s-eto-hub's picture
Commit de nettoyage des binaires
097eaf0
Raw History Blame Contribute Delete
21.6 kB
"""
Crawling functionality for Firecrawl v2 API.
"""
import time
from typing import Optional, Dict, Any, List
from ..types import (
CrawlRequest,
CrawlJob,
CrawlResponse, Document, CrawlParamsRequest, CrawlParamsResponse, CrawlParamsData,
WebhookConfig, CrawlErrorsResponse, ActiveCrawlsResponse, ActiveCrawl, PaginationConfig
)
from ..utils import HttpClient, handle_response_error, validate_scrape_options, prepare_scrape_options
from ..utils.normalize import normalize_document_input
def _validate_crawl_request(request: CrawlRequest) -> None:
"""
Validate crawl request parameters.
Args:
request: CrawlRequest to validate
Raises:
ValueError: If request is invalid
"""
if not request.url or not request.url.strip():
raise ValueError("URL cannot be empty")
if request.limit is not None and request.limit <= 0:
raise ValueError("Limit must be positive")
# Validate scrape_options (if provided)
if request.scrape_options is not None:
validate_scrape_options(request.scrape_options)
def _prepare_crawl_request(request: CrawlRequest) -> dict:
"""
Prepare crawl request for API submission.
Args:
request: CrawlRequest to prepare
Returns:
Dictionary ready for API submission
"""
# Validate request
_validate_crawl_request(request)
# Start with basic data
data = {"url": request.url}
# Add prompt if present
if request.prompt:
data["prompt"] = request.prompt
# Handle scrape_options conversion first (before model_dump)
if request.scrape_options is not None:
scrape_data = prepare_scrape_options(request.scrape_options)
if scrape_data:
data["scrapeOptions"] = scrape_data
# Convert request to dict
request_data = request.model_dump(exclude_none=True, exclude_unset=True)
# Remove url, prompt, and scrape_options (already handled)
request_data.pop("url", None)
request_data.pop("prompt", None)
request_data.pop("scrape_options", None)
# Handle webhook conversion first (before model_dump)
if request.webhook is not None:
if isinstance(request.webhook, str):
data["webhook"] = request.webhook
else:
# Convert WebhookConfig to dict
data["webhook"] = request.webhook.model_dump(exclude_none=True)
# Convert other snake_case fields to camelCase
field_mappings = {
"include_paths": "includePaths",
"exclude_paths": "excludePaths",
"max_discovery_depth": "maxDiscoveryDepth",
"sitemap": "sitemap",
"ignore_query_parameters": "ignoreQueryParameters",
"deduplicate_similar_urls": "deduplicateSimilarURLs",
"crawl_entire_domain": "crawlEntireDomain",
"allow_external_links": "allowExternalLinks",
"allow_subdomains": "allowSubdomains",
"ignore_robots_txt": "ignoreRobotsTxt",
"robots_user_agent": "robotsUserAgent",
"delay": "delay",
"max_concurrency": "maxConcurrency",
"regex_on_full_url": "regexOnFullURL",
"zero_data_retention": "zeroDataRetention"
}
# Apply field mappings
for snake_case, camel_case in field_mappings.items():
if snake_case in request_data:
data[camel_case] = request_data.pop(snake_case)
# Add any remaining fields that don't need conversion (like limit)
data.update(request_data)
# Trim integration if present
if "integration" in data and isinstance(data["integration"], str):
data["integration"] = data["integration"].strip()
return data
def _parse_crawl_documents(data_list: Optional[List[Any]]) -> List[Document]:
documents: List[Document] = []
for doc_data in data_list or []:
if isinstance(doc_data, dict):
documents.append(Document(**normalize_document_input(doc_data)))
return documents
def _parse_crawl_status_response(response_data: Dict[str, Any]) -> Dict[str, Any]:
if not response_data.get("success"):
raise Exception(response_data.get("error", "Unknown error occurred"))
return {
"status": response_data.get("status"),
"completed": response_data.get("completed", 0),
"total": response_data.get("total", 0),
"credits_used": response_data.get("creditsUsed", 0),
"expires_at": response_data.get("expiresAt"),
"next": response_data.get("next"),
"data": _parse_crawl_documents(response_data.get("data", [])),
}
def start_crawl(client: HttpClient, request: CrawlRequest) -> CrawlResponse:
"""
Start a crawl job for a website.
