vlmparse 0.1.8__py3-none-any.whl → 0.1.10__py3-none-any.whl

vlmparse/servers/utils.py

@@ -1,228 +1,9 @@
-import getpass
-import time
-from contextlib import contextmanager
-from pathlib import Path
 from urllib.parse import parse_qsl, urlparse
 
 import docker
 from loguru import logger
 
 
-def _ensure_image_exists(
-    client: docker.DockerClient,
-    image: str,
-    dockerfile_path: Path,
-):
-    """Check if image exists, build it if not."""
-    try:
-        client.images.get(image)
-        logger.info(f"Docker image {image} found")
-        return
-    except docker.errors.ImageNotFound:
-        logger.info(f"Docker image {image} not found, building...")
-
-        if not dockerfile_path.exists():
-            raise FileNotFoundError(
-                f"Dockerfile directory not found at {dockerfile_path}"
-            ) from None
-
-        logger.info(f"Building image from {dockerfile_path}")
-
-        # Use low-level API for real-time streaming
-        api_client = docker.APIClient(base_url="unix://var/run/docker.sock")
-
-        # Build the image with streaming
-        build_stream = api_client.build(
-            path=str(dockerfile_path),
-            tag=image,
-            rm=True,
-            decode=True,  # Automatically decode JSON responses to dict
-        )
-
-        # Stream build logs in real-time
-        for chunk in build_stream:
-            if "stream" in chunk:
-                for line in chunk["stream"].splitlines():
-                    logger.info(line)
-            elif "error" in chunk:
-                logger.error(chunk["error"])
-                raise docker.errors.BuildError(chunk["error"], build_stream) from None
-            elif "status" in chunk:
-                # Handle status updates (e.g., downloading layers)
-                logger.debug(chunk["status"])
-
-        logger.info(f"Successfully built image {image}")
-
-
-@contextmanager
-def docker_server(
-    config: "DockerServerConfig",  # noqa: F821
-    timeout: int = 1000,
-    cleanup: bool = True,
-):
-    """Generic context manager for Docker server deployment.
-
-    Args:
-        config: DockerServerConfig (can be VLLMDockerServerConfig or GenericDockerServerConfig)
-        timeout: Timeout in seconds to wait for server to be ready
-        cleanup: If True, stop and remove container on exit. If False, leave container running
-
-    Yields:
-        tuple: (base_url, container) - The base URL of the server and the Docker container object
-    """
-
-    client = docker.from_env()
-    container = None
-
-    try:
-        # Ensure image exists
-        logger.info(f"Checking for Docker image {config.docker_image}...")
-
-        if config.dockerfile_dir is not None:
-            _ensure_image_exists(
-                client, config.docker_image, Path(config.dockerfile_dir)
-            )
-        else:
-            # Pull pre-built image
-            try:
-                client.images.get(config.docker_image)
-                logger.info(f"Docker image {config.docker_image} found locally")
-            except docker.errors.ImageNotFound:
-                logger.info(
-                    f"Docker image {config.docker_image} not found locally, pulling..."
-                )
-                client.images.pull(config.docker_image)
-                logger.info(f"Successfully pulled {config.docker_image}")
-
-        logger.info(
-            f"Starting Docker container for {config.model_name} on port {config.docker_port}"
-        )
-
-        # Configure GPU access
-        device_requests = None
-
-        if config.gpu_device_ids is None:
-            # Default: Try to use all GPUs if available
-            device_requests = [
-                docker.types.DeviceRequest(count=-1, capabilities=[["gpu"]])
-            ]
-        elif len(config.gpu_device_ids) > 0 and config.gpu_device_ids[0] != "":
-            # Use specific GPU devices
-            device_requests = [
-                docker.types.DeviceRequest(
-                    device_ids=config.gpu_device_ids, capabilities=[["gpu"]]
-                )
-            ]
-        else:
-            # Empty list means CPU-only, no GPU
-            device_requests = None
-
-        # Use generic methods from config
-        command = config.get_command()
-        volumes = config.get_volumes()
-        environment = config.get_environment()
-        container_port = config.container_port
-        log_prefix = config.model_name
-
-        # Construct URI for label
-        uri = f"http://localhost:{config.docker_port}{config.get_base_url_suffix()}"
-
-        # Determine GPU label
-        if config.gpu_device_ids is None:
-            gpu_label = "all"
-        elif len(config.gpu_device_ids) == 0 or (
-            len(config.gpu_device_ids) == 1 and config.gpu_device_ids[0] == ""
-        ):
-            gpu_label = "cpu"
-        else:
-            gpu_label = ",".join(config.gpu_device_ids)
-
-        # Start container
-        container_kwargs = {
-            "image": config.docker_image,
-            "ports": {f"{container_port}/tcp": config.docker_port},
-            "detach": True,
-            "remove": True,
-            "name": f"vlmparse-{config.model_name.replace('/', '-')}-{getpass.getuser()}",
-            "labels": {
-                "vlmparse_model_name": config.model_name,
-                "vlmparse_uri": uri,
-                "vlmparse_gpus": gpu_label,
-            },
-        }
-
-        if device_requests is not None:
-            container_kwargs["device_requests"] = device_requests
-        if command:
-            container_kwargs["command"] = command
-        if environment:
-            container_kwargs["environment"] = environment
-        if volumes:
-            container_kwargs["volumes"] = volumes
-        if config.entrypoint:
-            container_kwargs["entrypoint"] = config.entrypoint
-
-        container = client.containers.run(**container_kwargs)
-
-        logger.info(
-            f"Container {container.short_id} started, waiting for server to be ready..."
-        )
-
-        # Wait for server to be ready
-        start_time = time.time()
-        server_ready = False
-        last_log_position = 0
-
-        while time.time() - start_time < timeout:
-            try:
-                container.reload()
-            except docker.errors.NotFound as e:
-                logger.error("Container stopped unexpectedly during startup")
-                raise RuntimeError(
-                    "Container crashed during initialization. Check Docker logs for details."
-                ) from e
-
-            if container.status == "running":
-                # Get all logs and display new ones
-                all_logs = container.logs().decode("utf-8")
-
-                # Display new log lines
-                if len(all_logs) > last_log_position:
-                    new_logs = all_logs[last_log_position:]
-                    for line in new_logs.splitlines():
-                        if line.strip():  # Only print non-empty lines
-                            logger.info(f"[{log_prefix}] {line}")
-                    last_log_position = len(all_logs)
-
-                # Check if server is ready
-                for indicator in config.server_ready_indicators:
-                    if indicator in all_logs:
-                        server_ready = True
-                if server_ready:
-                    logger.info(f"Server ready indicator '{indicator}' found in logs")
-                    break
-
-            time.sleep(2)
-
-        if not server_ready:
-            raise TimeoutError(f"Server did not become ready within {timeout} seconds")
-
-        # Build base URL using config's suffix method
-        base_url = (
-            f"http://localhost:{config.docker_port}{config.get_base_url_suffix()}"
-        )
-
-        logger.info(f"{log_prefix} server ready at {base_url}")
-
-        yield base_url, container
-
-    finally:
-        if cleanup and container:
-            logger.info(f"Stopping container {container.short_id}")
-            container.stop(timeout=10)
-            logger.info("Container stopped")
-
-
 def normalize_uri(uri: str) -> tuple:
     u = urlparse(uri)
 
