pi-lean-dimension 0.1.0 → 0.2.1

package/node_modules/pi-lean-portal/core/shared/accessibility-tree.ts

@@ -4,8 +4,13 @@
  * Parses Playwright's page.ariaSnapshot() YAML-like output into an
  * LLM-friendly text format with @e1, @e2 element references. Caches
  * parsed nodes so interactions (click, type) can map back via getByRole().
+ *
+ * Role sets and icons are loaded from the shared ``browser-data.json``
+ * — the same file used by the Python bridge so the two sides never drift.
  */
 
+import { ACCESSIBILITY } from "./browser-data.js";
+
 /** A single parsed node from the aria snapshot, cached for interaction */
 export interface AriaCachedNode {
 	ref: string;
@@ -31,84 +36,16 @@ export interface AriaParseResult {
 
 /**
  * Roles that get @e refs and can be used for interaction.
+ * Loaded from shared browser-data.json.
  */
-export const INTERACTIVE_ROLES = new Set([
-	"button",
-	"link",
-	"textbox",
-	"searchbox",
-	"combobox",
-	"checkbox",
-	"radio",
-	"heading",
-	"listbox",
-	"option",
-	"menuitem",
-	"menuitemcheckbox",
-	"menuitemradio",
-	"tab",
-	"treeitem",
-	"switch",
-	"slider",
-	"spinbutton",
-	"progressbar",
-	"meter",
-	"scrollbar",
-	"gridcell",
-	"cell",
-	"columnheader",
-	"rowheader",
-	"tabpanel",
-	"img",
-	"figure",
-	"listitem",
-	"dialog",
-	"alertdialog",
-	"tooltip",
-	"navigation",
-	"banner",
-	"form",
-	"search",
-	"toolbar",
-	"menu",
-	"menubar",
-	"note",
-	"alert",
-	"status",
-	"list",
-	"table",
-	"grid",
-	"treegrid",
-	"article",
-	"section",
-	"blockquote",
-	"code",
-]);
+const INTERACTIVE_ROLES = new Set(ACCESSIBILITY.interactiveRoles);
 
 /**
  * Roles that are shown in the tree but DON'T get @e refs
  * (informational only, not useful click targets).
+ * Loaded from shared browser-data.json.
  */
-export const INFORMATIONAL_ROLES = new Set([
-	"paragraph",
-	"text",
-	"group",
-	"region",
-	"main",
-	"complementary",
-	"contentinfo",
-	"definition",
-	"term",
-	"math",
-	"marquee",
-	"timer",
-	"log",
-	"deletion",
-	"insertion",
-	"mark",
-	"suggestion",
-	"comment",
-]);
+const INFORMATIONAL_ROLES = new Set(ACCESSIBILITY.informationalRoles);
 
 /**
  * Parse the YAML-like output of page.ariaSnapshot().
@@ -306,77 +243,7 @@ function countLeadingSpaces(s: string): number {
 // ─── Helpers ──────────────────────────────────────────────────────────
 
 export function roleIcon(role: string): string {
-	const icons: Record<string, string> = {
-		alert: "🔔 ",
-		alertdialog: "⚠ ",
-		article: "📰 ",
-		banner: "📰 ",
-		blockquote: "💬 ",
-		button: "🔘 ",
-		cell: "▫ ",
-		checkbox: "☑ ",
-		code: "💻 ",
-		columnheader: "📊 ",
-		combobox: "📋 ",
-		comment: "💬 ",
-		complementary: "📎 ",
-		contentinfo: "ℹ ",
-		definition: "📖 ",
-		deletion: "❌ ",
-		dialog: "💬 ",
-		figure: "🖼 ",
-		form: "📝 ",
-		grid: "📊 ",
-		gridcell: "▫ ",
-		group: "📦 ",
-		heading: "📌 ",
-		img: "🖼 ",
-		insertion: "➕ ",
-		link: "🔗 ",
-		list: "📋 ",
-		listbox: "📋 ",
-		listitem: "• ",
-		log: "📋 ",
-		main: "📄 ",
-		mark: "🖍️ ",
-		marquee: "📜 ",
-		math: "🧮 ",
-		menu: "📋 ",
-		menubar: "📋 ",
-		menuitem: "📋 ",
-		menuitemcheckbox: "☑ ",
-		menuitemradio: "○ ",
-		meter: "📊 ",
-		navigation: "🧭 ",
-		note: "📝 ",
-		option: "• ",
-		paragraph: "📃 ",
-		progressbar: "⏳ ",
-		radio: "○ ",
-		region: "📦 ",
-		rowheader: "📊 ",
-		search: "🔍 ",
-		searchbox: "🔍 ",
-		scrollbar: "📜 ",
-		section: "📄 ",
-		slider: "🔧 ",
-		spinbutton: "🔢 ",
-		status: "📊 ",
-		suggestion: "💡 ",
-		switch: "🔀 ",
-		tab: "📑 ",
-		table: "📊 ",
-		tabpanel: "📑 ",
-		term: "📖 ",
-		text: "📝 ",
-		textbox: "📝 ",
-		timer: "⏱️ ",
-		toolbar: "🔧 ",
-		tooltip: "💡 ",
-		treegrid: "📊 ",
-		treeitem: "• ",
-	};
-	return icons[role] || "";
+	return ACCESSIBILITY.roleIcons[role] ?? "";
 }
 
