firecrawl 4.0.0__py3-none-any.whl → 4.1.0__py3-none-any.whl

firecrawl/__init__.py

@@ -17,7 +17,7 @@ from .v1 import (
     V1ChangeTrackingOptions,
 )
 
-__version__ = "4.0.0"
+__version__ = "4.1.0"
 
 # Define the logger for the Firecrawl project
 logger: logging.Logger = logging.getLogger("firecrawl")

firecrawl/v2/client.py

@@ -18,6 +18,7 @@ from .types import (
     CrawlResponse,
     CrawlJob,
     CrawlParamsRequest,
+    PDFParser,
     CrawlParamsData,
     WebhookConfig,
     CrawlErrorsResponse,
@@ -105,7 +106,7 @@ class FirecrawlClient:
         timeout: Optional[int] = None,
         wait_for: Optional[int] = None,
         mobile: Optional[bool] = None,
-        parsers: Optional[List[str]] = None,
+        parsers: Optional[Union[List[str], List[Union[str, PDFParser]]]] = None,
         actions: Optional[List[Union['WaitAction', 'ScreenshotAction', 'ClickAction', 'WriteAction', 'PressAction', 'ScrollAction', 'ScrapeAction', 'ExecuteJavascriptAction', 'PDFAction']]] = None,
         location: Optional['Location'] = None,
         skip_tls_verification: Optional[bool] = None,
@@ -571,7 +572,7 @@ class FirecrawlClient:
         timeout: Optional[int] = None,
         wait_for: Optional[int] = None,
         mobile: Optional[bool] = None,
-        parsers: Optional[List[str]] = None,
+        parsers: Optional[Union[List[str], List[Union[str, PDFParser]]]] = None,
         actions: Optional[List[Union['WaitAction', 'ScreenshotAction', 'ClickAction', 'WriteAction', 'PressAction', 'ScrollAction', 'ScrapeAction', 'ExecuteJavascriptAction', 'PDFAction']]] = None,
         location: Optional['Location'] = None,
         skip_tls_verification: Optional[bool] = None,
@@ -759,7 +760,7 @@ class FirecrawlClient:
         timeout: Optional[int] = None,
         wait_for: Optional[int] = None,
         mobile: Optional[bool] = None,
-        parsers: Optional[List[str]] = None,
+        parsers: Optional[Union[List[str], List[Union[str, PDFParser]]]] = None,
         actions: Optional[List[Union['WaitAction', 'ScreenshotAction', 'ClickAction', 'WriteAction', 'PressAction', 'ScrollAction', 'ScrapeAction', 'ExecuteJavascriptAction', 'PDFAction']]] = None,
         location: Optional['Location'] = None,
         skip_tls_verification: Optional[bool] = None,

firecrawl/v2/types.py

@@ -278,7 +278,7 @@ class ScrapeOptions(BaseModel):
     timeout: Optional[int] = None
     wait_for: Optional[int] = None
     mobile: Optional[bool] = None
-    parsers: Optional[List[str]] = None
+    parsers: Optional[Union[List[str], List[Union[str, 'PDFParser']]]] = None
     actions: Optional[List[Union['WaitAction', 'ScreenshotAction', 'ClickAction', 'WriteAction', 'PressAction', 'ScrollAction', 'ScrapeAction', 'ExecuteJavascriptAction', 'PDFAction']]] = None
     location: Optional['Location'] = None
     skip_tls_verification: Optional[bool] = None
@@ -536,6 +536,11 @@ class PDFAction(BaseModel):
     landscape: Optional[bool] = None
     scale: Optional[float] = None
 
