feat: 深度研究支持独立新报告与原文改写,并完善沙箱、模型管理与全文改写

新报告不再覆盖当前文件卡片,写入时打开沙箱;步骤条仅在改写原文时位于卡片上方。同时补齐改写本地回退、模型管理与会商第三步相关改动。

Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
董宏凯andCursor committed 2026-08-21 21:46:04 +08:00
1 parent 11f928bc95
commit a5a9c952e8
63 files changed
+4113 -929

No files matched your search

+40 -4
View File
@@ -238,7 +238,7 @@ from deerflow.config import get_app_config
Lead-agent middlewares are assembled in strict append order across `packages/harness/deerflow/agents/middlewares/tool_error_handling_middleware.py` (`build_lead_runtime_middlewares`) and `packages/harness/deerflow/agents/lead_agent/agent.py` (`_build_middlewares`):
1. **ThreadDataMiddleware** - Creates per-thread directories under the user's isolation scope (`backend/.deer-flow/users/{user_id}/threads/{thread_id}/user-data/{workspace,uploads,outputs}`); resolves `user_id` via `get_effective_user_id()` (falls back to `"default"` in no-auth mode); Web UI thread deletion now follows LangGraph thread removal with Gateway cleanup of the local thread directory
1. **ThreadDataMiddleware** - Creates per-thread directories under the creator's stable isolation scope (`backend/.deer-flow/users/{user_id}/threads/{thread_id}/user-data/{workspace,uploads,outputs}`); resolves persisted threads through `resolve_path_user_id(thread_id)` (falls back to `get_effective_user_id()` / `"default"` when metadata is unavailable); Web UI thread deletion now follows LangGraph thread removal with Gateway cleanup of the local thread directory
2. **UploadsMiddleware** - Tracks and injects newly uploaded files into conversation
3. **SandboxMiddleware** - Acquires sandbox, stores `sandbox_id` in state
4. **DanglingToolCallMiddleware** - Injects placeholder ToolMessages for AIMessage tool_calls that lack responses (e.g., due to user interruption), including raw provider tool-call payloads preserved only in `additional_kwargs["tool_calls"]`
@@ -645,7 +645,11 @@ Operators upgrading an existing DB run the whole file (`USE <db>;` then execute)
- Physical: `backend/.deer-flow/users/{user_id}/threads/{thread_id}/user-data/...`, `deer-flow/skills/`
- Translation: `replace_virtual_path()` / `replace_virtual_paths_in_command()`
- Detection: `is_local_sandbox()` checks `sandbox_id == "local"`
- **`{user_id}` for path resolution = `resolve_path_user_id(thread_id)`, NOT the current login.** Task-deeplink conversations (rwfx/3qfx/xdfx, `threads_meta.metadata.taskId`) are shared across all users (see the `taskId` escape branches in `thread_meta/sql.py`), but generated files/uploads live under a *per-user* dir. Resolving by current login meant account B opening a shared task conversation 404'd on the md/artifacts account A generated ("切换账号后生成的文件找不到"). `deerflow/runtime/thread_paths.py` (`resolve_path_user_id` sync / `aresolve_path_user_id` async) resolves a task-shared thread to its **creator** (`threads_meta.user_id`), so every account reads/writes the same physical directory; personal threads resolve to the current login unchanged (no behavior change). Wired at every thread-path seam: write source `ThreadDataMiddleware.before_agent` (drives all sandbox writes via `thread_data` paths), `present_file_tool`, `uploads/manager.get_uploads_dir` + `uploads_middleware` + uploads router, artifact serving (`path_utils.{a,}resolve_thread_virtual_path` → artifacts/skills routers), embedded `client.get_artifact`. Positive (taskId→creator) results are cached process-wide (immutable mapping); any failure / memory-mode store / missing row falls back to the current login. **Caveat**: files a *non-creator* wrote into a shared thread before this fix stay orphaned under that writer's dir (no data loss); all new writes consolidate under the creator. **Extended to 圆桌会商 system threads** (`metadata.thread_type == "roundtable"` — Step 2 seats/coordinator + Step 3 report/summary/dashboard/structure, created by `multi_agent.py` with `_ROUNDTABLE_THREAD_METADATA`): their Step2/Step3 sandbox artifacts (md/流程图/报告) must stay viewable when a历史草稿 is reopened or a taskId会商 draft is opened by another account, so `thread_paths._is_creator_shared_thread(metadata)` treats both `taskId` **and** `thread_type=="roundtable"` as creator-shared, and `ThreadMetaRepository.check_access` (the artifacts router's `owner_check`) bypasses the owner filter for both — roundtable thread ids are only surfaced inside the owner's draft, so this does not let unrelated accounts *discover* others' 会商 files. Tests: `tests/test_thread_paths_owner.py`, `tests/test_thread_paths_roundtable.py`.