 function truncate(s: string, max: number): string {

package/node_modules/pi-lean-portal/core/shared/bot-detection.ts

@@ -6,93 +6,54 @@
  * the router flags the navigation as bot-blocked so the agent
  * can decide how to proceed — try web-fetch, try a different URL,
  * or switch to a stealth browser backend if one is configured.
+ *
+ * Signal lists are loaded from the shared ``browser-data.json`` —
+ * the same file used by the Python bridge so the two sides never drift.
  */
 
+import { BOT_SIGNALS } from "./browser-data.js";
+
 export interface BotDetectionResult {
 	/** True if the page appears to be a bot block/challenge page */
 	isBlocked: boolean;
-	/** Confidence score 0-1 */
-	confidence: number;
-	/** Signal that triggered the detection */
-	signal?: string;
 }
 
 /**
- * Content patterns that indicate a bot block.
- *
- * Only specific challenge phrases are included — generic single words
- * like "cloudflare", "captcha", "recaptcha", "hcaptcha", "enable javascript"
- * etc. are excluded because they cause false positives on legitimate
- * pages (Cloudflare's own site, web scraping articles, CAPTCHA service pages).
- * Real challenge pages always use these exact phrases.
+ * Content patterns that indicate a bot block — loaded from shared data.
  */
-const BLOCK_SIGNALS = [
-	"please verify you are human",
-	"attention required!",
-	"just a moment...",
-	"checking your browser",
-	"you have been blocked",
-	"sorry, you have been blocked",
-	"verify you are human",
-	"your request has been blocked",
-	"we are checking your browser",
-	"cf-challenge",
-	"_cf_chl_opt",
-	"cdn-cgi/challenge",
-];
+const BLOCK_SIGNALS = BOT_SIGNALS.blockSignals;
 
 /**
- * Body-only string patterns — checked in body text (not the title) to
- * avoid false matches on legitimate content.  Catches CDN-specific
- * block pages (Akamai "Access Denied", generic 403s).
+ * Body-only string patterns — loaded from shared data.
  */
-const BODY_ONLY_SIGNALS = [
-	"errors.edgesuite.net",
-	"you don't have permission to access",
-];
+const BODY_ONLY_SIGNALS = BOT_SIGNALS.bodyOnlySignals;
 
 /**
- * Body-only regex patterns — more specific than string inclusion.
- * These match the exact format of CDN error reference codes to
- * avoid false positives from generic "reference #123" in normal content.
+ * Body-only regex patterns — compiled from shared data at module load.
  */
-const BODY_ONLY_PATTERNS: RegExp[] = [
-	/reference\s*#[a-f0-9]+(?:\.[a-f0-9]+)+/i,
-];
+const BODY_ONLY_PATTERNS: RegExp[] = BOT_SIGNALS.bodyOnlyPatterns.map(
+	(src) => new RegExp(src, "i"),
+);
 
