refactor: Refactor TRPG NER model SDK: restructure codebase into base_model_trpgner package, implement training and inference modules, and add model download functionality. Remove legacy training and utils modules. Enhance documentation and examples for better usability.

8 files changed, 130 insertions, 64 deletions

diff --git a/.github/workflows/publish.yml b/.github/workflows/publish.yml
index 2004132..c11cd0e 100644
--- a/.github/workflows/publish.yml
+++ b/.github/workflows/publish.yml
@@ -4,6 +4,8 @@ on:
   push:
     tags:
       - 'v*.*.*'
+    branches:
+      - main
   workflow_dispatch:
     inputs:
       create_test:
@@ -11,6 +13,10 @@ on:
         required: false
         default: false
         type: boolean
+      tag_name:
+        description: 'Tag name to release (e.g., v1.0.0). Use this to re-run release with fixed code.'
+        required: true
+        default: 'v1.0.0'
 
 permissions:
   contents: write
@@ -25,14 +31,25 @@ jobs:
       version: ${{ steps.version.outputs.version }}
 
     steps:
-      - uses: actions/checkout@v6
+      - name: Checkout code
+        uses: actions/checkout@v6
         with:
           fetch-depth: 0
+          # 最佳实践: Tag 触发时使用 main 分支代码,而不是 Tag 快照
+          # 这样修复代码后可以手动重新触发 workflow 而无需重新打 tag
+          ref: main
 
       - name: Extract version from tag
         id: version
         run: |
-          VERSION=${{ github.ref_name }}
+          # Tag 触发时使用 github.ref_name, 手动触发时使用 inputs.tag_name
+          if [[ "${{ github.event_name }}" == "workflow_dispatch" ]]; then
+            VERSION="${{ github.event.inputs.tag_name }}"
+            echo "🔧 Manual trigger mode: Using tag $VERSION from input"
+          else
+            VERSION="${{ github.ref_name }}"
+            echo "🏷️  Tag trigger mode: Using tag $VERSION from GitHub"
+          fi
           echo "version=${VERSION#v}" >> $GITHUB_OUTPUT
           echo "Tag: $VERSION"
           echo "Version: ${VERSION#v}"
@@ -44,6 +61,7 @@ jobs:
 
       - name: Build with uv
         run: |
+          uv sync
           uv build
 
       - name: Check distribution
@@ -76,7 +94,7 @@ jobs:
 
     environment:
       name: test-pypi
-      url: https://test.pypi.org/p/base-model
+      url: https://test.pypi.org/p/base-model-trpgner
 
     permissions:
       id-token: write
@@ -95,11 +113,12 @@ jobs:
     name: Publish to PyPI
     needs: build
     runs-on: ubuntu-latest
-    if: github.event_name == 'push' && startsWith(github.ref, 'refs/tags/')
+    # Tag 推送时自动发布, 或手动触发且未指定测试模式时发布
+    if: (github.event_name == 'push' && startsWith(github.ref, 'refs/tags/')) || (github.event_name == 'workflow_dispatch' && inputs.create_test != true)
 
     environment:
       name: pypi
-      url: https://pypi.org/p/base-model
+      url: https://pypi.org/p/base-model-trpgner
 
     permissions:
       id-token: write
@@ -118,22 +137,27 @@ jobs:
     name: Create GitHub Release with ONNX
     needs: [build, publish-pypi]
     runs-on: ubuntu-latest
-    if: github.event_name == 'push' && startsWith(github.ref, 'refs/tags/')
+    # Tag 推送时自动创建, 或手动触发且未指定测试模式时创建
+    if: (github.event_name == 'push' && startsWith(github.ref, 'refs/tags/')) || (github.event_name == 'workflow_dispatch' && inputs.create_test != true)
 
     permissions:
       contents: write
 
     steps:
-      - uses: actions/checkout@v6
+      - name: Checkout code
+        uses: actions/checkout@v6
         with:
           fetch-depth: 0
+          # 使用 main 分支代码生成 CHANGELOG
+          ref: main
 
