@datalayer/agent-runtimes 1.3.15 → 1.3.17

package/scripts/codegen/generate_model_providers.py

@@ -0,0 +1,246 @@
+#!/usr/bin/env python3
+# Copyright (c) 2025-2026 Datalayer, Inc.
+# Distributed under the terms of the Modified BSD License.
+
+"""
+Generate Python and TypeScript code from YAML model-provider specifications.
+
+A model provider is who serves a model (Anthropic, Amazon Bedrock, Cloudflare,
+Ollama…): its site, its documentation, its terms of service, its privacy
+policy, what it says about the data a request carries, and whether it runs
+the model or the user's machine does. One YAML per provider under
+``agentspecs/model-providers``; a model spec's ``provider`` names one.
+
+Usage:
+    python generate_model_providers.py \\
+      --specs-dir agentspecs/agentspecs/model-providers \\
+      --python-output agent_runtimes/specs/model_providers.py \\
+      --typescript-output src/specs/modelProviders.ts
+"""
+
+import argparse
+import sys
+from pathlib import Path
+from typing import Any
+
+import yaml
+from versioning import ensure_spec_version, version_suffix
+
+FIELDS = ("website", "docs_url", "terms_url", "privacy_url", "data_usage_url")
+CAMEL = {
+    "website": "website",
+    "docs_url": "docsUrl",
+    "terms_url": "termsUrl",
+    "privacy_url": "privacyUrl",
+    "data_usage_url": "dataUsageUrl",
+}
+
+
+def load_provider_specs(specs_dir: Path) -> list[dict[str, Any]]:
+    """Load all model-provider YAML specifications from a directory."""
+    specs = []
+    for yaml_file in sorted(specs_dir.glob("*.yaml")):
+        with open(yaml_file) as f:
+            spec = yaml.safe_load(f)
+            ensure_spec_version(spec)
+            if spec["id"] != yaml_file.stem:
+                raise ValueError(
+                    f"{yaml_file.name}: the file is named for id {yaml_file.stem!r}, the spec says {spec['id']!r}"
+                )
+            specs.append(spec)
+    return specs
+
+
+def _const_name(provider_id: str) -> str:
+    return provider_id.replace("-", "_").replace(".", "_").upper()
+
+
+def _py(value: str) -> str:
+    return '"' + value.replace("\\", "\\\\").replace('"', '\\"') + '"'
+
+
+def _ts(value: str) -> str:
+    return "'" + value.replace("\\", "\\\\").replace("'", "\\'") + "'"
+
+
+def generate_python_code(specs: list[dict[str, Any]]) -> str:
+    """Generate Python code from model-provider specifications."""
+    lines = [
+        "# Copyright (c) 2025-2026 Datalayer, Inc.",
+        "# Distributed under the terms of the Modified BSD License.",
+        '"""',
+        "Model Provider Catalog.",
+        "",
+        "Who serves a model: the vendor's own API, a cloud that hosts it, or the",
+        "user's machine — with its site, documentation, terms, privacy policy and",
+        "what it says about the data a request carries.",
+        "",
+        "This file is AUTO-GENERATED from YAML specifications.",
+        "DO NOT EDIT MANUALLY - run 'make specs' to regenerate.",
+        '"""',
+        "",
+        "from typing import Dict",
+        "",
+        "from agent_runtimes.types import ModelProvider",
+        "",
+        "",
+        "# " + "=" * 76,
+        "# Model Provider Definitions",
+        "# " + "=" * 76,
+        "",
+    ]
+    for spec in specs:
+        const = _const_name(spec["id"]) + version_suffix(spec["version"])
+        lines.extend(
+            [
+                f"{const} = ModelProvider(",
+                f"    id={_py(spec['id'])},",
+                f"    version={_py(str(spec['version']))},",
+                f"    name={_py(spec['name'])},",
+                f"    description={_py(str(spec.get('description', '')).strip())},",
+            ]
+        )
+        for field in FIELDS:
+            value = spec.get(field)
+            if value:
+                lines.append(f"    {field}={_py(str(value))},")
+        lines.append(f"    hosting={_py(spec.get('hosting', 'cloud'))},")
+        lines.extend([")", ""])
+    lines.extend(
+        [
+            "",
+            "# " + "=" * 76,
+            "# Model Provider Catalog",
+            "# " + "=" * 76,
+            "",
+            "MODEL_PROVIDER_CATALOGUE: Dict[str, ModelProvider] = {",
+        ]
+    )
+    for spec in specs:
+        lines.append(
+            f"    {_py(spec['id'])}: {_const_name(spec['id']) + version_suffix(spec['version'])},"
+        )
+    lines.extend(
+        [
+            "}",
+            "",
+            "",
