ghostcrawl-langchain 2.1.0__py3-none-any.whl

ghostcrawl_langchain/__init__.py

@@ -0,0 +1,34 @@
+"""ghostcrawl-langchain — LangChain tool wrappers for the GhostCrawl SaaS API."""
+from .errors import (
+    GhostcrawlAuthError,
+    GhostcrawlError,
+    GhostcrawlQuotaError,
+    GhostcrawlRateLimitError,
+    GhostcrawlServerError,
+)
+from .tools import (
+    GhostcrawlCrawlTool,
+    GhostcrawlExtractTool,
+    GhostcrawlGoogleHotelsTool,
+    GhostcrawlGoogleSearchTool,
+    GhostcrawlGoogleSportsTool,
+    GhostcrawlScrapeTool,
+    GhostcrawlSearchTool,
+    GhostcrawlUsageTool,
+)
+
+__all__ = [
+    "GhostcrawlScrapeTool",
+    "GhostcrawlSearchTool",
+    "GhostcrawlGoogleSearchTool",
+    "GhostcrawlGoogleHotelsTool",
+    "GhostcrawlGoogleSportsTool",
+    "GhostcrawlExtractTool",
+    "GhostcrawlCrawlTool",
+    "GhostcrawlUsageTool",
+    "GhostcrawlError",
+    "GhostcrawlAuthError",
+    "GhostcrawlQuotaError",
+    "GhostcrawlRateLimitError",
+    "GhostcrawlServerError",
+]

ghostcrawl_langchain/_client.py

@@ -0,0 +1,84 @@
+"""Shared httpx client factory for ghostcrawl-langchain tools."""
+from __future__ import annotations
+import os
+import platform
+import sys
+from importlib.metadata import PackageNotFoundError, version
+
+import httpx
+
+
+def _ua() -> str:
+    try:
+        v = version("ghostcrawl-langchain")
+    except PackageNotFoundError:
+        v = "0.0.0"
+    return (
+        f"ghostcrawl-langchain/{v} "
+        f"Python/{sys.version_info.major}.{sys.version_info.minor} "
+        f"{platform.system()}"
+    )
+
+
+def get_client(
+    api_key: str | None = None,
+    base_url: str | None = None,
+) -> httpx.Client:
+    """Return a configured httpx.Client with Bearer auth and branded User-Agent.
+
+    Auth is read from the ``api_key`` argument first, then the
+    ``GHOSTCRAWL_API_KEY`` environment variable.  Raises ``ValueError`` if
+    neither is set.
+
+    The ``base_url`` argument overrides ``GHOSTCRAWL_BASE_URL`` (defaults to
+    ``https://api.ghostcrawl.io``).
+    """
+    key = (api_key or os.environ.get("GHOSTCRAWL_API_KEY", "")).strip()
+    if not key:
+        raise ValueError(
+            "ghostcrawl-langchain: GHOSTCRAWL_API_KEY is not set. "
+            "Set the GHOSTCRAWL_API_KEY environment variable or pass api_key to get_client()."
+        )
+    resolved_base = (
+        base_url
+        or os.environ.get("GHOSTCRAWL_BASE_URL", "https://api.ghostcrawl.io")
+    ).rstrip("/")
+    return httpx.Client(
+        base_url=resolved_base,
+        headers={
+            "Authorization": f"Bearer {key}",
+            "User-Agent": _ua(),
+            "Content-Type": "application/json",
+        },
+        timeout=60.0,
+    )
+
+
+def get_async_client(
+    api_key: str | None = None,
+    base_url: str | None = None,
+) -> httpx.AsyncClient:
+    """Return a configured ``httpx.AsyncClient`` mirroring :func:`get_client`.
+
+    Used by LangChain BaseTool ``_arun`` overrides so the async path does not
+    block the event loop on synchronous I/O.
+    """
+    key = (api_key or os.environ.get("GHOSTCRAWL_API_KEY", "")).strip()
+    if not key:
+        raise ValueError(
+            "ghostcrawl-langchain: GHOSTCRAWL_API_KEY is not set. "
+            "Set the GHOSTCRAWL_API_KEY environment variable or pass api_key to get_async_client()."
+        )
+    resolved_base = (
+        base_url
+        or os.environ.get("GHOSTCRAWL_BASE_URL", "https://api.ghostcrawl.io")
+    ).rstrip("/")
+    return httpx.AsyncClient(
+        base_url=resolved_base,
+        headers={
+            "Authorization": f"Bearer {key}",
+            "User-Agent": _ua(),
+            "Content-Type": "application/json",
+        },
+        timeout=60.0,
+    )

