vore-scenario-generator/src/client/scraper.ts

import {
	type InProgressGeneratedState,
	RolledValues,
	RollSelections,
	type RollTableAuthor,
	RollTableDatabase,
	type RollTableDetailsAndResults,
	type RollTableDetailsNoResults,
	type RollTableLimited,
	type RollTableResult,
	type RollTableResultFull,
	type RollTableResultSet
} from '../common/rolltable';
import { authorIdKey } from '../common/template';

export function asBoolean(s: string|undefined): boolean|undefined {
	if (typeof s === "undefined") {
		return
	}
	switch (s.toLowerCase()) {
		case 'true':
			return true
		case 'false':
			return false
		default:
			return
	}
}

export function asInteger(s: string|undefined): number|undefined {
	if (typeof s === "undefined") {
		return
	}
	const result = parseInt(s)
	if (Number.isNaN(result)) {
		return
	}
	return result
}

export function asTimestamp(s: string|undefined): Date|undefined {
	const i = asInteger(s)
	if (typeof i === "undefined") {
		return
	}
	const date = new Date(i)
	if (Number.isNaN(date.valueOf())) {
		return
	}
	return date
}

export function textFrom(e: HTMLElement|null): string|undefined {
	if (!e) {
		return
	}
	return e.innerText.trim()
}

export function hrefFrom(e: HTMLAnchorElement|null): string | null {
	if (!e) {
		return null
	}
	return e.href
}

export function checkedFrom(e: HTMLInputElement|null): boolean | null {
	if (!e) {
		return null
	}
	return e.checked
}

// element to find here is .author
export function scrapeAuthor(author: HTMLElement|null): RollTableAuthor|null|undefined {
	if (!author) {
		return null
	}
	const id = asInteger(author.dataset[authorIdKey])
	const name = textFrom(author.querySelector(`.authorName`))
	const url = hrefFrom(author.querySelector<HTMLAnchorElement>("a[href]"))
	const relation = textFrom(author.querySelector(`.authorRelation`))
	if (typeof id === "undefined" || typeof name === "undefined" || typeof relation === 'undefined') {
		return
	}
	return {
		id,
		name,
		url,
		relation
	}
}

// element to find here is .resultSet
export function scrapeResultSet(set: HTMLElement|null): RollTableResultSet|null|undefined {
	if (!set) {
		return null
	}
	const id = asInteger(set.dataset["id"])
	const name = textFrom(set.querySelector(`.setName`))
	const global = asBoolean(set.dataset["global"])
	if (typeof id === "undefined" || typeof global === "undefined") {
		return
	}
	return {
		id,
		name: name ?? null,
		description: null,
		global,
	}
}

// element to find here is .tableHeader
export function scrapeTableHeader(head: HTMLElement|null): RollTableLimited|RollTableDetailsNoResults|null|undefined {
	if (!head) {
		return null
	}
	const emoji = textFrom(head.querySelector(".tableEmoji"))
	const title = textFrom(head.querySelector(".tableTitle"))
	const ordinal = asInteger(head.dataset["ordinal"])
	const id = asInteger(head.dataset["id"])
	const identifier = head.dataset["identifier"]
	const name = head.dataset["name"]
	if (typeof emoji === 'undefined' || typeof title === 'undefined' || typeof ordinal === 'undefined') {
		return
	}
	const header = `${emoji} ${title}`
	if (typeof id === 'undefined' || typeof identifier === 'undefined' || typeof name === 'undefined') {
		return {
			full: false,
			emoji,
			title,
			header,
			ordinal,
		}
	}
	return {
		full: 'details',
		id,
		identifier,
		emoji,
		title,
		header,
		ordinal,
		name,
	}
}

export function scrapeGeneratedHead(head: HTMLElement|null): {table: RollTableLimited|RollTableDetailsNoResults, selected: boolean|null}|null|undefined {
	if (!head) {
		return null
	}
	const table = scrapeTableHeader(head.querySelector(`.tableHeader`))
	if (!table) {
		return
	}
	const selected = checkedFrom(head.querySelector(`input[type=checkbox].generatedSelect`))
	return {
		table,
		selected,
	}
}