+class PDFParser(BaseModel):
+    """PDF parser configuration with optional page limit."""
+    type: Literal["pdf"] = "pdf"
+    max_pages: Optional[int] = None
+
 # Location types
 class Location(BaseModel):
     """Location configuration for scraping."""
@@ -594,6 +599,26 @@ class SearchRequest(BaseModel):
         
         return normalized_categories
 
+    @field_validator('parsers')
+    @classmethod
+    def validate_parsers(cls, v):
+        """Validate and normalize parsers input."""
+        if v is None:
+            return v
+        
+        normalized_parsers = []
+        for parser in v:
+            if isinstance(parser, str):
+                normalized_parsers.append(parser)
+            elif isinstance(parser, dict):
+                normalized_parsers.append(PDFParser(**parser))
+            elif isinstance(parser, PDFParser):
+                normalized_parsers.append(parser)
+            else:
+                raise ValueError(f"Invalid parser format: {parser}")
+        
+        return normalized_parsers
+
 class LinkResult(BaseModel):
     """A generic link result with optional metadata (used by search and map)."""
     url: str
@@ -686,4 +711,4 @@ AnyResponse = Union[
     MapResponse,
     SearchResponse,
     ErrorResponse,
-]
+]

firecrawl/v2/utils/validation.py

@@ -311,6 +311,20 @@ def prepare_scrape_options(options: Optional[ScrapeOptions]) -> Optional[Dict[st
                                 converted_action[action_key] = action_value
                         converted_actions.append(converted_action)
                 scrape_data["actions"] = converted_actions
+            elif key == "parsers":
+                converted_parsers = []
+                for parser in value:
+                    if isinstance(parser, str):
+                        converted_parsers.append(parser)
+                    elif isinstance(parser, dict):
+                        converted_parsers.append(parser)
+                    else:
+                        parser_data = parser.model_dump(exclude_none=True)
+                        # Convert snake_case to camelCase for API
+                        if "max_pages" in parser_data:
+                            parser_data["maxPages"] = parser_data.pop("max_pages")
+                        converted_parsers.append(parser_data)
+                scrape_data["parsers"] = converted_parsers
             elif key == "location":
                 # Handle location conversion
                 if isinstance(value, dict):
@@ -321,4 +335,4 @@ def prepare_scrape_options(options: Optional[ScrapeOptions]) -> Optional[Dict[st
                 # For fields that don't need conversion, use as-is
                 scrape_data[key] = value
     
-    return scrape_data 
+    return scrape_data  

{firecrawl-4.0.0.dist-info → firecrawl-4.1.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: firecrawl
-Version: 4.0.0
+Version: 4.1.0
 Summary: Python SDK for Firecrawl API
 Home-page: https://github.com/firecrawl/firecrawl
 Author: Mendable.ai

