| 1 | #!/usr/bin/env python3 |
| 2 | """Check that docs/PROVIDERS.md tracks the shipped provider registry. |
| 3 | |
| 4 | This is intentionally lightweight. It does not try to generate prose; it checks |
| 5 | the stable identifiers and default strings that are easy for docs to drift from: |
| 6 | |
| 7 | - canonical ProviderKind IDs |
| 8 | - provider TOML tables |
| 9 | - live TUI ApiProvider IDs |
| 10 | - shipped-provider table rows |
| 11 | - static ModelRegistry provider rows |
| 12 | - default provider model/base URL constants |
| 13 | """ |
| 14 | |
| 15 | from __future__ import annotations |
| 16 | |
| 17 | import re |
| 18 | import sys |
| 19 | from pathlib import Path |
| 20 | |
| 21 | |
| 22 | ROOT = Path(__file__).resolve().parents[1] |
| 23 | CONFIG_RS = ROOT / "crates" / "config" / "src" / "lib.rs" |
| 24 | # ProviderKind's enum + identity impl were split out of lib.rs into this module. |
| 25 | PROVIDER_KIND_RS = ROOT / "crates" / "config" / "src" / "provider_kind.rs" |
| 26 | PROVIDER_RS = ROOT / "crates" / "config" / "src" / "provider.rs" |
| 27 | TUI_CONFIG_RS = ROOT / "crates" / "tui" / "src" / "config.rs" |
| 28 | # Default provider model/base-URL constants were split out of config.rs into |
| 29 | # this leaf module (#3311); read them from there for the default-string check. |
| 30 | TUI_CONFIG_MODELS_RS = ROOT / "crates" / "tui" / "src" / "config" / "models.rs" |
| 31 | AGENT_RS = ROOT / "crates" / "agent" / "src" / "lib.rs" |
| 32 | PROVIDERS_MD = ROOT / "docs" / "PROVIDERS.md" |
| 33 | |
| 34 | |
| 35 | API_PROVIDER_ONLY_IDS = {"deepseek-cn"} |
| 36 | |
| 37 | # `custom` is the dynamic OpenAI-compatible meta-provider (#1519): a single |
| 38 | # catch-all `[providers.custom]` table that backs arbitrary user-defined |
| 39 | # endpoints, not a canonical shipped provider with a docs row. It is excluded |
| 40 | # from the provider-table drift check. |
| 41 | META_PROVIDER_TABLES = {"custom"} |
| 42 | SHARED_PROVIDER_TABLES = { |
| 43 | "siliconflow-CN": "siliconflow_cn", |
| 44 | } |
| 45 | HUGGINGFACE_ALIASES = {"huggingface", "hugging-face", "hugging_face", "hf"} |
| 46 | HUGGINGFACE_API_KEY_ENV_ORDER = ["HUGGINGFACE_API_KEY", "HF_TOKEN"] |
| 47 | HUGGINGFACE_BASE_URL_ENV_ORDER = ["HUGGINGFACE_BASE_URL", "HF_BASE_URL"] |
| 48 | HUGGINGFACE_MODEL_ENV_ORDER = ["HUGGINGFACE_MODEL", "HF_MODEL"] |
| 49 | SENSITIVE_IDENTIFIER_RE = re.compile(r"(?i)(api[_-]?key|token|secret|password|credential)") |
| 50 | SENSITIVE_BEARER_RE = re.compile(r"(?i)(authorization:\s*bearer\s+)\S+") |
| 51 | SENSITIVE_ASSIGNMENT_RE = re.compile( |
| 52 | r"(?i)\b(api[_-]?key|token|secret|password|credential)(\s*[:=]\s*)\S+" |
| 53 | ) |
| 54 | |
| 55 | |
| 56 | def read(path: Path) -> str: |
| 57 | return path.read_text(encoding="utf-8") |
| 58 | |
| 59 | |
| 60 | def display_public_value(value: str) -> str: |
| 61 | if SENSITIVE_IDENTIFIER_RE.search(value): |
| 62 | return "<redacted sensitive identifier>" |
| 63 | return value |
| 64 | |
| 65 | |
| 66 | def redact_sensitive_text(value: str) -> str: |
| 67 | value = SENSITIVE_BEARER_RE.sub(r"\1<redacted>", value) |
| 68 | value = SENSITIVE_ASSIGNMENT_RE.sub(r"\1\2<redacted>", value) |
| 69 | return SENSITIVE_IDENTIFIER_RE.sub("<redacted sensitive identifier>", value) |
| 70 | |
| 71 | |
| 72 | def require_index(source: str, needle: str, context: str, start: int = 0) -> int: |
| 73 | try: |
| 74 | return source.index(needle, start) |
| 75 | except ValueError: |
| 76 | raise ValueError(f"{context}: missing {needle!r}") from None |
| 77 | |
| 78 | |
| 79 | def markdown_section(source: str, heading: str) -> str: |
| 80 | start = require_index(source, heading, "docs/PROVIDERS.md") |
| 81 | next_heading = source.find("\n## ", start + len(heading)) |
| 82 | end = len(source) if next_heading == -1 else next_heading |
| 83 | return source[start:end] |
| 84 | |
| 85 | |
| 86 | def extract_match_block( |
| 87 | source: str, signature: str, context: str, start: int = 0 |
| 88 | ) -> str: |
| 89 | start = require_index(source, signature, context, start) |
| 90 | match_start = require_index(source, "match", f"match block after {signature!r}", start) |
| 91 | brace_start = require_index(source, "{", f"match block after {signature!r}", match_start) |
| 92 | depth = 0 |
| 93 | for index in range(brace_start, len(source)): |
| 94 | char = source[index] |
| 95 | if char == "{": |
| 96 | depth += 1 |
| 97 | elif char == "}": |
| 98 | depth -= 1 |
| 99 | if depth == 0: |
| 100 | return source[brace_start + 1 : index] |
| 101 | raise ValueError(f"could not parse match block after {signature!r}") |
| 102 | |
| 103 | |
| 104 | def parse_aliases_for_variant(source: str, enum_name: str, variant: str, context: str) -> set[str]: |
| 105 | # `ProviderKind`'s enum + identity impl (incl. `parse`) live in |
| 106 | # provider_kind.rs after the config module split; read the impl from there |
| 107 | # regardless of the file the caller passed for other lookups. |
| 108 | if enum_name == "ProviderKind": |
| 109 | source = read(PROVIDER_KIND_RS) |
| 110 | context = "crates/config/src/provider_kind.rs" |
| 111 | impl_start = require_index(source, f"impl {enum_name}", context) |
| 112 | block = extract_match_block( |
| 113 | source, |
| 114 | "pub fn parse(value: &str) -> Option<Self>", |
| 115 | context, |
| 116 | impl_start, |
| 117 | ) |
| 118 | match_arm = re.search( |
| 119 | rf'((?:"[^"]+"\s*\|\s*)*"[^"]+")\s*=>\s*Some\(Self::{variant}\)', |
| 120 | block, |
| 121 | ) |
| 122 | if match_arm: |
| 123 | return set(re.findall(r'"([^"]+)"', match_arm.group(1))) |
| 124 | if enum_name in {"ProviderKind", "ApiProvider"}: |
| 125 | provider_rs = read(PROVIDER_RS) |
| 126 | provider_macro = re.search( |
| 127 | rf'provider!\(\s*\n\s*\w+,\s*\n\s*{variant},\s*\n\s*"([^"]+)".*?' |
| 128 | r"aliases:\s*\[(.*?)\]\s*\);", |
| 129 | provider_rs, |
| 130 | re.DOTALL, |
| 131 | ) |
| 132 | if provider_macro: |
| 133 | return {provider_macro.group(1)} | set( |
| 134 | re.findall(r'"([^"]+)"', provider_macro.group(2)) |
| 135 | ) |
| 136 | raise ValueError(f"{context}: missing parse arm for {variant}") |
| 137 | |
| 138 | |
| 139 | def provider_kind_ids(config_rs: str) -> dict[str, str]: |
| 140 | provider_rs = read(PROVIDER_RS) |
| 141 | pairs = re.findall( |
| 142 | r"provider!\(\s*\n\s*\w+,\s*\n\s*(\w+),\s*\n\s*\"([^\"]+)\"", |
| 143 | provider_rs, |
| 144 | ) |
| 145 | ids: dict[str, str] = {variant: provider_id for variant, provider_id in pairs} |
| 146 | # Providers with non-fixed wire policy or custom auth behavior use manual |
| 147 | # impls rather than the provider!() macro. |
| 148 | for variant_name, id_literal in [ |
| 149 | ("Deepseek", "deepseek"), |
| 150 | ("DeepseekAnthropic", "deepseek-anthropic"), |
| 151 | ("OpenaiCodex", "openai-codex"), |
| 152 | ("Anthropic", "anthropic"), |
| 153 | ("Openmodel", "openmodel"), |
| 154 | ("MinimaxAnthropic", "minimax-anthropic"), |
| 155 | ("OpencodeZen", "opencode-zen"), |
| 156 | # Alibaba Model Studio ships four plan/dialect identities, each with a |
| 157 | # hand-written impl Provider for the same reason as the rows above: |
| 158 | # the wire policy is not fixed, so provider!() cannot express them. |
| 159 | ("ModelstudioTokenPlan", "modelstudio-token-plan"), |
| 160 | ("ModelstudioTokenPlanAnthropic", "modelstudio-token-plan-anthropic"), |
| 161 | ("ModelstudioCodingPlan", "modelstudio-coding-plan"), |
| 162 | ("ModelstudioCodingPlanAnthropic", "modelstudio-coding-plan-anthropic"), |
| 163 | ]: |
| 164 | match = re.search( |
| 165 | rf'impl\s+Provider\s+for\s+{variant_name}.*?fn\s+id.*?\"({id_literal})\"', |
| 166 | provider_rs, re.DOTALL, |
| 167 | ) |
| 168 | if match: |
| 169 | ids[variant_name] = match.group(1) |
| 170 | if not ids: |
| 171 | raise ValueError("provider!() invocations returned no providers") |
| 172 | return ids |
| 173 | |
| 174 | |
| 175 | def api_provider_ids(tui_config_rs: str) -> dict[str, str]: |
| 176 | # ApiProvider ids derive from ProviderKind ids (via delegation to .kind().as_str()) |
| 177 | # plus the legacy "deepseek-cn" variant that exists only in ApiProvider. |
| 178 | variant_to_id = provider_kind_ids("") |
| 179 | # ApiProvider::SiliconflowCn maps to ProviderKind::SiliconflowCN |
| 180 | if "SiliconflowCN" in variant_to_id: |
| 181 | variant_to_id["SiliconflowCn"] = variant_to_id["SiliconflowCN"] |
| 182 | variant_to_id["DeepseekCN"] = "deepseek-cn" |
| 183 | return variant_to_id |
| 184 | |
| 185 | |
| 186 | def provider_tables(config_rs: str) -> set[str]: |
| 187 | struct_start = require_index( |
| 188 | config_rs, "pub struct ProvidersToml", "crates/config/src/lib.rs" |
| 189 | ) |
| 190 | struct_end = require_index(config_rs, "\n}", "ProvidersToml struct", struct_start) |
| 191 | fields = re.findall( |
| 192 | r"pub\s+([a-z0-9_]+)\s*:\s*ProviderConfigToml", |
| 193 | config_rs[struct_start:struct_end], |
| 194 | ) |
| 195 | if not fields: |
| 196 | raise ValueError("ProvidersToml returned no provider tables") |
| 197 | return set(fields) |
| 198 | |
| 199 | |
| 200 | def shipped_provider_rows(providers_md: str) -> set[str]: |
| 201 | table = markdown_section(providers_md, "## Shipped Providers") |
| 202 | return set(re.findall(r"^\|\s*`([^`]+)`\s*\|", table, flags=re.MULTILINE)) |
| 203 | |
| 204 | |
| 205 | def shipped_provider_tables(providers_md: str) -> set[str]: |
| 206 | table = markdown_section(providers_md, "## Shipped Providers") |
| 207 | return set(re.findall(r"\|\s*`\[providers\.([a-z0-9_]+)\]`\s*\|", table)) |
| 208 | |
| 209 | |
| 210 | def static_registry_provider_rows(providers_md: str) -> set[str]: |
| 211 | table = markdown_section(providers_md, "## Static Model Registry") |
| 212 | return set(re.findall(r"^\|\s*`([^`]+)`\s*\|", table, flags=re.MULTILINE)) |
| 213 | |
| 214 | |
| 215 | def model_registry_providers(agent_rs: str, variant_to_id: dict[str, str]) -> set[str]: |
| 216 | variants = set(re.findall(r"provider:\s*ProviderKind::(\w+)", agent_rs)) |
| 217 | missing = variants - set(variant_to_id) |
| 218 | if missing: |
| 219 | raise ValueError(f"ModelRegistry uses unknown provider variants: {sorted(missing)}") |
| 220 | return {variant_to_id[variant] for variant in variants} |
| 221 | |
| 222 | |
| 223 | def default_strings(tui_config_rs: str) -> set[str]: |
| 224 | # Model/base-URL constants now live in config/models.rs (#3311); scan it |
| 225 | # alongside config.rs so the check follows the leaf split. |
| 226 | sources = tui_config_rs + "\n" + read(TUI_CONFIG_MODELS_RS) |
| 227 | defaults = set() |
| 228 | for name, value in re.findall( |
| 229 | r'const\s+(DEFAULT_[A-Z0-9_]+(?:MODEL|BASE_URL)):\s*&str\s*=\s*"([^"]+)"', |
| 230 | sources, |
| 231 | ): |
| 232 | if name == "DEFAULT_DEEPSEEKCN_BASE_URL": |
| 233 | continue |
| 234 | defaults.add(value) |
| 235 | if not defaults: |
| 236 | raise ValueError("no default provider model/base URL constants found") |
| 237 | return defaults |
| 238 | |
| 239 | |
| 240 | def missing_default_strings(providers_md: str, defaults: set[str]) -> list[str]: |
| 241 | # Inline-code validation should not let fenced TOML/bash examples pair a |
| 242 | # stray backtick with later prose; strip fenced blocks before scanning. |
| 243 | inline_source = re.sub(r"```.*?```", "", providers_md, flags=re.DOTALL) |
| 244 | code_spans = set(re.findall(r"`([^`]+)`", inline_source)) |
| 245 | return sorted(defaults - code_spans) |
| 246 | |
| 247 | |
| 248 | def report_set(label: str, expected: set[str], actual: set[str]) -> list[str]: |
| 249 | errors = [] |
| 250 | missing = sorted(expected - actual) |
| 251 | extra = sorted(actual - expected) |
| 252 | if missing: |
| 253 | errors.append(f"{label} missing: {', '.join(missing)}") |
| 254 | if extra: |
| 255 | errors.append(f"{label} extra: {', '.join(extra)}") |
| 256 | return errors |
| 257 | |
| 258 | |
| 259 | def report_provider_enum_drift( |
| 260 | provider_kind_ids: set[str], api_provider_ids: set[str] |
| 261 | ) -> list[str]: |
| 262 | errors = [] |
| 263 | missing_from_api_provider = sorted(provider_kind_ids - api_provider_ids) |
| 264 | unexpected_api_provider_ids = sorted( |
| 265 | api_provider_ids - provider_kind_ids - API_PROVIDER_ONLY_IDS |
| 266 | ) |
| 267 | missing_allowlisted_ids = sorted(API_PROVIDER_ONLY_IDS - api_provider_ids) |
| 268 | |
| 269 | if missing_from_api_provider: |
| 270 | errors.append( |
| 271 | "ApiProvider missing ProviderKind IDs: " |
| 272 | + ", ".join(missing_from_api_provider) |
| 273 | ) |
| 274 | if unexpected_api_provider_ids: |
| 275 | errors.append( |
| 276 | "ApiProvider has non-whitelisted IDs absent from ProviderKind: " |
| 277 | + ", ".join(unexpected_api_provider_ids) |
| 278 | ) |
| 279 | if missing_allowlisted_ids: |
| 280 | errors.append( |
| 281 | "ApiProvider-only whitelist entries are absent from ApiProvider: " |
| 282 | + ", ".join(missing_allowlisted_ids) |
| 283 | ) |
| 284 | return errors |
| 285 | |
| 286 | |
| 287 | def report_huggingface_coverage( |
| 288 | config_rs: str, tui_config_rs: str, providers_md: str |
| 289 | ) -> list[str]: |
| 290 | errors = [] |
| 291 | |
| 292 | config_aliases = parse_aliases_for_variant( |
| 293 | config_rs, "ProviderKind", "Huggingface", "crates/config/src/lib.rs" |
| 294 | ) |
| 295 | tui_aliases = parse_aliases_for_variant( |
| 296 | tui_config_rs, "ApiProvider", "Huggingface", "crates/tui/src/config.rs" |
| 297 | ) |
| 298 | errors += report_set( |
| 299 | "ProviderKind Hugging Face aliases", |
| 300 | HUGGINGFACE_ALIASES, |
| 301 | config_aliases & HUGGINGFACE_ALIASES, |
| 302 | ) |
| 303 | errors += report_set( |
| 304 | "ApiProvider Hugging Face aliases", |
| 305 | HUGGINGFACE_ALIASES, |
| 306 | tui_aliases & HUGGINGFACE_ALIASES, |
| 307 | ) |
| 308 | |
| 309 | inline_source = re.sub(r"```.*?```", "", providers_md, flags=re.DOTALL) |
| 310 | code_spans = set(re.findall(r"`([^`]+)`", inline_source)) |
| 311 | errors += report_set( |
| 312 | "documented Hugging Face aliases", |
| 313 | HUGGINGFACE_ALIASES, |
| 314 | code_spans & HUGGINGFACE_ALIASES, |
| 315 | ) |
| 316 | |
| 317 | for label, env_order in [ |
| 318 | ("Hugging Face auth env precedence", HUGGINGFACE_API_KEY_ENV_ORDER), |
| 319 | ("Hugging Face base URL env precedence", HUGGINGFACE_BASE_URL_ENV_ORDER), |
| 320 | ("Hugging Face model env precedence", HUGGINGFACE_MODEL_ENV_ORDER), |
| 321 | ]: |
| 322 | errors += report_env_lookup_order( |
| 323 | label, config_rs, env_order, "crates/config/src/lib.rs" |
| 324 | ) |
| 325 | errors += report_env_lookup_order( |
| 326 | label, tui_config_rs, env_order, "crates/tui/src/config.rs" |
| 327 | ) |
| 328 | errors += report_string_order(label, providers_md, env_order, "docs/PROVIDERS.md") |
| 329 | |
| 330 | return errors |
| 331 | |
| 332 | |
| 333 | def report_env_lookup_order( |
| 334 | label: str, source: str, expected_order: list[str], context: str |
| 335 | ) -> list[str]: |
| 336 | lookup_needles = [f'std::env::var("{name}")' for name in expected_order] |
| 337 | return report_string_order(label, source, lookup_needles, context) |
| 338 | |
| 339 | |
| 340 | def report_string_order( |
| 341 | label: str, source: str, expected_order: list[str], context: str |
| 342 | ) -> list[str]: |
| 343 | contains_sensitive_expected_value = any( |
| 344 | SENSITIVE_IDENTIFIER_RE.search(value) for value in expected_order |
| 345 | ) |
| 346 | positions = [] |
| 347 | for needle in expected_order: |
| 348 | index = source.find(needle) |
| 349 | if index == -1: |
| 350 | if contains_sensitive_expected_value: |
| 351 | return [f"{label} missing required entry in {context}"] |
| 352 | return [f"{label} missing {display_public_value(needle)!r} in {context}"] |
| 353 | positions.append(index) |
| 354 | if positions != sorted(positions): |
| 355 | if contains_sensitive_expected_value: |
| 356 | return [f"{label} has wrong order in {context}"] |
| 357 | return [ |
| 358 | f"{label} has wrong order in {context}: expected " |
| 359 | + " before ".join(display_public_value(value) for value in expected_order) |
| 360 | ] |
| 361 | return [] |
| 362 | |
| 363 | |
| 364 | def provider_table_name(provider_id: str) -> str: |
| 365 | return SHARED_PROVIDER_TABLES.get(provider_id, provider_id.replace("-", "_")) |
| 366 | |
| 367 | |
| 368 | def main() -> int: |
| 369 | try: |
| 370 | config_rs = read(CONFIG_RS) |
| 371 | tui_config_rs = read(TUI_CONFIG_RS) |
| 372 | agent_rs = read(AGENT_RS) |
| 373 | providers_md = read(PROVIDERS_MD) |
| 374 | |
| 375 | variant_to_id = provider_kind_ids(config_rs) |
| 376 | canonical_ids = set(variant_to_id.values()) |
| 377 | live_api_provider_ids = set(api_provider_ids(tui_config_rs).values()) |
| 378 | expected_tables = {provider_table_name(provider_id) for provider_id in canonical_ids} |
| 379 | |
| 380 | errors: list[str] = [] |
| 381 | errors += report_provider_enum_drift(canonical_ids, live_api_provider_ids) |
| 382 | errors += report_huggingface_coverage(config_rs, tui_config_rs, providers_md) |
| 383 | errors += report_set( |
| 384 | "shipped provider rows", |
| 385 | canonical_ids, |
| 386 | shipped_provider_rows(providers_md), |
| 387 | ) |
| 388 | errors += report_set( |
| 389 | "provider TOML tables", |
| 390 | expected_tables, |
| 391 | provider_tables(config_rs) - META_PROVIDER_TABLES, |
| 392 | ) |
| 393 | errors += report_set( |
| 394 | "documented provider TOML tables", |
| 395 | expected_tables, |
| 396 | shipped_provider_tables(providers_md), |
| 397 | ) |
| 398 | errors += report_set( |
| 399 | "static ModelRegistry rows", |
| 400 | model_registry_providers(agent_rs, variant_to_id), |
| 401 | static_registry_provider_rows(providers_md), |
| 402 | ) |
| 403 | |
| 404 | missing_defaults = missing_default_strings(providers_md, default_strings(tui_config_rs)) |
| 405 | if missing_defaults: |
| 406 | errors.append( |
| 407 | "docs/PROVIDERS.md does not mention default strings as Markdown code spans: " |
| 408 | + ", ".join(missing_defaults) |
| 409 | ) |
| 410 | except ValueError as err: |
| 411 | errors = [str(err)] |
| 412 | |
| 413 | if errors: |
| 414 | print("Provider registry drift check failed:", file=sys.stderr) |
| 415 | for error in errors: |
| 416 | print(f"- {redact_sensitive_text(error)}", file=sys.stderr) |
| 417 | return 1 |
| 418 | |
| 419 | print("Provider registry drift check passed.") |
| 420 | return 0 |
| 421 | |
| 422 | |
| 423 | if __name__ == "__main__": |
| 424 | raise SystemExit(main()) |
| 425 |