deepdoc-lib 0.2.4__tar.gz → 0.2.5__tar.gz

{deepdoc_lib-0.2.4 → deepdoc_lib-0.2.5}/.env.example

@@ -29,7 +29,7 @@ DEEPDOC_VISION_BASE_URL=
 # openai    - OpenAI GPT-4 Vision
 # qwen      - 阿里云通义千问
 # zhipu     - 智谱AI
-# gemini    - Google Gemini
+# gemini    - Google Gemini (需安装: pip install deepdoc-lib[gemini])
 # ollama    - 本地 Ollama 服务
 # anthropic - Anthropic Claude (需额外配置)
 DEEPDOC_VISION_PROVIDER=qwen

{deepdoc_lib-0.2.4 → deepdoc_lib-0.2.5}/.github/workflows/ci.yml

@@ -22,6 +22,8 @@ jobs:
                   python-version: ${{ matrix.python-version }}
                   cache: "pip"
 
+            # Install without extras so the suite proves the default install
+            # imports and works without optional SDKs such as google-genai.
             - name: Install dependencies
               run: |
                   python -m pip install --upgrade pip
@@ -33,6 +35,12 @@ jobs:
                   python -m build --wheel
                   python -m pip install --dry-run --ignore-installed dist/*.whl 'PyJWT[crypto]>=2.15.1,<3'
 
+            - name: Resolve the built wheel with protobuf 6
+              run: |
+                  wheel=$(ls dist/*.whl)
+                  python -m pip install --dry-run --ignore-installed "$wheel" 'protobuf>=6.31,<7'
+                  python -m pip install --dry-run --ignore-installed "${wheel}[gemini]" 'protobuf>=6.31,<7'
+
             # Preprovision immutable official data instead of asking the NLTK
             # downloader to traverse the runner's proxy. Do not disable pathsec.
             - name: Check out pinned NLTK test resources
@@ -67,6 +75,11 @@ jobs:
                   NLTK_DATA: ${{ runner.temp }}/deepdoc-nltk-data
               run: python -m pytest tests
 
+            - name: Test Gemini with the latest google-genai
+              run: |
+                  python -m pip install -e ".[gemini]"
+                  python -m pytest tests/test_gemini_sdk_compatibility.py
+
     minimum-sdk:
         name: Minimum SDK contract (3.10)
         runs-on: ubuntu-latest
@@ -79,14 +92,14 @@ jobs:
                   python-version: "3.10"
                   cache: "pip"
 
-            - name: Build and resolve the wheel with the minimum SDK and current PyJWT
+            - name: Build and resolve the wheel with the minimum SDKs and current PyJWT
               run: |
                   python -m pip install --upgrade pip build
                   python -m build --wheel
-                  python -m pip install --dry-run --ignore-installed dist/*.whl 'zai-sdk==0.0.4' 'PyJWT[crypto]>=2.15.1,<3'
+                  python -m pip install --dry-run --ignore-installed dist/*.whl 'zai-sdk==0.0.4' 'google-genai==1.0.0' 'PyJWT[crypto]>=2.15.1,<3'
 
-            - name: Install and test the minimum SDK contract
+            - name: Install and test the minimum SDK contracts
               run: |
-                  python -m pip install dist/*.whl 'zai-sdk==0.0.4' 'PyJWT[crypto]>=2.15.1,<3' pytest
+                  python -m pip install dist/*.whl 'zai-sdk==0.0.4' 'google-genai==1.0.0' 'PyJWT[crypto]>=2.15.1,<3' pytest
                   python -m pip check
-                  python -m pytest tests/test_zhipu_sdk_compatibility.py
+                  python -m pytest tests/test_zhipu_sdk_compatibility.py tests/test_gemini_sdk_compatibility.py

{deepdoc_lib-0.2.4 → deepdoc_lib-0.2.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: deepdoc_lib
-Version: 0.2.4
+Version: 0.2.5
 Summary: [RAGFlow](https://ragflow.io/) is an open-source RAG (Retrieval-Augmented Generation) engine based on deep document understanding. It offers a streamlined RAG workflow for businesses of any scale, combining LLM (Large Language Models) to provide truthful question-answering capabilities, backed by well-founded citations from various complex formatted data.
 Author-email: Zhichang Yu <yuzhichang@gmail.com>
 Requires-Python: <3.14,>=3.10
@@ -43,13 +43,14 @@ Requires-Dist: tiktoken>=0.7.0
 Requires-Dist: xgboost<3.0.0,>=2.1.0
 Requires-Dist: xpinyin>=0.7.6
 Requires-Dist: zai-sdk>=0.0.4
-Requires-Dist: google-generativeai<0.9.0,>=0.8.1
 Requires-Dist: trio>=0.29.0
 Requires-Dist: setuptools<76.0.0,>=75.2.0
 Requires-Dist: huggingface-hub<1.0.0,>=0.25.0
 Provides-Extra: gpu
 Requires-Dist: onnxruntime-gpu>=1.19.2; (sys_platform != "darwin" and platform_machine == "x86_64") and extra == "gpu"
 Requires-Dist: torch; extra == "gpu"
+Provides-Extra: gemini
+Requires-Dist: google-genai>=1.0.0; extra == "gemini"
 Dynamic: license-file
 
