hyperbrowser 0.32.0__py3-none-any.whl → 0.33.0__py3-none-any.whl

hyperbrowser/tools/__init__.py

@@ -1,12 +1,16 @@
+import json
 from hyperbrowser.models.crawl import StartCrawlJobParams
+from hyperbrowser.models.extract import StartExtractJobParams
 from hyperbrowser.models.scrape import StartScrapeJobParams
 from hyperbrowser import Hyperbrowser, AsyncHyperbrowser
 
 from .openai import (
+    EXTRACT_TOOL_OPENAI,
     SCRAPE_TOOL_OPENAI,
     CRAWL_TOOL_OPENAI,
 )
 from .anthropic import (
+    EXTRACT_TOOL_ANTHROPIC,
     SCRAPE_TOOL_ANTHROPIC,
     CRAWL_TOOL_ANTHROPIC,
 )
@@ -56,7 +60,27 @@ class WebsiteCrawlTool:
         return markdown
 
 
+class WebsiteExtractTool:
+    openai_tool_definition = EXTRACT_TOOL_OPENAI
+    anthropic_tool_definition = EXTRACT_TOOL_ANTHROPIC
+
+    @staticmethod
+    def runnable(hb: Hyperbrowser, params: dict) -> str:
+        if params.get("schema") and isinstance(params.get("schema"), str):
+            params["schema"] = json.loads(params["schema"])
+        resp = hb.extract.start_and_wait(params=StartExtractJobParams(**params))
+        return json.dumps(resp.data) if resp.data else ""
+
+    @staticmethod
+    async def async_runnable(hb: AsyncHyperbrowser, params: dict) -> str:
+        if params.get("schema") and isinstance(params.get("schema"), str):
+            params["schema"] = json.loads(params["schema"])
+        resp = await hb.extract.start_and_wait(params=StartExtractJobParams(**params))
+        return json.dumps(resp.data) if resp.data else ""
+
+
 __all__ = [
     "WebsiteScrapeTool",
     "WebsiteCrawlTool",
+    "WebsiteExtractTool",
 ]

hyperbrowser/tools/anthropic.py

@@ -1,7 +1,7 @@
 from typing import Dict, Union, Optional
 from typing_extensions import Literal, Required, TypeAlias, TypedDict
 
-from hyperbrowser.tools.schema import CRAWL_SCHEMA, SCRAPE_SCHEMA
+from hyperbrowser.tools.schema import CRAWL_SCHEMA, EXTRACT_SCHEMA, SCRAPE_SCHEMA
 
 
 class CacheControlEphemeralParam(TypedDict, total=False):
@@ -54,3 +54,9 @@ CRAWL_TOOL_ANTHROPIC: ToolParam = {
     "name": "crawl_website",
     "description": "Crawl a website and return the content in markdown format",
 }
+
+EXTRACT_TOOL_ANTHROPIC: ToolParam = {
+    "input_schema": EXTRACT_SCHEMA,
+    "name": "extract_data",
+    "description": "Extract data in a structured format from multiple URLs in a single function call. IMPORTANT: When information must be gathered from multiple sources (such as comparing items, researching topics across sites, or answering questions that span multiple webpages), ALWAYS include all relevant URLs in ONE function call. This enables comprehensive answers with cross-referenced information. Returns data as a json string.",
+}

hyperbrowser/tools/openai.py

@@ -1,7 +1,7 @@
 from typing import Dict, Optional
 from typing_extensions import Literal, Required, TypedDict, TypeAlias
 
-from hyperbrowser.tools.schema import CRAWL_SCHEMA, SCRAPE_SCHEMA
+from hyperbrowser.tools.schema import CRAWL_SCHEMA, EXTRACT_SCHEMA, SCRAPE_SCHEMA
 
 FunctionParameters: TypeAlias = Dict[str, object]
 
@@ -67,3 +67,13 @@ CRAWL_TOOL_OPENAI: ChatCompletionToolParam = {
         "strict": True,
     },
 }
+
+EXTRACT_TOOL_OPENAI: ChatCompletionToolParam = {
+    "type": "function",
+    "function": {
+        "name": "extract_data",
+        "description": "Extract data in a structured format from multiple URLs in a single function call. IMPORTANT: When information must be gathered from multiple sources (such as comparing items, researching topics across sites, or answering questions that span multiple webpages), ALWAYS include all relevant URLs in ONE function call. This enables comprehensive answers with cross-referenced information. Returns data as a json string.",
+        "parameters": EXTRACT_SCHEMA,
+        "strict": True,
+    },
+}