+            "def get_model_provider(provider_id: str) -> ModelProvider | None:",
+            '    """The provider a model spec\'s `provider` names, or None."""',
+            "    return MODEL_PROVIDER_CATALOGUE.get(provider_id)",
+            "",
+            "",
+            "def list_model_providers() -> list[ModelProvider]:",
+            "    return list(MODEL_PROVIDER_CATALOGUE.values())",
+            "",
+        ]
+    )
+    return "\n".join(lines)
+
+
+def generate_typescript_code(specs: list[dict[str, Any]]) -> str:
+    """Generate TypeScript code from model-provider specifications."""
+    lines = [
+        "/*",
+        " * Copyright (c) 2025-2026 Datalayer, Inc.",
+        " * Distributed under the terms of the Modified BSD License.",
+        " */",
+        "",
+        "/**",
+        " * Model Provider Catalog.",
+        " *",
+        " * Who serves a model: the vendor's own API, a cloud that hosts it, or the",
+        " * user's machine — with its site, documentation, terms, privacy policy and",
+        " * what it says about the data a request carries.",
+        " *",
+        " * This file is AUTO-GENERATED from YAML specifications.",
+        " * DO NOT EDIT MANUALLY - run 'make specs' to regenerate.",
+        " */",
+        "",
+        "import type { ModelProvider } from '../types/models';",
+        "",
+    ]
+    for spec in specs:
+        const = _const_name(spec["id"]) + version_suffix(spec["version"])
+        lines.extend(
+            [
+                f"export const {const}: ModelProvider = {{",
+                f"  id: {_ts(spec['id'])},",
+                f"  version: {_ts(str(spec['version']))},",
+                f"  name: {_ts(spec['name'])},",
+                f"  description: {_ts(str(spec.get('description', '')).strip())},",
+            ]
+        )
+        for field in FIELDS:
+            value = spec.get(field)
+            if value:
+                lines.append(f"  {CAMEL[field]}: {_ts(str(value))},")
+        lines.append(f"  hosting: {_ts(spec.get('hosting', 'cloud'))},")
+        lines.extend(["};", ""])
+    lines.append(
+        "export const MODEL_PROVIDER_CATALOGUE: Record<string, ModelProvider> = {"
+    )
+    for spec in specs:
+        lines.append(
+            f"  {_ts(spec['id'])}: {_const_name(spec['id']) + version_suffix(spec['version'])},"
+        )
+    lines.extend(
+        [
+            "};",
+            "",
+            "/** The provider a model spec's `provider` names, or undefined. */",
+            "export function getModelProvider(providerId: string): ModelProvider | undefined {",
+            "  return MODEL_PROVIDER_CATALOGUE[providerId];",
+            "}",
+            "",
+            "export function listModelProviders(): ModelProvider[] {",
+            "  return Object.values(MODEL_PROVIDER_CATALOGUE);",
+            "}",
+            "",
+        ]
+    )
+    return "\n".join(lines)
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Generate Python and TypeScript code from YAML model-provider specifications"
+    )
+    parser.add_argument(
+        "--specs-dir",
+        type=Path,
+        required=True,
+        help="Directory containing model-provider YAML files",
+    )
+    parser.add_argument(
+        "--python-output", type=Path, required=True, help="Output path for Python file"
+    )
+    parser.add_argument(
+        "--typescript-output",
+        type=Path,
+        required=True,
+        help="Output path for TypeScript file",
+    )
+    args = parser.parse_args()
+
+    if not args.specs_dir.exists():
+        print(f"Error: Specs directory not found: {args.specs_dir}", file=sys.stderr)
+        sys.exit(1)
+    specs = load_provider_specs(args.specs_dir)
+    if not specs:
+        print(
+            f"Warning: No model-provider specifications found in {args.specs_dir}",
+            file=sys.stderr,
+        )
+        return
+    args.python_output.parent.mkdir(parents=True, exist_ok=True)
+    args.python_output.write_text(generate_python_code(specs))
+    print(f"Generated Python code: {args.python_output}")
+    args.typescript_output.parent.mkdir(parents=True, exist_ok=True)
+    args.typescript_output.write_text(generate_typescript_code(specs))
+    print(f"Generated TypeScript code: {args.typescript_output}")
+    print(f"\n✓ Successfully generated code from {len(specs)} model-provider specs")
+
+
+if __name__ == "__main__":
+    main()