- **`{user_id}` for path resolution = `resolve_path_user_id(thread_id)`, using the persisted creator for every tracked thread.** The route's regular owner check still authorizes access first; creator resolution only selects the stable physical directory (`threads_meta.user_id`) and avoids a lost request context falling back to `users/default`. This also preserves cross-account task-deeplink conversations (rwfx/3qfx/xdfx) and roundtable system threads. Positive creator mappings are cached process-wide and DB/memory/missing-row failures fall back to the current login. It is wired at the thread-path consumers including `ThreadDataMiddleware.before_agent`, `present_file_tool`, uploads, artifact serving, and embedded `client.get_artifact`. **Rewrite-history recovery is metadata-only and job-owner scoped** (`list_by_session(..., user_id=current_user)`), so its GET/list endpoints intentionally skip redundant thread-owner checks and return an empty history instead of a history-only 404; the start/commit endpoints retain strict thread write authorization. Tests: `tests/test_thread_paths_owner.py`, `tests/test_thread_paths_roundtable.py`.
- **Durable document-rewrite commit:** `document_rewrite_pipeline.atomic_write_text` writes through a deliberately short sibling temporary filename so long Windows artifact paths stay below the path-length limit, and guarantees the target `outputs` parent exists. If the sandbox recreates that directory in the narrow commit window, it repairs the directory and retries; the original file is otherwise only changed through `os.replace`.
- **Deep Research structural regeneration:** `POST /api/deep-research/sessions/{id}/report-variants` creates a distinct completed session from the source report and its selected evidence. It rebases both saved `[来源:source_id]` and legacy streamed `[[source:source_id]]` markers to cloned ids, and removes inherited markers with no selected clone. The frontend uses this endpoint only for “选择结构再生成”, then queues the durable writer with `operation=generate`; generation receives the confirmed outline and complete selected evidence set but never receives the inherited report as text to edit. The new report's completed card can subsequently start a separate in-place `AI 全文改写`. Requirement plans are deterministic and operation-specific, model reasoning is forwarded as live SSE before the first Markdown token, and generation prompts emit display-ready `[来源:id]` citations. Citation validation repairs only a unique near-complete `drs_src_*` prefix (provider-truncated suffix) and still rejects genuinely unknown ids. The report-only stream requests 8192 output tokens (except Codex Responses models, which reject token-limit kwargs) and rejects a malformed trailing citation instead of committing a provider-truncated report. Never restore the former second planning-model call unless its latency and streaming contract are redesigned together.
**Sandbox Tools** (in `packages/harness/deerflow/sandbox/tools.py`):
- `bash` - Execute commands with path translation and error handling
@@ -996,7 +1000,7 @@ The embedded detail proxy hides WeKnora's outer `.main > .aside_box` application
**Embedded native knowledge-base switching:** The original WeKnora breadcrumb picker remains available inside the detail iframe. A signed 15-minute embed cookie carries only the current DeerFlow user's visible mappings (their own plus published bases); the proxied `/api/v1/knowledge-bases` response is filtered to that allowlist. Choosing another entry performs a full iframe navigation, rechecks the DeerFlow mapping and read/write permission, then reissues the cookie for the selected base. Never expose the upstream service-admin list directly or permit an arbitrary remote knowledge-base ID.
