causaliq-knowledge 0.3.0__py3-none-any.whl → 0.4.0__py3-none-any.whl

causaliq_knowledge/llm/cache.py

@@ -10,6 +10,7 @@ The base cache infrastructure will migrate to causaliq-core.
 
 from __future__ import annotations
 
+import json
 from dataclasses import asdict, dataclass, field
 from datetime import datetime, timezone
 from pathlib import Path
@@ -47,6 +48,7 @@ class LLMMetadata:
         tokens: Token usage statistics.
         cost_usd: Estimated cost of the request in USD.
         cache_hit: Whether this was served from cache.
+        request_id: Optional identifier for the request (not in cache key).
     """
 
     provider: str = ""
@@ -55,6 +57,7 @@ class LLMMetadata:
     tokens: LLMTokenUsage = field(default_factory=LLMTokenUsage)
     cost_usd: float = 0.0
     cache_hit: bool = False
+    request_id: str = ""
 
     def to_dict(self) -> dict[str, Any]:
         """Convert to dictionary for JSON serialisation."""
@@ -65,6 +68,7 @@ class LLMMetadata:
             "tokens": asdict(self.tokens),
             "cost_usd": self.cost_usd,
             "cache_hit": self.cache_hit,
+            "request_id": self.request_id,
         }
 
     @classmethod
@@ -82,6 +86,7 @@ class LLMMetadata:
             ),
             cost_usd=data.get("cost_usd", 0.0),
             cache_hit=data.get("cache_hit", False),
+            request_id=data.get("request_id", ""),
         )
 
 
@@ -107,11 +112,33 @@ class LLMResponse:
             "model_version": self.model_version,
         }
 
+    def to_export_dict(self) -> dict[str, Any]:
+        """Convert to dictionary for export, parsing JSON content if valid.
+
+        Unlike to_dict(), this attempts to parse the content as JSON
+        for more readable exported files.
+        """
+        # Try to parse content as JSON for cleaner export
+        try:
+            parsed_content = json.loads(self.content)
+        except (json.JSONDecodeError, TypeError):
+            parsed_content = self.content
+
+        return {
+            "content": parsed_content,
+            "finish_reason": self.finish_reason,
+            "model_version": self.model_version,
+        }
+
     @classmethod
     def from_dict(cls, data: dict[str, Any]) -> LLMResponse:
         """Create from dictionary."""
+        content = data.get("content", "")
+        # Handle both string and parsed JSON content (from export files)
+        if isinstance(content, dict):
+            content = json.dumps(content)
         return cls(
-            content=data.get("content", ""),
+            content=content,
             finish_reason=data.get("finish_reason", "stop"),
             model_version=data.get("model_version", ""),
         )
@@ -137,6 +164,30 @@ class LLMCacheEntry:
     response: LLMResponse = field(default_factory=LLMResponse)
     metadata: LLMMetadata = field(default_factory=LLMMetadata)
 
+    @staticmethod
+    def _split_message_content(messages: list[dict[str, Any]]) -> list[Any]:
+        """Convert message content with newlines into arrays of lines."""
+        result = []
+        for msg in messages:
+            new_msg = dict(msg)
+            content = new_msg.get("content", "")
+            if isinstance(content, str) and "\n" in content:
+                new_msg["content"] = content.split("\n")
+            result.append(new_msg)
+        return result
+
+    @staticmethod
+    def _join_message_content(messages: list[Any]) -> list[dict[str, Any]]:
+        """Convert message content arrays back into strings with newlines."""
+        result = []
+        for msg in messages:
+            new_msg = dict(msg)
+            content = new_msg.get("content", "")
+            if isinstance(content, list):
+                new_msg["content"] = "\n".join(content)
+            result.append(new_msg)
+        return result
+
     def to_dict(self) -> dict[str, Any]:
         """Convert to dictionary for JSON serialisation."""
         return {
@@ -150,13 +201,37 @@ class LLMCacheEntry:
             "metadata": self.metadata.to_dict(),
         }
 
