Merging pull request #18601

* new components

* pnpm-lock.yaml

* version

Author: michelle0927

components/dataforseo/actions/create-onpage-task/create-onpage-task.mjs

@@ -0,0 +1,95 @@
+import dataforseo from "../../dataforseo.app.mjs";
+import { ConfigurationError } from "@pipedream/platform";
+
+export default {
+  key: "dataforseo-create-onpage-task",
+  name: "Create OnPage Task",
+  description: "Create an OnPage task to retrieve detailed information on how well your pages are optimized for search. [See the documentation](https://docs.dataforseo.com/v3/on_page/task_post/)",
+  version: "0.0.1",
+  type: "action",
+  annotations: {
+    destructiveHint: false,
+    openWorldHint: true,
+    readOnlyHint: false,
+  },
+  props: {
+    dataforseo,
+    target: {
+      propDefinition: [
+        dataforseo,
+        "target",
+      ],
+    },
+    maxCrawlPages: {
+      type: "integer",
+      label: "Max Crawl Pages",
+      description: "The number of pages to crawl on the specified domain",
+    },
+    startUrl: {
+      type: "string",
+      label: "Start URL",
+      description: "The URL to start crawling from",
+      optional: true,
+    },
+    loadResources: {
+      type: "boolean",
+      label: "Load Resources",
+      description: "Set to `true` if you want to load image, stylesheets, scripts, and broken resources on the page",
+      optional: true,
+    },
+    enableJavascript: {
+      type: "boolean",
+      label: "Enable Javascript",
+      description: "Set to `true` if you want to load the scripts available on a page",
+      optional: true,
+    },
+    tag: {
+      propDefinition: [
+        dataforseo,
+        "tag",
+      ],
+    },
+    pingbackUrl: {
+      type: "string",
+      label: "Pingback URL",
+      description: "When a task is completed you will be notified by a GET request sent to the URL you have specified",
+      optional: true,
+    },
+  },
+  methods: {
+    createOnpageTask(args = {}) {
+      return this.dataforseo._makeRequest({
+        path: "/on_page/task_post",
+        method: "post",
+        ...args,
+      });
+    },
+  },
+  async run({ $ }) {
+    const response = await this.createOnpageTask({
+      $,
+      data: [
+        {
+          target: this.target,
+          max_crawl_pages: this.maxCrawlPages,
+          start_url: this.startUrl,
+          load_resources: this.loadResources,
+          enable_javascript: this.enableJavascript,
+          tag: this.tag,
+          pingback_url: this.pingbackUrl,
+        },
+      ],
+    });
+
+    if (response.status_code !== 20000) {
+      throw new ConfigurationError(`Error: ${response.status_message}`);
+    }
+
+    if (response.tasks[0].status_code !== 20000 && response.tasks[0].status_code !== 20100) {
+      throw new ConfigurationError(`Error: ${response.tasks[0].status_message}`);
+    }
+
+    $.export("$summary", "Successfully created onpage task.");
+    return response;
+  },
+};

components/dataforseo/actions/get-crawled-pages/get-crawled-pages.mjs

@@ -0,0 +1,74 @@
+import dataforseo from "../../dataforseo.app.mjs";
+import { ConfigurationError } from "@pipedream/platform";
+
+export default {
+  key: "dataforseo-get-crawled-pages",
+  name: "Get Crawled Pages with OnPage",
+  description: "Get page-specific data with detailed information on how well your pages are optimized for search. [See the documentation](https://docs.dataforseo.com/v3/on_page/pages/)",
+  version: "0.0.1",
+  type: "action",
+  annotations: {
+    destructiveHint: false,
+    openWorldHint: true,
+    readOnlyHint: true,
+  },
+  props: {
+    dataforseo,
+    taskId: {
+      type: "string",
+      label: "Task ID",
+      description: "The ID of the task. You can get this ID in the response of the Task POST endpoint. Example: `07131248-1535-0216-1000-17384017ad04`",
+    },
+    limit: {
+      propDefinition: [
+        dataforseo,
+        "limit",
+      ],
+    },
+    searchAfterToken: {
+      type: "string",
+      label: "Search After Token",
+      description: "The token for subsequent requests",
+      optional: true,
+    },
+    tag: {
+      propDefinition: [
+        dataforseo,
+        "tag",
+      ],
+    },
+  },
+  methods: {
+    getCrawledPage(args = {}) {
+      return this.dataforseo._makeRequest({
+        path: "/on_page/pages",
+        method: "post",
+        ...args,
+      });
+    },
+  },
+  async run({ $ }) {
+    const response = await this.getCrawledPage({
+      $,
+      data: [
+        {
+          id: this.taskId,
+          limit: this.limit,
+          search_after_token: this.searchAfterToken,
+          tag: this.tag,
+        },
+      ],
+    });
+
+    if (response.status_code !== 20000) {
+      throw new ConfigurationError(`Error: ${response.status_message}`);
+    }
+
+    if (response.tasks[0].status_code !== 20000) {
+      throw new ConfigurationError(`Error: ${response.tasks[0].status_message}`);
+    }
+
+    $.export("$summary", "Successfully retrieved crawled pages.");
+    return response;
+  },
+};

components/dataforseo/actions/parse-page-content/parse-page-content.mjs

@@ -3,10 +3,10 @@ import dataforseo from "../../dataforseo.app.mjs";
 
 export default {
   key: "dataforseo-parse-page-content",
-  name: "Parse Page Content",
+  name: "Parse Page Content with OnPage",
   description:
     "Parse the content on any page and return its structured content. [See the documentation](https://docs.dataforseo.com/v3/on_page/content_parsing/live/)",
-  version: "0.0.3",
+  version: "0.0.4",
   annotations: {
     destructiveHint: false,
     openWorldHint: true,

components/dataforseo/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@pipedream/dataforseo",
-  "version": "0.3.0",
+  "version": "0.4.0",
   "description": "Pipedream DataForSEO Components",
   "main": "dataforseo.app.mjs",
   "keywords": [
@@ -13,6 +13,6 @@
     "access": "public"
   },
   "dependencies": {
-    "@pipedream/platform": "^3.0.3"
+    "@pipedream/platform": "^3.1.0"
   }
 }