package/scripts/codegen/generate_models.py

@@ -38,7 +38,14 @@ def _make_const_name(model_id: str) -> str:
 
     E.g. 'anthropic:claude-sonnet-4-5-20250514' -> 'ANTHROPIC_CLAUDE_SONNET_4_5_20250514'
     """
-    return model_id.upper().replace(":", "_").replace("-", "_").replace(".", "_")
+    # A Cloudflare id carries a vendor segment: "cloudflare:openai/gpt-oss-120b".
+    return (
+        model_id.upper()
+        .replace(":", "_")
+        .replace("-", "_")
+        .replace(".", "_")
+        .replace("/", "_")
+    )
 
 
 def _pick_default(specs: list[dict]) -> Optional[dict]:
@@ -89,7 +96,7 @@ def generate_python_code(specs: list[dict[str, Any]]) -> str:
         "from enum import Enum",
         "from typing import Dict, List, Optional",
         "",
-        "from agent_runtimes.types import AIModel",
+        "from agent_runtimes.types import AIModel, ModelPricing",
         "",
         "",
         "# " + "=" * 76,
@@ -102,10 +109,18 @@ def generate_python_code(specs: list[dict[str, Any]]) -> str:
         "",
     ]
 
-    # Generate enum members
+    # Generate enum members; an older id keeps its member, with its older value.
+    seen = set()
     for spec in specs:
         enum_name = _make_enum_name(spec["id"])
         lines.append(f'    {enum_name} = "{spec["id"]}"')
+        seen.add(enum_name)
+    for spec in specs:
+        for alias in spec.get("aliases") or []:
+            alias_name = _make_enum_name(alias)
+            if alias_name not in seen:
+                lines.append(f'    {alias_name} = "{alias}"  # before the id moved')
+                seen.add(alias_name)
 
     lines.extend(
         [
@@ -141,6 +156,11 @@ def generate_python_code(specs: list[dict[str, Any]]) -> str:
             f'    name="{spec["name"]}",',
             f'    description="{spec.get("description", "")}",',
             f'    provider="{spec["provider"]}",',
+            *(
+                [f'    provider_url="{spec["provider_url"]}",']
+                if spec.get("provider_url")
+                else []
+            ),
             f"    default={spec.get('default', False)},",
             # What is worth offering today, as distinct from what the platform
             # knows how to talk to.
@@ -160,6 +180,30 @@ def generate_python_code(specs: list[dict[str, Any]]) -> str:
         if capabilities:
             formatted = "[" + ", ".join(f'"{c}"' for c in capabilities) + "]"
             model_lines.append(f"    capabilities={formatted},")
+        if spec.get("billing"):
+            model_lines.append(f'    billing="{spec["billing"]}",')
+        if spec.get("route"):
+            model_lines.append(f'    route="{spec["route"]}",')
+        if spec.get("context_window") is not None:
+            model_lines.append(f"    context_window={int(spec['context_window'])},")
+        if spec.get("zero_data_retention") is not None:
+            model_lines.append(
+                f"    zero_data_retention={bool(spec['zero_data_retention'])},"
+            )
+        if spec.get("request_logging"):
+            model_lines.append(f'    request_logging="{spec["request_logging"]}",')
+        aliases = spec.get("aliases") or []
+        if aliases:
+            model_lines.append(
+                "    aliases=[" + ", ".join(f'"{a}"' for a in aliases) + "],"
+            )
+        pricing = spec.get("pricing") or {}
+        if pricing:
+            model_lines.append(
+                "    pricing=ModelPricing("
+                f"input_usd_per_million={float(pricing['input_usd_per_million'])}, "
+                f"output_usd_per_million={float(pricing['output_usd_per_million'])}),"
+            )
 
         model_lines.extend([")", ""])
         lines.extend(model_lines)
@@ -180,6 +224,18 @@ def generate_python_code(specs: list[dict[str, Any]]) -> str:
         const_name = _make_const_name(model_id) + version_suffix(spec["version"])
         lines.append(f'    "{model_id}": {const_name},')
 
+    lines.extend(
+        [
+            "}",
+            "",
+            "# The ids a model answered to before its id moved, to the id it has now.",
+            "# Kept apart from the catalogue, which lists each model once.",
+            "AI_MODEL_ALIASES: Dict[str, str] = {",
+        ]
+    )
+    for spec in specs:
+        for alias in spec.get("aliases") or []:
+            lines.append(f'    "{alias}": "{spec["id"]}",')
     lines.extend(
         [
             "}",
@@ -237,7 +293,7 @@ def generate_python_code(specs: list[dict[str, Any]]) -> str:
             "    Returns:",
             "        The AIModel specification, or None if not found.",
             '    """',
-            "    model = AI_MODEL_CATALOGUE.get(model_id)",
+            "    model = AI_MODEL_CATALOGUE.get(AI_MODEL_ALIASES.get(model_id, model_id))",
             "    if model is not None:",
             "        return model",
             "    base, _, ver = model_id.rpartition(':')",
@@ -269,6 +325,19 @@ def generate_python_code(specs: list[dict[str, Any]]) -> str:
             '    """',
             "    return list(AI_MODEL_CATALOGUE.values())",
             "",
+            "",
+            "def is_chat_model(model: AIModel) -> bool:",
+            '    """A model a chat can run on: not a typed-judgment model (Jev), which',
+            "    answers typed questions about a state and nothing else. A spec that",
+            "    states no capability is read as a chat model.",
+            '    """',
+            '    return "judgments" not in model.capabilities',
+            "",
+            "",
+            "def list_chat_models() -> list[AIModel]:",
+            '    """The models a person choosing a chat model is offered from."""',
+            "    return [model for model in AI_MODEL_CATALOGUE.values() if is_chat_model(model)]",
+            "",
         ]
     )
 
@@ -342,6 +411,11 @@ def generate_typescript_code(specs: list[dict[str, Any]]) -> str:
             f"  name: '{spec['name']}',",
             f"  description: '{description}',",
             f"  provider: '{spec['provider']}',",
+            *(
+                [f"  providerUrl: '{spec['provider_url']}',"]
+                if spec.get("provider_url")
+                else []
+            ),
             f"  default: {str(spec.get('default', False)).lower()},",
             f"  available: {str(spec.get('available', False)).lower()},",
             f"  requiredEnvVars: {env_vars_formatted},",
@@ -358,6 +432,30 @@ def generate_typescript_code(specs: list[dict[str, Any]]) -> str:
         if capabilities:
             formatted = "[" + ", ".join(f"'{c}'" for c in capabilities) + "]"
             model_lines.append(f"  capabilities: {formatted},")
+        if spec.get("billing"):
+            model_lines.append(f"  billing: '{spec['billing']}',")
+        if spec.get("route"):
+            model_lines.append(f"  route: '{spec['route']}',")
+        if spec.get("context_window") is not None:
+            model_lines.append(f"  contextWindow: {int(spec['context_window'])},")
+        if spec.get("zero_data_retention") is not None:
+            model_lines.append(
+                f"  zeroDataRetention: {'true' if spec['zero_data_retention'] else 'false'},"
+            )
+        if spec.get("request_logging"):
+            model_lines.append(f"  requestLogging: '{spec['request_logging']}',")
+        aliases = spec.get("aliases") or []
+        if aliases:
+            model_lines.append(
+                "  aliases: [" + ", ".join(f"'{a}'" for a in aliases) + "],"
+            )
+        pricing = spec.get("pricing") or {}
+        if pricing:
+            model_lines.append(
+                "  pricing: { "
+                f"inputUsdPerMillion: {float(pricing['input_usd_per_million'])}, "
+                f"outputUsdPerMillion: {float(pricing['output_usd_per_million'])} }},"
+            )
         model_lines.extend(
             [
                 "};",
@@ -386,6 +484,39 @@ def generate_typescript_code(specs: list[dict[str, Any]]) -> str:
         [
             "};",
             "",
+            "/**",
+            " * The ids a model answered to before its id moved, to the id it has now.",
+            " * Kept apart from the catalogue, which lists each model once.",
+            " */",
+            "export const AI_MODEL_ALIASES: Record<string, string> = {",
+        ]
+    )
+    for spec in specs:
+        for alias in spec.get("aliases") or []:
+            lines.append(f"  '{alias}': '{spec['id']}',")
+    lines.extend(
+        [
+            "};",
+            "",
+            "/** A model by its id, or by an id it had before; undefined when the catalogue has neither. */",
+            "export function getModel(modelId: string): AIModel | undefined {",
+            "  return AI_MODEL_CATALOGUE[AI_MODEL_ALIASES[modelId] ?? modelId];",
+            "}",
+            "",
+            "/**",
+            " * A model a chat can run on: not a typed-judgment model (Jev), which",
+            " * answers typed questions about a state and nothing else. A spec that",
+            " * states no capability is read as a chat model.",
+            " */",
+            "export function isChatModel(model: AIModel): boolean {",
+            "  return !(model.capabilities ?? []).includes('judgments');",
+            "}",
+            "",
+            "/** The models a person choosing a chat model is offered from. */",
+            "export function listChatModels(): AIModel[] {",
+            "  return Object.values(AI_MODEL_CATALOGUE).filter(isChatModel);",
+            "}",
+            "",
         ]
     )