@@ -277,3 +58,86 @@ def get_model_from_uri(uri: str) -> str:
     if model is None:
         raise ValueError(f"No model found for URI {uri}")
     return model
+
+
+def _get_container_labels(container) -> dict[str, str]:
+    labels: dict[str, str] = {}
+    try:
+        labels.update(getattr(container, "labels", None) or {})
+    except Exception:
+        pass
+
+    try:
+        labels.update((container.attrs or {}).get("Config", {}).get("Labels", {}) or {})
+    except Exception:
+        pass
+
+    return labels
+
+
+def _stop_compose_stack_for_container(target_container) -> bool:
+    """If container belongs to a docker-compose project, stop+remove the whole stack.
+
+    Returns True if a compose stack was detected and a stack stop was attempted.
+    """
+
+    import subprocess
+
+    labels = _get_container_labels(target_container)
+
+    project = labels.get("com.docker.compose.project") or labels.get(
+        "vlmparse_compose_project"
+    )
+    compose_file = labels.get("vlmparse_compose_file")
+
+    if not project:
+        return False
+
+    # Preferred: docker compose down (stops + removes all services/networks consistently)
+    if compose_file:
+        logger.info(
+            f"Detected docker-compose project '{project}'. Bringing stack down (stop + remove)..."
+        )
+        subprocess.run(
+            [
+                "docker",
+                "compose",
+                "-f",
+                compose_file,
+                "--project-name",
+                project,
+                "down",
+                "--remove-orphans",
+            ],
+            check=False,
+            capture_output=True,
+            text=True,
+        )
+        logger.info("✓ Compose stack brought down")
+        return True
+
+    # Fallback: remove all containers in the compose project via Docker labels
+    import docker
+
+    logger.info(
+        f"Detected docker-compose project '{project}' (compose file unknown). "
+        "Stopping + removing all project containers via Docker API..."
+    )
+    client = docker.from_env()
+    containers = client.containers.list(
+        all=True, filters={"label": [f"com.docker.compose.project={project}"]}
+    )
+    for c in containers:
+        try:
+            c.stop()
+        except Exception:
+            pass
+        try:
+            c.remove(force=True)
+        except Exception:
+            pass
+
+    logger.info(
+        f"✓ Removed {len(containers)} container(s) from compose project '{project}'"
+    )
+    return True

