pi-lean-dimension 0.1.0 → 0.2.0

package/node_modules/pi-lean-portal/backends/python-base/pi_browser_bridge/accessibility.py

@@ -7,91 +7,28 @@ parsed nodes so the TypeScript adapter can map interactions back via
 get_by_role().
 
 Mirrors the logic in core/shared/accessibility-tree.ts.
+
+Role sets and icons are loaded from the shared ``browser-data.json``
+— the same file used by the TypeScript side so the two
+implementations never drift.
 """
 
 import re
 from dataclasses import dataclass, field
 from typing import Optional
 
+from .browser_data import ACCESSIBILITY
+
 # ─── Role classification sets ─────────────────────────────────────────
 
-INTERACTIVE_ROLES: set[str] = {
-    "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",
-}
-
-INFORMATIONAL_ROLES: set[str] = {
-    "paragraph",
-    "text",
-    "group",
-    "region",
-    "main",
-    "complementary",
-    "contentinfo",
-    "definition",
-    "term",
-    "math",
-    "marquee",
-    "timer",
-    "log",
-    "deletion",
-    "insertion",
-    "mark",
-    "suggestion",
-    "comment",
-}
+INTERACTIVE_ROLES: set[str] = set(ACCESSIBILITY["interactiveRoles"])
+
+INFORMATIONAL_ROLES: set[str] = set(ACCESSIBILITY["informationalRoles"])
 
 
 # ─── Data structures ──────────────────────────────────────────────────
 
+
 @dataclass
 class AriaCachedNode:
     ref: str
@@ -113,82 +50,10 @@ class AriaParseResult:
     count: int = 0
 
 
-# ─── Role icons (same mapping as TypeScript) ───────────────—──────────
-
-_ROLE_ICONS: dict[str, str] = {
-    "button": "\U0001f518 ",        # 🔘
-    "link": "\U0001f517 ",          # 🔗
-    "textbox": "\U0001f4dd ",       # 📝
-    "searchbox": "\U0001f50d ",     # 🔍
-    "combobox": "\U0001f4cb ",      # 📋
-    "checkbox": "\u2611 ",          # ☑
-    "radio": "\u25cb ",             # ○
-    "heading": "\U0001f4cc ",       # 📌
-    "listbox": "\U0001f4cb ",       # 📋
-    "option": "\u2022 ",            # •
-    "tab": "\U0001f4d1 ",           # 📑
-    "switch": "\U0001f500 ",        # 🔀
-    "slider": "\U0001f527 ",        # 🔧
-    "spinbutton": "\U0001f522 ",    # 🔢
-    "img": "\U0001f5bc ",          # 🖼
-    "figure": "\U0001f5bc ",       # 🖼
-    "table": "\U0001f4ca ",         # 📊
-    "grid": "\U0001f4ca ",          # 📊
-    "gridcell": "\u25ab ",          # ▫
-    "cell": "\u25ab ",              # ▫
-    "dialog": "\U0001f4ac ",        # 💬
-    "alertdialog": "\u26a0 ",       # ⚠
-    "navigation": "\U0001f9ed ",    # 🧭
-    "banner": "\U0001f4f0 ",        # 📰
-    "main": "\U0001f4c4 ",          # 📄
-    "complementary": "\U0001f4ce ", # 📎
-    "contentinfo": "\u2139 ",       # ℹ
-    "form": "\U0001f4dd ",          # 📝
-    "search": "\U0001f50d ",        # 🔍
-    "group": "\U0001f4e6 ",         # 📦
-    "toolbar": "\U0001f527 ",       # 🔧
-    "treeitem": "\u2022 ",          # •
-    "menu": "\U0001f4cb ",          # 📋
-    "menubar": "\U0001f4cb ",       # 📋
-    "paragraph": "\U0001f4c3 ",     # 📃
-    "article": "\U0001f4f0 ",       # 📰
-    "section": "\U0001f4c4 ",       # 📄
-    "list": "\U0001f4cb ",          # 📋
-    "listitem": "\u2022 ",          # •
-    "note": "\U0001f4dd ",          # 📝
-    "alert": "\U0001f514 ",         # 🔔
-    "status": "\U0001f4ca ",        # 📊
-    "code": "\U0001f4bb ",          # 💻
-    "blockquote": "\U0001f4ac ",   # 💬
-    "columnheader": "\U0001f4ca ", # 📊
-    "comment": "\U0001f4ac ",      # 💬
-    "definition": "\U0001f4d6 ",   # 📖
-    "deletion": "\u274c ",         # ❌
-    "insertion": "\u2795 ",        # ➕
-    "log": "\U0001f4cb ",          # 📋
-    "mark": "\U0001f58d \ufe0f ", # 🖍️
-    "marquee": "\U0001f4dc ",      # 📜
-    "math": "\U0001f9ee ",         # 🧮
-    "menuitem": "\U0001f4cb ",    # 📋
-    "menuitemcheckbox": "\u2611 ",  # ☑
-    "menuitemradio": "\u25cb ",    # ○
-    "meter": "\U0001f4ca ",        # 📊
-    "progressbar": "\u23f3 ",      # ⏳
-    "region": "\U0001f4e6 ",        # 📦
-    "rowheader": "\U0001f4ca ",    # 📊
-    "scrollbar": "\U0001f4dc ",    # 📜
-    "suggestion": "\U0001f4a1 ",   # 💡
-    "tabpanel": "\U0001f4d1 ",     # 📑
-    "term": "\U0001f4d6 ",         # 📖
-    "text": "\U0001f4dd ",         # 📝
-    "timer": "\u23f1 \ufe0f ",    # ⏱️
-    "tooltip": "\U0001f4a1 ",      # 💡
-    "treegrid": "\U0001f4ca ",     # 📊
-}
-
+# ─── Role icons ──────────────────────────────────────────────────────
 
