superlocalmemory 2.8.6 → 3.0.0

package/src/superlocalmemory/llm/backbone.py

@@ -0,0 +1,316 @@
+# Copyright (c) 2026 Varun Pratap Bhardwaj / Qualixar
+# Licensed under the MIT License - see LICENSE file
+# Part of SuperLocalMemory V3 | https://qualixar.com | https://varunpratap.com
+
+"""LLM backbone — unified interface for LLM providers.
+
+Supports OpenAI, Anthropic, Azure OpenAI, and Ollama via raw HTTP (httpx).
+Falls back gracefully when no API key is configured — Mode A still works.
+
+Providers:
+- ``"ollama"``: Local Ollama (OpenAI-compatible, no auth needed).
+- ``"openai"``: OpenAI API (GPT-4o, etc.).
+- ``"anthropic"``: Anthropic API (Claude, etc.).
+- ``"azure"``: Azure OpenAI (via AI Foundry deployment).
+
+Part of Qualixar | Author: Varun Pratap Bhardwaj
+License: MIT
+"""
+
+from __future__ import annotations
+
+import logging
+import os
+import socket
+import time
+from dataclasses import dataclass
+from typing import Any
+
+import httpx
+
+from superlocalmemory.core.config import LLMConfig
+
+logger = logging.getLogger(__name__)
+
+# ---------------------------------------------------------------------------
+# Constants
+# ---------------------------------------------------------------------------
+
+_OPENAI_URL = "https://api.openai.com/v1/chat/completions"
+_OPENROUTER_URL = "https://openrouter.ai/api/v1/chat/completions"
+_ANTHROPIC_URL = "https://api.anthropic.com/v1/messages"
+_ANTHROPIC_API_VERSION = "2023-06-01"
+_AZURE_API_VERSION = "2024-12-01-preview"
+_OLLAMA_DEFAULT_BASE = "http://localhost:11434"
+
+_ENV_KEYS: dict[str, str] = {
+    "openai": "OPENAI_API_KEY",
+    "anthropic": "ANTHROPIC_API_KEY",
+    "azure": "AZURE_OPENAI_API_KEY",
+    "ollama": "OLLAMA_HOST",
+    "openrouter": "OPENROUTER_API_KEY",
+}
+
+_SUPPORTED_PROVIDERS = frozenset({"openai", "anthropic", "azure", "ollama", "openrouter"})
+
+_MAX_RETRIES = 3
+_RETRY_BASE_DELAY = 1.0  # seconds, doubles each retry
+
+
+# ---------------------------------------------------------------------------
+# Exceptions
+# ---------------------------------------------------------------------------
+
+class LLMUnavailableError(Exception):
+    """Raised when no API key is available for the configured provider."""
+
+
+# ---------------------------------------------------------------------------
+# Response dataclass
+# ---------------------------------------------------------------------------
+
+@dataclass(frozen=True)
+class LLMResponse:
+    """Immutable container for a single LLM generation result."""
+
+    text: str
+    model: str
+    tokens_used: int
+    latency_ms: float
+
+
+# ---------------------------------------------------------------------------
+# Main class
+# ---------------------------------------------------------------------------
+
+class LLMBackbone:
+    """Unified LLM interface driven by LLMConfig.
+
+    All HTTP via httpx — no provider SDKs needed.
+    Includes retry logic (3 attempts, exponential backoff) and
+    socket-level timeout as a hard backstop for SSL hangs (S15 lesson).
+    """
+
+    def __init__(self, config: LLMConfig) -> None:
+        if config.provider and config.provider not in _SUPPORTED_PROVIDERS:
+            raise ValueError(
+                f"Unsupported provider '{config.provider}'. "
+                f"Choose from {sorted(_SUPPORTED_PROVIDERS)}."
+            )
+        self._provider = config.provider
+        self._model = config.model
+        self._timeout = config.timeout_seconds
+        self._default_temperature = config.temperature
+        self._default_max_tokens = config.max_tokens
+
+        # Resolve API key: config > environment variable.