{firecrawl-4.0.0.dist-info → firecrawl-4.1.0.dist-info}/RECORD

@@ -1,4 +1,4 @@
-firecrawl/__init__.py,sha256=Fvnels2_mk87sESx-1GhfwBGCrRsN05RoUHsu4AswrY,2192
+firecrawl/__init__.py,sha256=BT5Sx5lBRhzEEFNivjIhZaBA8kAeZCPFixbO9n8Myxw,2192
 firecrawl/client.py,sha256=tp3mUo_3aGPuZ53kpU4bhM-5EtwD_IUWrJ7wm0GMuCc,11159
 firecrawl/firecrawl.backup.py,sha256=v1FEN3jR4g5Aupg4xp6SLkuFvYMQuUKND2YELbYjE6c,200430
 firecrawl/types.py,sha256=W9N2pqQuevEIIjYHN9rbDf31E-nwdCECqIn11Foz2T8,2836
@@ -44,9 +44,9 @@ firecrawl/__tests__/unit/v2/watcher/test_ws_watcher.py,sha256=87w47n0iOihtu4jTR4
 firecrawl/v1/__init__.py,sha256=aP1oisPeZVGGZynvENc07JySMOZfv_4zAlxQ0ecMJXA,481
 firecrawl/v1/client.py,sha256=sydurfEFTsXyowyaGryA1lkPxN_r9Nf6iQpM43OwJyM,201672
 firecrawl/v2/__init__.py,sha256=Jc6a8tBjYG5OPkjDM5pl-notyys-7DEj7PLEfepv3fc,137
-firecrawl/v2/client.py,sha256=aD4SDVKhh5glgzdnJ8JvCPCW_u8pv0BAfIg2Wffvjao,31137
+firecrawl/v2/client.py,sha256=AMAHQ8Uz9bsEIy2vmIDNNUIT0FOivhLyj6lesEr1Rbg,31260
 firecrawl/v2/client_async.py,sha256=XyzojIJlWatBGlAMish22H-XHkkH9zHsD6MGtAdtFg8,10487
-firecrawl/v2/types.py,sha256=RnFZf9CXBS3XkeB74L48sxWZ_ECZht4gVOKtyQLZz0o,22973
+firecrawl/v2/types.py,sha256=YnbEmskB4eyfSIBDQK_Kv6xNjID3McagGEoIUiIIbL8,23840
 firecrawl/v2/watcher.py,sha256=FOU71tqSKxgeuGycu4ye0SLc2dw7clIcoQjPsi-4Csc,14229
 firecrawl/v2/watcher_async.py,sha256=AVjW2mgABniolSsauK4u0FW8ya6WzRUdyEg2R-8vGCw,10278
 firecrawl/v2/methods/batch.py,sha256=jFSIPtvulUrPz3Y3zT1gDNwYEf8Botpfh4GOeYsVYRI,14852
@@ -70,11 +70,11 @@ firecrawl/v2/utils/get_version.py,sha256=0CxW_41q2hlzIxEWOivUCaYw3GFiSIH32RPUMcI
 firecrawl/v2/utils/http_client.py,sha256=gUrC1CvU5sj03w27Lbq-3-yH38Yi_OXiI01-piwA83w,6027
 firecrawl/v2/utils/http_client_async.py,sha256=iy89_bk2HS3afSRHZ8016eMCa9Fk-5MFTntcOHfbPgE,1936
 firecrawl/v2/utils/normalize.py,sha256=nlTU6QRghT1YKZzNZlIQj4STSRuSUGrS9cCErZIcY5w,3636
-firecrawl/v2/utils/validation.py,sha256=L8by7z-t6GuMGIYkK7il1BM8d-4_-sAdG9hDMF_LeG4,14518
-firecrawl-4.0.0.dist-info/licenses/LICENSE,sha256=nPCunEDwjRGHlmjvsiDUyIWbkqqyj3Ej84ntnh0g0zA,1084
+firecrawl/v2/utils/validation.py,sha256=qWWiWaVcvODmVxf9rxIVy1j_dyuJCvdMMUoYhvWUEIU,15269
+firecrawl-4.1.0.dist-info/licenses/LICENSE,sha256=nPCunEDwjRGHlmjvsiDUyIWbkqqyj3Ej84ntnh0g0zA,1084
 tests/test_change_tracking.py,sha256=_IJ5ShLcoj2fHDBaw-nE4I4lHdmDB617ocK_XMHhXps,4177
 tests/test_timeout_conversion.py,sha256=PWlIEMASQNhu4cp1OW_ebklnE9NCiigPnEFCtI5N3w0,3996
-firecrawl-4.0.0.dist-info/METADATA,sha256=axsFmxpij8pHewgXTyRI3s5Rj96KoRG0Gxf4zCm11RY,7392
-firecrawl-4.0.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-firecrawl-4.0.0.dist-info/top_level.txt,sha256=8T3jOaSN5mtLghO-R3MQ8KO290gIX8hmfxQmglBPdLE,16
-firecrawl-4.0.0.dist-info/RECORD,,
+firecrawl-4.1.0.dist-info/METADATA,sha256=IFmGv9ChnRIoKksvP9RebpAcqm8c5-SHZhN8LrF04l4,7392
+firecrawl-4.1.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+firecrawl-4.1.0.dist-info/top_level.txt,sha256=8T3jOaSN5mtLghO-R3MQ8KO290gIX8hmfxQmglBPdLE,16
+firecrawl-4.1.0.dist-info/RECORD,,