[DataProcessor]merge processor#7726
Conversation
|
Thanks for your contribution! |
This comment was marked as outdated.
This comment was marked as outdated.
Sorry, something went wrong.
CI报告基于以下代码生成（30分钟更新一次）: 1 任务总览❌ 有 1 个 required 任务失败，另有 4 个 required 任务运行中，需等待结果。
2 任务状态汇总2.1 Required任务 : 5/10 通过
2.2 可选任务 — 22/26 通过
3 失败详情（仅 required）Approval — 代码规范（置信度: 高）Approval
根因详情: 关键日志: 修复建议:
修复建议摘要: 请 xyxinyang 或 zyyzghb 对 PR 进行 Approve 关联变更: PR 新增了大量 链接: 查看日志 |
Codecov Report❌ Patch coverage is Additional details and impacted files@@ Coverage Diff @@
## develop #7726 +/- ##
==========================================
Coverage ? 70.17%
==========================================
Files ? 409
Lines ? 57420
Branches ? 9013
==========================================
Hits ? 40293
Misses ? 14328
Partials ? 2799
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
This comment was marked as outdated.
This comment was marked as outdated.
Sorry, something went wrong.
This comment was marked as outdated.
This comment was marked as outdated.
Sorry, something went wrong.
This comment was marked as outdated.
This comment was marked as outdated.
Sorry, something went wrong.
PaddlePaddle-bot
left a comment
PaddlePaddle-bot
left a comment
There was a problem hiding this comment.
🤖 Paddle-CI-Agent | pr_review |
2026-05-07 16:29:16
📋 Review 摘要
PR 概述：将多个多模态视觉语言处理器（QwenVL、Qwen3VL、ErnieVL、PaddleOCRVL）整合到统一的 fastdeploy/input/multimodal/ 模块，并重构 chat() 消息解析与 _add_request 类型检测逻辑。
变更范围：fastdeploy/entrypoints/、fastdeploy/input/multimodal/（新增）、tests/
影响面 Tag：[DataProcessor] [APIServer]
📝 PR 规范检查
PR 标题 Tag 后缺少空格（[DataProcessor]merge → [DataProcessor] Merge …）；PR body 所有必填 section（Motivation / Modifications / Usage or Command / Accuracy Tests）内容均为空，仅保留了模板注释。
标题建议（可直接复制）：
[DataProcessor] Merge multimodal processors for QwenVL/Qwen3VL/ErnieVL/PaddleOCRVL
PR 描述建议（可直接复制，必须复刻 checklist §D2 模板的完整结构）：
## Motivation
将多个多模态视觉语言处理器（QwenVL、Qwen3VL、ErnieVL、PaddleOCRVL）从分散实现统一整合到 `fastdeploy/input/multimodal/` 模块，减少代码重复，提高可维护性；同时修复 `_add_request` 对空列表的潜在越界访问，优化 chat 消息解析逻辑。
## Modifications
- `fastdeploy/entrypoints/chat_utils.py`：简化 `parse_chat_messages`，`str`/`None` 类型的 `content` 直接赋值，而非强制包装为列表
- `fastdeploy/entrypoints/llm.py`：
- `chat()` 直接传递 `messages`，不再包装为 `{"messages": msg}` 字典列表
- `_add_request` 新增对 list-of-dict（messages 格式）的类型检测分支
- 修复 `prompts[i][0]` 在空列表时的潜在 IndexError（增加 `len > 0` 守卫）
- `fastdeploy/input/multimodal/`（新增模块）：
- `common.py`：共享图像缩放工具函数（`smart_resize_qwen`、`smart_resize_paddleocr`）
- `image_processors/ernie.py`：`AdaptiveImageProcessor`（ERNIE 4.5 VL）
- `image_processors/paddleocr.py`：`PaddleOCRImageProcessor`
- `image_processors/qwen.py`、`qwen3.py`：Qwen / Qwen3 图像处理器
- `ernie_vl.py`：`ErnieVLProcessor` 多模态处理器（3D 位置 ID、视频处理）
- `paddleocr_vl.py`：`PaddleOCRVLProcessor`
- `qwen_vl.py`、`qwen3_vl.py`：`QwenVLProcessor` / `Qwen3VLProcessor`
- `mm_processor.py`：`MMProcessor` 基类
- `fastdeploy/input/preprocess.py`、`processor.py`：适配新模块结构
- `tests/entrypoints/test_chat.py`、`test_generation.py`：更新 entrypoints 测试
- `tests/input/test_preprocess.py`：更新 input 处理测试
## Usage or Command
N/A
## Accuracy Tests
N/A
## Checklist
- [x] Add at least a tag in the PR title.