+        if self._provider == "ollama":
+            self._api_key = ""
+            host = config.api_base or os.environ.get(
+                "OLLAMA_HOST", _OLLAMA_DEFAULT_BASE,
+            )
+            self._base_url = f"{host.rstrip('/')}/v1/chat/completions"
+        elif self._provider == "openrouter":
+            self._api_key = config.api_key or os.environ.get(
+                _ENV_KEYS.get(self._provider, ""), "",
+            )
+            self._base_url = config.api_base or _OPENROUTER_URL
+        elif self._provider:
+            self._api_key = config.api_key or os.environ.get(
+                _ENV_KEYS.get(self._provider, ""), "",
+            )
+            self._base_url = config.api_base
+        else:
+            self._api_key = ""
+            self._base_url = ""
+
+    # -- Properties ---------------------------------------------------------
+
+    def is_available(self) -> bool:
+        """True when the provider is ready for requests."""
+        if not self._provider:
+            return False
+        if self._provider == "ollama":
+            return True
+        return bool(self._api_key)
+
+    @property
+    def provider(self) -> str:
+        return self._provider
+
+    @property
+    def model(self) -> str:
+        return self._model
+
+    # -- Core generation ----------------------------------------------------
+
+    def generate(
+        self,
+        prompt: str,
+        system: str = "",
+        temperature: float | None = None,
+        max_tokens: int | None = None,
+    ) -> str:
+        """Send prompt to the LLM and return generated text.
+
+        Returns empty string on content-filter errors (Azure 400)
+        instead of crashing — lets callers continue gracefully.
+        Retries up to 3 times with exponential backoff on transient errors.
+        """
+        if not self.is_available():
+            raise LLMUnavailableError(
+                f"No API key for provider '{self._provider}'. "
+                f"Set {_ENV_KEYS.get(self._provider, 'API_KEY')} or pass api_key=."
+            )
+
+        temp = temperature if temperature is not None else self._default_temperature
+        tokens = max_tokens if max_tokens is not None else self._default_max_tokens
+        url, headers, payload = self._build_request(prompt, system, tokens, temp)
+
+        last_error: Exception | None = None
+        for attempt in range(_MAX_RETRIES):
+            try:
+                response = self._send(url, headers, payload)
+                return self._extract_text(response)
+            except httpx.HTTPStatusError as exc:
+                # Azure content filter returns 400 — not retryable.
+                if exc.response.status_code == 400:
+                    logger.warning("Content filter or bad request (400). Returning empty.")
+                    return ""
+                last_error = exc
+            except (httpx.TimeoutException, httpx.ConnectError) as exc:
+                last_error = exc
+
+            if attempt < _MAX_RETRIES - 1:
+                delay = _RETRY_BASE_DELAY * (2 ** attempt)
+                logger.info("Retry %d/%d after %.1fs", attempt + 1, _MAX_RETRIES, delay)
+                time.sleep(delay)
+
+        logger.error("All %d retries exhausted: %s", _MAX_RETRIES, last_error)
+        return ""
+
+    # -- HTTP transport -----------------------------------------------------
+
+    def _send(self, url: str, headers: dict, payload: dict) -> dict:
+        """Execute HTTP POST with socket-level SSL backstop."""
+        old_default = socket.getdefaulttimeout()
+        socket.setdefaulttimeout(self._timeout + 30)
+        try:
+            timeout = httpx.Timeout(
+                connect=10.0,
+                read=self._timeout,
+                write=10.0,
+                pool=10.0,
+            )
+            with httpx.Client(timeout=timeout) as client:
+                resp = client.post(url, headers=headers, json=payload)
+                resp.raise_for_status()
+                return resp.json()
+        finally:
+            socket.setdefaulttimeout(old_default)
+
+    # -- Request builders ---------------------------------------------------
+
+    def _build_request(
+        self, prompt: str, system: str, max_tokens: int, temperature: float,
+    ) -> tuple[str, dict[str, str], dict]:
+        """Build provider-specific (url, headers, payload)."""
