@epsilon-asi/actors 0.0.7 → 0.0.10

package/src/sites/upwork-com/util/scrapeJobListing.ts

@@ -0,0 +1,300 @@
+import type {ElementHandle, Page} from "puppeteer-core";
+import {z} from "zod/v4";
+
+/* -------------------------------------------------------------------------------------------------
+ * Zod Schemas
+ * ------------------------------------------------------------------------------------------------- */
+export enum JobCategoryType {
+    "hourly" = "hourly",
+    "fixed" = "fixed",
+}
+
+export const UpworkJobListingSchema = z.object({
+    id: z.string(),
+
+    title: z.string(),
+    url: z.string(),
+
+    postedAtText: z.string().nullable(),
+    proposalsText: z.string().nullable(),
+
+    description: z.string(),
+
+    skills: z.array(z.string()),
+
+    category: z.object({
+        type: z.enum(JobCategoryType, {}).nullable(),
+        raw: z.string().nullable(),
+    }),
+
+    hourlyRate: z
+        .object({
+            min: z.number().nullable(),
+            max: z.number().nullable(),
+            raw: z.string().nullable(),
+        })
+        .nullable(),
+
+    fixedBudget: z
+        .object({
+            amount: z.number().nullable(),
+            raw: z.string().nullable(),
+        })
+        .nullable(),
+
+    experienceLevel: z.string().nullable(),
+
+    estimatedTime: z.string().nullable(),
+
+    client: z.object({
+        paymentVerified: z.boolean(),
+
+        rating: z.number().nullable(),
+
+        totalSpent: z.string().nullable(),
+
+        location: z.string().nullable(),
+    }),
+
+    rawText: z.string(),
+});
+
+export type UpworkJobListing = z.infer<typeof UpworkJobListingSchema>;
+
+/* -------------------------------------------------------------------------------------------------
+ * Helpers
+ * ------------------------------------------------------------------------------------------------- */
+
+function cleanText(input?: string | null): string {
+    return (input ?? "")
+        .replace(/\s+/g, " ")
+        .replace(/\u00a0/g, " ")
+        .trim();
+}
+
+function parseMoney(value?: string): number | null {
+    if (!value) return null;
+    const normalized = value.replace(/[^0-9.]/g, "");
+
+    if (!normalized) {
+        return null;
+    }
+
+    const parsed = Number(normalized);
+
+    return Number.isNaN(parsed) ? null : parsed;
+}
+
+function parseHourlyRate(
+    raw: string | null,
+): UpworkJobListing["hourlyRate"] {
+    if (!raw) {
+        return null;
+    }
+
+    const matches = raw.match(/\$([\d,.]+)\s*-\s*\$([\d,.]+)/);
+
+    if (!matches || matches.length < 3) {
+        return {
+            min: null,
+            max: null,
+            raw,
+        };
+    }
+    return {
+        min: parseMoney(matches[1]),
+        max: parseMoney(matches[2]),
+        raw,
+    };
+}
+
+function parseFixedBudget(
+    raw: string | null,
+): UpworkJobListing["fixedBudget"] {
+    if (!raw) {
+        return null;
+    }
+
+    const match = raw.match(/\$([\d,.]+)/);
+
+    return {
+        amount: match ? parseMoney(match[1]) : null,
+        raw,
+    };
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * Core Parser
+ * ------------------------------------------------------------------------------------------------- */
+
+export async function parseUpworkJobListing(
+    article: ElementHandle<Element>,
+): Promise<UpworkJobListing> {
+    const extracted = await article.evaluate((node) => {
+        const getText = (selector: string): string | null => {
+            const element = node.querySelector(selector);
+
+            return element?.textContent?.trim() ?? null;
+        };
+
+        const getTexts = (selector: string): string[] => {
+            return Array.from(node.querySelectorAll(selector))
+                .map((el) => el.textContent?.trim() ?? "")
+                .filter(Boolean);
+        };
+
+        const titleAnchor = node.querySelector(
+            '[data-test="job-tile-title-link UpLink"]',
+        ) as HTMLAnchorElement | null;
+
+        const jobInfoItems = Array.from(
+            node.querySelectorAll('[data-test="JobInfo"] li'),
+        ).map((li) => li.textContent?.trim() ?? "");
+
+        const clientInfoItems = Array.from(
+            node.querySelectorAll('[data-test="JobInfoClient"] li'),
+        ).map((li) => li.textContent?.trim() ?? "");
+
+        const hourlyItem =
+            jobInfoItems.find((x) => x.toLowerCase().includes("hourly")) ?? null;
+
+        const fixedItem =
+            jobInfoItems.find((x) => x.toLowerCase().includes("fixed")) ?? null;
+
+        const experienceLevel =
+            node
+                .querySelector('[data-test="experience-level"]')
+                ?.textContent?.trim() ?? null;
+
+        const estimatedTime =
+            node
+                .querySelector('[data-test="duration-label"]')
+                ?.textContent?.trim() ?? null;
+
+        const ratingText =
+            node
+                .querySelector('[data-test="feedback-rating UpCRating"]')
+                ?.textContent?.trim() ?? null;
+
+        const spentText =
+            node
+                .querySelector('[data-test="total-spent"]')
+                ?.textContent?.trim() ?? null;
+
+        const locationText =
+            node
+                .querySelector('[data-test="location"]')
+                ?.textContent?.trim() ?? null;
+
+        return {
+            id:
+                node.getAttribute("data-ev-job-uid") ??
