prompture 0.0.38__tar.gz → 0.0.38.dev1__tar.gz

{prompture-0.0.38/prompture.egg-info → prompture-0.0.38.dev1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: prompture
-Version: 0.0.38
+Version: 0.0.38.dev1
 Summary: Ask LLMs to return structured JSON and run cross-model tests. API-first.
 Author-email: Juan Denis <juan@vene.co>
 License-Expression: MIT

{prompture-0.0.38 → prompture-0.0.38.dev1}/docs/source/conf.py

@@ -14,7 +14,7 @@ sys.path.insert(0, os.path.abspath("../../"))
 # https://www.sphinx-doc.org/en/master/usage/configuration.html#project-information
 
 project = "Prompture"
-copyright = '2026, Juan Denis'
+copyright = '2026, <a href="https://juandenis.com">Juan Denis</a>'
 author = "Juan Denis"
 
 # Read version dynamically: VERSION file > setuptools_scm > fallback

{prompture-0.0.38 → prompture-0.0.38.dev1}/prompture/_version.py

@@ -28,7 +28,7 @@ version_tuple: VERSION_TUPLE
 commit_id: COMMIT_ID
 __commit_id__: COMMIT_ID
 
-__version__ = version = '0.0.38'
-__version_tuple__ = version_tuple = (0, 0, 38)
+__version__ = version = '0.0.38.dev1'
+__version_tuple__ = version_tuple = (0, 0, 38, 'dev1')
 
 __commit_id__ = commit_id = None

{prompture-0.0.38 → prompture-0.0.38.dev1}/prompture/drivers/async_azure_driver.py

@@ -113,7 +113,7 @@ class AsyncAzureDriver(CostMixin, AsyncDriver):
             "prompt_tokens": prompt_tokens,
             "completion_tokens": completion_tokens,
             "total_tokens": total_tokens,
-            "cost": round(total_cost, 6),
+            "cost": total_cost,
             "raw_response": resp.model_dump(),
             "model_name": model,
             "deployment_id": self.deployment_id,

prompture-0.0.38.dev1/prompture/drivers/async_claude_driver.py

@@ -0,0 +1,113 @@
+"""Async Anthropic Claude driver. Requires the ``anthropic`` package."""
+
+from __future__ import annotations
+
+import json
+import os
+from typing import Any
+
+try:
+    import anthropic
+except Exception:
+    anthropic = None
+
+from ..async_driver import AsyncDriver
+from ..cost_mixin import CostMixin
+from .claude_driver import ClaudeDriver
+
+
+class AsyncClaudeDriver(CostMixin, AsyncDriver):
+    supports_json_mode = True
+    supports_json_schema = True
+    supports_vision = True
+
+    MODEL_PRICING = ClaudeDriver.MODEL_PRICING
+
+    def __init__(self, api_key: str | None = None, model: str = "claude-3-5-haiku-20241022"):
+        self.api_key = api_key or os.getenv("CLAUDE_API_KEY")
+        self.model = model or os.getenv("CLAUDE_MODEL_NAME", "claude-3-5-haiku-20241022")
+
+    supports_messages = True
+
+    def _prepare_messages(self, messages: list[dict[str, Any]]) -> list[dict[str, Any]]:
+        from .vision_helpers import _prepare_claude_vision_messages
+
+        return _prepare_claude_vision_messages(messages)
+
+    async def generate(self, prompt: str, options: dict[str, Any]) -> dict[str, Any]:
+        messages = [{"role": "user", "content": prompt}]
+        return await self._do_generate(messages, options)
+
+    async def generate_messages(self, messages: list[dict[str, str]], options: dict[str, Any]) -> dict[str, Any]:
+        return await self._do_generate(self._prepare_messages(messages), options)
+
+    async def _do_generate(self, messages: list[dict[str, str]], options: dict[str, Any]) -> dict[str, Any]:
+        if anthropic is None:
+            raise RuntimeError("anthropic package not installed")
+
+        opts = {**{"temperature": 0.0, "max_tokens": 512}, **options}
+        model = options.get("model", self.model)
+
+        client = anthropic.AsyncAnthropic(api_key=self.api_key)
+
+        # Anthropic requires system messages as a top-level parameter
+        system_content = None
+        api_messages = []
+        for msg in messages:
+            if msg.get("role") == "system":
+                system_content = msg.get("content", "")
+            else:
+                api_messages.append(msg)
+
+        # Build common kwargs
+        common_kwargs: dict[str, Any] = {
+            "model": model,
+            "messages": api_messages,
+            "temperature": opts["temperature"],
+            "max_tokens": opts["max_tokens"],
+        }
+        if system_content:
+            common_kwargs["system"] = system_content
+
+        # Native JSON mode: use tool-use for schema enforcement
+        if options.get("json_mode"):
+            json_schema = options.get("json_schema")
+            if json_schema:
+                tool_def = {
+                    "name": "extract_json",
+                    "description": "Extract structured data matching the schema",
+                    "input_schema": json_schema,
+                }
+                resp = await client.messages.create(
+                    **common_kwargs,
+                    tools=[tool_def],
+                    tool_choice={"type": "tool", "name": "extract_json"},
+                )
+                text = ""
+                for block in resp.content:
+                    if block.type == "tool_use":
+                        text = json.dumps(block.input)
+                        break
+            else:
+                resp = await client.messages.create(**common_kwargs)
+                text = resp.content[0].text
+        else:
+            resp = await client.messages.create(**common_kwargs)
+            text = resp.content[0].text
+
+        prompt_tokens = resp.usage.input_tokens
+        completion_tokens = resp.usage.output_tokens
+        total_tokens = prompt_tokens + completion_tokens
+
+        total_cost = self._calculate_cost("claude", model, prompt_tokens, completion_tokens)
+
+        meta = {
+            "prompt_tokens": prompt_tokens,
+            "completion_tokens": completion_tokens,
+            "total_tokens": total_tokens,
+            "cost": total_cost,
+            "raw_response": dict(resp),
+            "model_name": model,
+        }
+
+        return {"text": text, "meta": meta}