+        builders = {
+            "ollama": self._build_ollama,
+            "anthropic": self._build_anthropic,
+            "azure": self._build_azure,
+        }
+        builder = builders.get(self._provider, self._build_openai)
+        return builder(prompt, system, max_tokens, temperature)
+
+    def _build_openai(
+        self, prompt: str, system: str, max_tokens: int, temperature: float,
+    ) -> tuple[str, dict[str, str], dict]:
+        messages = self._make_messages(system, prompt)
+        headers = {
+            "Authorization": f"Bearer {self._api_key}",
+            "Content-Type": "application/json",
+        }
+        payload = {
+            "model": self._model,
+            "messages": messages,
+            "max_tokens": max_tokens,
+            "temperature": temperature,
+        }
+        url = self._base_url or _OPENAI_URL
+        return url, headers, payload
+
+    def _build_ollama(
+        self, prompt: str, system: str, max_tokens: int, temperature: float,
+    ) -> tuple[str, dict[str, str], dict]:
+        messages = self._make_messages(system, prompt)
+        headers = {"Content-Type": "application/json"}
+        payload = {
+            "model": self._model,
+            "messages": messages,
+            "max_tokens": max_tokens,
+            "temperature": temperature,
+        }
+        return self._base_url, headers, payload
+
+    def _build_anthropic(
+        self, prompt: str, system: str, max_tokens: int, temperature: float,
+    ) -> tuple[str, dict[str, str], dict]:
+        headers = {
+            "x-api-key": self._api_key,
+            "anthropic-version": _ANTHROPIC_API_VERSION,
+            "Content-Type": "application/json",
+        }
+        payload: dict[str, Any] = {
+            "model": self._model,
+            "max_tokens": max_tokens,
+            "temperature": temperature,
+            "messages": [{"role": "user", "content": prompt}],
+        }
+        if system:
+            payload["system"] = system
+        return _ANTHROPIC_URL, headers, payload
+
+    def _build_azure(
+        self, prompt: str, system: str, max_tokens: int, temperature: float,
+    ) -> tuple[str, dict[str, str], dict]:
+        if not self._base_url:
+            raise ValueError("Azure provider requires api_base URL.")
+        url = (
+            f"{self._base_url.rstrip('/')}/openai/deployments/"
+            f"{self._model}/chat/completions"
+            f"?api-version={_AZURE_API_VERSION}"
+        )
+        messages = self._make_messages(system, prompt)
+        headers = {"api-key": self._api_key, "Content-Type": "application/json"}
+        payload: dict[str, Any] = {"messages": messages}
+        if "gpt-5" in self._model.lower():
+            payload["max_completion_tokens"] = max(max_tokens, 200)
+            payload["reasoning_effort"] = "none"
+            if temperature > 0:
+                payload["temperature"] = temperature
+        else:
+            payload["max_tokens"] = max_tokens
+            payload["temperature"] = temperature
+        return url, headers, payload
+
+    # -- Response parsing ---------------------------------------------------
+
+    def _extract_text(self, data: dict) -> str:
+        """Extract text from provider-specific JSON response."""
+        if self._provider == "anthropic":
+            return data.get("content", [{}])[0].get("text", "").strip()
+        # OpenAI / Azure / Ollama share response format.
+        choices = data.get("choices", [{}])
+        return choices[0].get("message", {}).get("content", "").strip()
+
+    # -- Helpers ------------------------------------------------------------
+
+    @staticmethod
+    def _make_messages(system: str, prompt: str) -> list[dict[str, str]]:
+        """Build messages array with optional system message."""
+        messages: list[dict[str, str]] = []
+        if system:
+            messages.append({"role": "system", "content": system})
+        messages.append({"role": "user", "content": prompt})
+        return messages

package/src/superlocalmemory/math/fisher.py

@@ -0,0 +1,356 @@
+# Copyright (c) 2026 Varun Pratap Bhardwaj / Qualixar
+# Licensed under the MIT License - see LICENSE file
+# Part of SuperLocalMemory V3 | https://qualixar.com | https://varunpratap.com
+
+"""Fisher-Rao geodesic metric with Bayesian variance update.
+
+Information-geometric similarity on the statistical manifold of diagonal
+Gaussians.  Two embeddings with the same mean but different uncertainty
+have zero cosine distance but nonzero Fisher distance --- this is the
+core insight that differentiates Fisher-Rao from cosine similarity.
+
+Geodesic distance (Atkinson & Mitchell 1981, Pinele et al. 2020):
+
+    Univariate component i:
+        delta_i = ((mu1_i - mu2_i)^2 + 2*(sigma1_i - sigma2_i)^2)
+                  / (4 * sigma1_i * sigma2_i)
+        d_i     = sqrt(2) * arccosh(1 + delta_i)
+
+    Diagonal multivariate (product-manifold decomposition):
+        d_FR(p, q) = sqrt( sum_i  d_i^2 )
+
+Bayesian variance update (NEW in Innovation Wave 4):
+
+    V1 bug: query always received UNIFORM variance, so Fisher degenerated
+    to a monotonic transform of cosine.  FIX: every fact maintains its own
+    variance vector that narrows on each observation.