- Tag list: [`[FDConfig]`,`[APIServer]`,`[Engine]`, `[Scheduler]`, `[PD Disaggregation]`, `[Executor]`, `[Graph Optimization]`, `[Speculative Decoding]`, `[RL]`, `[Models]`, `[Quantization]`, `[Loader]`, `[OP]`, `[KVCache]`, `[DataProcessor]`, `[BugFix]`, `[Docs]`, `[CI]`, `[Optimization]`, `[Feature]`, `[Benchmark]`, `[Others]`, `[XPU]`, `[HPU]`, `[GCU]`, `[DCU]`, `[Iluvatar]`, `[Metax]`]
- You can add new tags based on the PR content, but the semantics must be clear.
- [ ] Format your code, run `pre-commit` before commit.
- [x] Add unit tests. Please write the reason in this PR if no unit tests.
- [ ] Provide accuracy results.
- [ ] If the current PR is submitting to the `release` branch, make sure the PR has been submitted to the `develop` branch, then cherry-pick it to the `release` branch with the `[Cherry-Pick]` PR tag.问题
| 级别 | 文件 | 概述 |
|---|---|---|
| 📝 PR 规范 | 标题 / body | Tag 后缺少空格；Motivation / Modifications / Usage / Accuracy Tests 所有 section 内容为空 |
| 🟡 建议 | fastdeploy/entrypoints/llm.py（_add_request） |
prompts[i] 为空列表 [] 时，len > 0 守卫使两个 list 分支均不命中，isinstance(dict) 也不匹配，tasks 变量处于未定义状态（潜在 NameError 或复用上轮循环旧值） |
| ❓ 疑问 | fastdeploy/input/multimodal/image_processors/ernie.py（set_pixels） |
assert isinstance(min_pixels, int) and min_pixels >= 0 在 python -O 优化模式下会被静默跳过，导致非法参数流入后续计算；建议改为 raise ValueError |
🟡 建议修复（
llm.py_add_request）在现有所有类型分支末尾增加兜底，避免
tasks未定义：else: raise ValueError( f"Unsupported prompt type at index {i}: {type(prompts[i])}, value={prompts[i]!r}" )
❓ 疑问（
ernie.pyset_pixels）# 当前（在 python -O 下会被跳过） assert isinstance(min_pixels, int) and min_pixels >= 0, "min_pixels must be positive int" # 建议改为 if not (isinstance(min_pixels, int) and min_pixels >= 0): raise ValueError("min_pixels must be a non-negative int")
总体评价
PR 整合了多个多模态处理器，整体架构清晰、抽象合理，新增测试覆盖基本功能。主要待改进点：PR 描述为空需补充、_add_request 存在空列表未处理的边界情况、ernie.py 中 assert 用于运行时校验违反最佳实践。
3 participants
Motivation
Modifications
Usage or Command
Accuracy Tests
Checklist
[FDConfig],[APIServer],[Engine],[Scheduler],[PD Disaggregation],[Executor],[Graph Optimization],[Speculative Decoding],[RL],[Models],[Quantization],[Loader],[OP],[KVCache],[DataProcessor],[BugFix],[Docs],[CI],[Optimization],[Feature],[Benchmark],[Others],[XPU],[HPU],[GCU],[DCU],[Iluvatar],[Metax]]pre-commitbefore commit.releasebranch, make sure the PR has been submitted to thedevelopbranch, then cherry-pick it to thereleasebranch with the[Cherry-Pick]PR tag.