+    def to_export_dict(self) -> dict[str, Any]:
+        """Convert to dictionary for export with readable formatting.
+
+        - Message content with newlines is split into arrays of lines
+        - Response JSON content is parsed into a proper JSON structure
+        """
+        return {
+            "cache_key": {
+                "model": self.model,
+                "messages": self._split_message_content(self.messages),
+                "temperature": self.temperature,
+                "max_tokens": self.max_tokens,
+            },
+            "response": self.response.to_export_dict(),
+            "metadata": self.metadata.to_dict(),
+        }
+
     @classmethod
     def from_dict(cls, data: dict[str, Any]) -> LLMCacheEntry:
-        """Create from dictionary."""
+        """Create from dictionary.
+
+        Handles both internal format (string content) and export format
+        (array of lines for content).
+        """
         cache_key = data.get("cache_key", {})
+        messages = cache_key.get("messages", [])
+        # Handle export format where content is array of lines
+        messages = cls._join_message_content(messages)
         return cls(
             model=cache_key.get("model", ""),
-            messages=cache_key.get("messages", []),
+            messages=messages,
             temperature=cache_key.get("temperature", 0.0),
             max_tokens=cache_key.get("max_tokens"),
             response=LLMResponse.from_dict(data.get("response", {})),
@@ -179,6 +254,7 @@ class LLMCacheEntry:
         input_tokens: int = 0,
         output_tokens: int = 0,
         cost_usd: float = 0.0,
+        request_id: str = "",
     ) -> LLMCacheEntry:
         """Create a cache entry with common parameters.
 
@@ -195,6 +271,7 @@ class LLMCacheEntry:
             input_tokens: Number of input tokens.
             output_tokens: Number of output tokens.
             cost_usd: Estimated cost in USD.
+            request_id: Optional identifier for the request (not part of hash).
 
         Returns:
             Configured LLMCacheEntry.
@@ -220,6 +297,7 @@ class LLMCacheEntry:
                 ),
                 cost_usd=cost_usd,
                 cache_hit=False,
+                request_id=request_id,
             ),
         )
 
@@ -285,18 +363,14 @@ class LLMEntryEncoder(JsonEncoder):
     ) -> str:
         """Generate a human-readable filename for export.
 
-        Creates a filename from model name and query details, with a
-        short hash suffix for uniqueness.
+        Creates a filename using request_id, timestamp, and provider:
+            {request_id}_{yyyy-mm-dd-hhmmss}_{provider}.json
 
-        For edge queries, extracts node names for format:
-            {model}_{node_a}_{node_b}_edge_{hash}.json
-
-        For other queries, uses prompt excerpt:
-            {model}_{prompt_excerpt}_{hash}.json
+        If request_id is not set, falls back to a short hash prefix.
 
         Args:
             entry: The cache entry to generate filename for.
-            cache_key: The cache key (hash) for uniqueness suffix.
+            cache_key: The cache key (hash) for fallback uniqueness.
 
         Returns:
             Human-readable filename with .json extension.
@@ -305,67 +379,56 @@ class LLMEntryEncoder(JsonEncoder):
             >>> encoder = LLMEntryEncoder()
             >>> entry = LLMCacheEntry.create(
             ...     model="gpt-4",
-            ...     messages=[{"role": "user", "content": "smoking and lung"}],
-            ...     content="Yes...",
+            ...     messages=[{"role": "user", "content": "test"}],
+            ...     content="Response",
+            ...     provider="openai",
+            ...     request_id="expt23",
             ... )
-            >>> encoder.generate_export_filename(entry, "a1b2c3d4e5f6")
-            'gpt4_smoking_lung_edge_a1b2.json'
+            >>> # Returns something like: expt23_2026-01-29-143052_openai.json
         """
         import re
-
-        # Sanitize model name (alphanumeric only, lowercase)
-        model = re.sub(r"[^a-z0-9]", "", entry.model.lower())
-        if len(model) > 15:
-            model = model[:15]
-
-        # Extract user message content
-        prompt = ""
-        for msg in entry.messages:
-            if msg.get("role") == "user":
-                prompt = msg.get("content", "")
-                break
-
-        # Try to extract node names for edge queries
-        # Look for patterns like "X and Y", "X cause Y", "between X and Y"
-        prompt_lower = prompt.lower()
-        slug = ""
-
-        # Pattern: "between X and Y" or "X and Y"
-        match = re.search(r"(?:between\s+)?(\w+)\s+and\s+(\w+)", prompt_lower)
-        if match:
-            node_a = match.group(1)[:15]
-            node_b = match.group(2)[:15]
-            slug = f"{node_a}_{node_b}_edge"
-
-        # Fallback: extract first significant words from prompt
-        if not slug:
-            # Remove common words, keep alphanumeric
-            cleaned = re.sub(r"[^a-z0-9\s]", "", prompt_lower)
-            words = [
-                w
-                for w in cleaned.split()
-                if w
-                not in ("the", "a", "an", "is", "are", "does", "do", "can")
-            ]
-            slug = "_".join(words[:4])
-            if len(slug) > 30:
-                slug = slug[:30].rstrip("_")
-
-        # Short hash suffix for uniqueness (4 chars)
-        hash_suffix = cache_key[:4] if cache_key else "0000"
-
-        # Build filename
-        parts = [p for p in [model, slug, hash_suffix] if p]
-        return "_".join(parts) + ".json"
+        from datetime import datetime
+
+        # Get request_id or use hash prefix as fallback
+        request_id = entry.metadata.request_id or cache_key[:8]
+        # Sanitise request_id (alphanumeric, hyphens, underscores only)
+        request_id = re.sub(r"[^a-zA-Z0-9_-]", "", request_id)
+        if not request_id:
+            request_id = cache_key[:8] if cache_key else "unknown"
+
+        # Parse timestamp and format as yyyy-mm-dd-hhmmss
+        timestamp_str = entry.metadata.timestamp
+        if timestamp_str:
+            try:
+                # Parse ISO format timestamp
+                dt = datetime.fromisoformat(
+                    timestamp_str.replace("Z", "+00:00")
+                )
+                formatted_ts = dt.strftime("%Y-%m-%d-%H%M%S")
+            except ValueError:
+                formatted_ts = "unknown"
+        else:
+            formatted_ts = "unknown"
+
+        # Get provider, sanitised
+        provider = entry.metadata.provider or "unknown"
+        provider = re.sub(r"[^a-z0-9]", "", provider.lower())
+        if not provider:
+            provider = "unknown"
+
+        # Build filename: id_timestamp_provider.json
+        return f"{request_id}_{formatted_ts}_{provider}.json"
 
