vlmparse 0.1.4__py3-none-any.whl → 0.1.6__py3-none-any.whl

vlmparse/data_model/document.py

@@ -41,6 +41,10 @@ class Page(VLMParseBaseModel):
     buffer_image: Optional[Image.Image | str | dict] = None
     latency: Optional[float] = None
     """Time taken to process the page in seconds."""
+    prompt_tokens: Optional[int] = None
+    completion_tokens: Optional[int] = None
+    """Include reasoning tokens"""
+    reasoning_tokens: Optional[int] = None
 
     @property
     def image(self):
@@ -66,7 +70,7 @@ class Page(VLMParseBaseModel):
 
         image = self.image
 
-        if layout:
+        if layout and image is not None:
             if self.items is None:
                 return image
             items = self.items
@@ -85,6 +89,9 @@ class Page(VLMParseBaseModel):
                 )
         return image
 
+    def to_markdown(self, **kwargs):
+        return self.text if self.text is not None else ""
+
 
 class Document(VLMParseBaseModel):
     file_path: str
@@ -104,6 +111,9 @@ class Document(VLMParseBaseModel):
             page.error is not None for page in self.pages
         )
 
+    def to_markdown(self, **kwargs):
+        return "\n\n".join([page.to_markdown(**kwargs) for page in self.pages])
+
     def to_zip(
         self,
         file_path,

vlmparse/registries.py

@@ -77,9 +77,7 @@ class ConverterConfigRegistry:
         """Register a config factory for a model name."""
         self._registry[model_name] = config_factory
 
-    def get(
-        self, model_name: str, uri: str | None = None
-    ) -> OpenAIConverterConfig | None:
+    def get(self, model_name: str, uri: str | None = None) -> OpenAIConverterConfig:
         """Get config for a model name. Returns default if not registered."""
         if model_name in self._registry:
             return self._registry[model_name](uri=uri)
@@ -108,6 +106,7 @@ for gemini_model in [
     "gemini-2.5-flash",
     "gemini-2.5-flash-lite",
     "gemini-3-pro-preview",
+    "gemini-3-flash-preview",
 ]:
     converter_config_registry.register(
         gemini_model,
@@ -120,12 +119,9 @@ for gemini_model in [
         ),
     )
 for openai_model in [
-    "gpt-5.1",
-    "gpt-5.1-mini",
-    "gpt-5.1-nano",
+    "gpt-5.2",
     "gpt-5",
     "gpt-5-mini",
-    "gpt-5-nano",
 ]:
     converter_config_registry.register(
         openai_model,

vlmparse/servers/docker_server.py

@@ -47,6 +47,20 @@ class DockerServerConfig(BaseModel):
         """Build command for container. Override in subclasses for specific logic."""
         return self.command_args if self.command_args else None
 
+    def update_command_args(
+        self,
+        vllm_kwargs: dict | None = None,
+        forget_predefined_vllm_kwargs: bool = False,
+    ) -> list[str]:
+        if vllm_kwargs is not None:
+            new_kwargs = [f"--{k}={v}" for k, v in vllm_kwargs.items()]
+            if forget_predefined_vllm_kwargs:
+                self.command_args = new_kwargs
+            else:
+                self.command_args.extend(new_kwargs)
+
+        return self.command_args
+
     def get_volumes(self) -> dict | None:
         """Setup volumes for container. Override in subclasses for specific logic."""
         return self.volumes
@@ -78,7 +92,7 @@ class VLLMDockerServerConfig(DockerServerConfig):
         from vlmparse.clients.openai_converter import LLMParams
 
         return LLMParams(
-            base_url=f"http://localhost:{self.docker_port}/v1",
+            base_url=f"http://localhost:{self.docker_port}{self.get_base_url_suffix()}",
             model_name=self.default_model_name,
         )
 
@@ -144,7 +158,7 @@ class ConverterServer:
         """Start the Docker server."""
         if self._server_context is not None:
             logger.warning("Server already started")
-            return self.base_url
+            return self.base_url, self._container
 
         # Use the generic docker_server for all server types
         self._server_context = docker_server(config=self.config, cleanup=self.auto_stop)

vlmparse/servers/utils.py

@@ -3,9 +3,8 @@ import time
 from contextlib import contextmanager
 from pathlib import Path
 
