firecrawl 3.2.1__tar.gz → 3.3.2__tar.gz

{firecrawl-3.2.1 → firecrawl-3.3.2}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.4
+Metadata-Version: 2.1
 Name: firecrawl
-Version: 3.2.1
+Version: 3.3.2
 Summary: Python SDK for Firecrawl API
 Home-page: https://github.com/firecrawl/firecrawl
 Author: Mendable.ai
@@ -40,10 +40,6 @@ Requires-Dist: websockets
 Requires-Dist: nest-asyncio
 Requires-Dist: pydantic>=2.0
 Requires-Dist: aiohttp
-Dynamic: author
-Dynamic: home-page
-Dynamic: license-file
-Dynamic: requires-python
 
 # Firecrawl Python SDK
 

{firecrawl-3.2.1 → firecrawl-3.3.2}/firecrawl/__init__.py

@@ -17,7 +17,7 @@ from .v1 import (
     V1ChangeTrackingOptions,
 )
 
-__version__ = "3.2.1"
+__version__ = "3.3.2"
 
 # Define the logger for the Firecrawl project
 logger: logging.Logger = logging.getLogger("firecrawl")

{firecrawl-3.2.1 → firecrawl-3.3.2}/firecrawl/__tests__/unit/v2/methods/aio/test_aio_crawl_request_preparation.py

@@ -14,7 +14,7 @@ class TestAsyncCrawlRequestPreparation:
             include_paths=["/docs/*"],
             exclude_paths=["/admin/*"],
             max_discovery_depth=2,
-            ignore_sitemap=True,
+            sitemap="skip",
             ignore_query_parameters=True,
             crawl_entire_domain=True,
             allow_external_links=False,
@@ -26,7 +26,7 @@ class TestAsyncCrawlRequestPreparation:
         assert payload["includePaths"] == ["/docs/*"]
         assert payload["excludePaths"] == ["/admin/*"]
         assert payload["maxDiscoveryDepth"] == 2
-        assert payload["ignoreSitemap"] is True
+        assert payload["sitemap"] == "skip"
         assert payload["ignoreQueryParameters"] is True
         assert payload["crawlEntireDomain"] is True
         assert payload["allowExternalLinks"] is False

{firecrawl-3.2.1 → firecrawl-3.3.2}/firecrawl/__tests__/unit/v2/methods/test_crawl_request_preparation.py

@@ -24,7 +24,7 @@ class TestCrawlRequestPreparation:
             url="https://example.com",
             limit=10,
             max_discovery_depth=3,
-            ignore_sitemap=True,
+            sitemap="skip",
             crawl_entire_domain=False,
             allow_external_links=True
         )
@@ -39,8 +39,8 @@ class TestCrawlRequestPreparation:
         assert data["limit"] == 10
         assert "maxDiscoveryDepth" in data
         assert data["maxDiscoveryDepth"] == 3
-        assert "ignoreSitemap" in data
-        assert data["ignoreSitemap"] is True
+        assert "sitemap" in data
+        assert data["sitemap"] == "skip"
         assert "crawlEntireDomain" in data
         assert data["crawlEntireDomain"] is False
         assert "allowExternalLinks" in data
@@ -106,7 +106,7 @@ class TestCrawlRequestPreparation:
             include_paths=["/blog/*", "/docs/*"],
             exclude_paths=["/admin/*"],
             max_discovery_depth=3,
-            ignore_sitemap=False,
+            sitemap="include",
             limit=100,
             crawl_entire_domain=True,
             allow_external_links=False,
@@ -126,8 +126,8 @@ class TestCrawlRequestPreparation:
         assert data["excludePaths"] == ["/admin/*"]
         assert "maxDiscoveryDepth" in data
         assert data["maxDiscoveryDepth"] == 3
-        assert "ignoreSitemap" in data
-        assert data["ignoreSitemap"] is False
+        assert "sitemap" in data
+        assert data["sitemap"] == "include"
         assert "limit" in data
         assert data["limit"] == 100
         assert "crawlEntireDomain" in data

{firecrawl-3.2.1 → firecrawl-3.3.2}/firecrawl/v2/client.py

@@ -13,6 +13,7 @@ from .types import (
     SearchRequest,
     SearchData,
     SourceOption,
+    CategoryOption,
     CrawlRequest,
     CrawlResponse,
     CrawlJob,
@@ -171,6 +172,7 @@ class FirecrawlClient:
         query: str,
         *,
         sources: Optional[List[SourceOption]] = None,
+        categories: Optional[List[CategoryOption]] = None,
         limit: Optional[int] = None,
         tbs: Optional[str] = None,
         location: Optional[str] = None,
@@ -195,6 +197,7 @@ class FirecrawlClient:
         request = SearchRequest(
             query=query,
             sources=sources,
+            categories=categories,
             limit=limit,
             tbs=tbs,
             location=location,