 /**
- * HTML-level signals — checked against ``document.documentElement.innerHTML``
- * rather than visible body text.  These look for CAPTCHA widget embed codes
- * in the raw HTML source, which is a stronger signal than visible text
- * (generic mentions of "captcha" in body copy are excluded to avoid false
- * positives, but embed codes in HTML attributes are unambiguous).
+ * HTML-level signals — loaded from shared data.
  */
-const HTML_SIGNALS = [
-	"recaptcha",
-	"hcaptcha",
-	"turnstile",
-	"g-recaptcha",
-	"data-sitekey",
-];
+const HTML_SIGNALS = BOT_SIGNALS.htmlSignals;
 
 /**
  * Check if page text content suggests a bot block.
  */
-export function checkBodyText(bodyText: string): BotDetectionResult {
-	if (!bodyText) return { isBlocked: false, confidence: 0 };
+function checkBodyText(bodyText: string): BotDetectionResult {
+	if (!bodyText) return { isBlocked: false };
 
 	const lower = bodyText.toLowerCase();
 	for (const signal of BLOCK_SIGNALS) {
 		if (lower.includes(signal)) {
-			return {
-				isBlocked: true,
-				confidence: signal.length > 20 ? 0.9 : 0.7,
-				signal,
-			};
+			return { isBlocked: true };
 		}
 	}
 
-	return { isBlocked: false, confidence: 0 };
+	return { isBlocked: false };
 }
 
 /**
@@ -102,14 +63,14 @@ export function checkBodyText(bodyText: string): BotDetectionResult {
  * (for Akamai reference codes, etc.).
  */
 function checkBodyOnlyText(bodyText: string): BotDetectionResult {
-	if (!bodyText) return { isBlocked: false, confidence: 0 };
+	if (!bodyText) return { isBlocked: false };
 
 	const lower = bodyText.toLowerCase();
 
 	// Check string signals first
 	for (const signal of BODY_ONLY_SIGNALS) {
 		if (lower.includes(signal)) {
-			return { isBlocked: true, confidence: 0.85, signal };
+			return { isBlocked: true };
 		}
 	}
 
@@ -117,39 +78,32 @@ function checkBodyOnlyText(bodyText: string): BotDetectionResult {
 	for (const pattern of BODY_ONLY_PATTERNS) {
 		const match = bodyText.match(pattern);
 		if (match) {
-			return {
-				isBlocked: true,
-				confidence: 0.9,
-				signal: `regex: ${match[0].slice(0, 50)}`,
-			};
+			return { isBlocked: true };
 		}
 	}
 
-	return { isBlocked: false, confidence: 0 };
+	return { isBlocked: false };
 }
 
 /**
  * Check raw HTML content for CAPTCHA widget embed codes.
  *
  * Looks for specific widget identifiers in the HTML source (not visible text),
- * so false positive risk from generic mentions is low.  Mirrors the Python
- * bridge's ``_HTML_SIGNALS`` in ``backends/chromium-py/bridge.py``.
+ * so false positive risk from generic mentions is low.  The signal list is
+ * loaded from the shared ``browser-data.json`` so TypeScript and Python
+ * bridge implementations share a single source of truth.
  */
-export function checkHtmlContent(html: string): BotDetectionResult {
-	if (!html) return { isBlocked: false, confidence: 0 };
+function checkHtmlContent(html: string): BotDetectionResult {
+	if (!html) return { isBlocked: false };
 
 	const lower = html.toLowerCase();
 	for (const signal of HTML_SIGNALS) {
 		if (lower.includes(signal)) {
-			return {
-				isBlocked: true,
-				confidence: 0.85,
-				signal: `html:${signal}`,
-			};
+			return { isBlocked: true };
 		}
 	}
 
-	return { isBlocked: false, confidence: 0 };
+	return { isBlocked: false };
 }
 
 /**
@@ -183,5 +137,5 @@ export function checkPage(
 		if (htmlResult.isBlocked) return htmlResult;
 	}
 