-from loguru import logger
-
 import docker
+from loguru import logger
 
 
 def _ensure_image_exists(
@@ -230,6 +229,8 @@ def get_model_from_uri(uri: str) -> str:
     for container in containers:
         c_uri = container.labels.get("vlmparse_uri")
         c_model = container.labels.get("vlmparse_model_name")
+        if c_uri is not None:
+            c_uri = c_uri.replace("localhost", "0.0.0.0")
 
         # Check if user URI matches container URI (ignoring /v1 suffix if missing)
         if c_uri and (

{vlmparse-0.1.4.dist-info → vlmparse-0.1.6.dist-info}/METADATA

@@ -1,7 +1,7 @@
 Metadata-Version: 2.4
 Name: vlmparse
-Version: 0.1.4
-Requires-Python: >=3.12.0
+Version: 0.1.6
+Requires-Python: >=3.11.0
 Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: devtools>=0.12.2
@@ -72,6 +72,19 @@ Supported Converters:
 
 ## Installation
 
+Simplest solution with only the cli: 
+
+```bash
+uv tool install vlmparse
+```
+
+If you want to run the granite-docling model or use the streamlit viewing app:
+
+```bash
+uv tool install vlmparse[docling_core,st_app]
+```
+
+If you prefer cloning the repository and using the local version:
 ```bash
 uv sync
 ```
@@ -86,10 +99,11 @@ Activate the virtual environment:
 ```bash
 source .venv/bin/activate
 ```
-Other solution: append uv run to all the commands below.
 
 ## CLI Usage
 
+Note that you can bypass the previous installation step and just add uvx before each of the commands below.
+
 ### Convert PDFs
 
 With a general VLM (requires setting your api key as an environment variable):
@@ -185,11 +199,13 @@ server.stop()
 ```
 
 
-Converter with automatic server deployment:
+Converter with automatic server management:
 