Args:
client: HTTP client instance
request: CrawlRequest containing URL and options
Returns:
CrawlResponse with job information
Raises:
ValueError: If request is invalid
Exception: If the crawl operation fails to start
"""
request_data = _prepare_crawl_request(request)
response = client.post("/v2/crawl", request_data)
if not response.ok:
handle_response_error(response, "start crawl")
response_data = response.json()
if response_data.get("success"):
job_data = {
"id": response_data.get("id"),
"url": response_data.get("url")
}
return CrawlResponse(**job_data)
else:
raise Exception(response_data.get("error", "Unknown error occurred"))
def get_crawl_status(
client: HttpClient,
job_id: str,
pagination_config: Optional[PaginationConfig] = None,
*,
request_timeout: Optional[float] = None,
) -> CrawlJob:
"""
Get the status of a crawl job.
Args:
client: HTTP client instance
job_id: ID of the crawl job
pagination_config: Optional configuration for pagination behavior
request_timeout: Timeout (in seconds) for each individual HTTP request. When auto-pagination
is enabled (default) and there are multiple pages of results, this timeout applies to
each page request separately, not to the entire operation
Returns:
CrawlJob with current status and data
Raises:
Exception: If the status check fails
"""
# Make the API request
response = client.get(f"/v2/crawl/{job_id}", timeout=request_timeout)
# Handle errors
if not response.ok:
handle_response_error(response, "get crawl status")
# Parse response
response_data = response.json()
payload = _parse_crawl_status_response(response_data)
documents = payload["data"]
# Handle pagination if requested
auto_paginate = pagination_config.auto_paginate if pagination_config else True
if auto_paginate and payload["next"] and not (
pagination_config
and pagination_config.max_results is not None
and len(documents) >= pagination_config.max_results
):
documents = _fetch_all_pages(
client,
payload["next"],
documents,
pagination_config,
request_timeout=request_timeout,
)
# Create CrawlJob with current status and data
return CrawlJob(
status=payload["status"],
completed=payload["completed"],
total=payload["total"],
credits_used=payload["credits_used"],
expires_at=payload["expires_at"],
next=payload["next"] if not auto_paginate else None,
data=documents,
)
def get_crawl_status_page(
client: HttpClient,
next_url: str,
*,
request_timeout: Optional[float] = None,
) -> CrawlJob:
"""
Fetch a single page of crawl results using the provided next URL.
Args:
client: HTTP client instance
next_url: Opaque next URL from a prior crawl status response
request_timeout: Timeout (in seconds) for the HTTP request
Returns:
CrawlJob with the page data and next URL (if any)
Raises:
Exception: If the request fails or returns an error response
"""
response = client.get(next_url, timeout=request_timeout)
if not response.ok:
handle_response_error(response, "get crawl status page")
response_data = response.json()
payload = _parse_crawl_status_response(response_data)
return CrawlJob(
status=payload["status"],
completed=payload["completed"],
total=payload["total"],
credits_used=payload["credits_used"],
expires_at=payload["expires_at"],
next=payload["next"],
data=payload["data"],
)
def _fetch_all_pages(
client: HttpClient,
next_url: str,
initial_documents: List[Document],
pagination_config: Optional[PaginationConfig] = None,
*,
request_timeout: Optional[float] = None,
) -> List[Document]:
"""
Fetch all pages of crawl results.
Args:
client: HTTP client instance
next_url: URL for the next page
initial_documents: Documents from the first page
pagination_config: Optional configuration for pagination limits
request_timeout: Optional timeout (in seconds) for the underlying HTTP request
Returns:
List of all documents from all pages
"""
documents = initial_documents.copy()
current_url = next_url
page_count = 0
# Apply pagination limits
max_pages = pagination_config.max_pages if pagination_config else None
max_results = pagination_config.max_results if pagination_config else None
max_wait_time = pagination_config.max_wait_time if pagination_config else None
start_time = time.monotonic()
while current_url:
# Check pagination limits (treat 0 as a valid limit)
if (max_pages is not None) and page_count >= max_pages:
break
if (max_wait_time is not None) and (time.monotonic() - start_time) > max_wait_time:
break
# Fetch next page
response = client.get(current_url, timeout=request_timeout)
if not response.ok:
# Log error but continue with what we have
import logging
logger = logging.getLogger("firecrawl")
logger.warning("Failed to fetch next page", extra={"status_code": response.status_code})
break
page_data = response.json()
try:
page_payload = _parse_crawl_status_response(page_data)
except Exception:
break
# Add documents from this page
for document in page_payload["data"]:
# Check max_results limit BEFORE adding each document
if max_results is not None and len(documents) >= max_results:
break
documents.append(document)
# Check if we hit max_results limit
if max_results is not None and len(documents) >= max_results:
break
# Get next URL
current_url = page_payload["next"]
page_count += 1
return documents
def cancel_crawl(client: HttpClient, job_id: str) -> bool:
"""
Cancel a running crawl job.