+
+        new_var_i = 1 / (1/old_var_i + 1/obs_var_i)
+
+    Multiple consistent observations reduce variance, increasing confidence.
+    This gives Fisher-Rao a genuine ranking advantage over cosine: well-
+    confirmed memories have tighter distributions and score higher when the
+    query matches.
+
+References:
+    Rao C R (1945). Information and the accuracy attainable in the
+        estimation of statistical parameters. Bull. Calcutta Math. Soc.
+    Atkinson C & Mitchell A (1981). Rao's distance measure.
+        Sankhya: The Indian Journal of Statistics, Series A.
+    Pinele J, Strapasson J E & Costa S I R (2020). The Fisher-Rao
+        distance between multivariate normal distributions: special
+        cases, bounds and applications. Entropy 22(4):404.
+
+Part of Qualixar | Author: Varun Pratap Bhardwaj
+License: MIT
+"""
+
+from __future__ import annotations
+
+import math
+
+import numpy as np
+
+# ---------------------------------------------------------------------------
+# Constants
+# ---------------------------------------------------------------------------
+
+_VARIANCE_FLOOR: float = 0.05  # Minimum per-dimension variance (40x dynamic range)
+_VARIANCE_CEIL: float = 2.0    # Maximum per-dimension variance (initial uncertainty)
+_DEFAULT_TEMPERATURE: float = 15.0
+_SMALL_DELTA_THRESHOLD: float = 1e-7
+
+
+# ---------------------------------------------------------------------------
+# Public API
+# ---------------------------------------------------------------------------
+
+class FisherRaoMetric:
+    """Fisher-Rao geodesic metric with Bayesian variance tracking.
+
+    Each fact is modelled as a diagonal Gaussian N(mu, diag(sigma^2)).
+    Distance is the Fisher-Rao geodesic on this statistical manifold.
+    Variance starts high (uncertain) and NARROWS on repeated access via
+    Bayesian update, giving well-confirmed memories a ranking advantage.
+
+    Attributes:
+        temperature: Softmax scaling for similarity conversion.
+    """
+
+    __slots__ = ("temperature",)
+
+    def __init__(self, temperature: float = _DEFAULT_TEMPERATURE) -> None:
+        if temperature <= 0:
+            raise ValueError(f"temperature must be positive, got {temperature}")
+        self.temperature = temperature
+
+    # ------------------------------------------------------------------
+    # Parameter derivation
+    # ------------------------------------------------------------------
+
+    def compute_params(
+        self,
+        embedding: list[float],
+    ) -> tuple[list[float], list[float]]:
+        """Derive Fisher parameters (mean, variance) from a raw embedding.
+
+        Mean is the L2-normalised embedding.  Variance is content-derived
+        and *heterogeneous*: dimensions with strong signal get low variance,
+        weak-signal dimensions get high variance.
+
+        Mapping: ``var_i = CEIL - (CEIL - FLOOR) * |normed_i| / max_abs``
+        Strong signal (large |normed_i|) -> low variance (high confidence).
+
+        Args:
+            embedding: Raw embedding vector, any dimensionality.
+
+        Returns:
+            (mean, variance) as plain Python lists, same length as input.
+        """
+        arr = np.asarray(embedding, dtype=np.float64)
+        norm = np.linalg.norm(arr)
+        if norm < 1e-12:
+            mean = np.zeros_like(arr)
+            variance = np.full_like(arr, _VARIANCE_CEIL)
+            return mean.tolist(), variance.tolist()
+
+        mean = arr / norm
+
+        abs_vals = np.abs(mean)
+        max_abs = float(np.max(abs_vals))
+        if max_abs < 1e-12:
+            max_abs = 1.0
+
+        normalised_signal = abs_vals / max_abs  # in [0, 1]
+        variance = _VARIANCE_CEIL - (_VARIANCE_CEIL - _VARIANCE_FLOOR) * normalised_signal
+        variance = np.clip(variance, _VARIANCE_FLOOR, _VARIANCE_CEIL)
+
+        return mean.tolist(), variance.tolist()
+
+    # ------------------------------------------------------------------
+    # Geodesic distance
+    # ------------------------------------------------------------------
+
+    def distance(
+        self,
+        mean_a: list[float],
+        var_a: list[float],
+        mean_b: list[float],
+        var_b: list[float],
+    ) -> float:
+        """Exact Fisher-Rao geodesic distance between diagonal Gaussians.