 ```python
 from vlmparse.converter_with_server import ConverterWithServer
 
-converter_with_server = ConverterWithServer(model="mineru2.5")
-documents = converter_with_server.parse(inputs=["file1.pdf", "file2.pdf"], out_folder="./output")
+with ConverterWithServer(model="mineru2.5") as converter_with_server:
+    documents = converter_with_server.parse(inputs=["file1.pdf", "file2.pdf"], out_folder="./output")
 ```
+
+Note that if you pass an uri of a vllm server to `ConverterWithServer`, the model name is inferred automatically and no server is started.

vlmparse-0.1.6.dist-info/RECORD

@@ -0,0 +1,36 @@
+vlmparse/base_model.py,sha256=4U4UPe8SNArliKnUf8pp8zQugWYsnhg9okylt7mrW1U,381
+vlmparse/build_doc.py,sha256=LAWrnFrqamN5PwJo57AUtQOPrMFGnCGw4gBjEKZ6pYo,2127
+vlmparse/cli.py,sha256=gY45YZe5LanN-ozG2vVtOaB2qyNWpjO2DvPmJeBi_wA,13045
+vlmparse/constants.py,sha256=7-47S01n4MI2ebR09bpdOo3_P16d-z-NVGsm6KJP8ls,110
+vlmparse/converter.py,sha256=F0JSY9sFYUggCvaUCb27kKGJJpnZKW2FStMDVJoIOeQ,7383
+vlmparse/converter_with_server.py,sha256=62kcEp0NjzDR2vVmEfCeeLlwbb8E3sWcseb2jjK7DpM,4861
+vlmparse/registries.py,sha256=yBVrrhy61rSoLwdNV-z0C4lqIpTbLoWab3V6u7aSyNM,5797
+vlmparse/utils.py,sha256=rcVrtPiQVj_8HAmFQOu___72uYIapp_X89yxrMNCBow,1236
+vlmparse/clients/chandra.py,sha256=EulsCZdwOtm0pQ6CDm320U96k8aWFN4wKqCm1Xo7VCE,9775
+vlmparse/clients/deepseekocr.py,sha256=Uw6tPvP2KVsPDlz1ZUgYdbgQSjmFPuYeFDrGMMOTBAo,6501
+vlmparse/clients/docling.py,sha256=SAkLsqseuWfkuiel8FWR1G0Z5s-SZU3dE2JbsOvF4SA,5328
+vlmparse/clients/dotsocr.py,sha256=uGJoYEiDkP3-rmfdkAnMeAX-T4RZyEPoh6jmow5_-J8,10336
+vlmparse/clients/granite_docling.py,sha256=EQpsv5qSJG0HtMSacmJStER2sq4TGf1EMU5_NmJsl4g,4634
+vlmparse/clients/hunyuanocr.py,sha256=UFqaS4b8UM9EtizyrZIxlqcYlESmxm8xrQZP7lL6tkE,1857
+vlmparse/clients/lightonocr.py,sha256=wx1Im8Z3wlRWwYbPqnSd3LqTtdAU8CnX5mzu1BuCUY8,1314
+vlmparse/clients/mineru.py,sha256=6jZ1sKn2kGwUvD8gVs4PqEDH7uUXYK8pAB5Fr1JeqnY,3617
+vlmparse/clients/nanonetocr.py,sha256=BT5vaeerCsK5agvOaHK3NvLUqWd1FfDmrMmDYbp646I,1543
+vlmparse/clients/olmocr.py,sha256=A4Vl0meYpU5QPTML_OxyyRM07xCxtfrMZedgGMYEcuU,1851
+vlmparse/clients/openai_converter.py,sha256=nMKJeWH43UxHMMLns3wjX0pYjU5Xnai6IYxFmS9I63s,6193
+vlmparse/clients/paddleocrvl.py,sha256=qFBDj_UQocyq3WCh24tUOx9Ud7S9DfSm-1n3ztikY2s,1402
+vlmparse/clients/prompts.py,sha256=-J60lqxgRzlkQ9VsQLxmWsIMaDt-gNqWqWoqHIw9CLc,4228
+vlmparse/clients/pipe_utils/cleaner.py,sha256=oxBkBTOkluN1lmeNbzajRIe0_D__ZGwUOBaI_Ph0uxE,2396
+vlmparse/clients/pipe_utils/html_to_md_conversion.py,sha256=cFFqzD2jCNw_968_eu3Wt--Ox7iJj2Rn5UoP_DZWosU,4112
+vlmparse/clients/pipe_utils/utils.py,sha256=935ecIO446I0pstszE_1nrIPHn1Ffrxunq7fVd0dsd8,315
+vlmparse/data_model/box.py,sha256=lJsh4qhjgYXZF5vTSJ1qMXD5GVlBi2_SBedBMlfJikU,16868
+vlmparse/data_model/document.py,sha256=xheaMeStOj2c9GZKmdtxcEl_Dj44V5JyVp6JnTrSpH0,4615
+vlmparse/servers/docker_server.py,sha256=FBW2TvtUHUQNwj0rBP92shvoiJCGlc_oAKQBXN8260E,7114
+vlmparse/servers/utils.py,sha256=qy2-rnQTCQKt6CeTV5H74tvRTXyzBV2KswQiYW8Tf-k,8908
+vlmparse/st_viewer/fs_nav.py,sha256=7GNH68h2Loh5pQ64Pe72-D2cs2BLhqRXevEmKdFmPX0,1616
+vlmparse/st_viewer/st_viewer.py,sha256=m2rQTtk5rlwErNmivNAg-4rkHkvNkvLhoJZxFQi7Dwk,2105
+vlmparse-0.1.6.dist-info/licenses/LICENSE,sha256=3TKJHk8hPBR5dbLWZ3IpfCftl-_m-iyBwpYQGZYxj14,1080
+vlmparse-0.1.6.dist-info/METADATA,sha256=Xad3SjAYvCzUvPo6A6GKvc3daxtf5XNs1AQjDlF7RmI,5597
+vlmparse-0.1.6.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+vlmparse-0.1.6.dist-info/entry_points.txt,sha256=gD5berP6HwE2wNIkls-Lw5goiceA8uMgPEd7ifnFJXs,47
+vlmparse-0.1.6.dist-info/top_level.txt,sha256=k4ni-GNH_iAX7liQEsk_KY_c3xgZgt8k9fsSs9IXLXs,9
+vlmparse-0.1.6.dist-info/RECORD,,