ghostcrawl_langchain/errors.py

@@ -0,0 +1,33 @@
+"""ghostcrawl-langchain error classes.
+
+This is the canonical Python status_code → exception class mapping; the Node SDK
+(sdks/node/src/errors.ts) and the other official SDK surfaces are kept in lockstep
+with it.
+
+Third-party attribution: see sdks/python-langchain/LICENSE-NOTICE.md.
+"""
+from __future__ import annotations
+
+
+class GhostcrawlError(Exception):
+    """Base class for all ghostcrawl-langchain errors."""
+
+
+class GhostcrawlAuthError(GhostcrawlError):
+    """401 — Invalid API key or missing Authorization header."""
+
+
+class GhostcrawlQuotaError(GhostcrawlError):
+    """402 — Quota exceeded; top up at https://ghostcrawl.io/billing."""
+
+
+class GhostcrawlRateLimitError(GhostcrawlError):
+    """429 — Rate limited; check retry_after attribute (seconds)."""
+
+    def __init__(self, message: str, retry_after: int | None = None):
+        super().__init__(message)
+        self.retry_after = retry_after
+
+
+class GhostcrawlServerError(GhostcrawlError):
+    """5xx — Retryable server-side error."""

ghostcrawl_langchain/tools.py

@@ -0,0 +1,778 @@
+"""LangChain BaseTool subclasses for GhostCrawl REST endpoints.
+
+Auth: Authorization: Bearer <token>.
+
+- _handle_error maps httpx.HTTPStatusError to Ghostcrawl* domain subclasses.
+- All _run() bodies wrap the httpx call in try/except -> _handle_error.
+- All _arun() bodies use httpx.AsyncClient.
+"""
+from __future__ import annotations
+
+import ast
+import json
+from typing import Any, Dict, NoReturn, Optional, Type
+from urllib.parse import parse_qs
+
+import httpx
+from langchain_core.tools import BaseTool
+from pydantic import BaseModel, Field, field_validator
+
+from ._client import get_async_client, get_client
+from .errors import (
+    GhostcrawlAuthError,
+    GhostcrawlError,
+    GhostcrawlQuotaError,
+    GhostcrawlRateLimitError,
+    GhostcrawlServerError,
+)
+
+
+# ---------------------------------------------------------------------------
+# Error mapping
+# ---------------------------------------------------------------------------
+
+def _handle_error(exc: httpx.HTTPStatusError) -> NoReturn:
+    """Map an ``httpx.HTTPStatusError`` to the right ``Ghostcrawl*`` subclass.
+
+    Mapping is kept lock-step across the official SDKs (see ``errors.py`` — the
+    canonical Python mapping, and ``sdks/node/src/errors.ts``):
+    401 → auth, 402 → quota, 429 → rate-limit (with ``retry_after``),
+    5xx → retryable server error, anything else → generic ``GhostcrawlError``.
+    """
+    status = exc.response.status_code
+    body = exc.response.text
+    if status == 401:
+        raise GhostcrawlAuthError(f"401 Unauthorized: {body}") from exc
+    if status == 402:
+        raise GhostcrawlQuotaError(f"402 Payment Required: {body}") from exc
+    if status == 429:
+        retry_after_raw = exc.response.headers.get("Retry-After")
+        try:
+            retry_after = int(retry_after_raw) if retry_after_raw is not None else None
+        except (TypeError, ValueError):
+            retry_after = None
+        raise GhostcrawlRateLimitError(
+            f"429 Too Many Requests: {body}", retry_after=retry_after
+        ) from exc
+    if 500 <= status < 600:
+        raise GhostcrawlServerError(f"{status} Server Error: {body}") from exc
+    raise GhostcrawlError(f"{status} HTTP error: {body}") from exc
+
+
+# ---------------------------------------------------------------------------
+# Input schemas
+# ---------------------------------------------------------------------------
+
+def _coerce_scalar(raw: Any) -> Any:
+    """Narrow a URL ``k=v`` value to bool/int when it matches exactly.
+
+    Pure value-narrowing — no ``eval``, no attribute access. Lists (repeated
+    keys) and anything that is not an exact true/false/digit-string pass
+    through unchanged as inert values.
+    """
+    if isinstance(raw, list):
+        return raw
+    if isinstance(raw, str):
+        low = raw.lower()
+        if low == "true":
+            return True
+        if low == "false":
+            return False
+        if raw.isdigit():
+            return int(raw)
+    return raw
+
+
+class _GhostcrawlBaseInput(BaseModel):
+    """Shared base carrying the single 3-stage ``params`` coercion validator.
+
+    Technique adapted from the Bee ``str_to_dict_validator`` reference
+    (cited by path in 140.5-13-SUMMARY.md), rewritten ghostcrawl-style and
+    forward-merged into ONE base class so every params-bearing tool input
+    inherits the same battle-tested coercion — replacing the five independent
+    JSON-only ``_coerce_params`` validators (D-05a / D-09 / D-10).
+
+    Stage 2 uses ``ast.literal_eval`` which is literal-only (dict/list/str/
+    num/bool/None): it never executes code, calls functions, or imports. A
+    call-expression string is rejected by the literal stage and (lacking
+    ``=``) falls through stage 3 to ``{}`` — never executing attacker input.
+
+    The validator is declared ``check_fields=False`` so this base (and any
