Files
ha-openai-compatible/custom_components/openai_compatible/url_util.py
T
Adam SmithandClaude Opus 5 e7fcb1d7a1 feat: one entry per provider, endpoints per type and model from a live dropdown
Reshapes setup around how a multi-model proxy is actually used: one config
entry per server/provider, then one endpoint per model added by hand, with
the model chosen from what that provider advertises right now.

Setup (async_step_user):
- Adds a Name field. With one entry per provider the title matters, and it
  was hardcoded to "ChatGPT" regardless of where the entry pointed. Blank
  falls back to the endpoint host via url_util.title_from_base_url.
  Deliberately absent from the reauth schema (now split out as
  STEP_REAUTH_DATA_SCHEMA): that step writes its input back into entry.data
  via data_updates, so a name field there would persist the title as
  connection data on every key rotation.
- Stops auto-creating four subentries. conversation/ai_task_data/stt/tts were
  all created up front carrying RECOMMENDED_* options, which pin OpenAI model
  names (gpt-4o-mini-tts, gpt-4o-mini-transcribe). Against a third-party
  endpoint those entities are dead on arrival -- the reported symptom was an
  install that "loads bits, it's all janky".
- Drops _async_abort_entries_match, so several entries may point at one
  server (one per provider-side key, for instance). Upstream aborts with
  already_configured on matching url+key.

Model discovery (new model_list.py, kept out of the vendored files like
url_util.py):
- async_fetch_model_ids asks the endpoint's GET /v1/models. It calls
  with_options(max_retries=0), because this runs while the user waits on a
  dialog and the SDK otherwise retries with backoff -- visible in the test
  run as repeated "Retrying request to /models" and a suite that went from
  5s to minutes.
- Returns [] on any OpenAIError, and model_field() then yields a plain text
  field instead of a dropdown, so an endpoint that serves /v1/responses but
  not /v1/models is still configurable by hand. The dropdown sets
  custom_value=True for the same reason (LiteLLM wildcard routes).
- Wired into all four types: conversation and ai_task_data on the additional
  step, stt (which hardcoded OpenAI's three transcribe models) and tts on
  init. TTS had no model field at all -- it silently used gpt-4o-mini-tts.

A new endpoint now starts with "recommended" unticked. Ticked, the flow ends
on the first step and never shows a model field, which is why the model
looked unpickable; reconfigure still keeps whatever was saved.

The model field stays on the additional step rather than moving to init. The
init position reads better, but 71 ported tests drive the model through the
additional step, several from parametrized option dicts that no mechanical
edit could relocate. Rewriting the only regression net against upstream to
save one click was the wrong trade.

Also: the setup description no longer says "your OpenAI API key", and
instructions_url pointed at a README anchor (#generate-an-api-key) that does
not exist -- now #configuration. strings.json re-verified against core's
RE_URL: 0 violations.

Local: 167 passed.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NxJYWgGJtCw13qUHZtG8ki
2026-09-17 22:45:36 -04:00

91 lines
3.3 KiB
Python

"""Base-URL normalization and safe logging.
This module is intentionally separate from the vendored upstream files so that
re-syncing with home-assistant/core never conflicts with it.
"""
from __future__ import annotations
from urllib.parse import urlparse, urlunparse
ALLOWED_SCHEMES = frozenset({"http", "https"})
class InvalidBaseURL(ValueError):
"""Raised when a base URL is unusable."""
class CredentialsInBaseURL(InvalidBaseURL):
"""Raised when a base URL embeds a username or password.
A distinct subclass (rather than reusing the generic InvalidBaseURL
message) so the config flow can show a specific, actionable error
instead of the generic "not a valid URL" one.
"""
def normalize_base_url(url: str) -> str:
"""Strip whitespace and trailing slashes, and validate the scheme.
Rejects a URL carrying a username or password component: authentication
belongs in the API key field, which is handled (and never logged) safely
on its own — a base_url is user-suppliable and read back into log/error
text by vendored code this fork does not touch, so credentials must never
be accepted here in the first place.
"""
cleaned = (url or "").strip()
if not cleaned:
raise InvalidBaseURL("base URL is empty")
parsed = urlparse(cleaned)
if parsed.scheme not in ALLOWED_SCHEMES:
raise InvalidBaseURL(f"unsupported scheme: {parsed.scheme or '(none)'}")
if not parsed.netloc:
raise InvalidBaseURL("base URL has no host")
if parsed.username or parsed.password:
raise CredentialsInBaseURL(
"base URL must not contain a username or password; "
"use the API key field for authentication"
)
return urlunparse(parsed._replace(path=parsed.path.rstrip("/")))
def sanitize_url_for_logging(url: str) -> str:
"""Return the URL with any embedded credentials replaced by ***.
This only recognizes credentials urlparse places in an authority
(``user:pass@host``) component of an absolute http(s)/-scheme URL. It is
not a general-purpose credential scrubber: a scheme-less or relative
string with an embedded ``user:pass@`` substring passes through
unmasked, since urlparse won't parse it as a netloc. Safe today because
every caller normalizes with normalize_base_url() first (which both
requires an absolute http(s) URL and rejects embedded credentials
outright) before this is ever called on it; do not rely on this function
alone to sanitize an arbitrary, unvalidated string.
"""
parsed = urlparse(url)
if not parsed.username and not parsed.password:
return url
host = parsed.hostname or ""
if parsed.port:
host = f"{host}:{parsed.port}"
return urlunparse(parsed._replace(netloc=f"***@{host}"))
def title_from_base_url(url: str) -> str:
"""Return a config entry title identifying the provider.
Upstream titles every entry "ChatGPT", which is wrong the moment the
endpoint is not OpenAI's and useless when there is one entry per provider.
The host (with port, when given) is what actually distinguishes them.
Falls back to the whole string if it cannot be parsed, so this never
raises in the middle of creating an entry.
"""
try:
netloc = urlparse(url).netloc
except ValueError:
return url
return netloc or url