     def export_entry(self, entry: LLMCacheEntry, path: Path) -> None:
         """Export an LLMCacheEntry to a JSON file.
 
+        Uses to_export_dict() to parse JSON content for readability.
+
         Args:
             entry: The cache entry to export.
             path: Destination file path.
         """
-        self.export(entry.to_dict(), path)
+        self.export(entry.to_export_dict(), path)
 
     def import_entry(self, path: Path) -> LLMCacheEntry:
         """Import an LLMCacheEntry from a JSON file.

{causaliq_knowledge-0.3.0.dist-info → causaliq_knowledge-0.4.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: causaliq-knowledge
-Version: 0.3.0
+Version: 0.4.0
 Summary: Incorporating LLM and human knowledge into causal discovery
 Author-email: CausalIQ <info@causaliq.com>
 Maintainer-email: CausalIQ <info@causaliq.com>
@@ -24,6 +24,7 @@ Classifier: Topic :: Software Development :: Libraries :: Python Modules
 Requires-Python: >=3.9
 Description-Content-Type: text/markdown
 License-File: LICENSE
+Requires-Dist: causaliq-workflow>=0.1.1.dev3
 Requires-Dist: click>=8.0.0
 Requires-Dist: httpx>=0.24.0
 Requires-Dist: pydantic>=2.0.0
@@ -32,7 +33,7 @@ Requires-Dist: causaliq-core>=0.3.0; extra == "dev"
 Requires-Dist: pytest>=7.0.0; extra == "dev"
 Requires-Dist: pytest-cov>=4.0.0; extra == "dev"
 Requires-Dist: pytest-mock>=3.10.0; extra == "dev"
-Requires-Dist: black>=22.0.0; extra == "dev"
+Requires-Dist: black<26.0.0,>=25.0.0; extra == "dev"
 Requires-Dist: isort>=5.10.0; extra == "dev"
 Requires-Dist: flake8>=5.0.0; extra == "dev"
 Requires-Dist: mypy>=1.0.0; extra == "dev"
@@ -42,6 +43,7 @@ Requires-Dist: build>=0.8.0; extra == "dev"
 Requires-Dist: twine>=4.0.0; extra == "dev"
 Provides-Extra: test
 Requires-Dist: causaliq-core>=0.3.0; extra == "test"
+Requires-Dist: causaliq-workflow>=0.1.1.dev3; extra == "test"
 Requires-Dist: pytest>=7.0.0; extra == "test"
 Requires-Dist: pytest-cov>=4.0.0; extra == "test"
 Requires-Dist: pytest-mock>=3.10.0; extra == "test"
@@ -89,13 +91,15 @@ Currently implemented releases:
 