hyperbrowser/tools/schema.py

@@ -84,3 +84,30 @@ CRAWL_SCHEMA = {
     ],
     "additionalProperties": False,
 }
+
+EXTRACT_SCHEMA = {
+    "type": "object",
+    "properties": {
+        "urls": {
+            "type": "array",
+            "items": {
+                "type": "string",
+            },
+            "description": "A required list of up to 10 urls you want to process IN A SINGLE EXTRACTION. When answering questions that involve multiple sources or topics, ALWAYS include ALL relevant URLs in this single array rather than making separate function calls. This enables cross-referencing information across multiple sources to provide comprehensive answers. To allow crawling for any of the urls provided in the list, simply add /* to the end of the url (https://hyperbrowser.ai/*). This will crawl other pages on the site with the same origin and find relevant pages to use for the extraction context.",
+        },
+        "prompt": {
+            "type": "string",
+            "description": "A prompt describing how you want the data structured, or what you want to extract from the urls provided. Can also be used to guide the extraction process. For multi-source queries, structure this prompt to request unified, comparative, or aggregated information across all provided URLs.",
+        },
+        "schema": {
+            "type": "string",
+            "description": "A strict json schema you want the returned data to be structured as. For multi-source extraction, design this schema to accommodate information from all URLs in a single structure. Ensure that this is a proper json schema, and the root level should be of type 'object'.",
+        },
+        "max_links": {
+            "type": "number",
+            "description": "The maximum number of links to look for if performing a crawl for any given url in the urls list.",
+        },
+    },
+    "required": ["urls", "prompt", "schema", "max_links"],
+    "additionalProperties": False,
+}

{hyperbrowser-0.32.0.dist-info → hyperbrowser-0.33.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: hyperbrowser
-Version: 0.32.0
+Version: 0.33.0
 Summary: Python SDK for hyperbrowser
 License: MIT
 Author: Nikhil Shahi

{hyperbrowser-0.32.0.dist-info → hyperbrowser-0.33.0.dist-info}/RECORD

@@ -31,14 +31,14 @@ hyperbrowser/models/extract.py,sha256=kkcM7DXkwnFJ7d_JZMDAb4VAlQdh-snJktFpJs92z8
 hyperbrowser/models/profile.py,sha256=KRb_LNxxW00AsD_thzzthFS51vInJawt1RcoNz4Q9i8,1322
 hyperbrowser/models/scrape.py,sha256=ZdVNN4rh7EKLh-O1Ou5aU_OgTNxC9pFQlkG-Q2fR-Z4,4611
 hyperbrowser/models/session.py,sha256=i1NkrQWNlKziDd98ySdrUUH7XSv6qOa2cmiK5vV7VeI,6730
-hyperbrowser/tools/__init__.py,sha256=OUaTUM-kiigYmzfbpx3XQhzMK1xT1wd8cqXgR4znsAY,2021
-hyperbrowser/tools/anthropic.py,sha256=5pEkJm1H-26GToTwXsDjo4GGqVy1hATws4Pg59mumow,1667
-hyperbrowser/tools/openai.py,sha256=4-71IIWSxc_ByhywcfWj9-QI9iYNEe0xO6B2spE8WG0,2200
-hyperbrowser/tools/schema.py,sha256=cR2MUX8TvUyN8TnCyeX0pccp4AmPjrdaKzuAXRThOJo,3075
+hyperbrowser/tools/__init__.py,sha256=Wkxk3PKjXs8A2-oN1fDlv5C5nggVznxSITaDIuZbWK4,3029
+hyperbrowser/tools/anthropic.py,sha256=Ij-DcyGwmV9fYhKyZcuEZijr3A5Pr6Tetxh6TMJEZls,2227
+hyperbrowser/tools/openai.py,sha256=yb1BtTNUnkN3_hcsWQDiHwHg8vA4126klEo2R1M1mYc,2854
+hyperbrowser/tools/schema.py,sha256=UsirWHaRpDIlc1yusI_l26yizzEtQugmrOlovsCeC6E,4920
 hyperbrowser/transport/async_transport.py,sha256=6HKoeM5TutIqraEscEWobvSPWF3iVKh2hPflGNKwykw,4128
 hyperbrowser/transport/base.py,sha256=ildpMrDiM8nvrSGrH2LTOafmB17T7PQB_NQ1ODA378U,1703
 hyperbrowser/transport/sync.py,sha256=aUVpxWF8sqSycLNKxVNEZvlsZSoqc1eHgPK1Y1QA1u8,3422
-hyperbrowser-0.32.0.dist-info/LICENSE,sha256=6rUGKlyKb_1ZAH7h7YITYAAUNFN3MNGGKCyfrw49NLE,1071
-hyperbrowser-0.32.0.dist-info/METADATA,sha256=rxEInDkCZjDhAMsDvmHUe-VwXSEBimHmfhxAF6XrHQQ,3438
-hyperbrowser-0.32.0.dist-info/WHEEL,sha256=XbeZDeTWKc1w7CSIyre5aMDU_-PohRwTQceYnisIYYY,88
-hyperbrowser-0.32.0.dist-info/RECORD,,
+hyperbrowser-0.33.0.dist-info/LICENSE,sha256=6rUGKlyKb_1ZAH7h7YITYAAUNFN3MNGGKCyfrw49NLE,1071
+hyperbrowser-0.33.0.dist-info/METADATA,sha256=twHJeCFloGz_L7USASXaAuHXNXLLjVVPZSXs6Zkrq7o,3438
+hyperbrowser-0.33.0.dist-info/WHEEL,sha256=XbeZDeTWKc1w7CSIyre5aMDU_-PohRwTQceYnisIYYY,88
+hyperbrowser-0.33.0.dist-info/RECORD,,