Merge pull request #4 from oxylabs/feature/new-output-types

Feature/new output types

Author: andriuskuk

agentic_code_guide.md

@@ -78,7 +78,7 @@ Output (result):
 
     ```python
     class DataModel(BaseModel):
-        type: Literal["json", "markdown", "html", "screenshot"]
+        type: Literal["json", "markdown", "html", "screenshot", "csv"]
         content: dict[str, Any] | str | None
 
     class BrowserAgentJob(BaseModel):
@@ -138,7 +138,7 @@ if __name__ == "__main__":
 Parameters:
 
 - url (str): Target URL to scrape (required)
-- output_format (Literal["json", "markdown"]): Output format (default: "markdown")
+- output_format (Literal["json", "markdown", "csv", "screenshot"]): Output format (default: "markdown")
 - schema (dict | None): OpenAPI schema for structured extraction (required if output_format is "json")
 - render_javascript (bool): Render JavaScript (default: False)
 - geo_location (str): proxy location in ISO2 format.
@@ -156,6 +156,8 @@ Output (result):
 
     If output_format is "json", data will be a dictionary.
     If output_format is "markdown", data will be a string.
+    If output_format is "csv", data will be a string formatted in a form of csv.
+    If output_format is "screenshot", data will be a string.
 
 
 ## Use Cases Examples

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "oxylabs-ai-studio"
-version = "0.2.14"
+version = "0.2.15"
 description = "Oxylabs studio python sdk"
 readme = "README.md"
 keywords = ["oxylabs", "ai", "studio"]

src/oxylabs_ai_studio/apps/ai_crawler.py

@@ -31,14 +31,16 @@ def crawl(
         self,
         url: str,
         user_prompt: str,
-        output_format: Literal["json", "markdown"] = "markdown",
+        output_format: Literal["json", "markdown", "csv"] = "markdown",
         schema: dict[str, Any] | None = None,
         render_javascript: bool = False,
         return_sources_limit: int = 25,
         geo_location: str | None = None,
     ) -> AiCrawlerJob:
-        if output_format == "json" and schema is None:
-            raise ValueError("openapi_schema is required when output_format is json")
+        if output_format in ["json", "csv"] and schema is None:
+            raise ValueError(
+                "openapi_schema is required when output_format is json or csv.",
+            )
 
         body = {
             "domain": url,
@@ -119,15 +121,17 @@ async def crawl_async(
         self,
         url: str,
         user_prompt: str = "",
-        output_format: Literal["json", "markdown"] = "markdown",
+        output_format: Literal["json", "markdown", "csv"] = "markdown",
         schema: dict[str, Any] | None = None,
         render_javascript: bool = False,
         return_sources_limit: int = 25,
         geo_location: str | None = None,
     ) -> AiCrawlerJob:
         """Async version of crawl."""
-        if output_format == "json" and schema is None:
-            raise ValueError("openapi_schema is required when output_format is json")
+        if output_format in ["json", "csv"] and schema is None:
+            raise ValueError(
+                "openapi_schema is required when output_format is json or csv.",
+            )
 
         body = {
             "domain": url,

src/oxylabs_ai_studio/apps/ai_scraper.py

@@ -31,13 +31,15 @@ def __init__(self, api_key: str | None = None):
     def scrape(
         self,
         url: str,
-        output_format: Literal["json", "markdown"] = "markdown",
+        output_format: Literal["json", "markdown", "csv", "screenshot"] = "markdown",
         schema: dict[str, Any] | None = None,
         render_javascript: bool = False,
         geo_location: str | None = None,
     ) -> AiScraperJob:
-        if output_format == "json" and schema is None:
-            raise ValueError("openapi_schema is required when output_format is json")
+        if output_format in ["json", "csv"] and schema is None:
+            raise ValueError(
+                "openapi_schema is required when output_format is json or csv.",
+            )
 
         body = {
             "url": url,
@@ -117,14 +119,16 @@ def generate_schema(self, prompt: str) -> dict[str, Any] | None:
     async def scrape_async(
         self,
         url: str,
-        output_format: Literal["json", "markdown"] = "markdown",
+        output_format: Literal["json", "markdown", "csv", "screenshot"] = "markdown",
         schema: dict[str, Any] | None = None,
         render_javascript: bool = False,
         geo_location: str | None = None,
     ) -> AiScraperJob:
         """Async version of scrape."""
-        if output_format == "json" and schema is None:
-            raise ValueError("openapi_schema is required when output_format is json")
+        if output_format in ["json", "csv"] and schema is None:
+            raise ValueError(
+                "openapi_schema is required when output_format is json or csv.",
+            )
 
         body = {
             "url": url,

src/oxylabs_ai_studio/apps/browser_agent.py

@@ -16,7 +16,7 @@
 
 
 class DataModel(BaseModel):
-    type: Literal["json", "markdown", "html", "screenshot"]
+    type: Literal["json", "markdown", "html", "screenshot", "csv"]
     content: dict[str, Any] | str | None
 
 
@@ -34,12 +34,16 @@ def run(
         self,
         url: str,
         user_prompt: str = "",
-        output_format: Literal["json", "markdown", "html", "screenshot"] = "markdown",
+        output_format: Literal[
+            "json", "markdown", "html", "screenshot", "csv"
+        ] = "markdown",
         schema: dict[str, Any] | None = None,
         geo_location: str | None = None,
     ) -> BrowserAgentJob:
-        if output_format == "json" and schema is None:
-            raise ValueError("openapi_schema is required when output_format is json")
+        if output_format in ["json", "csv"] and schema is None:
+            raise ValueError(
+                "openapi_schema is required when output_format is json or csv.",
+            )
 
         body = {
             "url": url,
@@ -115,13 +119,21 @@ async def run_async(
         self,
         url: str,
         user_prompt: str = "",
-        output_format: Literal["json", "markdown", "html", "screenshot"] = "markdown",
+        output_format: Literal[
+            "json",
+            "markdown",
+            "html",
+            "screenshot",
+            "csv",
+        ] = "markdown",
         schema: dict[str, Any] | None = None,
         geo_location: str | None = None,
     ) -> BrowserAgentJob:
         """Async version of run."""
-        if output_format == "json" and schema is None:
-            raise ValueError("openapi_schema is required when output_format is json")
+        if output_format in ["json", "csv"] and schema is None:
+            raise ValueError(
+                "openapi_schema is required when output_format is json or csv.",
+            )
 
         body = {
             "url": url,