firecrawl-py 2.7.1__py3-none-any.whl → 2.8.0__py3-none-any.whl

build/lib/build/lib/tests/test_change_tracking.py

@@ -0,0 +1,98 @@
+import unittest
+from unittest.mock import patch, MagicMock
+import json
+import os
+from firecrawl import FirecrawlApp
+
+class TestChangeTracking(unittest.TestCase):
+    @patch('requests.post')
+    def test_change_tracking_format(self, mock_post):
+        mock_response = MagicMock()
+        mock_response.status_code = 200
+        mock_response.json.return_value = {
+            'success': True,
+            'data': {
+                'markdown': 'Test markdown content',
+                'changeTracking': {
+                    'previousScrapeAt': '2023-01-01T00:00:00Z',
+                    'changeStatus': 'changed',
+                    'visibility': 'visible'
+                }
+            }
+        }
+        mock_post.return_value = mock_response
+
+        app = FirecrawlApp(api_key=os.environ.get('TEST_API_KEY', 'dummy-api-key-for-testing'))
+        result = app.scrape_url('https://example.com', {
+            'formats': ['markdown', 'changeTracking']
+        })
+
+        args, kwargs = mock_post.call_args
+        self.assertEqual(kwargs['json']['formats'], ['markdown', 'changeTracking'])
+        
+        self.assertEqual(result['changeTracking']['previousScrapeAt'], '2023-01-01T00:00:00Z')
+        self.assertEqual(result['changeTracking']['changeStatus'], 'changed')
+        self.assertEqual(result['changeTracking']['visibility'], 'visible')
+
+    @patch('requests.post')
+    def test_change_tracking_options(self, mock_post):
+        mock_response = MagicMock()
+        mock_response.status_code = 200
+        mock_response.json.return_value = {
+            'success': True,
+            'data': {
+                'markdown': 'Test markdown content',
+                'changeTracking': {
+                    'previousScrapeAt': '2023-01-01T00:00:00Z',
+                    'changeStatus': 'changed',
+                    'visibility': 'visible',
+                    'diff': {
+                        'text': '@@ -1,1 +1,1 @@\n-old content\n+new content',
+                        'json': {
+                            'files': [{
+                                'from': None,
+                                'to': None,
+                                'chunks': [{
+                                    'content': '@@ -1,1 +1,1 @@',
+                                    'changes': [{
+                                        'type': 'del',
+                                        'content': '-old content',
+                                        'del': True,
+                                        'ln': 1
+                                    }, {
+                                        'type': 'add',
+                                        'content': '+new content',
+                                        'add': True,
+                                        'ln': 1
+                                    }]
+                                }]
+                            }]
+                        }
+                    },
+                    'json': {
+                        'title': {
+                            'previous': 'Old Title',
+                            'current': 'New Title'
+                        }
+                    }
+                }
+            }
+        }
+        mock_post.return_value = mock_response
+
+        app = FirecrawlApp(api_key=os.environ.get('TEST_API_KEY', 'dummy-api-key-for-testing'))
+        result = app.scrape_url('https://example.com', {
+            'formats': ['markdown', 'changeTracking'],
+            'changeTrackingOptions': {
+                'modes': ['git-diff', 'json'],
+                'schema': {'type': 'object', 'properties': {'title': {'type': 'string'}}}
+            }
+        })
+
+        args, kwargs = mock_post.call_args
+        self.assertEqual(kwargs['json']['formats'], ['markdown', 'changeTracking'])
+        self.assertEqual(kwargs['json']['changeTrackingOptions']['modes'], ['git-diff', 'json'])
+        
+        self.assertEqual(result['changeTracking']['diff']['text'], '@@ -1,1 +1,1 @@\n-old content\n+new content')
+        self.assertEqual(result['changeTracking']['json']['title']['previous'], 'Old Title')
+        self.assertEqual(result['changeTracking']['json']['title']['current'], 'New Title')

build/lib/firecrawl/__init__.py

@@ -13,7 +13,7 @@ import os
 
 from .firecrawl import FirecrawlApp, AsyncFirecrawlApp, JsonConfig, ScrapeOptions, ChangeTrackingOptions # noqa
 