The independent enterprise-research workbench (`/api/enterprise-research/tasks`) is deliberately separate from `/api/deep-research`: it stores an owner-scoped task plus a bounded, deduplicated evidence snapshot in `enterprise_research_tasks`, then calls WeKnora through the existing DeerFlow LLMWiki mapping and visibility checks. Never expose remote WeKnora IDs or credentials to its browser API, and do not make this workbench mutate a knowledge base or an existing Deep Research session/report. After collection, its plan endpoints can save an editable Markdown outline and an explicitly selected source subset; validate every selected chunk ID against the saved snapshot. The AI-plan variant may call a configured DeerFlow model but must not retrieve new evidence. `enterprise_research_report_jobs` freezes that plan/source subset plus action/model/style/length and before-image before calling the shared `DeerFlowCompletionBackend`, so the writer cannot run a fresh uncontrolled retrieval or accidentally use later task edits. Actions are `initial`, `rewrite`, and report-only `followup`; only successful initial/rewrite jobs may replace the task report. Its SSE route sends the persisted draft first and then only live `report_delta`/`reasoning_delta` events; reasoning must never be appended to report Markdown. Active rows carry a nullable owner/task dedupe key with a database unique constraint; terminal transitions clear it so history remains unlimited while simultaneous workers cannot create two active jobs for one task. Startup recovery must requeue the exact previously-read running snapshot and then use an atomic queued→running claim. Cancellation, completion, and failure are conditional state transitions so cross-process races cannot turn cancelled work into a committed report. Restore only accepts a completed rewrite before-image, rejects active jobs, and writes a new reversible history row before changing the task. The persistence contract is `deerflow.persistence.enterprise_research`; large evidence/report payloads use `PortableLongText` because MySQL `TEXT` cannot safely hold them. Regression coverage: `tests/test_enterprise_research_tasks.py`.
The independent enterprise-research workbench is **temporarily disabled**. Its source remains under `app/gateway/routers/enterprise_research.py` and `deerflow.persistence.enterprise_research`, but Gateway does not mount `/api/enterprise-research`, initialize its stores/executor, or register its ORM rows in `Base.metadata`. This ensures startup does not create or alter `enterprise_research_tasks` / `enterprise_research_report_jobs` on MySQL. It remains separate from `/api/deep-research`; do not re-enable it incidentally while changing the active Deep Research workflow. Regression coverage for its preserved implementation remains in `tests/test_enterprise_research_tasks.py`.
### Embedded Client (`packages/harness/deerflow/client.py`)
@@ -1128,6 +1132,15 @@ The router re-orders rows deterministically and only permits an enabled custom
theme as `default_theme`; removing or disabling the default safely falls back
to `light`. Regression coverage: `tests/test_appearance_theme.py`.
Compact top-bar shortcuts live in the same `appearance` setting. Their route
parameters support `login`, `theme`, and keyless `path`. A `path` parameter has
no query-key input; the operator selects its value source from the right-hand
dropdown and the frontend appends the resolved, URL-encoded value to the
external URL path (or Hash route) in configured order. Password stays an
ordinary `login` query parameter with the key `password`, so an external login
link can become `#/login/<user>?password=…`. Regression coverage:
`tests/test_appearance_theme.py`.
### File Upload
Multi-file upload with automatic document conversion:
@@ -1289,7 +1302,10 @@ token. Models that expose thinking inside ordinary `<think>...</think>` text
must be split statefully before a report delta, checkpoint, or Markdown
projection is created; an optional live-only `report_thinking` frame can drive
the message-list thought trace, while the sandbox receives answer Markdown
only. The Vite
only. The post-report digest (`summarize_report`) must likewise forward
provider reasoning through live-only `summary_thinking` frames so the
chat timeline can render thought while waiting for the first `summary_delta`.
The Vite
page owns Word export through
`frontend-web/src/core/scheduled-tasks/markdown-docx.ts`; do not add Python
document-conversion dependencies merely to export text.
@@ -1375,3 +1391,23 @@ candidates fail with 409 rather than overwrite a later edit. Ambiguous or
ordinary evidence questions must remain non-mutating, and cancellation or a
stream error must never save a partial report. The rewrite prompt receives only
the selected range plus the selected evidence set, never a new web search.
The sandbox whole-report rewrite may carry an optional `reportOutline` Markdown
template from the shared report-structure library. Persist the original template
in the durable input snapshot and append it as an explicit chapter-structure
constraint before requirements planning and prose generation; a template must
never be used as permission to invent facts absent from the current report.
`DocumentRewriteVersionRepository.create` backs the reversible before-image for
both artifact and whole-report rewrites. It must flush and serialize the new row
on the writer before commit, and must never call `session.refresh()` after
commit: a read/write-splitting MySQL endpoint can route that refresh to a
lagging replica, falsely report a version-save failure, and trigger the caller's
intentional restore of the rewritten document.
Version snapshots are an undo enhancement, not a condition for preserving an
already committed rewrite. If snapshot creation fails after the artifact/report
write, log the exception, complete the rewrite with
`versionSnapshotFailed: true`, and do not emit a user-facing failure or restore
the original. The sandbox uses that signal only to enable its ordinary manual
save fallback.