Parse Page Content

Author: GTFalcao

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

@@ -0,0 +1,71 @@
+import { parseObjectEntries } from "../../common/utils.mjs";
+import dataforseo from "../../dataforseo.app.mjs";
+
+export default {
+  key: "dataforseo-parse-page-content",
+  name: "Parse Page Content",
+  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: "0.0.1",
+  type: "action",
+  methods: {
+    parsePageContent(args = {}) {
+      return this._makeRequest({
+        path: "/on_page/content_parsing/live",
+        method: "post",
+        ...args,
+      });
+    },
+  },
+  props: {
+    dataforseo,
+    url: {
+      type: "string",
+      label: "URL",
+      description:
+        "The URL of the page to parse, e.g. `https://pipedream.com/`",
+    },
+    customUserAgent: {
+      type: "string",
+      label: "Custom User Agent",
+      description: "Custom user agent for crawling a website. Default is `Mozilla/5.0 (compatible; RSiteAuditor)`",
+      optional: true,
+    },
+    storeRawHtml: {
+      type: "boolean",
+      label: "Store Raw HTML",
+      description: "Set to `true` if you want to get the HTML of the page using the [https://docs.dataforseo.com/v3/on_page/raw_html/](OnPage Raw HTML endpoint)",
+      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,
+    },
+    additionalOptions: {
+      propDefinition: [
+        dataforseo,
+        "additionalOptions",
+      ],
+      description:
+        "Additional parameters to send in the request. [See the documentation](https://docs.dataforseo.com/v3/on_page/content_parsing/live/) for all available parameters. Values will be parsed as JSON where applicable.",
+    },
+  },
+  async run({ $ }) {
+    const response = await this.parsePageContent({
+      $,
+      data: [
+        {
+          url: this.url,
+          custom_user_agent: this.customUserAgent,
+          store_raw_html: this.storeRawHtml,
+          enable_javascript: this.enableJavascript,
+          ...parseObjectEntries(this.additionalOptions),
+        },
+      ],
+    });
+    $.export("$summary", "Successfully parsed page content");
+    return response;
+  },
+};

components/dataforseo/dataforseo.app.mjs

@@ -25,6 +25,7 @@ export default {
       label: "Location Coordinate",
       description:
         "The location to search, in the format `latitude,longitude,radius` where radius is specified in kilometers. Example: `53.476225,-2.243572,200`",
+
     },
     targetType: {
       type: "string",