pryscraper/llm_providers/base.py
cryptorugmunch bb77eb5f35 chore(license): re-license to dual MIT (core) + BSL 1.1 (stealth)
Re-license Pry from full Proprietary to a dual-license model:

- Core engine, extraction, templates (80+), MCP server, x402 payment rail,
  CLI, SDK, browser extension, WordPress plugin, Shopify app, and
  llm_providers: MIT (see LICENSE)
- Anti-detection / stealth subset (15 files): BSL 1.1 with Change Date
  2029-01-01 (see LICENSE-BSL-STEALTH)

BSL files (anti-detection moat):
  ultimate_scraper.py, stealth_engine.py, stealth_scripts/*.js (6),
  camoufox_integration.py, tls_fingerprint.py, cookie_warmer.py,
  behavioral_biometrics.py, adaptive.py, browser_pool.py, network.py,
  captcha_solver.py, shadow_dom.py, lazy_load.py, signup_automator.py,
  auth_connector.py

This enables community contributions to the core engine (templates,
integrations, MCP tools) while protecting the anti-detection techniques
that constitute the actual competitive moat. BSL Additional Use Grant
permits free non-production use; production deployment requires a
commercial license from enterprise@rugmunch.io.

Changes:
- Replace proprietary LICENSE with MIT LICENSE + new LICENSE-BSL-STEALTH
- Add SPDX-License-Identifier headers to 300+ source files
- Add docs/adr/0002-dual-licensing.md (ADR documenting the decision)
- Update README.md: new License section with BSL Additional Use Grant
- Update LICENSING_PRICING_STRATEGY.md: Section 3 (PryScraper) for dual license
- Update AGENTS.md: license line in header + new rule 8 (PRs touching BSL rejected)
- Update pyproject.toml: license = "MIT AND BSL-1.1" + classifiers + license-files
- Update DECISIONS.md index with ADR-0002
- Update STATUS.md (2026-07-03) and PLAN.md sprint goals

Refs: ADR-0002
2026-07-02 19:49:21 +02:00

72 lines
2.3 KiB
Python

"""Pry — LLM Provider abstraction with referral revenue tracking.
Supports pluggable providers: OpenAI, Anthropic, Google, Cohere, Mistral, Ollama, OpenRouter.
Includes referral/affiliate link tracking for revenue sharing."""
# SPDX-License-Identifier: MIT
# Copyright (c) 2026 Rug Munch Media LLC
#
# Part of Pry — https://git.rugmunch.io/RugMunchMedia/pryscraper
# Licensed under MIT. See LICENSE.
import logging
from abc import ABC, abstractmethod
from dataclasses import dataclass, field
from typing import Any
logger = logging.getLogger(__name__)
@dataclass
class LLMResponse:
"""Standard response from any LLM provider."""
text: str
model: str
provider: str
input_tokens: int = 0
output_tokens: int = 0
cost_usd: float = 0.0
referral_id: str = ""
latency_ms: int = 0
raw: dict[str, Any] = field(default_factory=dict)
@dataclass
class ReferralConfig:
"""Referral/affiliate config for revenue sharing."""
enabled: bool = True
program_id: str = "pry-default"
# Provider-specific referral links (with our affiliate ID)
referral_links: dict[str, str] = field(default_factory=dict)
# NEW: link to the full provider catalog
catalog: dict = field(default_factory=dict)
def __post_init__(self):
if not self.referral_links:
from referrals import PROVIDER_CATALOG
for _category, providers in PROVIDER_CATALOG.items():
for p in providers:
self.referral_links[p["tag"]] = p["url"]
self.catalog = PROVIDER_CATALOG
class LLMProvider(ABC):
"""Abstract base class for LLM providers."""
name: str = ""
cost_per_1k_input: float = 0.0
cost_per_1k_output: float = 0.0
referral_url: str = ""
@abstractmethod
async def complete(self, prompt: str, system: str = "", max_tokens: int = 1024,
temperature: float = 0.7, model: str = "") -> LLMResponse:
"""Send completion request to provider."""
raise NotImplementedError
@abstractmethod
async def embed(self, text: str, model: str = "") -> list[float]:
"""Generate embedding for text."""
raise NotImplementedError
def estimate_cost(self, input_tokens: int, output_tokens: int) -> float:
return (input_tokens / 1000) * self.cost_per_1k_input + (output_tokens / 1000) * self.cost_per_1k_output