vlmparse/st_viewer/st_viewer.py

@@ -40,7 +40,7 @@ def run_streamlit(folder: str) -> None:
     col1, col2 = st.columns(2)
     with col1:
         with st.container(height=700):
-            st.write(doc.pages[settings["page_no"]].text)
+            st.markdown(doc.pages[settings["page_no"]].text, unsafe_allow_html=True)
 
     with col2:
         if settings["plot_layouts"]:

{vlmparse-0.1.8.dist-info → vlmparse-0.1.10.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: vlmparse
-Version: 0.1.8
+Version: 0.1.10
 Requires-Python: >=3.11.0
 Description-Content-Type: text/markdown
 License-File: LICENSE
@@ -56,7 +56,7 @@ Dynamic: license-file
 
 <div align="center">
 
-[\[📜 arXiv coming soon\]] | [[Dataset (🤗Hugging Face)]](https://huggingface.co/datasets/pulsia/fr-bench-pdf2md) | [[pypi]](https://pypi.org/project/vlmparse/) | [[vlmparse]](https://github.com/ld-lab-pulsia/vlmparse) | [[Benchmark]](https://github.com/ld-lab-pulsia/benchpdf2md)
+[\[📜 arXiv coming soon\]] | [[Dataset (🤗Hugging Face)]](https://huggingface.co/datasets/pulsia/fr-bench-pdf2md) | [[pypi]](https://pypi.org/project/vlmparse/) | [[vlmparse]](https://github.com/ld-lab-pulsia/vlmparse) | [[Benchmark]](https://github.com/ld-lab-pulsia/benchpdf2md) | [[Leaderboard]](https://huggingface.co/spaces/pulsia/fr-bench-pdf2md)
 
 </div>
 
@@ -71,7 +71,7 @@ Features:
 
 Supported Converters:
 
-- **Open Source Small VLMs**: `lightonocr`, `mineru2.5`, `hunyuanocr`, `paddleocrvl`, `granite-docling`, `olmocr2-fp8`, `dotsocr`, `chandra`, `deepseekocr`, `nanonets/Nanonets-OCR2-3B`
+- **Open Source Small VLMs**: `lightonocr2`, `mineru2.5`, `hunyuanocr`, `paddleocrvl-1.5`, `granite-docling`, `olmocr2-fp8`, `dotsocr`, `chandra`, `deepseekocr2`, `nanonets/Nanonets-OCR2-3B`
 - **Open Source Generalist VLMs**: such as the Qwen family.
 - **Pipelines**: `docling`
 - **Proprietary LLMs**: `gemini`, `gpt`
@@ -115,13 +115,13 @@ Note that you can bypass the previous installation step and just add uvx before
 With a general VLM (requires setting your api key as an environment variable):
 
 ```bash
-vlmparse convert --input "*.pdf" --out_folder ./output --model gemini-2.5-flash-lite
+vlmparse convert "*.pdf" --out_folder ./output --model gemini-2.5-flash-lite
 ```
 