+                node.getAttribute("data-test-key") ??
+                "",
+
+            title: titleAnchor?.textContent?.trim() ?? "",
+
+            url: titleAnchor?.href ?? "",
+
+            postedAtText:
+                getText('[data-test="job-pubilshed-date"] span:first-child') ?? null,
+
+            proposalsText:
+                getText('[data-test="proposals-tier"]') ?? null,
+
+            description:
+                getText('[data-test="UpCLineClamp JobDescription"] p') ?? "",
+
+            skills: getTexts('[data-test="token"] span'),
+
+            hourlyRaw: hourlyItem,
+            fixedRaw: fixedItem,
+
+            experienceLevel,
+            estimatedTime,
+
+            paymentVerified:
+                node.querySelector('[data-test="payment-verified"]') !== null,
+
+            ratingText,
+            spentText,
+            locationText,
+
+            rawText: node.textContent ?? "",
+
+            categoryType: hourlyItem
+                ? "hourly"
+                : fixedItem
+                    ? "fixed"
+                    : null,
+        };
+    });
+
+    const hourlyRate = parseHourlyRate(extracted.hourlyRaw);
+    const fixedBudget = parseFixedBudget(extracted.fixedRaw);
+
+    const ratingMatch = extracted.ratingText?.match(/(\d+(\.\d+)?)/);
+
+    const parsed: UpworkJobListing = {
+        id: cleanText(extracted.id),
+
+        title: cleanText(extracted.title),
+
+        url: cleanText(extracted.url),
+
+        postedAtText: cleanText(extracted.postedAtText),
+
+        proposalsText: cleanText(extracted.proposalsText),
+
+        description: cleanText(extracted.description),
+
+        skills: extracted.skills.map(cleanText),
+
+        category: {
+            type: extracted.categoryType as (JobCategoryType | null),
+            raw: cleanText(extracted.hourlyRaw ?? extracted.fixedRaw),
+        },
+
+        hourlyRate,
+
+        fixedBudget,
+
+        experienceLevel: cleanText(extracted.experienceLevel),
+
+        estimatedTime: cleanText(extracted.estimatedTime),
+
+        client: {
+            paymentVerified: extracted.paymentVerified,
+
+            rating: ratingMatch ? Number(ratingMatch[1]) : null,
+
+            totalSpent: cleanText(extracted.spentText),
+
+            location: cleanText(extracted.locationText),
+        },
+
+        rawText: cleanText(extracted.rawText),
+    };
+
+    return UpworkJobListingSchema.parse(parsed);
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * Multi-Listing Parser
+ * ------------------------------------------------------------------------------------------------- */
+
+export async function parseUpworkSearchResults(
+    page: Page,
+): Promise<UpworkJobListing[]> {
+    await page.waitForSelector('#main > section > article[data-ev-sublocation="search_results"]');
+
+    const articles = await page.$$(
+        '#main > section > article[data-ev-sublocation="search_results"]',
+    );
+
+    const results = await Promise.all(
+        articles.map((article) => parseUpworkJobListing(article)),
+    );
+
+    return results;
+}