 # Deepdoc
@@ -68,6 +69,12 @@ GPU (Linux x86_64 only):
 pip install deepdoc-lib[gpu]
 ```
 
+Gemini vision provider (optional, installs the `google-genai` SDK):
+
+```bash
+pip install deepdoc-lib[gemini]
+```
+
 Note: `onnxruntime` (CPU) and `onnxruntime-gpu` should not be installed together. If you're switching an existing environment to GPU, uninstall CPU ORT first:
 
 ```bash
@@ -222,6 +229,8 @@ vision_model = create_vision_model("qwen")
 
 Supported providers: ["openai", "qwen", "zhipu", "ollama", "gemini", "anthropic"]
 
+The `gemini` provider needs the `gemini` extra: `pip install deepdoc-lib[gemini]`.
+
 - Use Configuration File
 
 Create `deepdoc_config.yaml`:

{deepdoc_lib-0.2.4 → deepdoc_lib-0.2.5}/README.md

@@ -14,6 +14,12 @@ GPU (Linux x86_64 only):
 pip install deepdoc-lib[gpu]
 ```
 
+Gemini vision provider (optional, installs the `google-genai` SDK):
+
+```bash
+pip install deepdoc-lib[gemini]
+```
+
 Note: `onnxruntime` (CPU) and `onnxruntime-gpu` should not be installed together. If you're switching an existing environment to GPU, uninstall CPU ORT first:
 
 ```bash
@@ -168,6 +174,8 @@ vision_model = create_vision_model("qwen")
 
 Supported providers: ["openai", "qwen", "zhipu", "ollama", "gemini", "anthropic"]
 
+The `gemini` provider needs the `gemini` extra: `pip install deepdoc-lib[gemini]`.
+
 - Use Configuration File
 
 Create `deepdoc_config.yaml`:

{deepdoc_lib-0.2.4 → deepdoc_lib-0.2.5}/deepdoc/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '0.2.4'
-__version_tuple__ = version_tuple = (0, 2, 4)
+__version__ = version = '0.2.5'
+__version_tuple__ = version_tuple = (0, 2, 5)
 
-__commit_id__ = commit_id = 'g7403fae36'
+__commit_id__ = commit_id = 'g8ef70a65b'

{deepdoc_lib-0.2.4 → deepdoc_lib-0.2.5}/deepdoc/depend/simple_cv_model.py

@@ -306,9 +306,16 @@ class GeminiCV(Base):
 
     def __init__(self, key, model_name="gemini-pro-vision", lang="Chinese", **kwargs):
         super().__init__(key, model_name, lang)
-        import google.generativeai as genai
-        genai.configure(api_key=key)
-        self.model = genai.GenerativeModel(model_name)
+        # The legacy google-generativeai SDK caps protobuf below 6, so Gemini
+        # uses the maintained google-genai SDK as an optional extra instead.
+        try:
+            from google import genai
+        except ImportError as e:
+            raise ImportError(
+                "Gemini vision requires the google-genai package. "
+                "Install it with: pip install 'deepdoc-lib[gemini]'"
+            ) from e
+        self.client = genai.Client(api_key=key)
 
     def describe_with_prompt(self, image, prompt=None):
         """使用Google Gemini模型描述图片"""
@@ -323,11 +330,19 @@ class GeminiCV(Base):
             
             pil_image = Image.open(BytesIO(img_data))
             
-            response = self.model.generate_content([
-                prompt or self.prompt(image),
-                pil_image
-            ])
+            response = self.client.models.generate_content(
+                model=self.model_name,
+                contents=[prompt or self.prompt(image), pil_image],
+            )
 
+            # google-genai returns None (the legacy SDK raised) when the
+            # response has no text, e.g. a safety block; keep the error path.
+            # A block can be reported on the prompt or on the candidate.
+            if response.text is None:
+                candidates = response.candidates or []
+                finish_reason = candidates[0].finish_reason if candidates else None
+                finish_reason = getattr(finish_reason, "name", finish_reason)
+                raise ValueError(f"Gemini returned no text (finish_reason={finish_reason}, prompt_feedback={response.prompt_feedback})")
             return response.text.strip()
         except Exception as e:
             return "**ERROR**: " + str(e)

