ff-automationv2 1.0.0 → 2.0.1-beta.0

package/src/domAnalysis/simplifyAndFlatten.ts

@@ -1,118 +1,118 @@
-import * as cheerio from "cheerio";
-import { Element } from "domhandler";
-
-export interface FlatNode {
-    node: Record<string, any>;
-    search: string;
-}
-
-export class DomSimplifier {
-
-    public simplify(rawDom: string): FlatNode[] {
-
-        const $ = cheerio.load(rawDom);
-        const flatNodes: FlatNode[] = [];
-
-        $("*").each((_, node) => {
-
-            if (node.type !== "tag") return;
-
-            const el = node as Element;
-            const tag = el.tagName.toLowerCase();
-            if (!tag || tag === "meta") return;
-
-            const attribs = el.attribs ?? {};   // ✅ declared only once
-
-            if (tag !== "select" && tag !== "option") {
-                if (
-                    (attribs["element-visibility"] === "hidden" ||
-                        attribs["element-covered"] === "true") &&
-                    attribs["element-covered-by-tag"] !== "iframe"
-                ) {
-                    return;
-                }
-            }
-
-            const simplified: Record<string, any> = { tag };
-
-            if (attribs["ff-inspect"])
-                simplified["FF-inspecter"] = attribs["ff-inspect"];
-
-            if (attribs["ff-xpath"])
-                simplified["ff-xpath"] = attribs["ff-xpath"];
-
-            // -----------------
-            // Direct text only
-            // -----------------
-            let directText = "";
-
-            $(el).contents().each((_, child) => {
-                if (child.type === "text") {
-                    const txt = child.data?.trim();
-                    if (txt) directText += txt;
-                }
-            });
-
-            const cleanText = directText
-                .replace(/[^a-zA-Z0-9\s]/g, " ")
-                .trim();
-
-            if (cleanText)
-                simplified["text"] = cleanText.slice(0, 200);
-
-            // -----------------
-            // Copy attributes
-            // -----------------
-            const EXCLUDED_ATTRS = new Set([
-                "style", "onclick", "onchange", "href",
-                "oninput", "onmouseover", "onfocus",
-                "ff-inspect", "element-visibility",
-                "element-pointer-events", "element-zindex", "element-covered"
-            ]);
-
-            Object.entries(attribs).forEach(([key, val]) => {
-                if (!EXCLUDED_ATTRS.has(key.toLowerCase())) {
-                    simplified[key] = val;
-                }
-            });
-
-            // -----------------
-            // Build search
-            // -----------------
-            const parts = new Set<string>();
-
-            if (cleanText && cleanText.length <= 70) {
-                parts.add(cleanText.toLowerCase());
-            }
-
-            const SEARCH_EXCLUDED = new Set([
-                "class", "style", "onclick", "onchange",
-                "oninput", "onmouseover", "onfocus",
-                "tabindex", "role", "href"
-            ]);
-
-            Object.entries(simplified).forEach(([key, value]) => {
-                if (!value) return;
-                if (SEARCH_EXCLUDED.has(key.toLowerCase())) return;
-                if (["tag", "FF-inspecter", "search", "ff-xpath"].includes(key)) return;
-
-                const v = String(value)
-                    .replace(/[^a-zA-Z0-9\s]/g, " ")
-                    .trim()
-                    .toLowerCase();
-
-                if (v) parts.add(v);
-            });
-
-            simplified["search"] = Array.from(parts).join(" ").trim();
-
-            flatNodes.push({
-                node: simplified,
-                search: simplified["search"]
-            });
-
-        });
-
-        return flatNodes;
-    }
-}
+import * as cheerio from "cheerio";
+import { Element } from "domhandler";
+
+export interface FlatNode {
+    node: Record<string, any>;
+    search: string;
+}
+
+export class DomSimplifier {
+
+    public simplify(rawDom: string): FlatNode[] {
+
+        const $ = cheerio.load(rawDom);
+        const flatNodes: FlatNode[] = [];
+
+        $("*").each((_, node) => {
+
+            if (node.type !== "tag") return;
+
+            const el = node as Element;
+            const tag = el.tagName.toLowerCase();
+            if (!tag || tag === "meta") return;
+
+            const attribs = el.attribs ?? {};   // ✅ declared only once
+
+            if (tag !== "select" && tag !== "option") {
+                if (
+                    (attribs["element-visibility"] === "hidden" ||
+                        attribs["element-covered"] === "true") &&
+                    attribs["element-covered-by-tag"] !== "iframe"
+                ) {
+                    return;
+                }
+            }
+
+            const simplified: Record<string, any> = { tag };
+
+            if (attribs["ff-inspect"])
+                simplified["FF-inspecter"] = attribs["ff-inspect"];
+
+            if (attribs["ff-xpath"])
+                simplified["ff-xpath"] = attribs["ff-xpath"];
+
+            // -----------------
+            // Direct text only
+            // -----------------
+            let directText = "";
+
+            $(el).contents().each((_, child) => {
+                if (child.type === "text") {
+                    const txt = child.data?.trim();
+                    if (txt) directText += txt;
+                }
+            });
+
+            const cleanText = directText
+                .replace(/[^a-zA-Z0-9\s]/g, " ")
+                .trim();
+
+            if (cleanText)
+                simplified["text"] = cleanText.slice(0, 200);
+
+            // -----------------
+            // Copy attributes
+            // -----------------
+            const EXCLUDED_ATTRS = new Set([
+                "style", "onclick", "onchange", "href",
+                "oninput", "onmouseover", "onfocus",
+                "ff-inspect", "element-visibility",
+                "element-pointer-events", "element-zindex", "element-covered"
+            ]);
+
+            Object.entries(attribs).forEach(([key, val]) => {
+                if (!EXCLUDED_ATTRS.has(key.toLowerCase())) {
+                    simplified[key] = val;
+                }
+            });
+
+            // -----------------
+            // Build search
+            // -----------------
+            const parts = new Set<string>();
+
+            if (cleanText && cleanText.length <= 70) {
+                parts.add(cleanText.toLowerCase());
+            }
+
+            const SEARCH_EXCLUDED = new Set([
+                "class", "style", "onclick", "onchange",
+                "oninput", "onmouseover", "onfocus",
+                "tabindex", "role", "href"
+            ]);
+
+            Object.entries(simplified).forEach(([key, value]) => {
+                if (!value) return;
+                if (SEARCH_EXCLUDED.has(key.toLowerCase())) return;
+                if (["tag", "FF-inspecter", "search", "ff-xpath"].includes(key)) return;
+
+                const v = String(value)
+                    .replace(/[^a-zA-Z0-9\s]/g, " ")
+                    .trim()
+                    .toLowerCase();
+
+                if (v) parts.add(v);
+            });
+
+            simplified["search"] = Array.from(parts).join(" ").trim();
+
+            flatNodes.push({
+                node: simplified,
+                search: simplified["search"]
+            });
+
+        });
+
+        return flatNodes;
+    }
+}