Args:
client: HTTP client instance
job_id: ID of the crawl job to cancel
Returns:
bool: True if the crawl was cancelled, False otherwise
Raises:
Exception: If the cancellation fails
"""
response = client.delete(f"/v2/crawl/{job_id}")
if not response.ok:
handle_response_error(response, "cancel crawl")
response_data = response.json()
return response_data.get("status") == "cancelled"
def wait_for_crawl_completion(
client: HttpClient,
job_id: str,
poll_interval: int = 2,
timeout: Optional[int] = None,
*,
request_timeout: Optional[float] = None,
) -> CrawlJob:
"""
Wait for a crawl job to complete, polling for status updates.
Args:
client: HTTP client instance
job_id: ID of the crawl job
poll_interval: Seconds between status checks
timeout: Maximum seconds to wait (None for no timeout)
request_timeout: Optional timeout (in seconds) for each status request
Returns:
CrawlJob when job completes
Raises:
Exception: If the job fails
TimeoutError: If timeout is reached
"""
start_time = time.monotonic()
while True:
crawl_job = get_crawl_status(
client,
job_id,
request_timeout=request_timeout,
)
# Check if job is complete
if crawl_job.status in ["completed", "failed", "cancelled"]:
return crawl_job
# Check timeout
if timeout is not None and (time.monotonic() - start_time) > timeout:
raise TimeoutError(f"Crawl job {job_id} did not complete within {timeout} seconds")
# Wait before next poll
time.sleep(poll_interval)
def crawl(
client: HttpClient,
request: CrawlRequest,
poll_interval: int = 2,
timeout: Optional[int] = None,
*,
request_timeout: Optional[float] = None,
) -> CrawlJob:
"""
Start a crawl job and wait for it to complete.
Args:
client: HTTP client instance
request: CrawlRequest containing URL and options
poll_interval: Seconds between status checks
timeout: Maximum seconds to wait for the entire crawl job to complete (None for no timeout)
request_timeout: Timeout (in seconds) for each individual HTTP request, including pagination
requests when fetching results. If there are multiple pages, each page request gets this timeout
Returns:
CrawlJob when job completes
Raises:
ValueError: If request is invalid
Exception: If the crawl fails to start or complete
TimeoutError: If timeout is reached
"""
# Start the crawl
crawl_job = start_crawl(client, request)
job_id = crawl_job.id
# Determine the per-request timeout. If not provided, reuse the overall timeout value.
effective_request_timeout = request_timeout if request_timeout is not None else timeout
# Wait for completion
return wait_for_crawl_completion(
client,
job_id,
poll_interval,
timeout,
request_timeout=effective_request_timeout,
)
def crawl_params_preview(client: HttpClient, request: CrawlParamsRequest) -> CrawlParamsData:
"""
Get crawl parameters from LLM based on URL and prompt.