{deepdoc_lib-0.2.4 → deepdoc_lib-0.2.5}/deepdoc_lib.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: deepdoc_lib
-Version: 0.2.4
+Version: 0.2.5
 Summary: [RAGFlow](https://ragflow.io/) is an open-source RAG (Retrieval-Augmented Generation) engine based on deep document understanding. It offers a streamlined RAG workflow for businesses of any scale, combining LLM (Large Language Models) to provide truthful question-answering capabilities, backed by well-founded citations from various complex formatted data.
 Author-email: Zhichang Yu <yuzhichang@gmail.com>
 Requires-Python: <3.14,>=3.10
@@ -43,13 +43,14 @@ Requires-Dist: tiktoken>=0.7.0
 Requires-Dist: xgboost<3.0.0,>=2.1.0
 Requires-Dist: xpinyin>=0.7.6
 Requires-Dist: zai-sdk>=0.0.4
-Requires-Dist: google-generativeai<0.9.0,>=0.8.1
 Requires-Dist: trio>=0.29.0
 Requires-Dist: setuptools<76.0.0,>=75.2.0
 Requires-Dist: huggingface-hub<1.0.0,>=0.25.0
 Provides-Extra: gpu
 Requires-Dist: onnxruntime-gpu>=1.19.2; (sys_platform != "darwin" and platform_machine == "x86_64") and extra == "gpu"
 Requires-Dist: torch; extra == "gpu"
+Provides-Extra: gemini
+Requires-Dist: google-genai>=1.0.0; extra == "gemini"
 Dynamic: license-file
 
 # Deepdoc
@@ -68,6 +69,12 @@ GPU (Linux x86_64 only):
 pip install deepdoc-lib[gpu]
 ```
 
+Gemini vision provider (optional, installs the `google-genai` SDK):
+
+```bash
+pip install deepdoc-lib[gemini]
+```
+
 Note: `onnxruntime` (CPU) and `onnxruntime-gpu` should not be installed together. If you're switching an existing environment to GPU, uninstall CPU ORT first:
 
 ```bash
@@ -222,6 +229,8 @@ vision_model = create_vision_model("qwen")
 
 Supported providers: ["openai", "qwen", "zhipu", "ollama", "gemini", "anthropic"]
 
+The `gemini` provider needs the `gemini` extra: `pip install deepdoc-lib[gemini]`.
+
 - Use Configuration File
 
 Create `deepdoc_config.yaml`:

{deepdoc_lib-0.2.4 → deepdoc_lib-0.2.5}/deepdoc_lib.egg-info/requires.txt

@@ -35,11 +35,13 @@ tiktoken>=0.7.0
 xgboost<3.0.0,>=2.1.0
 xpinyin>=0.7.6
 zai-sdk>=0.0.4
-google-generativeai<0.9.0,>=0.8.1
 trio>=0.29.0
 setuptools<76.0.0,>=75.2.0
 huggingface-hub<1.0.0,>=0.25.0
 
+[gemini]
+google-genai>=1.0.0
+
 [gpu]
 torch
 

{deepdoc_lib-0.2.4 → deepdoc_lib-0.2.5}/deepdoc_lib.egg-info/scm_file_list.json

@@ -102,6 +102,7 @@
     "tests/test_data/test5.csv",
     "tests/test_data/test_gbk.txt",
     "tests/test_data/test_html.csv",
+    "tests/test_gemini_sdk_compatibility.py",
     "tests/test_model_store.py",
     "tests/test_parser.py",
     "tests/test_real_files.py",

deepdoc_lib-0.2.5/deepdoc_lib.egg-info/scm_version.json

@@ -0,0 +1,8 @@
+{
+  "tag": "0.2.5",
+  "distance": 0,
+  "node": "g8ef70a65b970bf74e5fffbd15ba7d111fd5a24f5",
+  "dirty": false,
+  "branch": "HEAD",
+  "node_date": "2026-11-09"
+}

{deepdoc_lib-0.2.4 → deepdoc_lib-0.2.5}/pyproject.toml

@@ -48,7 +48,6 @@ dependencies = [
     "xgboost>=2.1.0,<3.0.0",
     "xpinyin>=0.7.6",
     "zai-sdk>=0.0.4",
-    "google-generativeai>=0.8.1,<0.9.0",
     "trio>=0.29.0",
     "setuptools>=75.2.0,<76.0.0",
     "huggingface-hub>=0.25.0,<1.0.0",
@@ -59,6 +58,7 @@ gpu = [
     "onnxruntime-gpu>=1.19.2; sys_platform != 'darwin' and platform_machine == 'x86_64'",
     "torch",
 ]
+gemini = ["google-genai>=1.0.0"]
 
 [project.scripts]
 deepdoc-download-models = "deepdoc.download_models:main"

deepdoc_lib-0.2.4/deepdoc_lib.egg-info/scm_version.json

@@ -1,8 +0,0 @@
-{
-  "tag": "0.2.4",
-  "distance": 0,
-  "node": "g7403fae3679433d12d3c7cc805d8792247f99747",
-  "dirty": false,
-  "branch": "HEAD",
-  "node_date": "2026-10-17"
-}