-__version__ = "2.7.1"
+__version__ = "2.8.0"
 
 # Define the logger for the Firecrawl project
 logger: logging.Logger = logging.getLogger("firecrawl")

build/lib/firecrawl/firecrawl.py

@@ -140,6 +140,7 @@ class ChangeTrackingOptions(pydantic.BaseModel):
     modes: Optional[List[Literal["git-diff", "json"]]] = None
     schema: Optional[Any] = None
     prompt: Optional[str] = None
+    tag: Optional[str] = None
 
 class ScrapeOptions(pydantic.BaseModel):
     """Parameters for scraping operations."""
@@ -157,6 +158,8 @@ class ScrapeOptions(pydantic.BaseModel):
     blockAds: Optional[bool] = None
     proxy: Optional[Literal["basic", "stealth", "auto"]] = None
     changeTrackingOptions: Optional[ChangeTrackingOptions] = None
+    maxAge: Optional[int] = None
+    storeInCache: Optional[bool] = None
 
 class WaitAction(pydantic.BaseModel):
     """Wait action to perform during scraping."""
@@ -292,6 +295,7 @@ class MapParams(pydantic.BaseModel):
     sitemapOnly: Optional[bool] = None
     limit: Optional[int] = None
     timeout: Optional[int] = None
+    useIndex: Optional[bool] = None
 
 class MapResponse(pydantic.BaseModel):
     """Response from mapping operations."""
@@ -464,6 +468,8 @@ class FirecrawlApp:
             json_options: Optional[JsonConfig] = None,
             actions: Optional[List[Union[WaitAction, ScreenshotAction, ClickAction, WriteAction, PressAction, ScrollAction, ScrapeAction, ExecuteJavascriptAction]]] = None,
             change_tracking_options: Optional[ChangeTrackingOptions] = None,
+            max_age: Optional[int] = None,
+            store_in_cache: Optional[bool] = None,
             **kwargs) -> ScrapeResponse[Any]:
         """
         Scrape and extract content from a URL.
@@ -545,6 +551,10 @@ class FirecrawlApp:
             scrape_params['actions'] = [action if isinstance(action, dict) else action.dict(exclude_none=True) for action in actions]
         if change_tracking_options:
             scrape_params['changeTrackingOptions'] = change_tracking_options if isinstance(change_tracking_options, dict) else change_tracking_options.dict(exclude_none=True)
+        if max_age is not None:
+            scrape_params['maxAge'] = max_age
+        if store_in_cache is not None:
+            scrape_params['storeInCache'] = store_in_cache
         
         scrape_params.update(kwargs)
 
@@ -1102,6 +1112,7 @@ class FirecrawlApp:
             sitemap_only: Optional[bool] = None,
             limit: Optional[int] = None,
             timeout: Optional[int] = None,
+            use_index: Optional[bool] = None,
             **kwargs) -> MapResponse:
         """
         Map and discover links from a URL.
@@ -1144,7 +1155,9 @@ class FirecrawlApp:
             map_params['limit'] = limit
         if timeout is not None:
             map_params['timeout'] = timeout
-
+        if use_index is not None:
+            map_params['useIndex'] = use_index
+        
         # Add any additional kwargs
         map_params.update(kwargs)
 

firecrawl/__init__.py

@@ -13,7 +13,7 @@ import os
 
 from .firecrawl import FirecrawlApp, AsyncFirecrawlApp, JsonConfig, ScrapeOptions, ChangeTrackingOptions # noqa
 
-__version__ = "2.7.1"
+__version__ = "2.8.0"
 