Args:
client: HTTP client instance
request: CrawlParamsRequest containing URL and prompt
Returns:
CrawlParamsData containing suggested crawl options
Raises:
ValueError: If request is invalid
Exception: If the operation fails
"""
# Validate request
if not request.url or not request.url.strip():
raise ValueError("URL cannot be empty")
if not request.prompt or not request.prompt.strip():
raise ValueError("Prompt cannot be empty")
# Prepare request data
request_data = {
"url": request.url,
"prompt": request.prompt
}
# Make the API request
response = client.post("/v2/crawl/params-preview", request_data)
# Handle errors
if not response.ok:
handle_response_error(response, "crawl params preview")
# Parse response
response_data = response.json()
if response_data.get("success"):
params_data = response_data.get("data", {})
# Convert camelCase to snake_case for CrawlParamsData
converted_params = {}
field_mappings = {
"includePaths": "include_paths",
"excludePaths": "exclude_paths",
"maxDiscoveryDepth": "max_discovery_depth",
"sitemap": "sitemap",
"ignoreQueryParameters": "ignore_query_parameters",
"deduplicateSimilarURLs": "deduplicate_similar_urls",
"crawlEntireDomain": "crawl_entire_domain",
"allowExternalLinks": "allow_external_links",
"allowSubdomains": "allow_subdomains",
"ignoreRobotsTxt": "ignore_robots_txt",
"robotsUserAgent": "robots_user_agent",
"maxConcurrency": "max_concurrency",
"scrapeOptions": "scrape_options",
"zeroDataRetention": "zero_data_retention"
}
# Handle webhook conversion
if "webhook" in params_data:
webhook_data = params_data["webhook"]
if isinstance(webhook_data, dict):
converted_params["webhook"] = WebhookConfig(**webhook_data)
else:
converted_params["webhook"] = webhook_data
for camel_case, snake_case in field_mappings.items():
if camel_case in params_data:
if camel_case == "scrapeOptions" and params_data[camel_case] is not None:
# Handle nested scrapeOptions conversion
scrape_opts_data = params_data[camel_case]
converted_scrape_opts = {}
scrape_field_mappings = {
"includeTags": "include_tags",
"excludeTags": "exclude_tags",
"onlyMainContent": "only_main_content",
"waitFor": "wait_for",
"skipTlsVerification": "skip_tls_verification",
"removeBase64Images": "remove_base64_images"
}
for scrape_camel, scrape_snake in scrape_field_mappings.items():
if scrape_camel in scrape_opts_data:
converted_scrape_opts[scrape_snake] = scrape_opts_data[scrape_camel]
# Handle formats field - if it's a list, convert to ScrapeFormats
if "formats" in scrape_opts_data:
formats_data = scrape_opts_data["formats"]
if isinstance(formats_data, list):
# Convert list to ScrapeFormats object
from ..types import ScrapeFormats
converted_scrape_opts["formats"] = ScrapeFormats(formats=formats_data)
else:
converted_scrape_opts["formats"] = formats_data
# Add fields that don't need conversion
for key, value in scrape_opts_data.items():
if key not in scrape_field_mappings and key != "formats":
converted_scrape_opts[key] = value
converted_params[snake_case] = converted_scrape_opts
else:
converted_params[snake_case] = params_data[camel_case]
# Add fields that don't need conversion
for key, value in params_data.items():
if key not in field_mappings:
converted_params[key] = value
# Add warning if present
if "warning" in response_data:
converted_params["warning"] = response_data["warning"]
return CrawlParamsData(**converted_params)
else:
raise Exception(response_data.get("error", "Unknown error occurred"))
def get_crawl_errors(http_client: HttpClient, crawl_id: str) -> CrawlErrorsResponse:
"""
Get errors from a crawl job.
Args:
http_client: HTTP client for making requests
crawl_id: The ID of the crawl job
Returns:
CrawlErrorsResponse containing errors and robots blocked URLs
Raises:
Exception: If the request fails
"""
response = http_client.get(f"/v2/crawl/{crawl_id}/errors")
if not response.ok:
handle_response_error(response, "check crawl errors")
try:
body = response.json()
payload = body.get("data", body)
# Manual key normalization since we avoid Pydantic aliases
normalized = {
"errors": payload.get("errors", []),
"robots_blocked": payload.get("robotsBlocked", payload.get("robots_blocked", [])),
}
return CrawlErrorsResponse(**normalized)
except Exception as e:
raise Exception(f"Failed to parse crawl errors response: {e}")
def get_active_crawls(client: HttpClient) -> ActiveCrawlsResponse:
"""
Get a list of currently active crawl jobs.
Args:
client: HTTP client instance
Returns:
ActiveCrawlsResponse containing a list of active crawl jobs
Raises:
Exception: If the request fails
"""
response = client.get("/v2/crawl/active")
if not response.ok:
handle_response_error(response, "get active crawls")
body = response.json()
if not body.get("success"):
raise Exception(body.get("error", "Unknown error occurred"))
crawls_in = body.get("crawls", [])
normalized_crawls = []
for c in crawls_in:
if isinstance(c, dict):
normalized_crawls.append({
"id": c.get("id"),
"team_id": c.get("teamId", c.get("team_id")),
"url": c.get("url"),
"options": c.get("options"),
})
return ActiveCrawlsResponse(success=True, crawls=[ActiveCrawl(**nc) for nc in normalized_crawls])