+
+        Uses Atkinson & Mitchell (1981) per-component closed form with
+        Pinele et al. (2020) product-manifold decomposition.
+
+        Args:
+            mean_a:  Mean of distribution A.
+            var_a:   Per-dimension variance of A (strictly positive).
+            mean_b:  Mean of distribution B.
+            var_b:   Per-dimension variance of B (strictly positive).
+
+        Returns:
+            Non-negative geodesic distance.
+
+        Raises:
+            ValueError: On NaN, non-positive variance, or length mismatch.
+        """
+        mu1 = np.asarray(mean_a, dtype=np.float64)
+        sig1 = np.asarray(var_a, dtype=np.float64)
+        mu2 = np.asarray(mean_b, dtype=np.float64)
+        sig2 = np.asarray(var_b, dtype=np.float64)
+
+        _validate(mu1, sig1, mu2, sig2)
+
+        # Per-component: delta_i = ((mu1-mu2)^2 + 2*(sigma1-sigma2)^2) / (4*s1*s2)
+        mu_diff_sq = (mu1 - mu2) ** 2
+        sig_diff_sq = (sig1 - sig2) ** 2
+        product = sig1 * sig2
+
+        delta = (mu_diff_sq + 2.0 * sig_diff_sq) / (4.0 * product)
+
+        # Per-component squared distance: 2 * arccosh(1 + delta_i)^2
+        acosh_vals = _stable_arccosh_1p_vec(delta)
+        total_dist_sq = float(np.sum(2.0 * acosh_vals ** 2))
+
+        return math.sqrt(max(total_dist_sq, 0.0))
+
+    # ------------------------------------------------------------------
+    # Similarity (exponential kernel)
+    # ------------------------------------------------------------------
+
+    def similarity(
+        self,
+        mean_a: list[float],
+        var_a: list[float],
+        mean_b: list[float],
+        var_b: list[float],
+    ) -> float:
+        """Convert Fisher-Rao distance to similarity in [0, 1].
+
+        sim = exp(-distance / temperature)
+
+        Args:
+            mean_a:  Mean of distribution A.
+            var_a:   Per-dimension variance of A.
+            mean_b:  Mean of distribution B.
+            var_b:   Per-dimension variance of B.
+
+        Returns:
+            Similarity in [0, 1].  1 = identical, 0 = maximally different.
+        """
+        d = self.distance(mean_a, var_a, mean_b, var_b)
+        return float(np.exp(-d / self.temperature))
+
+    # ------------------------------------------------------------------
+    # Bayesian variance update (THE key V1 fix)
+    # ------------------------------------------------------------------
+
+    def bayesian_update(
+        self,
+        old_var: list[float],
+        observation_var: list[float],
+    ) -> list[float]:
+        """Bayesian precision-additive variance update.
+
+        Each observation tightens the posterior variance:
+
+            1/new_var_i = 1/old_var_i + 1/obs_var_i
+
+        This is the standard conjugate update for a Gaussian likelihood
+        with known mean and unknown precision.  After *k* observations
+        with identical variance sigma^2:
+
+            var_k = sigma^2 / k
+
+        So variance shrinks as 1/k --- giving well-confirmed memories
+        significantly tighter distributions and higher Fisher similarity
+        to matching queries.
+
+        Args:
+            old_var:         Current per-dimension variance.
+            observation_var: Variance of the new observation (derived from
+                             the embedding of the confirming content).
+
+        Returns:
+            Updated variance (strictly within [FLOOR, CEIL]).