 # Define the logger for the Firecrawl project
 logger: logging.Logger = logging.getLogger("firecrawl")

firecrawl/firecrawl.py

@@ -140,6 +140,7 @@ class ChangeTrackingOptions(pydantic.BaseModel):
     modes: Optional[List[Literal["git-diff", "json"]]] = None
     schema: Optional[Any] = None
     prompt: Optional[str] = None
+    tag: Optional[str] = None
 
 class ScrapeOptions(pydantic.BaseModel):
     """Parameters for scraping operations."""
@@ -157,6 +158,8 @@ class ScrapeOptions(pydantic.BaseModel):
     blockAds: Optional[bool] = None
     proxy: Optional[Literal["basic", "stealth", "auto"]] = None
     changeTrackingOptions: Optional[ChangeTrackingOptions] = None
+    maxAge: Optional[int] = None
+    storeInCache: Optional[bool] = None
 
 class WaitAction(pydantic.BaseModel):
     """Wait action to perform during scraping."""
@@ -292,6 +295,7 @@ class MapParams(pydantic.BaseModel):
     sitemapOnly: Optional[bool] = None
     limit: Optional[int] = None
     timeout: Optional[int] = None
+    useIndex: Optional[bool] = None
 
 class MapResponse(pydantic.BaseModel):
     """Response from mapping operations."""
@@ -464,6 +468,8 @@ class FirecrawlApp:
             json_options: Optional[JsonConfig] = None,
             actions: Optional[List[Union[WaitAction, ScreenshotAction, ClickAction, WriteAction, PressAction, ScrollAction, ScrapeAction, ExecuteJavascriptAction]]] = None,
             change_tracking_options: Optional[ChangeTrackingOptions] = None,
+            max_age: Optional[int] = None,
+            store_in_cache: Optional[bool] = None,
             **kwargs) -> ScrapeResponse[Any]:
         """
         Scrape and extract content from a URL.
@@ -545,6 +551,10 @@ class FirecrawlApp:
             scrape_params['actions'] = [action if isinstance(action, dict) else action.dict(exclude_none=True) for action in actions]
         if change_tracking_options:
             scrape_params['changeTrackingOptions'] = change_tracking_options if isinstance(change_tracking_options, dict) else change_tracking_options.dict(exclude_none=True)
+        if max_age is not None:
+            scrape_params['maxAge'] = max_age
+        if store_in_cache is not None:
+            scrape_params['storeInCache'] = store_in_cache
         
         scrape_params.update(kwargs)
 
@@ -1102,6 +1112,7 @@ class FirecrawlApp:
             sitemap_only: Optional[bool] = None,
             limit: Optional[int] = None,
             timeout: Optional[int] = None,
+            use_index: Optional[bool] = None,
             **kwargs) -> MapResponse:
         """
         Map and discover links from a URL.
@@ -1144,7 +1155,9 @@ class FirecrawlApp:
             map_params['limit'] = limit
         if timeout is not None:
             map_params['timeout'] = timeout
-
+        if use_index is not None:
+            map_params['useIndex'] = use_index
+        
         # Add any additional kwargs
         map_params.update(kwargs)
 

{firecrawl_py-2.7.1.dist-info → firecrawl_py-2.8.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: firecrawl-py
-Version: 2.7.1
+Version: 2.8.0
 Summary: Python SDK for Firecrawl API
 Home-page: https://github.com/mendableai/firecrawl
 Author: Mendable.ai