       - name: Generate CHANGELOG
         id: changelog
         uses: requarks/changelog-action@v1
         with:
           token: ${{ github.token }}
-          tag: ${{ github.ref_name }}
+          # 根据触发类型选择正确的 tag
+          tag: ${{ github.event_name == 'workflow_dispatch' && inputs.tag_name || github.ref_name }}
           includeInvalidCommits: true
           changelogFilePath: CHANGELOG.md
           writeToFile: true
@@ -150,29 +174,34 @@ jobs:
         env:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         run: |
-          VERSION="${{ github.ref_name }}"
+          # 根据触发类型选择正确的 tag
+          if [[ "${{ github.event_name }}" == "workflow_dispatch" ]]; then
+            VERSION="${{ github.event.inputs.tag_name }}"
+          else
+            VERSION="${{ github.ref_name }}"
+          fi
 
           cat > release_notes.md << 'EOF'
           ## 📦 安装
 
           ### pip 安装
           ```bash
-          pip install base-model
+          pip install base-model-trpgner
           ```
 
           ### 使用 uv（推荐）
           ```bash
-          uv pip install base-model
+          uv pip install base-model-trpgner
           ```
 
           ### 训练模式
           ```bash
-          pip install base-model[train]
+          pip install base-model-trpgner[train]
           ```
 
           ## 🚀 快速开始
           ```python
-          from basemodel import TRPGParser
+          from basemodeltrpgner import TRPGParser
 
           parser = TRPGParser()
           result = parser.parse("风雨 2024-06-08 21:44:59 剧烈的疼痛...")
@@ -184,18 +213,43 @@ jobs:
           ${{ steps.changelog.outputs.changes }}
           EOF
 
-          # 上传 Python 包和 ONNX 模型
-          gh release create "${VERSION}" \
-            --repo "${{ github.repository }}" \
-            --notes-file release_notes.md \
-            --title "🚀 ${VERSION}" \
-            artifacts/dist/* \
-            artifacts/onnx-artifact/* || true
+          # 检查 release 是否已存在
+          if gh release view "${VERSION}" --repo "${{ github.repository }}" >/dev/null 2>&1; then
+            echo "📝 Release ${VERSION} 已存在，更新 release 资源..."
+
+            # 删除旧的 assets 以便上传新的
+            ASSETS=$(gh release view "${VERSION}" --repo "${{ github.repository }}" --json assets -q '.assets[].name')
+            for asset in $ASSETS; do
+              echo "  删除旧资源: $asset"
+              gh release delete-asset "${VERSION}" "$asset" --repo "${{ github.repository }}" || true
+            done
+
+            # 更新 release notes 和 assets
+            gh release edit "${VERSION}" \
+              --repo "${{ github.repository }}" \
+              --notes-file release_notes.md \
+              --title "🚀 ${VERSION}"
+
+            # 上传新的资源
+            gh release upload "${VERSION}" \
+              artifacts/dist/* \
+              artifacts/onnx-artifact/* \
+              --repo "${{ github.repository }}" --clobber
+          else
+            echo "✨ 创建新 release ${VERSION}..."
+            gh release create "${VERSION}" \
+              --repo "${{ github.repository }}" \
+              --notes-file release_notes.md \
+              --title "🚀 ${VERSION}" \
+              artifacts/dist/* \
+              artifacts/onnx-artifact/*
+          fi
 
       - name: Commit CHANGELOG.md
         if: hashFiles('CHANGELOG.md') != ''
         uses: stefanzweifel/git-auto-commit-action@v7
         with:
           branch: main
-          commit_message: "docs: update CHANGELOG.md for ${{ github.ref_name }} [skip ci]"
+          # 根据触发类型选择正确的 tag
+          commit_message: "docs: update CHANGELOG.md for ${{ github.event_name == 'workflow_dispatch' && inputs.tag_name || github.ref_name }} [skip ci]"
           file_pattern: CHANGELOG.md
diff --git a/pyproject.toml b/pyproject.toml
index d9b6d38..a1dcc20 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -6,13 +6,19 @@ build-backend = "hatchling.build"
 name = "base-model-trpgner"
 version = "0.1.1"
 description = "HydroRoll TRPG NER 模型 - 桌上角色扮演游戏日志命名实体识别"
-authors = [
-    { name = "HsiangNianian", email = "leader@hydroroll.team" }
-]
+authors = [{ name = "HsiangNianian", email = "leader@hydroroll.team" }]
 readme = "README.md"
 requires-python = ">=3.12"
 license = { text = "AFL-3.0" }
-keywords = ["hydroroll", "trpg", "nlp", "ner", "chinese", "onnx", "robot framework"]
+keywords = [
+    "hydroroll",
+    "trpg",
+    "nlp",
+    "ner",
+    "chinese",
+    "onnx",
+    "robot framework",
+]
 classifiers = [
     "Development Status :: 3 - Alpha",
     "Intended Audience :: Developers",
@@ -24,11 +30,7 @@ classifiers = [
     "Topic :: Text Processing :: Linguistic",
 ]
 
