Worked on parsing

This commit is contained in:
Lukas Wölfer
2025-05-05 22:52:11 +02:00
parent 81ae6c81b4
commit 92629d4bf9

95
main.ts
View File

@@ -15,47 +15,98 @@ async function getWorkshopVideos(bot: Mwn): Promise<string[]> {
// Extract and print the file titles
return response.query.allimages.map(image => image.title);
console.log('Files starting with "ABC ":', files);
const file_content = await bot.read(files[0])
console.dir(file_content.revisions[0].content)
console.dir(await bot.parseTitle(files[0]))
}
interface VideoDescription {
teachers: string[]
location: string
event: string
level: string
date: string
patterns: string
notes: string
}
function parseSummary(description: string): VideoDescription {
function parseTemplate(description: string): Record<string, string> {
const templateRegex = /{{(.*DanceWorkshopDescription.*)}}/s;
const match = description.match(templateRegex);
if (!match) {
throw new Error("Could not find template")
}
const propertyRegex = /\|\s*(?<key>[^|]*?)\s*=(?<value>[^|]*?)\s*$/gm
const properties = match[1].matchAll(propertyRegex).map(v => ({ [v.groups['key']]: v.groups['value'] }))
return properties
}
function parseSummary(description: string, name: string): VideoDescription {
const properties = parseTemplate(description)
const b: Partial<VideoDescription> = {};
if (Object.hasOwn(properties, 'teachers')) {
console.warn(`Page ${name} has old template usage [contains 'teachers' key]`)
const t = properties['teachers'].split('&').map(item => item.trim());
b.teachers = t
} else {
b.teachers = [properties['leader'], properties['follower']].filter(v => v !== undefined)
}
b.location = properties['location']
b.date = properties['date']
b.level = properties['level']
b.event = properties['event'] || b.location
if (!Object.hasOwn(properties, 'patterns')) {
console.warn(`Page ${name} has old template usage [no 'patterns' key]`)
const patternsRegex = /===\s*Shown Patterns\s*===\s*(?<patterns>(^\s*\*+.*$\s+)+)/gm
const match = description.match(patternsRegex)
if (!match) {
throw new Error("Could not find \"Shown Patterns\"")
}
b.patterns = match.groups['patterns']
}
if (!Object.hasOwn(properties, 'notes')) {
console.warn(`Page ${name} has old template usage [no 'patterns' key]`)
const notesRegex = /===\s*Notes\s*===\s*(?<notes>(^\s*:+.*$\s+)+)/gm
const match = description.match(notesRegex)
if (!match) {
throw new Error("Could not find \"Notes\"")
}
b.notes = match.groups['notes']
}
return b
}
async function main() {
const bot = new Mwn({
apiUrl: 'https://dancing.thasky.one/api.php',
username: process.env.BOTNAME,
async function main() {
const bot = new Mwn({
apiUrl: 'https://dancing.thasky.one/api.php',
username: process.env.BOTNAME,
password: process.env.BOTPW,
password: process.env.BOTPW,
userAgent: 'mwn bot',
userAgent: 'mwn bot',
});
});
try {
await bot.login();
const relevantFiles = await getWorkshopVideos(bot)
} catch (error) {
console.error('Error:', error);
} finally {
await bot.logout();
}
try {
await bot.login();
const relevantFiles = await getWorkshopVideos(bot)
const file_content = await bot.read(relevantFiles[0])
const content = file_content.revisions[0].content
console.log(content)
parseSummary(content, relevantFiles[0])
} catch (error) {
console.error('Error:', error);
} finally {
await bot.logout();
}
}
main();