prompture-0.0.38.dev1/prompture/drivers/async_google_driver.py

@@ -0,0 +1,152 @@
+"""Async Google Generative AI (Gemini) driver."""
+
+from __future__ import annotations
+
+import logging
+import os
+from typing import Any
+
+import google.generativeai as genai
+
+from ..async_driver import AsyncDriver
+from ..cost_mixin import CostMixin
+from .google_driver import GoogleDriver
+
+logger = logging.getLogger(__name__)
+
+
+class AsyncGoogleDriver(CostMixin, AsyncDriver):
+    """Async driver for Google's Generative AI API (Gemini)."""
+
+    supports_json_mode = True
+    supports_json_schema = True
+    supports_vision = True
+
+    MODEL_PRICING = GoogleDriver.MODEL_PRICING
+    _PRICING_UNIT = 1_000_000
+
+    def __init__(self, api_key: str | None = None, model: str = "gemini-1.5-pro"):
+        self.api_key = api_key or os.getenv("GOOGLE_API_KEY")
+        if not self.api_key:
+            raise ValueError("Google API key not found. Set GOOGLE_API_KEY env var or pass api_key to constructor")
+        self.model = model
+        genai.configure(api_key=self.api_key)
+        self.options: dict[str, Any] = {}
+
+    def _calculate_cost_chars(self, prompt_chars: int, completion_chars: int) -> float:
+        """Calculate cost from character counts (same logic as sync GoogleDriver)."""
+        from ..model_rates import get_model_rates
+
+        live_rates = get_model_rates("google", self.model)
+        if live_rates:
+            est_prompt_tokens = prompt_chars / 4
+            est_completion_tokens = completion_chars / 4
+            prompt_cost = (est_prompt_tokens / 1_000_000) * live_rates["input"]
+            completion_cost = (est_completion_tokens / 1_000_000) * live_rates["output"]
+        else:
+            model_pricing = self.MODEL_PRICING.get(self.model, {"prompt": 0, "completion": 0})
+            prompt_cost = (prompt_chars / 1_000_000) * model_pricing["prompt"]
+            completion_cost = (completion_chars / 1_000_000) * model_pricing["completion"]
+        return round(prompt_cost + completion_cost, 6)
+
+    supports_messages = True
+
+    def _prepare_messages(self, messages: list[dict[str, Any]]) -> list[dict[str, Any]]:
+        from .vision_helpers import _prepare_google_vision_messages
+
+        return _prepare_google_vision_messages(messages)
+
+    async def generate(self, prompt: str, options: dict[str, Any] | None = None) -> dict[str, Any]:
+        messages = [{"role": "user", "content": prompt}]
+        return await self._do_generate(messages, options)
+
+    async def generate_messages(self, messages: list[dict[str, str]], options: dict[str, Any]) -> dict[str, Any]:
+        return await self._do_generate(self._prepare_messages(messages), options)
+
+    async def _do_generate(
+        self, messages: list[dict[str, str]], options: dict[str, Any] | None = None
+    ) -> dict[str, Any]:
+        merged_options = self.options.copy()
+        if options:
+            merged_options.update(options)
+
+        generation_config = merged_options.get("generation_config", {})
+        safety_settings = merged_options.get("safety_settings", {})
+
+        if "temperature" in merged_options and "temperature" not in generation_config:
+            generation_config["temperature"] = merged_options["temperature"]
+        if "max_tokens" in merged_options and "max_output_tokens" not in generation_config:
+            generation_config["max_output_tokens"] = merged_options["max_tokens"]
+        if "top_p" in merged_options and "top_p" not in generation_config:
+            generation_config["top_p"] = merged_options["top_p"]
+        if "top_k" in merged_options and "top_k" not in generation_config:
+            generation_config["top_k"] = merged_options["top_k"]
+
+        # Native JSON mode support
+        if merged_options.get("json_mode"):
+            generation_config["response_mime_type"] = "application/json"
+            json_schema = merged_options.get("json_schema")
+            if json_schema:
+                generation_config["response_schema"] = json_schema
+
+        # Convert messages to Gemini format
+        system_instruction = None
+        contents: list[dict[str, Any]] = []
+        for msg in messages:
+            role = msg.get("role", "user")
+            content = msg.get("content", "")
+            if role == "system":
+                system_instruction = content if isinstance(content, str) else str(content)
+            else:
+                gemini_role = "model" if role == "assistant" else "user"
+                if msg.get("_vision_parts"):
+                    # Already converted to Gemini parts by _prepare_messages
+                    contents.append({"role": gemini_role, "parts": content})
+                else:
+                    contents.append({"role": gemini_role, "parts": [content]})
+
+        try:
+            model_kwargs: dict[str, Any] = {}
+            if system_instruction:
+                model_kwargs["system_instruction"] = system_instruction
+            model = genai.GenerativeModel(self.model, **model_kwargs)
+
+            gen_input: Any = contents if len(contents) != 1 else contents[0]["parts"][0]
+            response = await model.generate_content_async(
+                gen_input,
+                generation_config=generation_config if generation_config else None,
+                safety_settings=safety_settings if safety_settings else None,
+            )
+
+            if not response.text:
+                raise ValueError("Empty response from model")
+
+            total_prompt_chars = 0
+            for msg in messages:
+                c = msg.get("content", "")
+                if isinstance(c, str):
+                    total_prompt_chars += len(c)
+                elif isinstance(c, list):
+                    for part in c:
+                        if isinstance(part, str):
+                            total_prompt_chars += len(part)
+                        elif isinstance(part, dict) and "text" in part:
+                            total_prompt_chars += len(part["text"])
+            completion_chars = len(response.text)
+
+            total_cost = self._calculate_cost_chars(total_prompt_chars, completion_chars)
+
+            meta = {
+                "prompt_chars": total_prompt_chars,
+                "completion_chars": completion_chars,
+                "total_chars": total_prompt_chars + completion_chars,
+                "cost": total_cost,
+                "raw_response": response.prompt_feedback if hasattr(response, "prompt_feedback") else None,
+                "model_name": self.model,
+            }
+
+            return {"text": response.text, "meta": meta}
+
+        except Exception as e:
+            logger.error(f"Google API request failed: {e}")
+            raise RuntimeError(f"Google API request failed: {e}") from e

{prompture-0.0.38 → prompture-0.0.38.dev1}/prompture/drivers/async_grok_driver.py

@@ -88,7 +88,7 @@ class AsyncGrokDriver(CostMixin, AsyncDriver):
             "prompt_tokens": prompt_tokens,
             "completion_tokens": completion_tokens,
             "total_tokens": total_tokens,
-            "cost": round(total_cost, 6),
+            "cost": total_cost,
             "raw_response": resp,
             "model_name": model,
         }

{prompture-0.0.38 → prompture-0.0.38.dev1}/prompture/drivers/async_groq_driver.py