// element to find here is .resultText
export function scrapeResultText(result: HTMLElement|null): {full: false, text: string}|{full: true, mappingId: number, textId: number, updated: Date, text: string}|undefined|null {
	if (!result) {
		return null
	}
	const text = textFrom(result)
	const mappingId = asInteger(result.dataset["mappingid"])
	const textId = asInteger(result.dataset["textid"])
	const updated = asTimestamp(result.dataset["updated"])
	if (typeof text === 'undefined') {
		return
	}
	if (typeof mappingId === 'undefined' || typeof textId === 'undefined' || typeof updated == 'undefined') {
		return {
			full: false,
			text,
		}
	}
	return {
		full: true,
		text,
		textId,
		mappingId,
		updated: new Date(updated)
	}
}

// element to find here is .generatedElement
export function scrapeGeneratedElement(generated: HTMLElement|null): {result: RollTableResult, selected: boolean|null}|null|undefined {
	if (!generated) {
		return null
	}
	const result = scrapeResultText(generated.querySelector(`.resultText`))
	const author = scrapeAuthor(generated.querySelector(`.author`))
	const set = scrapeResultSet(generated.querySelector(`.resultSet`))
	const header = scrapeGeneratedHead(generated.querySelector(`.generatedHead`))
	if (!header || !result) {
		return
	}
	const {table, selected} = header
	if (!set || typeof author === "undefined" || !result.full) {
		return {
			result: {
				full: false,
				table,
				text: result.text,
			},
			selected
		}
	}
	return {
		result: {
			...result,
			author,
			set,
			table,
		},
		selected,
	}
}

export function scrapeGeneratedScenario(scenario: HTMLElement): InProgressGeneratedState|undefined
export function scrapeGeneratedScenario(scenario: null): null
// element to find here is #generatedScenario
export function scrapeGeneratedScenario(scenario: HTMLElement|null): InProgressGeneratedState|null|undefined {
	if (!scenario) {
		return null
	}
	const rolls = new RolledValues()
	const selection = new RollSelections()
	for (const item of scenario.querySelectorAll<HTMLElement>(`.generatedElement`)) {
		const element = scrapeGeneratedElement(item)
		if (!element) {
			return
		}
		const {result, selected} = element
		rolls.add(result)
		if (selected) {
			selection.add(result.table)
		}
	}
	return {
		final: false,
		rolled: rolls,
		selected: selection,
	}
}

export function scrapeResponseList(responseTypeElement: HTMLElement|null, db: RollTableDatabase): [table: RollTableDetailsAndResults, active: RollTableResultFull<RollTableDetailsAndResults>|null]|null|undefined {
	if (!responseTypeElement) {
		return null
	}
	const table = scrapeTableHeader(responseTypeElement.querySelector(`.tableHeader`))
	if (!table || !table.full) {
		return
	}
	const resultTable = db.addTable(table)
	let activeResult: RollTableResultFull<RollTableDetailsAndResults>|null = null
	for (const resultElement of responseTypeElement.querySelectorAll<HTMLElement>(`.response`)) {
		const partialResult = scrapeResultText(resultElement.querySelector(`.resultText`))
		const author = scrapeAuthor(resultElement.querySelector(`.author`))
		const set = scrapeResultSet(resultElement.querySelector(`.resultSet`))
		const active = resultElement.classList.contains("active")

		if (!partialResult || !partialResult.full || typeof author === "undefined" || !set) {
			return
		}

		const result = db.addResult({
			...partialResult,
			set,
			author,
			table: resultTable,
		})
		if (active) {
			activeResult = result
		}
	}
	return [resultTable, activeResult]
}

export function scrapeResponseLists(lists: HTMLElement): {db: RollTableDatabase, active: ReadonlyMap<RollTableDetailsAndResults, RollTableResultFull<RollTableDetailsAndResults>|null>}|undefined
export function scrapeResponseLists(lists: null): null
export function scrapeResponseLists(lists: HTMLElement|null): {db: RollTableDatabase, active: ReadonlyMap<RollTableDetailsAndResults, RollTableResultFull<RollTableDetailsAndResults>|null>}|null|undefined {
	if (!lists) {
		return null
	}
	const db = new RollTableDatabase()
	const active = new Map<RollTableDetailsAndResults, RollTableResultFull<RollTableDetailsAndResults>|null>
	for (const responseTypeElement of lists.querySelectorAll<HTMLElement>(`.responseType`)) {
		const responseType = scrapeResponseList(responseTypeElement, db)
		if (!responseType) {
			return
		}
		const [table, activeResult] = responseType
		active.set(table, activeResult)
	}
	return {
		db,
		active,
	}
}