 - **Release v0.1.0 - Foundation LLM**: Simple LLM queries to 1 or 2 LLMs about edge existence and orientation to support graph averaging
 - **Release v0.2.0 - Additional LLMs**: Support for 7 LLM providers (Groq, Gemini, OpenAI, Anthropic, DeepSeek, Mistral, Ollama)
-- **Release v0.3.0 - LLM Caching** *(in development)*: SQLite-based response caching with CLI tools for cache management
+- **Release v0.3.0 - LLM Caching**: SQLite-based response caching with CLI tools for cache management
+- **Release v0.4.0 - Graph Generation**: CLI and CausalIQ workflow action for LLM-generated causal graphs
 
 Planned:
 
-- **Release v0.4.0 - LLM Context**: Variable/role/literature etc context
-- **Release v0.5.0 - Algorithm integration**: Integration into structure learning algorithms
-- **Release v0.6.0 - Legacy Reference**: Support for legacy approaches of deriving knowledge from reference networks
+- **Release v0.5.0 - Graph Caching**: save generated graphs to Workflow caches
+- **Release v0.6.0 - LLM Cost Tracking**: Query LLM provider APIs for usage and cost statistics
+- **Release v0.7.0 - LLM Context**: Variable/role/literature etc context
+- **Release v0.8.0 - Algorithm integration**: Integration into structure learning algorithms
 