@@ -81,7 +81,7 @@ class AsyncGroqDriver(CostMixin, AsyncDriver):
             "prompt_tokens": prompt_tokens,
             "completion_tokens": completion_tokens,
             "total_tokens": total_tokens,
-            "cost": round(total_cost, 6),
+            "cost": total_cost,
             "raw_response": resp.model_dump(),
             "model_name": model,
         }

prompture-0.0.38.dev1/prompture/drivers/async_openai_driver.py

@@ -0,0 +1,102 @@
+"""Async OpenAI driver. Requires the ``openai`` package (>=1.0.0)."""
+
+from __future__ import annotations
+
+import os
+from typing import Any
+
+try:
+    from openai import AsyncOpenAI
+except Exception:
+    AsyncOpenAI = None
+
+from ..async_driver import AsyncDriver
+from ..cost_mixin import CostMixin
+from .openai_driver import OpenAIDriver
+
+
+class AsyncOpenAIDriver(CostMixin, AsyncDriver):
+    supports_json_mode = True
+    supports_json_schema = True
+    supports_vision = True
+
+    MODEL_PRICING = OpenAIDriver.MODEL_PRICING
+
+    def __init__(self, api_key: str | None = None, model: str = "gpt-4o-mini"):
+        self.api_key = api_key or os.getenv("OPENAI_API_KEY")
+        self.model = model
+        if AsyncOpenAI:
+            self.client = AsyncOpenAI(api_key=self.api_key)
+        else:
+            self.client = None
+
+    supports_messages = True
+
+    def _prepare_messages(self, messages: list[dict[str, Any]]) -> list[dict[str, Any]]:
+        from .vision_helpers import _prepare_openai_vision_messages
+
+        return _prepare_openai_vision_messages(messages)
+
+    async def generate(self, prompt: str, options: dict[str, Any]) -> dict[str, Any]:
+        messages = [{"role": "user", "content": prompt}]
+        return await self._do_generate(messages, options)
+
+    async def generate_messages(self, messages: list[dict[str, str]], options: dict[str, Any]) -> dict[str, Any]:
+        return await self._do_generate(self._prepare_messages(messages), options)
+
+    async def _do_generate(self, messages: list[dict[str, str]], options: dict[str, Any]) -> dict[str, Any]:
+        if self.client is None:
+            raise RuntimeError("openai package (>=1.0.0) is not installed")
+
+        model = options.get("model", self.model)
+
+        model_info = self.MODEL_PRICING.get(model, {})
+        tokens_param = model_info.get("tokens_param", "max_tokens")
+        supports_temperature = model_info.get("supports_temperature", True)
+
+        opts = {"temperature": 1.0, "max_tokens": 512, **options}
+
+        kwargs = {
+            "model": model,
+            "messages": messages,
+        }
+        kwargs[tokens_param] = opts.get("max_tokens", 512)
+
+        if supports_temperature and "temperature" in opts:
+            kwargs["temperature"] = opts["temperature"]
+
+        # Native JSON mode support
+        if options.get("json_mode"):
+            json_schema = options.get("json_schema")
+            if json_schema:
+                kwargs["response_format"] = {
+                    "type": "json_schema",
+                    "json_schema": {
+                        "name": "extraction",
+                        "strict": True,
+                        "schema": json_schema,
+                    },
+                }
+            else:
+                kwargs["response_format"] = {"type": "json_object"}
+
+        resp = await self.client.chat.completions.create(**kwargs)
+
+        usage = getattr(resp, "usage", None)
+        prompt_tokens = getattr(usage, "prompt_tokens", 0)
+        completion_tokens = getattr(usage, "completion_tokens", 0)
+        total_tokens = getattr(usage, "total_tokens", 0)
+
+        total_cost = self._calculate_cost("openai", model, prompt_tokens, completion_tokens)
+
+        meta = {
+            "prompt_tokens": prompt_tokens,
+            "completion_tokens": completion_tokens,
+            "total_tokens": total_tokens,
+            "cost": total_cost,
+            "raw_response": resp.model_dump(),
+            "model_name": model,
+        }
+
+        text = resp.choices[0].message.content
+        return {"text": text, "meta": meta}

{prompture-0.0.38 → prompture-0.0.38.dev1}/prompture/drivers/async_openrouter_driver.py

@@ -93,7 +93,7 @@ class AsyncOpenRouterDriver(CostMixin, AsyncDriver):
             "prompt_tokens": prompt_tokens,
             "completion_tokens": completion_tokens,
             "total_tokens": total_tokens,
-            "cost": round(total_cost, 6),
+            "cost": total_cost,
             "raw_response": resp,
             "model_name": model,
         }