firecrawl_py-2.8.0.dist-info/RECORD

@@ -0,0 +1,40 @@
+build/lib/build/lib/build/lib/build/lib/firecrawl/__init__.py,sha256=yjca32tLszqh8QmT5UsMFmlU7KupIvrfvXHFcNaoGFQ,2612
+build/lib/build/lib/build/lib/build/lib/firecrawl/firecrawl.py,sha256=tWGrSt1_G0CJrAbzOwkDAgV2txPSjU5oCzcltRvbgm0,190740
+build/lib/build/lib/build/lib/build/lib/firecrawl/__tests__/e2e_withAuth/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+build/lib/build/lib/build/lib/build/lib/firecrawl/__tests__/e2e_withAuth/test.py,sha256=-Fq2vPcMo0iQi4dwsUkkCd931ybDaTxMBnZbRfGdDcA,7931
+build/lib/build/lib/build/lib/build/lib/firecrawl/__tests__/v1/e2e_withAuth/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+build/lib/build/lib/build/lib/build/lib/firecrawl/__tests__/v1/e2e_withAuth/test.py,sha256=DcCw-cohtnL-t9XPekUtRoQrgg3UCWu8Ikqudf9ory8,19880
+build/lib/build/lib/build/lib/build/lib/tests/test_change_tracking.py,sha256=_IJ5ShLcoj2fHDBaw-nE4I4lHdmDB617ocK_XMHhXps,4177
+build/lib/build/lib/build/lib/firecrawl/__init__.py,sha256=yjca32tLszqh8QmT5UsMFmlU7KupIvrfvXHFcNaoGFQ,2612
+build/lib/build/lib/build/lib/firecrawl/firecrawl.py,sha256=tWGrSt1_G0CJrAbzOwkDAgV2txPSjU5oCzcltRvbgm0,190740
+build/lib/build/lib/build/lib/firecrawl/__tests__/e2e_withAuth/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+build/lib/build/lib/build/lib/firecrawl/__tests__/e2e_withAuth/test.py,sha256=-Fq2vPcMo0iQi4dwsUkkCd931ybDaTxMBnZbRfGdDcA,7931
+build/lib/build/lib/build/lib/firecrawl/__tests__/v1/e2e_withAuth/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+build/lib/build/lib/build/lib/firecrawl/__tests__/v1/e2e_withAuth/test.py,sha256=DcCw-cohtnL-t9XPekUtRoQrgg3UCWu8Ikqudf9ory8,19880
+build/lib/build/lib/build/lib/tests/test_change_tracking.py,sha256=_IJ5ShLcoj2fHDBaw-nE4I4lHdmDB617ocK_XMHhXps,4177
+build/lib/build/lib/firecrawl/__init__.py,sha256=yjca32tLszqh8QmT5UsMFmlU7KupIvrfvXHFcNaoGFQ,2612
+build/lib/build/lib/firecrawl/firecrawl.py,sha256=tWGrSt1_G0CJrAbzOwkDAgV2txPSjU5oCzcltRvbgm0,190740
+build/lib/build/lib/firecrawl/__tests__/e2e_withAuth/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+build/lib/build/lib/firecrawl/__tests__/e2e_withAuth/test.py,sha256=-Fq2vPcMo0iQi4dwsUkkCd931ybDaTxMBnZbRfGdDcA,7931
+build/lib/build/lib/firecrawl/__tests__/v1/e2e_withAuth/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+build/lib/build/lib/firecrawl/__tests__/v1/e2e_withAuth/test.py,sha256=DcCw-cohtnL-t9XPekUtRoQrgg3UCWu8Ikqudf9ory8,19880
+build/lib/build/lib/tests/test_change_tracking.py,sha256=_IJ5ShLcoj2fHDBaw-nE4I4lHdmDB617ocK_XMHhXps,4177
+build/lib/firecrawl/__init__.py,sha256=yjca32tLszqh8QmT5UsMFmlU7KupIvrfvXHFcNaoGFQ,2612
+build/lib/firecrawl/firecrawl.py,sha256=tWGrSt1_G0CJrAbzOwkDAgV2txPSjU5oCzcltRvbgm0,190740
+build/lib/firecrawl/__tests__/e2e_withAuth/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+build/lib/firecrawl/__tests__/e2e_withAuth/test.py,sha256=-Fq2vPcMo0iQi4dwsUkkCd931ybDaTxMBnZbRfGdDcA,7931
+build/lib/firecrawl/__tests__/v1/e2e_withAuth/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+build/lib/firecrawl/__tests__/v1/e2e_withAuth/test.py,sha256=DcCw-cohtnL-t9XPekUtRoQrgg3UCWu8Ikqudf9ory8,19880
+build/lib/tests/test_change_tracking.py,sha256=_IJ5ShLcoj2fHDBaw-nE4I4lHdmDB617ocK_XMHhXps,4177
+firecrawl/__init__.py,sha256=yjca32tLszqh8QmT5UsMFmlU7KupIvrfvXHFcNaoGFQ,2612
+firecrawl/firecrawl.py,sha256=tWGrSt1_G0CJrAbzOwkDAgV2txPSjU5oCzcltRvbgm0,190740
+firecrawl/__tests__/e2e_withAuth/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+firecrawl/__tests__/e2e_withAuth/test.py,sha256=-Fq2vPcMo0iQi4dwsUkkCd931ybDaTxMBnZbRfGdDcA,7931
+firecrawl/__tests__/v1/e2e_withAuth/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+firecrawl/__tests__/v1/e2e_withAuth/test.py,sha256=DcCw-cohtnL-t9XPekUtRoQrgg3UCWu8Ikqudf9ory8,19880
+tests/test_change_tracking.py,sha256=_IJ5ShLcoj2fHDBaw-nE4I4lHdmDB617ocK_XMHhXps,4177
+firecrawl_py-2.8.0.dist-info/LICENSE,sha256=nPCunEDwjRGHlmjvsiDUyIWbkqqyj3Ej84ntnh0g0zA,1084
+firecrawl_py-2.8.0.dist-info/METADATA,sha256=oMrtp1WndHvC_k4bwXxhayCuToEEMFwxpQB9vjid3JM,7168
+firecrawl_py-2.8.0.dist-info/WHEEL,sha256=2wepM1nk4DS4eFpYrW1TTqPcoGNfHhhO_i5m4cOimbo,92
+firecrawl_py-2.8.0.dist-info/top_level.txt,sha256=ytN_R30g2U2qZYFyIm710Z8QeK9FO1Uwa-WPGHXyqjE,27
+firecrawl_py-2.8.0.dist-info/RECORD,,