+    subclass without a ``params`` field, e.g. ``_EmptyInput``) imports without
+    a PydanticUserError.
+    """
+
+    @field_validator("params", mode="before", check_fields=False)
+    @classmethod
+    def _coerce_params(cls, v: Any) -> dict:
+        if isinstance(v, dict):
+            return v
+        if not isinstance(v, str):
+            return v or {}
+        s = v.strip()
+        if not s:
+            return {}
+
+        # Stage 1 — JSON object string ({"render_js": true}).
+        try:
+            parsed = json.loads(s)
+            if isinstance(parsed, dict):
+                return parsed
+        except (json.JSONDecodeError, ValueError):
+            pass
+
+        # Stage 2 — Python-literal dict string ({'render_js': True}).
+        # literal-only via ast.literal_eval: no code execution / no imports.
+        try:
+            literal = ast.literal_eval(s)
+            if isinstance(literal, dict):
+                return literal
+        except (ValueError, SyntaxError, TypeError):
+            pass
+
+        # Stage 3 — URL k=v string (screenshot=true&wait=3000).
+        if "=" in s:
+            parsed_qs = parse_qs(s, keep_blank_values=True)
+            return {
+                key: _coerce_scalar(vals[0] if len(vals) == 1 else vals)
+                for key, vals in parsed_qs.items()
+            }
+        return {}
+
+
+class _ScrapeInput(_GhostcrawlBaseInput):
+    url: str = Field(..., description="URL to scrape (http/https).")
+    render_js: bool = Field(
+        False,
+        description="Execute JavaScript before extraction. Required for SPAs and lazy-loaded content.",
+    )
+    output_format: str = Field(
+        "markdown",
+        description="Content format to return: 'html' returns raw HTML; 'markdown' returns cleaned Markdown.",
+    )
+    params: dict = Field(
+        default_factory=dict,
+        description=(
+            "Extra /v1/scrape parameters as a dict or JSON string. "
+            "Supported keys: wait_for (CSS selector), extract_schema (JSON schema object), "
+            "country (ISO 3166-1 alpha-2), screenshot (bool), full_page (bool), "
+            "screenshot_selector (CSS selector for element-scoped capture)."
+        ),
+    )
+
+
+class _SearchInput(_GhostcrawlBaseInput):
+    query: str = Field(..., description="Search query string.")
+    engine: str = Field(
+        "brave",
+        description="Search engine: 'brave' or 'tavily'. Requires your own provider API key via X-Provider-Authorization.",
+    )
+    limit: int = Field(
+        10,
+        ge=1,
+        le=20,
+        description="Maximum number of results to return (1–20).",
+    )
+    params: dict = Field(
+        default_factory=dict,
+        description="Extra /v1/search parameters as a dict or JSON string.",
+    )
+
+
+class _ExtractInput(_GhostcrawlBaseInput):
+    url: str = Field(..., description="URL to extract structured data from (http/https).")
+    schema_: dict = Field(
+        default_factory=dict,
+        description=(
+            "JSON Schema object describing the fields to extract. "
+            "Example: {\"type\": \"object\", \"properties\": {\"title\": {\"type\": \"string\"}}}. "
+            "When omitted, ghostcrawl returns the full page content."
+        ),
+    )
+    params: dict = Field(
+        default_factory=dict,
+        description="Extra /v1/extract parameters as a dict or JSON string.",
+    )
+
+    @field_validator("schema_", mode="before")
+    @classmethod
+    def _coerce_schema(cls, v: Any) -> dict:
+        if isinstance(v, str):
+            try:
+                return json.loads(v)
+            except (json.JSONDecodeError, ValueError) as e:
+                raise ValueError(f"schema must be valid JSON: {e}") from e
+        return v or {}
+
+
+class _CrawlInput(_GhostcrawlBaseInput):
+    url: str = Field(..., description="Seed URL to start crawling from (http/https).")
+    max_pages: int = Field(
+        10,
+        ge=1,
+        le=500,
+        description="Maximum number of pages to crawl from the seed URL (1–500).",
+    )
+    same_origin: bool = Field(
+        True,
+        description="Restrict crawl to the same origin (scheme + host + port). Disable to follow cross-origin links.",
+    )
+    params: dict = Field(
+        default_factory=dict,
+        description="Extra /v1/crawl parameters as a dict or JSON string.",
+    )
+
+
+class _GoogleSearchInput(_GhostcrawlBaseInput):
+    query: str = Field(..., description="Google search query string.")
+    country: str = Field(
+        "us",
+        description="ISO 3166-1 alpha-2 country code for regional Google SERP results.",
+    )
+    params: dict = Field(
+        default_factory=dict,
+        description="Extra /v1/google/search parameters as a dict or JSON string.",
+    )
+
+
+class _GoogleHotelsInput(_GhostcrawlBaseInput):
+    """Args for the Google Hotels SERP tool (Plan 140.5-18 / D-07a / D-01 parity).
+
+    Inherits ``_GhostcrawlBaseInput`` (D-05a / plan 140.5-13) so the shared
+    3-stage ``params`` coercion validator applies — no per-class validator.
