_url} /><p>${json.author}</p>${json.content}`
|
|
|
|
|
|
const jsonPath = `${screenshotPath}-parsed.json`;
|
|
const fixtureArtifactPath = `tmp/artifacts/${fixture}`;
|
|
const previewPath = `tmp/artifacts/${fixture}.preview.html`;
|
|
|
|
fs.writeFileSync(previewPath, previewHtml);
|
|
fs.writeFileSync(jsonPath, JSON.stringify(json));
|
|
fs.writeFileSync(fixtureArtifactPath, html);
|
|
|
|
bot.comment(process.env.GH_AUTH_TOKEN, `### 🤖 Automated Parsing Preview 🤖
|
|
**Commit:** \`${bot.env.commitMessage}\`
|
|
|
|
})
|
|
|
|
[Original Article](${url}) | ${bot.artifactLink(fixtureArtifactPath, 'HTML Fixture')} | ${bot.artifactLink(previewPath, 'Parsed Content Preview')}
|
|
|
|
<details>
|
|
<summary><b>Parsed JSON</b></summary>
|
|
|
|
\`\`\`json
|
|
${JSON.stringify(json, null, 2)}
|
|
\`\`\`
|
|
|
|
</details>
|
|
<br />
|
|
|
|
|
|
**\`null\` fields**
|
|
|
|
${Object.keys(json).map(key => json[key] !== null ? '' : ` * \`${key}\n\``).join('') || 'None'}
|
|
|
|
`
|
|
);
|
|
});
|
|
});
|
|
}
|
|
|
|
run();
|