 def _role_icon(role: str) -> str:
-    return _ROLE_ICONS.get(role, "")
+    return ACCESSIBILITY["roleIcons"].get(role, "")
 
 
 def _truncate(s: str, max_len: int) -> str:

package/node_modules/pi-lean-portal/backends/python-base/pi_browser_bridge/bot_detection.py

@@ -6,54 +6,29 @@ Mirrors the logic in ``core/shared/bot-detection.ts`` ``checkPage()``.
 The matching *must* run in the Python process because it needs
 ``page.title()`` and ``page.evaluate()`` for body/HTML — shipping the
 full HTML across JSON-RPC would be prohibitively expensive.  However,
-the signal lists are pure data and live here as a single source of
-truth for all Python-based browser backends.
+the signal lists are pure data and are loaded from the shared
+``browser-data.json`` file — the same file used by the TypeScript side
+so the two implementations never drift.
 """
 
 import re
 from typing import Any
 
+from .browser_data import BOT_SIGNALS
+
 #: Block-level signals — checked against BOTH title and body text.
-#: Mirror of TypeScript bot-detection.ts BLOCK_SIGNALS.
-#: Only specific challenge phrases are included — generic single words
-#: like "captcha", "cloudflare", "recaptcha" are excluded because they
-#: cause false positives on legitimate pages mentioning them in passing.
-_BLOCK_SIGNALS: tuple[str, ...] = (
-    "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",
-)
+_BLOCK_SIGNALS: tuple[str, ...] = tuple(BOT_SIGNALS["blockSignals"])
 
 #: Body-only string signals — high-specificity CDN patterns.
-#: Mirror of TypeScript bot-detection.ts BODY_ONLY_SIGNALS.
-_BODY_ONLY_SIGNALS: tuple[str, ...] = (
-    "errors.edgesuite.net",
-    "you don't have permission to access",
-)
+_BODY_ONLY_SIGNALS: tuple[str, ...] = tuple(BOT_SIGNALS["bodyOnlySignals"])
 
-#: Body-only regex patterns — checked against raw body text.
-#: Mirror of TypeScript bot-detection.ts BODY_ONLY_PATTERNS.
-_BODY_ONLY_PATTERNS: tuple[re.Pattern, ...] = (
-    re.compile(r"reference\s*#[a-f0-9]+(?:\.[a-f0-9]+)+", re.IGNORECASE),
+#: Body-only regex patterns — compiled from shared source strings.
+_BODY_ONLY_PATTERNS: tuple[re.Pattern, ...] = tuple(
+    re.compile(p, re.IGNORECASE) for p in BOT_SIGNALS["bodyOnlyPatterns"]
 )
 
-#: HTML-level CAPTCHA/widget signals (Python-only enhancement).
-_HTML_SIGNALS: tuple[str, ...] = (
-    "recaptcha",
-    "hcaptcha",
-    "turnstile",
-    "g-recaptcha",
-    "data-sitekey",
-)
+#: HTML-level CAPTCHA/widget signals.
+_HTML_SIGNALS: tuple[str, ...] = tuple(BOT_SIGNALS["htmlSignals"])
 
 
 def check_bot_detection(page: Any) -> bool: