[Bug]: BaiduYiyanEmbed provider crashes with AttributeError when key is a JSON non-object (embed-side companion of #17389; same pattern as #17204 / #17373 / #17456 / #17458 / #17463)

[Harsh23Kashyap]

Self Checks

• I have searched for existing issues, including closed ones.
• I confirm that I am using English to submit this report.
• Non-English title submissions will be closed directly.
• I have not modified the template and have filled in all the required fields.

RAGFlow workspace code commit ID

main @ ee388b0fa (also reproducible on any v0.26.x since the BaiduYiyanEmbed class was introduced).

RAGFlow image version

v0.26.4 (also reproducible on the current dev image).

Other environment information

• Provider: BaiduYiyan (factory name) — see conf/llm_factories.json and conf/models/baidu.json.
• Affected class: BaiduYiyanEmbed (the embedding side of the BaiduYiyan / Qianfan provider) — rag/llm/embedding_model.py:1019-1029
• Companion class (already fixed by PR fix(chat_model): raise clear ModelException when BaiduYiyan key is not a JSON object (closes #17389) #17390 / commit c6d1af4b1): BaiduYiyanChat — rag/llm/chat_model.py:1189-1196. PR fix(chat_model): raise clear ModelException when BaiduYiyan key is not a JSON object (closes #17389) #17390 introduced the _resolve_qianfan_credentials helper in rag/llm/key_utils.py and wired BaiduYiyanChat.__init__ through it. However, PR fix(chat_model): raise clear ModelException when BaiduYiyan key is not a JSON object (closes #17389) #17390 only fixed the chat side — the embed side has the same shape of bug and was not touched.

Actual behavior

The BaiduYiyanEmbed.__init__ method parses its API key as JSON to extract the yiyan_ak and yiyan_sk fields:

# rag/llm/embedding_model.py:1025-1027
key = json.loads(key)
ak = key.get("yiyan_ak", "")
sk = key.get("yiyan_sk", "")

There is no try/except around the json.loads(key) call. A user pasting any input that is NOT a JSON object — for example:

• key = '[1, 2, 3]' parses to a list → .get("yiyan_ak", "") raises AttributeError: 'list' object has no attribute 'get'
• key = '"a plain string"' parses to a string → same AttributeError
• key = 'null' parses to None → same AttributeError
• key = '42' parses to int → same AttributeError
• key = 'true' parses to bool → same AttributeError
• key = '3.14' parses to float → same AttributeError

In every case the call site raises an unhandled AttributeError that bubbles up as a 500 in the API and a stack trace in the server log. This is the same class of bug that PR #17215 fixed for Azure, PR #17377 fixed for Bedrock, PR #17457 fixed for VolcEngine/Ark, PR #17459 fixed for OpenRouter, and PR #17464 fixed for GoogleCV.

The BaiduYiyan / Qianfan provider REQUIRES a JSON object key (per conf/models/baidu.json). A user pasting a plain Baidu API key like "bce-v3/ALTAK-.../..." (the most common mistake: copying from the Qianfan console) OR a JSON non-object (e.g. a JSON list) gets a 500 from inside rag/llm internals — no indication of what they did wrong.

Expected behavior

A user pasting any non-BaiduYiyan-shaped key (a plain API key, a JSON list, a JSON number, a malformed JSON object) should get a clear, actionable error at provider-init time, not a 500. The provider REQUIRES a JSON object, so a non-object input should raise a clear ModelException(retryable=False) with a message that names the required fields (yiyan_ak, yiyan_sk) and points at conf/models/baidu.json.

Steps to reproduce

1. Start RAGFlow v0.26.4 (any deployment).
2. Try to add a BaiduYiyan / Qianfan embedding model provider via the admin UI (or POST /api/v1/llm/factories).
3. Paste any of these into the API key field:
   • A JSON array like ["a", "b"] — crashes with AttributeError: 'list' object has no attribute 'get'.
   • A JSON number like 42 — same crash.
   • A JSON string like "a plain string" (with quotes) — same crash.
   • A JSON null null — same crash.
   • A JSON bool true — same crash.
4. The server log shows an unhandled AttributeError. The admin UI shows a generic 500 with no actionable hint.

Expected error surface

After the fix, a JSON non-object should raise the same clear error as the chat-side helper (e.g. ModelException(retryable=False) with a message like: "BaiduYiyan key must be a JSON object, got <type>. See conf/models/baidu.json for the model class.").

Coordination note

PR #17390 (commit c6d1af4b1) added a _resolve_qianfan_credentials helper in rag/llm/key_utils.py and wired BaiduYiyanChat.__init__ through it. This fix uses the same helper for the embed side. If PR #17390 lands first, this PR can be rebased to drop the helper re-addition. If this PR lands first, PR #17390 can be rebased to drop its helper addition (just keeping the chat-side wire-up). Either order resolves to the same final state. The maintainers can decide the merge order.

Additional information

Affected files and line numbers (in commit ee388b0fa):

File	Class	Lines	Current state
rag/llm/embedding_model.py	BaiduYiyanEmbed	1019-1029	CRASHES on non-JSON or JSON non-object

The fix is a single-line wire-up in BaiduYiyanEmbed.__init__ to use the existing _resolve_qianfan_credentials helper from rag/llm/key_utils.py (the same helper PR #17390 added for the chat side). If the helper is not yet on upstream main when this PR is opened, this PR also re-adds the helper (mirroring PR #17390 verbatim) so the embed site is self-contained.

Regression tests covering:

1. Plain string key → ModelException (the provider REQUIRES a JSON object; no plain-key fallback)
2. JSON dict with both fields → return parsed fields
3. JSON dict missing yiyan_ak or yiyan_sk → default to empty string (preserves tolerance)
4. JSON list, number, string, null, bool → ModelException (the new fix)
5. Bad JSON string → ModelException (the new fix)
6. Helper propagates the right fields to BaiduYiyanEmbed.__init__ so the class gets exactly what it needs

A PR fixing this is in progress.

[Harsh23Kashyap]

Fix opened in #17505.

Summary of the fix (embed-side companion of #17389 / PR #17390):

• Re-added the _resolve_qianfan_credentials(key) helper in rag/llm/key_utils.py (PR fix(chat_model): raise clear ModelException when BaiduYiyan key is not a JSON object (closes #17389) #17390 introduced the same helper for the chat side; since PR fix(chat_model): raise clear ModelException when BaiduYiyan key is not a JSON object (closes #17389) #17390 has not been merged to upstream main yet, the helper is re-added here so the embed-side wire-up is self-contained).
• Wired BaiduYiyanEmbed.__init__ in rag/llm/embedding_model.py:1025 through the helper.
• 14 p0 regression tests in test/unit_test/rag/llm/test_baiduyiyan_embed_json_key_fallback.py (9 on the helper, 5 on the embed constructor). The qianfan SDK is stubbed at sys.modules["qianfan"] level to sidestep a pre-existing baidubce/litellm Python 3.13 incompatibility in the test venv.

The helper raises ModelException(retryable=False) on any non-JSON or JSON non-object input (plain string, malformed JSON, JSON list/number/string/null/bool) with a message that names the required fields (yiyan_ak, yiyan_sk) and points at conf/models/baidu.json.

Coordination note: this PR adds the same helper that PR #17390 added. Two valid merge orderings:

1. If PR fix(chat_model): raise clear ModelException when BaiduYiyan key is not a JSON object (closes #17389) #17390 lands first, this PR can be rebased to drop the helper re-addition (keeping only the embed-side wire-up).
2. If this PR lands first, PR fix(chat_model): raise clear ModelException when BaiduYiyan key is not a JSON object (closes #17389) #17390 can be rebased to drop its helper addition (just keeping the chat-side wire-up).

Either ordering resolves to the same final state.

After this PR + PR #17390, no remaining unfixed LLM-provider JSON-decode sites in rag/llm/ that we own -- the only remaining site is the chat-side GoogleChat (rag/llm/chat_model.py:1284), which is owned by @glu000 via #15994 and is intentionally OUT OF SCOPE.

Closing this once the PR lands.