+    """
+
+    query: str = Field(..., description="Hotel search query, e.g. 'hotels in san francisco'.")
+    check_in: str = Field(..., description="Check-in date, ISO 8601 (YYYY-MM-DD).")
+    check_out: str = Field(..., description="Check-out date, ISO 8601 (YYYY-MM-DD).")
+    adults: int = Field(2, description="Number of adult guests (default 2).")
+    rooms: int = Field(1, description="Number of rooms (default 1).")
+    currency: str = Field("USD", description="ISO 4217 currency code (default USD).")
+    country: str = Field("us", description="ISO 3166-1 alpha-2 region code (default us).")
+    params: dict = Field(
+        default_factory=dict,
+        description="Extra /v1/google/hotels parameters as a dict or JSON string.",
+    )
+
+
+class _GoogleSportsInput(_GhostcrawlBaseInput):
+    """Args for the Google Sports SERP tool (Plan 140.5-19 / D-07b / D-01 parity).
+
+    Inherits ``_GhostcrawlBaseInput`` (D-05a / plan 140.5-13) so the shared
+    3-stage ``params`` coercion validator applies — no per-class validator.
+    """
+
+    query: str = Field(..., description="Sports query, e.g. 'lakers score' or 'premier league'.")
+    country: str = Field("us", description="ISO 3166-1 alpha-2 region code (default us).")
+    params: dict = Field(
+        default_factory=dict,
+        description="Extra /v1/google/sports parameters as a dict or JSON string.",
+    )
+
+
+class _EmptyInput(BaseModel):
+    """Empty args schema for tools that take no parameters.
+
+    Stays on bare ``BaseModel`` (NOT ``_GhostcrawlBaseInput``): it carries no
+    ``params`` field, so inheriting the validator buys nothing. Leaving it on
+    ``BaseModel`` keeps the empty-tool args schema minimal and decoupled.
+    """
+
+
+# ---------------------------------------------------------------------------
+# BaseTool subclasses
+# ---------------------------------------------------------------------------
+
+class GhostcrawlScrapeTool(BaseTool):
+    """LangChain tool: scrape a single URL via ghostcrawl's stealth browser fleet."""
+
+    name: str = "ghostcrawl_scrape"
+    description: str = (
+        "Scrape a webpage using ghostcrawl's stealth browser fleet and return the content "
+        "as HTML or Markdown. Suitable for news articles, landing pages, product pages, "
+        "and any publicly accessible URL. "
+        "SUPPORTED: render_js (bool — execute JavaScript before extraction), "
+        "output_format ('html' | 'markdown'), wait_for (CSS selector to await), "
+        "extract_schema (JSON schema for structured extraction), country (ISO 3166-1 alpha-2 "
+        "for geo-targeted content). "
+        "UNSUPPORTED: CAPTCHA solving, multi-step login sessions (use ghostcrawl_crawl with "
+        "agent-mode for session-based flows). "
+        "Returns: page content as HTML or Markdown plus metadata including title and HTTP status_code. "
+        "Uses POST /v1/scrape with Authorization: Bearer token auth."
+    )
+    args_schema: Type[BaseModel] = _ScrapeInput
+
+    def _run(
+        self,
+        url: str,
+        render_js: bool = False,
+        output_format: str = "markdown",
+        params: Optional[dict] = None,
+        **_: Any,
+    ) -> str:
+        body: Dict[str, Any] = {
+            "url": url,
+            "render_js": render_js,
+            "output_format": output_format,
+            **(params or {}),
+        }
+        with get_client() as c:
+            r = c.post("/v1/scrape", json=body)
+            try:
+                r.raise_for_status()
+            except httpx.HTTPStatusError as e:
+                _handle_error(e)
+            return r.text
+
+    async def _arun(
+        self,
+        url: str,
+        render_js: bool = False,
+        output_format: str = "markdown",
+        params: Optional[dict] = None,
+        **kwargs: Any,
+    ) -> str:
+        body: Dict[str, Any] = {
+            "url": url,
+            "render_js": render_js,
+            "output_format": output_format,
+            **(params or {}),
+        }
+        async with get_async_client() as c:
+            r = await c.post("/v1/scrape", json=body)
+            try:
+                r.raise_for_status()
+            except httpx.HTTPStatusError as e:
+                _handle_error(e)
+            return r.text
+
+
+class GhostcrawlSearchTool(BaseTool):
+    """LangChain tool: run a web search via Brave or Tavily through ghostcrawl."""
+
+    name: str = "ghostcrawl_search"
+    description: str = (
+        "Run a web search through ghostcrawl and return normalized results. "
+        "Supports multiple search backends: 'brave' (default) or 'tavily'. "
+        "SUPPORTED: query (the search string), engine ('brave' | 'tavily'), "
+        "limit (1–20 results), country (ISO 3166-1 alpha-2 for regional results). "
+        "UNSUPPORTED: image search, news-only verticals (use ghostcrawl_scrape on Google News). "
+        "Returns: list of results with url, title, snippet, published_at, and relevance score. "
+        "ghostcrawl charges no markup — you pay Brave or Tavily directly via your provider key "