 Convert with auto deployment of a small vlm (or any huggingface VLM model, requires a gpu + docker installation):
 
 ```bash
-vlmparse convert --input "*.pdf" --out_folder ./output --model nanonets/Nanonets-OCR2-3B
+vlmparse convert "*.pdf" --out_folder ./output --model nanonets/Nanonets-OCR2-3B
 ```
 
 ### Deploy a local model server
@@ -131,13 +131,13 @@ Deployment (requires a gpu + docker installation):
 - Check that the port is not used by another service.
 
 ```bash
-vlmparse serve --model lightonocr --port 8000 --gpus 1
+vlmparse serve lightonocr2 --port 8000 --gpus 1
 ```
 
 then convert:
 
 ```bash
-vlmparse convert --input "*.pdf" --out_folder ./output --model lightonocr --uri http://localhost:8000/v1
+vlmparse convert "*.pdf" --out_folder ./output --uri http://localhost:8000/v1
 ```
 
 You can also list all running servers:

{vlmparse-0.1.8.dist-info → vlmparse-0.1.10.dist-info}/RECORD

@@ -1,15 +1,16 @@
 vlmparse/base_model.py,sha256=4U4UPe8SNArliKnUf8pp8zQugWYsnhg9okylt7mrW1U,381
 vlmparse/build_doc.py,sha256=fb7awoqVN-6NBlKVkMFb1v1iTWcxne5QAyNaKYTyvM4,2275
-vlmparse/cli.py,sha256=asew0JdpbgFZrZqnG-Bqh5A_DrXcP0XomLB3y3AgG6Y,12855
+vlmparse/cli.py,sha256=jP_BnFaeW1rm3iTcdw5WFRfQUgDYd6HC1Zh-5JbE9_4,18285
 vlmparse/constants.py,sha256=DYaK7KtTW8p9MPb3iPvoP5H1r7ICRuIFo89P01q4uCI,184
 vlmparse/converter.py,sha256=KKcXqrp3nJo3d7DXjHn3O2SklbsJ489rDY4NJ9O42Fs,8795
-vlmparse/converter_with_server.py,sha256=A84l3YNal-hs2mMlER1sB29rddsO8MNOP2j9ts0ujtE,7280
-vlmparse/registries.py,sha256=B4kxibP7XYbhL9bZ5gn21LQCPhHCYftAM4i0-xD9fRs,6469
+vlmparse/converter_with_server.py,sha256=nDGF-FEqskAECam_Sm8GbPMGPdI2Iua4lHaHbpMZx_k,8872
+vlmparse/registries.py,sha256=4xiDKyIzAW68ZWyOtUmBOvzcXVqTPPdeoxD2s9RbjZ0,6714
 vlmparse/utils.py,sha256=6Ff9OfAIVR-4_37QD5sifoNt_GmB3YUqgFwmIjuemtc,1727
 vlmparse/clients/chandra.py,sha256=zAHjgI_MJ5FVGANHCG8KJQByaw6-zTS6CHXsCBA8TJI,13025
-vlmparse/clients/deepseekocr.py,sha256=pKdNJD9v86BRn7YrXE6PGk_jQxnbZ_6UjgSUxgd3Su4,6859
+vlmparse/clients/deepseekocr.py,sha256=4NiW-the4JHPqI0rNF2xG3juGZJX4tiI7doCk1jyYec,12772
 vlmparse/clients/docling.py,sha256=BLtNAxVJR6qvPip4ZBP-se8IMNFSbJ-fWEGlTSwimK8,5310
 vlmparse/clients/dotsocr.py,sha256=oAUzDMTObeW0sTy5sFl08O6GQPSTic5ITbJYh_45Z54,10414
+vlmparse/clients/glmocr.py,sha256=WzRntPKx3BPDAZEjfAHQj2OukBEEOEKbtpZag0me53g,7727
 vlmparse/clients/granite_docling.py,sha256=KYaEdgk3oD0TuYDKqTQ4o6IkXC-E3AIYJ2KYxnEsjWM,3595