+        """
+        old = np.asarray(old_var, dtype=np.float64)
+        obs = np.asarray(observation_var, dtype=np.float64)
+
+        if old.shape != obs.shape:
+            raise ValueError(
+                f"Variance shape mismatch: {old.shape} vs {obs.shape}"
+            )
+
+        # Clamp inputs to valid range before update
+        old = np.clip(old, _VARIANCE_FLOOR, _VARIANCE_CEIL)
+        obs = np.clip(obs, _VARIANCE_FLOOR, _VARIANCE_CEIL)
+
+        # Precision-additive update: 1/new = 1/old + 1/obs
+        new_precision = (1.0 / old) + (1.0 / obs)
+        new_var = 1.0 / new_precision
+
+        new_var = np.clip(new_var, _VARIANCE_FLOOR, _VARIANCE_CEIL)
+        return new_var.tolist()
+
+    # ------------------------------------------------------------------
+    # Adaptive temperature
+    # ------------------------------------------------------------------
+
+    def adaptive_temperature(
+        self,
+        variances: list[list[float]],
+    ) -> float:
+        """Compute data-driven temperature from corpus variance statistics.
+
+        Instead of a fixed temperature, adapt to the actual spread of
+        variances in the memory store.  High average variance (uncertain
+        corpus) -> higher temperature (softer discrimination).  Low average
+        variance (well-confirmed corpus) -> lower temperature (sharper).
+
+        Formula:
+            T = base_T * (1 + avg_variance) / 2
+
+        This ensures T stays close to base_T when avg_variance ~ 1.0
+        (the typical midpoint) and scales proportionally.
+
+        Args:
+            variances: List of per-fact variance vectors.
+
+        Returns:
+            Adapted temperature (always positive).
+        """
+        if not variances:
+            return self.temperature
+
+        all_vars = np.array(variances, dtype=np.float64)
+        avg = float(np.mean(all_vars))
+
+        adapted = self.temperature * (1.0 + avg) / 2.0
+        return max(adapted, 0.1)  # floor to prevent division issues
+
+
+# ---------------------------------------------------------------------------
+# Validation
+# ---------------------------------------------------------------------------
+
+def _validate(
+    mu1: np.ndarray,
+    sig1: np.ndarray,
+    mu2: np.ndarray,
+    sig2: np.ndarray,
+) -> None:
+    """Validate Fisher metric inputs."""
+    for name, arr in [("mu1", mu1), ("sig1", sig1), ("mu2", mu2), ("sig2", sig2)]:
+        if np.any(np.isnan(arr)):
+            raise ValueError(f"{name} contains NaN")
+
+    if mu1.shape != mu2.shape:
+        raise ValueError(f"Mean shape mismatch: {mu1.shape} vs {mu2.shape}")
+    if sig1.shape != sig2.shape:
+        raise ValueError(f"Sigma shape mismatch: {sig1.shape} vs {sig2.shape}")
+    if mu1.shape[0] != sig1.shape[0]:
+        raise ValueError(
+            f"Mean/sigma length mismatch: {mu1.shape[0]} vs {sig1.shape[0]}"
+        )
+
+    if np.any(sig1 <= 0):
+        raise ValueError("sig1 must be strictly positive")
+    if np.any(sig2 <= 0):
+        raise ValueError("sig2 must be strictly positive")
+
+
+# ---------------------------------------------------------------------------
+# Numerically stable arccosh
+# ---------------------------------------------------------------------------
+
+def _stable_arccosh_1p_vec(delta: np.ndarray) -> np.ndarray:
+    """Vectorised stable arccosh(1 + delta).
+
+    For small delta, uses Taylor expansion to avoid catastrophic
+    cancellation: arccosh(1+d) ~ sqrt(2d) * (1 - d/12).
+    For larger delta, uses the identity:
+        arccosh(1+d) = log(1 + d + sqrt(d*(d+2))).
+
+    Args:
+        delta: Non-negative array.
+
+    Returns:
+        arccosh(1 + delta) element-wise.
+    """
+    delta = np.maximum(delta, 0.0)
+    result = np.empty_like(delta)
+
+    small = delta < _SMALL_DELTA_THRESHOLD
+    large = ~small
+
+    if np.any(small):
+        d_s = delta[small]
+        result[small] = np.sqrt(2.0 * d_s) * (1.0 - d_s / 12.0)
+
+    if np.any(large):
+        d_l = delta[large]
+        result[large] = np.log1p(d_l + np.sqrt(d_l * (d_l + 2.0)))
+
+    return result