+        "passed as X-Provider-Authorization: Bearer <YOUR_KEY>. "
+        "Uses POST /v1/search with Authorization: Bearer token auth."
+    )
+    args_schema: Type[BaseModel] = _SearchInput
+
+    def _run(
+        self,
+        query: str,
+        engine: str = "brave",
+        limit: int = 10,
+        params: Optional[dict] = None,
+        **_: Any,
+    ) -> str:
+        body: Dict[str, Any] = {
+            "query": query,
+            "engine": engine,
+            "limit": limit,
+            **(params or {}),
+        }
+        with get_client() as c:
+            r = c.post("/v1/search", json=body)
+            try:
+                r.raise_for_status()
+            except httpx.HTTPStatusError as e:
+                _handle_error(e)
+            return r.text
+
+    async def _arun(
+        self,
+        query: str,
+        engine: str = "brave",
+        limit: int = 10,
+        params: Optional[dict] = None,
+        **kwargs: Any,
+    ) -> str:
+        body: Dict[str, Any] = {
+            "query": query,
+            "engine": engine,
+            "limit": limit,
+            **(params or {}),
+        }
+        async with get_async_client() as c:
+            r = await c.post("/v1/search", json=body)
+            try:
+                r.raise_for_status()
+            except httpx.HTTPStatusError as e:
+                _handle_error(e)
+            return r.text
+
+
+class GhostcrawlGoogleSearchTool(BaseTool):
+    """LangChain tool: run a Google SERP query via ghostcrawl's /v1/google/search.
+
+    Lets LangChain users reach the Google SERP surface without dropping to
+    raw HTTP.
+    """
+
+    name: str = "ghostcrawl_google_search"
+    description: str = (
+        "Run a Google SERP query through ghostcrawl and return normalized results. "
+        "SUPPORTED: query (search string), country (ISO 3166-1 alpha-2 code for "
+        "regional results, e.g. 'us', 'gb', 'de'). "
+        "UNSUPPORTED: image search, news verticals, knowledge-panel deep extraction "
+        "(use ghostcrawl_scrape on the relevant Google result page). "
+        "Returns: list of organic results with url, title, snippet, position; plus "
+        "any featured snippet and related-searches metadata. "
+        "Differs from ghostcrawl_search (Brave/Tavily) — this hits Google directly "
+        "via /v1/google/search and does NOT require a third-party provider key. "
+        "Uses POST /v1/google/search with Authorization: Bearer token auth."
+    )
+    args_schema: Type[BaseModel] = _GoogleSearchInput
+
+    def _run(
+        self,
+        query: str,
+        country: str = "us",
+        params: Optional[dict] = None,
+        **_: Any,
+    ) -> str:
+        with get_client() as c:
+            body: Dict[str, Any] = {
+                "q": query,
+                "country_code": country,
+                **(params or {}),
+            }
+            r = c.post("/v1/google/search", json=body)
+            try:
+                r.raise_for_status()
+            except httpx.HTTPStatusError as e:
+                _handle_error(e)
+            return r.text
+
+    async def _arun(
+        self,
+        query: str,
+        country: str = "us",
+        params: Optional[dict] = None,
+        **kwargs: Any,
+    ) -> str:
+        async with get_async_client() as c:
+            body: Dict[str, Any] = {
+                "q": query,
+                "country_code": country,
+                **(params or {}),
+            }
+            r = await c.post("/v1/google/search", json=body)
+            try:
+                r.raise_for_status()
+            except httpx.HTTPStatusError as e:
+                _handle_error(e)
+            return r.text
+
+
+class GhostcrawlGoogleHotelsTool(BaseTool):
+    """LangChain tool: fetch Google Hotels listings via ghostcrawl's /v1/google/hotels.
+
+    Parity peer of ``GhostcrawlGoogleSearchTool`` (D-01 / D-07a, Plan 140.5-18).
+    Reaches the Google Travel (hotels) SERP without dropping to raw HTTP.
+    """
+
+    name: str = "ghostcrawl_google_hotels"
+    description: str = (
+        "Fetch Google Hotels (Travel) listings through ghostcrawl. "
+        "SUPPORTED: query (hotel search string), check_in / check_out (ISO 8601 "
+        "YYYY-MM-DD; check_out must be after check_in), adults (default 2), rooms "
+        "(default 1), currency (ISO 4217, default USD), country (ISO 3166-1 alpha-2). "
+        "Returns: hotels_results with name, price, total_price, rating, amenities, "
+        "booking_providers. HIGH BRITTLENESS (obfuscated Google Travel SPA classes). "
+        "Uses POST /v1/google/hotels with Authorization: Bearer token auth."
+    )
+    args_schema: Type[BaseModel] = _GoogleHotelsInput
+
+    def _run(
+        self,
+        query: str,
+        check_in: str,
+        check_out: str,
+        adults: int = 2,
+        rooms: int = 1,
+        currency: str = "USD",
+        country: str = "us",
+        params: Optional[dict] = None,
+        **_: Any,
+    ) -> str:
+        with get_client() as c:
+            body: Dict[str, Any] = {
+                "q": query,
+                "check_in": check_in,
+                "check_out": check_out,
+                "adults": adults,
+                "rooms": rooms,
+                "currency": currency,
+                "country_code": country,
+                **(params or {}),
+            }
+            r = c.post("/v1/google/hotels", json=body)
+            try:
+                r.raise_for_status()
+            except httpx.HTTPStatusError as e:
+                _handle_error(e)
+            return r.text
+
+    async def _arun(
+        self,
+        query: str,
+        check_in: str,
+        check_out: str,
+        adults: int = 2,
+        rooms: int = 1,
+        currency: str = "USD",
+        country: str = "us",
+        params: Optional[dict] = None,
+        **_: Any,
+    ) -> str:
+        async with get_async_client() as c:
+            body: Dict[str, Any] = {
+                "q": query,
+                "check_in": check_in,
+                "check_out": check_out,
+                "adults": adults,
+                "rooms": rooms,
+                "currency": currency,
+                "country_code": country,
+                **(params or {}),
+            }
+            r = await c.post("/v1/google/hotels", json=body)
+            try:
+                r.raise_for_status()
+            except httpx.HTTPStatusError as e:
+                _handle_error(e)
+            return r.text
+
+
+class GhostcrawlGoogleSportsTool(BaseTool):
+    """LangChain tool: fetch the Google sports knowledge panel via /v1/google/sports.
+
+    Parity peer of ``GhostcrawlGoogleHotelsTool`` (D-01 / D-07b, Plan 140.5-19).
+    Reaches the Google sports SERP knowledge panel without dropping to raw HTTP.
+    """
+
+    name: str = "ghostcrawl_google_sports"
+    description: str = (
+        "Fetch the Google sports knowledge panel (match summary + standings) through ghostcrawl. "
+        "SUPPORTED: query (sports search string, e.g. 'lakers score'), country (ISO 3166-1 "
+        "alpha-2, default us). "
+        "Returns: SearchResult with extras.sports_results = {match: {home_team, away_team, "
+        "scores, status}, standings: [...]}. HIGH BRITTLENESS (Google knowledge-panel classes "
+        "drift on deploy cadence). "
+        "Uses POST /v1/google/sports with Authorization: Bearer token auth."
+    )
+    args_schema: Type[BaseModel] = _GoogleSportsInput
+
+    def _run(
+        self,
+        query: str,
+        country: str = "us",
+        params: Optional[dict] = None,
+        **_: Any,
+    ) -> str:
+        with get_client() as c:
+            body: Dict[str, Any] = {
+                "q": query,
+                "country_code": country,
+                **(params or {}),
+            }
+            r = c.post("/v1/google/sports", json=body)
+            try:
+                r.raise_for_status()
+            except httpx.HTTPStatusError as e:
+                _handle_error(e)
+            return r.text
+
+    async def _arun(
+        self,
+        query: str,
+        country: str = "us",
+        params: Optional[dict] = None,
+        **_: Any,
+    ) -> str:
+        async with get_async_client() as c:
+            body: Dict[str, Any] = {
+                "q": query,
+                "country_code": country,
+                **(params or {}),
+            }
+            r = await c.post("/v1/google/sports", json=body)
+            try:
+                r.raise_for_status()
+            except httpx.HTTPStatusError as e:
+                _handle_error(e)
+            return r.text
+
+
+class GhostcrawlExtractTool(BaseTool):
+    """LangChain tool: extract structured data from a URL via ghostcrawl."""
+
+    name: str = "ghostcrawl_extract"
+    description: str = (
+        "Extract structured data from a URL by providing a JSON Schema. "
+        "ghostcrawl fetches the page through its stealth fleet, then uses the schema "
+        "to identify and return matching fields as a structured JSON object. "
+        "SUPPORTED: url (the target page), schema (JSON Schema object describing the fields "
+        "to extract — e.g. product title, price, reviews), render_js (bool), "
+        "session_id (existing session for stateful extraction). "
+        "UNSUPPORTED: extracting from binary files (PDF/DOCX) — convert to HTML first. "
+        "Returns: data dict keyed by the schema's property names, plus metadata. "
+        "Use this tool instead of ghostcrawl_scrape when you need structured output "
+        "rather than raw page content. "
+        "Uses POST /v1/extract with Authorization: Bearer token auth."
+    )
+    args_schema: Type[BaseModel] = _ExtractInput
+
+    def _run(
+        self,
+        url: str,
+        schema_: Optional[dict] = None,
+        schema: Optional[dict] = None,
+        params: Optional[dict] = None,
+        **_: Any,
+    ) -> str:
+        resolved_schema = schema_ or schema or {}
+        body: Dict[str, Any] = {
+            "url": url,