-	return { isBlocked: false, confidence: 0 };
+	return { isBlocked: false };
 }

package/node_modules/pi-lean-portal/core/shared/browser-data.json

@@ -0,0 +1,183 @@
+{
+  "version": 1,
+  "botSignals": {
+    "blockSignals": [
+      "please verify you are human",
+      "attention required!",
+      "just a moment...",
+      "checking your browser",
+      "you have been blocked",
+      "sorry, you have been blocked",
+      "verify you are human",
+      "your request has been blocked",
+      "we are checking your browser",
+      "cf-challenge",
+      "_cf_chl_opt",
+      "cdn-cgi/challenge"
+    ],
+    "bodyOnlySignals": [
+      "errors.edgesuite.net",
+      "you don't have permission to access"
+    ],
+    "bodyOnlyPatterns": [
+      "reference\\s*#[a-f0-9]+(?:\\.[a-f0-9]+)+"
+    ],
+    "htmlSignals": [
+      "recaptcha",
+      "hcaptcha",
+      "turnstile",
+      "g-recaptcha",
+      "data-sitekey"
+    ]
+  },
+  "accessibility": {
+    "interactiveRoles": [
+      "button",
+      "link",
+      "textbox",
+      "searchbox",
+      "combobox",
+      "checkbox",
+      "radio",
+      "heading",
+      "listbox",
+      "option",
+      "menuitem",
+      "menuitemcheckbox",
+      "menuitemradio",
+      "tab",
+      "treeitem",
+      "switch",
+      "slider",
+      "spinbutton",
+      "progressbar",
+      "meter",
+      "scrollbar",
+      "gridcell",
+      "cell",
+      "columnheader",
+      "rowheader",
+      "tabpanel",
+      "img",
+      "figure",
+      "listitem",
+      "dialog",
+      "alertdialog",
+      "tooltip",
+      "navigation",
+      "banner",
+      "form",
+      "search",
+      "toolbar",
+      "menu",
+      "menubar",
+      "note",
+      "alert",
+      "status",
+      "list",
+      "table",
+      "grid",
+      "treegrid",
+      "article",
+      "section",
+      "blockquote",
+      "code"
+    ],
+    "informationalRoles": [
+      "paragraph",
+      "text",
+      "group",
+      "region",
+      "main",
+      "complementary",
+      "contentinfo",
+      "definition",
+      "term",
+      "math",
+      "marquee",
+      "timer",
+      "log",
+      "deletion",
+      "insertion",
+      "mark",
+      "suggestion",
+      "comment"
+    ],
+    "roleIcons": {
+      "alert": "🔔 ",
+      "alertdialog": "⚠ ",
+      "article": "📰 ",
+      "banner": "📰 ",
+      "blockquote": "💬 ",
+      "button": "🔘 ",
+      "cell": "▫ ",
+      "checkbox": "☑ ",
+      "code": "💻 ",
+      "columnheader": "📊 ",
+      "combobox": "📋 ",
+      "comment": "💬 ",
+      "complementary": "📎 ",
+      "contentinfo": "ℹ ",
+      "definition": "📖 ",
+      "deletion": "❌ ",
+      "dialog": "💬 ",
+      "figure": "🖼 ",
+      "form": "📝 ",
+      "grid": "📊 ",
+      "gridcell": "▫ ",
+      "group": "📦 ",
+      "heading": "📌 ",
+      "img": "🖼 ",
+      "insertion": "➕ ",
+      "link": "🔗 ",
+      "list": "📋 ",
+      "listbox": "📋 ",
+      "listitem": "• ",
+      "log": "📋 ",
+      "main": "📄 ",
+      "mark": "🖍️ ",
+      "marquee": "📜 ",
+      "math": "🧮 ",
+      "menu": "📋 ",
+      "menubar": "📋 ",
+      "menuitem": "📋 ",
+      "menuitemcheckbox": "☑ ",
+      "menuitemradio": "○ ",
+      "meter": "📊 ",
+      "navigation": "🧭 ",
+      "note": "📝 ",
+      "option": "• ",
+      "paragraph": "📃 ",
+      "progressbar": "⏳ ",
+      "radio": "○ ",
+      "region": "📦 ",
+      "rowheader": "📊 ",
+      "search": "🔍 ",
+      "searchbox": "🔍 ",
+      "scrollbar": "📜 ",
+      "section": "📄 ",
+      "slider": "🔧 ",
+      "spinbutton": "🔢 ",
+      "status": "📊 ",
+      "suggestion": "💡 ",
+      "switch": "🔀 ",
+      "tab": "📑 ",
+      "table": "📊 ",
+      "tabpanel": "📑 ",
+      "term": "📖 ",
+      "text": "📝 ",
+      "textbox": "📝 ",
+      "timer": "⏱️ ",
+      "toolbar": "🔧 ",
+      "tooltip": "💡 ",
+      "treegrid": "📊 ",
+      "treeitem": "• "
+    }
+  },
+  "navSettle": {
+    "navTimeoutMs": 5000,
+    "settleTimeoutMs": 400,
+    "settleRaceMs": 150,
+    "domStabilizationJs": "new Promise((resolve) => {\n    const count = document.querySelectorAll(\"*\").length;\n    setTimeout(() => {\n        resolve(document.querySelectorAll(\"*\").length === count || count > 5000);\n    }, 400);\n})"
+  }
+}