-dependencies = [
-    "numpy>=1.24.0",
-    "onnxruntime>=1.23.2",
-    "transformers>=4.57.3",
-]
+dependencies = ["numpy>=1.24.0", "onnxruntime>=1.23.2", "transformers>=4.57.3"]
 
 [project.optional-dependencies]
 train = [
@@ -43,14 +45,8 @@ dev = [
     "black>=24.0.0",
     "ruff>=0.1.0",
 ]
-webui = [
-    "base-model-trpgner[train]",
-    "gradio>=6.2.0",
-    "scikit-learn>=1.4.0",
-]
-all = [
-    "base-model-trpgner[train,webui,dev]",
-]
+webui = ["base-model-trpgner[train]", "gradio>=6.2.0", "scikit-learn>=1.4.0"]
+all = ["base-model-trpgner[train,webui,dev]"]
 
 [project.urls]
 Homepage = "https://ailab.hydroroll.team/"
@@ -59,10 +55,9 @@ Documentation = "https://ailab.hydroroll.team/"
 Issues = "https://github.com/HydroRoll-Team/base-model/issues"
 
 [tool.hatch.build.targets.wheel]
-packages = ["src/basemodel"]
-# 只包含 ONNX 推理需要的文件（约 41MB）
+packages = ["src/base_model_trpgner"]
 artifacts = [
-    "src/basemodel/**/*.py",
+    "src/base_model_trpgner/**/*.py",
     "models/trpg-final/model.onnx",
     "models/trpg-final/model.onnx.data",
     "models/trpg-final/config.json",
@@ -73,23 +68,39 @@ artifacts = [
 ]
 
 [tool.hatch.build.targets.wheel.shared-data]
-"models/trpg-final" = "basemodel/models/trpg-final"
+"models/trpg-final" = "base_model_trpgner/models/trpg-final"
 
 [tool.hatch.build.targets.sdist]
-include = [
-    "/src",
-    "/models/trpg-final",
-    "/README.md",
-    "/COPYING",
-]
+include = ["/src", "/models/trpg-final", "/README.md", "/COPYING"]
 
 [tool.ruff]
 exclude = [
-    ".bzr", ".direnv", ".eggs", ".git", ".git-rewrite", ".hg",
-    ".ipynb_checkpoints", ".mypy_cache", ".nox", ".pants.d",
-    ".pyenv", ".pytest_cache", ".pytype", ".ruff_cache", ".svn",
-    ".tox", ".venv", ".vscode", "__pypackages__", "_build",
-    "buck-out", "build", "dist", "node_modules", "site-packages", "venv",
+    ".bzr",
+    ".direnv",
+    ".eggs",
+    ".git",
+    ".git-rewrite",
+    ".hg",
+    ".ipynb_checkpoints",
+    ".mypy_cache",
+    ".nox",
+    ".pants.d",
+    ".pyenv",
+    ".pytest_cache",
+    ".pytype",
+    ".ruff_cache",
+    ".svn",
+    ".tox",
+    ".venv",
+    ".vscode",
+    "__pypackages__",
+    "_build",
+    "buck-out",
+    "build",
+    "dist",
+    "node_modules",
+    "site-packages",
+    "venv",
 ]
 line-length = 100
 indent-width = 4
@@ -112,4 +123,4 @@ line-length = 100
 target-version = ["py312"]
 