+            **({"schema": resolved_schema} if resolved_schema else {}),
+            **(params or {}),
+        }
+        with get_client() as c:
+            r = c.post("/v1/extract", json=body)
+            try:
+                r.raise_for_status()
+            except httpx.HTTPStatusError as e:
+                _handle_error(e)
+            return r.text
+
+    async def _arun(
+        self,
+        url: str,
+        schema_: Optional[dict] = None,
+        schema: Optional[dict] = None,
+        params: Optional[dict] = None,
+        **kwargs: Any,
+    ) -> str:
+        resolved_schema = schema_ or schema or {}
+        body: Dict[str, Any] = {
+            "url": url,
+            **({"schema": resolved_schema} if resolved_schema else {}),
+            **(params or {}),
+        }
+        async with get_async_client() as c:
+            r = await c.post("/v1/extract", json=body)
+            try:
+                r.raise_for_status()
+            except httpx.HTTPStatusError as e:
+                _handle_error(e)
+            return r.text
+
+
+class GhostcrawlCrawlTool(BaseTool):
+    """LangChain tool: crawl a site starting from a seed URL via ghostcrawl."""
+
+    name: str = "ghostcrawl_crawl"
+    description: str = (
+        "Crawl a website starting from a seed URL and return content from multiple pages. "
+        "ghostcrawl follows links using its stealth fleet, respecting same-origin and "
+        "max_pages bounds to avoid unbounded crawls. "
+        "SUPPORTED: url (seed URL), max_pages (1–500, default 10), same_origin (bool — "
+        "restrict to same scheme+host+port, default true), render_js (bool for JS-heavy sites). "
+        "UNSUPPORTED: authenticated crawls requiring session cookies (use ghostcrawl_scrape "
+        "in a loop with session_id for authenticated pages). "
+        "Returns: list of crawled pages with url, content, title, and status_code per page. "
+        "QUOTA NOTE: each page counts as one ghostcrawl credit. Check ghostcrawl_usage "
+        "before launching large crawl jobs. "
+        "Uses POST /v1/crawl with Authorization: Bearer token auth."
+    )
+    args_schema: Type[BaseModel] = _CrawlInput
+
+    def _run(
+        self,
+        url: str,
+        max_pages: int = 10,
+        same_origin: bool = True,
+        params: Optional[dict] = None,
+        **_: Any,
+    ) -> str:
+        body: Dict[str, Any] = {
+            "url": url,
+            "max_pages": max_pages,
+            "same_origin": same_origin,
+            **(params or {}),
+        }
+        with get_client() as c:
+            r = c.post("/v1/crawl", json=body)
+            try:
+                r.raise_for_status()
+            except httpx.HTTPStatusError as e:
+                _handle_error(e)
+            return r.text
+
+    async def _arun(
+        self,
+        url: str,
+        max_pages: int = 10,
+        same_origin: bool = True,
+        params: Optional[dict] = None,
+        **kwargs: Any,
+    ) -> str:
+        body: Dict[str, Any] = {
+            "url": url,
+            "max_pages": max_pages,
+            "same_origin": same_origin,
+            **(params or {}),
+        }
+        async with get_async_client() as c:
+            r = await c.post("/v1/crawl", json=body)
+            try:
+                r.raise_for_status()
+            except httpx.HTTPStatusError as e:
+                _handle_error(e)
+            return r.text
+
+
+class GhostcrawlUsageTool(BaseTool):
+    """LangChain tool: check ghostcrawl API quota and current usage."""
+
+    name: str = "ghostcrawl_usage"
+    description: str = (
+        "Check your ghostcrawl API quota and current usage statistics. "
+        "Use this tool BEFORE launching large crawl or scrape jobs to confirm "
+        "you have sufficient credit headroom. Also useful for cost-control loops "
+        "that should pause when credits drop below a threshold. "
+        "SUPPORTED: no input parameters required — returns current period stats. "
+        "Returns: JSON object with requests_used, requests_quota, credits_remaining, "
+        "and period_end_iso (ISO-8601 timestamp of current billing period end). "
+        "Uses GET /v1/usage with Authorization: Bearer token auth. "
+        "For a metered-dim-only breakdown, the billing endpoint "
+        "GET /v1/billing/usage?metered_only=true (D-05e item 10) narrows to "
+        "billable dims (request, identity_served, proxy_bytes)."
+    )
+    args_schema: Type[BaseModel] = _EmptyInput
+
+    def _run(self, **_: Any) -> str:
+        with get_client() as c:
+            r = c.get("/v1/usage")
+            try:
+                r.raise_for_status()
+            except httpx.HTTPStatusError as e:
+                _handle_error(e)
+            return r.text
+
+    async def _arun(self, **_: Any) -> str:
+        async with get_async_client() as c:
+            r = await c.get("/v1/usage")
+            try:
+                r.raise_for_status()
+            except httpx.HTTPStatusError as e:
+                _handle_error(e)
+            return r.text