{firecrawl-3.2.1 → firecrawl-3.3.2}/firecrawl/v2/methods/search.py

@@ -121,6 +121,17 @@ def _validate_search_request(request: SearchRequest) -> SearchRequest:
                 if source.type not in valid_sources:
                     raise ValueError(f"Invalid source type: {source.type}. Valid types: {valid_sources}")
     
+    # Validate categories (if provided)
+    if request.categories is not None:
+        valid_categories = {"github", "research"}
+        for category in request.categories:
+            if isinstance(category, str):
+                if category not in valid_categories:
+                    raise ValueError(f"Invalid category type: {category}. Valid types: {valid_categories}")
+            elif hasattr(category, 'type'):
+                if category.type not in valid_categories:
+                    raise ValueError(f"Invalid category type: {category.type}. Valid types: {valid_categories}")
+    
     # Validate location (if provided)
     if request.location is not None:
         if not isinstance(request.location, str) or len(request.location.strip()) == 0:

{firecrawl-3.2.1 → firecrawl-3.3.2}/firecrawl/v2/types.py

@@ -174,6 +174,12 @@ class Source(BaseModel):
 
 SourceOption = Union[str, Source]
 
+class Category(BaseModel):
+    """Configuration for a search category."""
+    type: str
+
+CategoryOption = Union[str, Category]
+
 FormatString = Literal[
     # camelCase versions (API format)
     "markdown", "html", "rawHtml", "links", "screenshot", "summary", "changeTracking", "json",
@@ -331,7 +337,8 @@ class SearchResultWeb(BaseModel):
     """A web search result with URL, title, and description."""
     url: str
     title: Optional[str] = None
-    description: Optional[str] = None 
+    description: Optional[str] = None
+    category: Optional[str] = None
 
 class SearchResultNews(BaseModel):
   """A news search result with URL, title, snippet, date, image URL, and position."""
@@ -341,6 +348,7 @@ class SearchResultNews(BaseModel):
   date: Optional[str] = None
   image_url: Optional[str] = None
   position: Optional[int] = None
+  category: Optional[str] = None
 
 class SearchResultImages(BaseModel):
   """An image search result with URL, title, image URL, image width, image height, and position."""
@@ -521,6 +529,7 @@ class SearchRequest(BaseModel):
     """Request for search operations."""
     query: str
     sources: Optional[List[SourceOption]] = None
+    categories: Optional[List[CategoryOption]] = None
     limit: Optional[int] = 5
     tbs: Optional[str] = None
     location: Optional[str] = None
@@ -547,6 +556,26 @@ class SearchRequest(BaseModel):
                 raise ValueError(f"Invalid source format: {source}")
         
         return normalized_sources
+    
+    @field_validator('categories')
+    @classmethod
+    def validate_categories(cls, v):
+        """Validate and normalize categories input."""
+        if v is None:
+            return v
+        
+        normalized_categories = []
+        for category in v:
+            if isinstance(category, str):
+                normalized_categories.append(Category(type=category))
+            elif isinstance(category, dict):
+                normalized_categories.append(Category(**category))
+            elif isinstance(category, Category):
+                normalized_categories.append(category)
+            else:
+                raise ValueError(f"Invalid category format: {category}")
+        
+        return normalized_categories
 
 class LinkResult(BaseModel):
     """A generic link result with optional metadata (used by search and map)."""

{firecrawl-3.2.1 → firecrawl-3.3.2}/firecrawl.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.4
+Metadata-Version: 2.1
 Name: firecrawl
-Version: 3.2.1
+Version: 3.3.2
 Summary: Python SDK for Firecrawl API
 Home-page: https://github.com/firecrawl/firecrawl
 Author: Mendable.ai
@@ -40,10 +40,6 @@ Requires-Dist: websockets
 Requires-Dist: nest-asyncio
 Requires-Dist: pydantic>=2.0
 Requires-Dist: aiohttp
-Dynamic: author
-Dynamic: home-page
-Dynamic: license-file
-Dynamic: requires-python
 
 # Firecrawl Python SDK
 

{firecrawl-3.2.1 → firecrawl-3.3.2}/firecrawl.egg-info/top_level.txt

@@ -1,3 +1,2 @@
-dist
 firecrawl
 tests