firecrawl_py-2.7.1.dist-info/RECORD

@@ -1,19 +0,0 @@
-build/lib/firecrawl/__init__.py,sha256=jAxgyVgi4Aq94lwkcicF2_Ba2Y9u51-KfskU9d7ynRQ,2612
-build/lib/firecrawl/firecrawl.py,sha256=fsKXa1cHcIIsGUbFAENLlKYZaW349CHQ6O353Uq6ypg,190196
-build/lib/firecrawl/__tests__/e2e_withAuth/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-build/lib/firecrawl/__tests__/e2e_withAuth/test.py,sha256=-Fq2vPcMo0iQi4dwsUkkCd931ybDaTxMBnZbRfGdDcA,7931
-build/lib/firecrawl/__tests__/v1/e2e_withAuth/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-build/lib/firecrawl/__tests__/v1/e2e_withAuth/test.py,sha256=DcCw-cohtnL-t9XPekUtRoQrgg3UCWu8Ikqudf9ory8,19880
-build/lib/tests/test_change_tracking.py,sha256=_IJ5ShLcoj2fHDBaw-nE4I4lHdmDB617ocK_XMHhXps,4177
-firecrawl/__init__.py,sha256=jAxgyVgi4Aq94lwkcicF2_Ba2Y9u51-KfskU9d7ynRQ,2612
-firecrawl/firecrawl.py,sha256=fsKXa1cHcIIsGUbFAENLlKYZaW349CHQ6O353Uq6ypg,190196
-firecrawl/__tests__/e2e_withAuth/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-firecrawl/__tests__/e2e_withAuth/test.py,sha256=-Fq2vPcMo0iQi4dwsUkkCd931ybDaTxMBnZbRfGdDcA,7931
-firecrawl/__tests__/v1/e2e_withAuth/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-firecrawl/__tests__/v1/e2e_withAuth/test.py,sha256=DcCw-cohtnL-t9XPekUtRoQrgg3UCWu8Ikqudf9ory8,19880
-tests/test_change_tracking.py,sha256=_IJ5ShLcoj2fHDBaw-nE4I4lHdmDB617ocK_XMHhXps,4177
-firecrawl_py-2.7.1.dist-info/LICENSE,sha256=nPCunEDwjRGHlmjvsiDUyIWbkqqyj3Ej84ntnh0g0zA,1084
-firecrawl_py-2.7.1.dist-info/METADATA,sha256=Y987r2PX3wds2_T6J47titJBzWVKNLBvLvd_d9CVQKw,7168
-firecrawl_py-2.7.1.dist-info/WHEEL,sha256=2wepM1nk4DS4eFpYrW1TTqPcoGNfHhhO_i5m4cOimbo,92
-firecrawl_py-2.7.1.dist-info/top_level.txt,sha256=ytN_R30g2U2qZYFyIm710Z8QeK9FO1Uwa-WPGHXyqjE,27
-firecrawl_py-2.7.1.dist-info/RECORD,,