package/node_modules/pi-lean-portal/core/shared/browser-data.ts

@@ -0,0 +1,50 @@
+/**
+ * Typed loader for shared browser data (bot-detection signals,
+ * accessibility roles/icons, nav-settle constants).
+ *
+ * Imports from the sibling JSON file so the same source of truth is
+ * available to both TypeScript and Python consumers.  The import
+ * assertion ensures the JSON is loaded directly — no runtime path
+ * resolution needed, which keeps vitest and the pi runtime happy.
+ *
+ * @module
+ */
+
+import _raw from "./browser-data.json" with { type: "json" };
+
+// ─── Types ─────────────────────────────────────────────────────────────
+
+export interface BotSignals {
+	blockSignals: string[];
+	bodyOnlySignals: string[];
+	bodyOnlyPatterns: string[];
+	htmlSignals: string[];
+}
+
+export interface AccessibilityData {
+	interactiveRoles: string[];
+	informationalRoles: string[];
+	roleIcons: Record<string, string>;
+}
+
+export interface NavSettleData {
+	navTimeoutMs: number;
+	settleTimeoutMs: number;
+	settleRaceMs: number;
+	domStabilizationJs: string;
+}
+
+export interface BrowserDataFile {
+	version: number;
+	botSignals: BotSignals;
+	accessibility: AccessibilityData;
+	navSettle: NavSettleData;
+}
+
+const _data = _raw as BrowserDataFile;
+
+// ─── Exports ───────────────────────────────────────────────────────────
+
+export const BOT_SIGNALS: BotSignals = _data.botSignals;
+export const ACCESSIBILITY: AccessibilityData = _data.accessibility;
+export const NAV_SETTLE: NavSettleData = _data.navSettle;

package/node_modules/pi-lean-portal/core/shared/browser-events.ts

@@ -10,7 +10,7 @@
 
 import type { Page } from "playwright";
 
-export interface DialogEvent {
+export interface DialogLogEntry {
 	type: "alert" | "confirm" | "prompt" | "beforeunload";
 	message: string;
 	/** Default value for prompt dialogs */
@@ -35,13 +35,13 @@ export interface ConsoleEvent {
 }
 
 /** Dialogs logged per task, for reporting to user */
-const _dialogLog = new Map<string, DialogEvent[]>();
+const _dialogLog = new Map<string, DialogLogEntry[]>();
 
 /** Console messages logged per task */
 const _consoleLog = new Map<string, ConsoleEvent[]>();
 
 /** Get logged dialogs for a task */
-export function getDialogLog(taskId: string): DialogEvent[] {
+export function getDialogLog(taskId: string): DialogLogEntry[] {
 	return _dialogLog.get(taskId) ?? [];
 }
 
@@ -61,7 +61,7 @@ export function clearConsoleLog(taskId: string): void {
  * Captures console messages (log, warn, error, info, debug) for retrieval.
  */
 export function installDialogHandlers(taskId: string, page: Page): void {
-	const dialogLog: DialogEvent[] = [];
+	const dialogLog: DialogLogEntry[] = [];
 	_dialogLog.set(taskId, dialogLog);
 
 	const consoleLog: ConsoleEvent[] = [];
@@ -69,8 +69,8 @@ export function installDialogHandlers(taskId: string, page: Page): void {
 
 	// Auto-accept JavaScript dialogs
 	page.on("dialog", async (dialog) => {
-		const entry: DialogEvent = {
-			type: dialog.type() as DialogEvent["type"],
+		const entry: DialogLogEntry = {
+			type: dialog.type() as DialogLogEntry["type"],
 			message: dialog.message(),
 			defaultValue: dialog.defaultValue(),
 			handledAs: "accepted",