 ## Implementation Approach
 

causaliq_knowledge-0.4.0.dist-info/RECORD

@@ -0,0 +1,42 @@
+causaliq_knowledge/__init__.py,sha256=IVlm0G1g-xxJS13SFeC0h1D6LL7rfzX96F7rjfU-wqA,982
+causaliq_knowledge/action.py,sha256=X7EGSTV7IiwnO8cTcz5-ExXpRuwamSaC1jWpAc86i6I,16416
+causaliq_knowledge/base.py,sha256=GBG-sftOKkmUoQzTpm6anDTjP-2nInRZN_36dxoYhvk,2917
+causaliq_knowledge/models.py,sha256=tWGf186ASwO8NHiN97pEOLuBJmJI6Q9jvpU0mYZNdS0,4058
+causaliq_knowledge/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+causaliq_knowledge/cache/__init__.py,sha256=Av92YdCdVTRt9TmB2edRsIFDxq3f1Qi0daq0sFV1rp0,549
+causaliq_knowledge/cache/token_cache.py,sha256=o3qYGnc1e7nSJm9BsM6pmp6cbsPzhaCnEM1utCY682E,23085
+causaliq_knowledge/cache/encoders/__init__.py,sha256=gZ7gw96paFDbnJuc4v1aJsEJfVinI4zc03tXyFvfZxo,461
+causaliq_knowledge/cache/encoders/base.py,sha256=jK7--Or3lVp1UkKghKYFo_gKJp0HsMxosL_8eYL7RQQ,2679
+causaliq_knowledge/cache/encoders/json_encoder.py,sha256=7zN0qRdpVa8EZS08F22buyAYoIpnx5lngK5p1wK-3WI,15689
+causaliq_knowledge/cli/__init__.py,sha256=worFcurYE_T5_uYvnM4oL3aP3v_fSWLUhggbCP9TZMc,434
+causaliq_knowledge/cli/cache.py,sha256=w_fF3e8Ru7Sxl3FMgab_x2UWOfPMVq7BDvTUD-2Kayg,17863
+causaliq_knowledge/cli/generate.py,sha256=78c1GUv4w42qtPjy9NV0X1q5kw9ATl2yKcdl-KBTxhI,13187
+causaliq_knowledge/cli/main.py,sha256=MwUmokX9x3bL7E7pZMquWvVnKg3b-qdVk8VMR0ejK5o,4665
+causaliq_knowledge/cli/models.py,sha256=2ga5PWhOOo2vE6e3A3oxvO2FB88zztuRoUMPGlhyE6M,9587
+causaliq_knowledge/graph/__init__.py,sha256=920si3oBsuYIBW8gzHBYQnHCt9KupDdkPqVxTsj_py0,1952
+causaliq_knowledge/graph/generator.py,sha256=tM1KKKgpsiLLziCUKKnAiH9n1yO8zUnSFZ-QbFZKdJU,15971
+causaliq_knowledge/graph/loader.py,sha256=EO5Yj02qRrPY22rvfVk-LfXSZMVNEn37-H4u5kHCY0M,6615
+causaliq_knowledge/graph/models.py,sha256=4f9kaHHs9J_ma95EgV0GItliY-G4BLNNyIwBq8yTiVk,14924
+causaliq_knowledge/graph/params.py,sha256=RPviCO3ZOsOrm_rsysST4Y4hhWDN6jcJt46ajDvSY0M,5828
+causaliq_knowledge/graph/prompts.py,sha256=C29w5LQDf2tF9JeFADRrKSjkP6dVzjsa1FNX_6ndt70,15399
+causaliq_knowledge/graph/response.py,sha256=UaYbnVpfkWDZWMS9wQbEU4QP5In1YAqId2EuJ1V2kho,12437
+causaliq_knowledge/graph/view_filter.py,sha256=-ebhj8cXxgLimAeAZ023YgW6kI-c8jTp_LDKjYf1Kow,5297
+causaliq_knowledge/llm/__init__.py,sha256=30AL0h64zIkXoiqhMY7gjaf7mrtwtwMW38vzhns0My4,1663
+causaliq_knowledge/llm/anthropic_client.py,sha256=dPFHYGWL4xwQCtmQuGwGY4DBKSINOgOS-11ekznaiXo,8719
+causaliq_knowledge/llm/base_client.py,sha256=FJGX5QYawcelc3UScSMwvBJnKrUVR3PrBIY72KYthTU,12544
+causaliq_knowledge/llm/cache.py,sha256=6bpCyBv_bUorKceYc5qpgXi30A0tDRwAtlhxS3TQklE,15404
+causaliq_knowledge/llm/deepseek_client.py,sha256=ZcOpgnYa66XHjiTaF5ekR_BtosRYvVmzlIafp_Gsx_A,3543
+causaliq_knowledge/llm/gemini_client.py,sha256=XJMq9sPo7zExrALSr2rIRHLheSPqKo8ENG0KtdJ1cjw,9924
+causaliq_knowledge/llm/groq_client.py,sha256=PnTXqtMF1Km9DY4HiCZXQ6LeOzdjZtQJaeuGe1GbeME,7531
+causaliq_knowledge/llm/mistral_client.py,sha256=dTAOtymffCM1AJp5-JcfizofYrUA-jhKfHWrhZe2DDI,4187
+causaliq_knowledge/llm/ollama_client.py,sha256=PPU3g-nD8D546zcYB3uGxZ9yVbU4Gngo3snM2tRFeTc,8612
+causaliq_knowledge/llm/openai_client.py,sha256=MJmB6P32TZESMlXhn9d0-b3vFWXmf7ojHQ5CY8mCENI,3835
+causaliq_knowledge/llm/openai_compat_client.py,sha256=L8ZW5csuhUePq4mt3EGOUqhR3tleFmM72UlhPBsgIMQ,9518
+causaliq_knowledge/llm/prompts.py,sha256=bJ9iVGKUfTfLi2eWh-FFM4cNzk5Ux4Z0x8R6Ia27Dbo,6598
+causaliq_knowledge/llm/provider.py,sha256=VDEv-1esT_EgJk_Gwlfl4423ojglOxzPCBCFbOFE4DQ,15184
+causaliq_knowledge-0.4.0.dist-info/licenses/LICENSE,sha256=vUFUzQnti-D-MLSi9NxFlsFYOKwU25sxxH7WgJOQFIs,1084
+causaliq_knowledge-0.4.0.dist-info/METADATA,sha256=ZWEguAYGAWwk73VlIvb2KFXEyf37pTctjYZNlUgHWZM,9038
+causaliq_knowledge-0.4.0.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+causaliq_knowledge-0.4.0.dist-info/entry_points.txt,sha256=tuHaj0XNw9KySBpHOZeAC5Q08G96ftxocOy2POV1DdA,179
+causaliq_knowledge-0.4.0.dist-info/top_level.txt,sha256=GcxQf4BQAGa38i2-j8ylk2FmnBHtEZ9-8bSt-7Uka7k,19
+causaliq_knowledge-0.4.0.dist-info/RECORD,,

{causaliq_knowledge-0.3.0.dist-info → causaliq_knowledge-0.4.0.dist-info}/entry_points.txt

@@ -1,3 +1,6 @@
+[causaliq.actions]
+causaliq-knowledge = causaliq_knowledge:CausalIQAction
+
 [console_scripts]
 causaliq-knowledge = causaliq_knowledge.cli:main
 cqknow = causaliq_knowledge.cli:main