 [tool.uv]
-dev-dependencies = []
\ No newline at end of file
+dev-dependencies = []
diff --git a/src/basemodel/__init__.py b/src/base_model_trpgner/__init__.py
index 7287df4..9796c83 100644
--- a/src/basemodel/__init__.py
+++ b/src/base_model_trpgner/__init__.py
@@ -1,28 +1,28 @@
 """
-base-model - HydroRoll TRPG NER 模型 SDK
+base-model-trpgner - HydroRoll TRPG NER 模型 SDK
 
 这是一个用于 TRPG（桌上角色扮演游戏）日志命名实体识别的 Python SDK。
 
 基本用法:
-    >>> from basemodel import TRPGParser
+    >>> from base_model_trpgner import TRPGParser
     >>> parser = TRPGParser()
     >>> result = parser.parse("风雨 2024-06-08 21:44:59 剧烈的疼痛...")
     >>> print(result)
     {'metadata': {'speaker': '风雨', 'timestamp': '2024-06-08 21:44:59'}, 'content': [...]}
 
 训练功能（需要额外安装）:
-    >>> pip install base-model[train]
-    >>> from basemodel.training import train_ner_model
+    >>> pip install base-model-trpgner[train]
+    >>> from base_model_trpgner.training import train_ner_model
     >>> train_ner_model(conll_data="./data", output_dir="./model")
 """
 
-from basemodel.inference import TRPGParser, parse_line, parse_lines
+from base_model_trpgner.inference import TRPGParser, parse_line, parse_lines
 
 try:
     from importlib.metadata import version
-    __version__ = version("base-model")
+    __version__ = version("base_model_trpgner")
 except Exception:
-    __version__ = "0.1.0.dev"
+    __version__ = "0.1.1.dev"
 
 __all__ = [
     "__version__",
diff --git a/src/basemodel/download_model.py b/src/base_model_trpgner/download_model.py
index 2d65099..2d65099 100644
--- a/src/basemodel/download_model.py
+++ b/src/base_model_trpgner/download_model.py
diff --git a/src/basemodel/inference/__init__.py b/src/base_model_trpgner/inference/__init__.py
index 93a185f..93a185f 100644
--- a/src/basemodel/inference/__init__.py
+++ b/src/base_model_trpgner/inference/__init__.py
diff --git a/src/basemodel/training/__init__.py b/src/base_model_trpgner/training/__init__.py
index 5671c42..ccf3c03 100644
--- a/src/basemodel/training/__init__.py
+++ b/src/base_model_trpgner/training/__init__.py
@@ -36,7 +36,7 @@ def train_ner_model(
         resume_from_checkpoint: 恢复检查点路径
 
     Examples:
-        >>> from basemodel.training import train_ner_model
+        >>> from basemodeltrpgner.training import train_ner_model
         >>> train_ner_model(
         ...     conll_data="./data",
         ...     output_dir="./my_model",
@@ -59,7 +59,7 @@ def train_ner_model(
         ) from e
 
     # 导入数据处理函数
-    from basemodel.utils.conll import load_conll_dataset, tokenize_and_align_labels
+    from base_model_trpgner.utils.conll import load_conll_dataset, tokenize_and_align_labels
 
     print(f"🚀 Starting training...")
 
diff --git a/src/basemodel/utils/__init__.py b/src/base_model_trpgner/utils/__init__.py
index 12a3ef4..12a3ef4 100644
--- a/src/basemodel/utils/__init__.py
+++ b/src/base_model_trpgner/utils/__init__.py
diff --git a/utils/word_conll_to_char_conll.py b/utils/word_conll_to_char_conll.py
index e52405f..9f4217a 100644
--- a/utils/word_conll_to_char_conll.py
+++ b/utils/word_conll_to_char_conll.py
@@ -52,4 +52,5 @@ if __name__ == "__main__":
     with open(output_fp, "w", encoding="utf-8") as f:
         f.write("\n".join(char_conll_lines) + "\n")
 
-    print(f"Converted {input_fp} to character-level CoNLL format at {output_fp}")
\ No newline at end of file
+    print(f"Converted {input_fp} to character-level CoNLL format at {output_fp}")
+