ghostcrawl_langchain-2.1.0.dist-info/METADATA

@@ -0,0 +1,33 @@
+Metadata-Version: 2.4
+Name: ghostcrawl-langchain
+Version: 2.1.0
+Summary: LangChain tool wrappers for the ghostcrawl SaaS stealth-scraping API.
+License-Expression: MIT
+License-File: LICENSE
+Requires-Python: >=3.9
+Requires-Dist: httpx>=0.27
+Requires-Dist: langchain-core>=0.3
+Requires-Dist: pydantic>=2.7
+Provides-Extra: dev
+Requires-Dist: pytest>=8; extra == 'dev'
+Requires-Dist: respx>=0.21; extra == 'dev'
+Description-Content-Type: text/markdown
+
+# ghostcrawl-langchain
+
+LangChain tool integration for the [ghostcrawl](https://pypi.org/project/ghostcrawl/)
+stealth-scraping API.
+
+Exposes the ghostcrawl scrape / search / extract / crawl / Google-vertical
+surfaces as LangChain `BaseTool` subclasses with shared Pydantic params
+validation.
+
+## Install
+
+```bash
+pip install ghostcrawl-langchain
+```
+
+## License
+
+MIT

ghostcrawl_langchain-2.1.0.dist-info/RECORD

@@ -0,0 +1,8 @@
+ghostcrawl_langchain/__init__.py,sha256=4iC7bkRIIZfPMNKbLeKH7IQdhiiaflr7y_154-_i9qw,880
+ghostcrawl_langchain/_client.py,sha256=ZGC0YIoS37rMBP2wYyRYZOc8oj1slnZBwUYxvStmMno,2620
+ghostcrawl_langchain/errors.py,sha256=URGsTZ6eBMgz9KlOX5uu4hr1MHG-I0a0A6EafXuDM1A,1015
+ghostcrawl_langchain/tools.py,sha256=zzr5sRKxg6sitQgDNqL3hlsq9ol1EcE28EpRCt8ckw8,28690
+ghostcrawl_langchain-2.1.0.dist-info/METADATA,sha256=WJSKEHDBhUTEtWt1YIuXP0c3TBiffPQKMQbUxt7q94s,815
+ghostcrawl_langchain-2.1.0.dist-info/WHEEL,sha256=mffPy8wBnZQn2VnJUU5jE99KsxaSfiyMHV9Yt0aLVxs,87
+ghostcrawl_langchain-2.1.0.dist-info/licenses/LICENSE,sha256=gXqpnCmrqHuj-JhZIbRbwPENTHdyeKnoeQNBGaooEPg,1067
+ghostcrawl_langchain-2.1.0.dist-info/RECORD,,

ghostcrawl_langchain-2.1.0.dist-info/WHEEL

@@ -0,0 +1,4 @@
+Wheel-Version: 1.0
+Generator: hatchling 1.30.1
+Root-Is-Purelib: true
+Tag: py3-none-any

ghostcrawl_langchain-2.1.0.dist-info/licenses/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 GhostCrawl
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.