 vlmparse/clients/hunyuanocr.py,sha256=etpIiA28OoGW-o5pOGeBxOlUDjUQ4zcKXWnJ8ba44DU,1979
 vlmparse/clients/lightonocr.py,sha256=ZWC12U6myDr_2EuOPYGyJYxpBachjOUtLrxS62A8mzg,2048
@@ -18,21 +19,26 @@ vlmparse/clients/mistral_converter.py,sha256=_hEyK_2vM5LEwbt30bFodMrWJtavLsBDxCk
 vlmparse/clients/nanonetocr.py,sha256=gTbD4OtuHiWd6Ack6Bx-anZM9P_aErfSHXwtymETvqM,1665
 vlmparse/clients/olmocr.py,sha256=V4638WftLCTr5Q6ZRgWKKSPAhFYdpBw3izeuda6EKDQ,1966
 vlmparse/clients/openai_converter.py,sha256=bckm33Pkvqul--DjfEEEI3evn4_va0CoQcigdpCCMGc,7746
-vlmparse/clients/paddleocrvl.py,sha256=q3AgEWj0UyXGpSEVZISdfqv2PV_qY-uF498bL8U1tpg,1596
+vlmparse/clients/paddleocrvl.py,sha256=50HuQm5aOH7xi0vdJ5isowsw6eBtEBNV8DWY4wbaTGA,7094
 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=FOIHU0_CDfyZ9UA285BrnUFuEMJRxbu-OzlByBa-P9s,7951
+vlmparse/servers/base_server.py,sha256=NLoGaXu8tapAm2dKjzZSBNY0EOXAK4Zo-zMtTx_VvdA,4343
+vlmparse/servers/docker_compose_deployment.py,sha256=sGmZd7TRkOwZyh8raZhE2udxCBhKhxyn0hJcjCn9_DE,16999
+vlmparse/servers/docker_compose_server.py,sha256=UsAokpmGeTHaaS6MVVlM3KxpILDUzeWuS4GSlr_2gT0,1394
+vlmparse/servers/docker_run_deployment.py,sha256=34vHVZwRcw2Z5i5kPj8r2GCc7OmgVJXaSPctlXDWgvU,7898
+vlmparse/servers/docker_server.py,sha256=KrTTcwN1wxhK0bpMuZtJZ5cQz4MuaTWLTlh9nMG87Fg,3777
 vlmparse/servers/model_identity.py,sha256=DkH7KQAAZA9Sn7eJEnaKfH54XSEI17aqD1ScqqkTBEk,1711
-vlmparse/servers/utils.py,sha256=tIXhgbF9EVOJy2nYEguVq69gn9ATxtya_1F4wZSt68o,9454
+vlmparse/servers/server_registry.py,sha256=FUF_XnN8872vKnc8-TrEBntwBS5i3ZYVJvTHrHfI7IM,1315
+vlmparse/servers/utils.py,sha256=rbqn9i6XB1YOEFluP4Ur0Ma40_6riUxJ1eMS8LSWbKs,3998
 vlmparse/st_viewer/fs_nav.py,sha256=7GNH68h2Loh5pQ64Pe72-D2cs2BLhqRXevEmKdFmPX0,1616
-vlmparse/st_viewer/st_viewer.py,sha256=m2rQTtk5rlwErNmivNAg-4rkHkvNkvLhoJZxFQi7Dwk,2105
-vlmparse-0.1.8.dist-info/licenses/LICENSE,sha256=3TKJHk8hPBR5dbLWZ3IpfCftl-_m-iyBwpYQGZYxj14,1080
-vlmparse-0.1.8.dist-info/METADATA,sha256=dwu5tiTLuhVMYL-ZQCMNYW_MNlJu84V2us0aeRfrSpU,6048
-vlmparse-0.1.8.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
-vlmparse-0.1.8.dist-info/entry_points.txt,sha256=gD5berP6HwE2wNIkls-Lw5goiceA8uMgPEd7ifnFJXs,47
-vlmparse-0.1.8.dist-info/top_level.txt,sha256=k4ni-GNH_iAX7liQEsk_KY_c3xgZgt8k9fsSs9IXLXs,9
-vlmparse-0.1.8.dist-info/RECORD,,
+vlmparse/st_viewer/st_viewer.py,sha256=wg0qfhAKdvnkpc3xDK8QnWP9adjEThzeS-I5vHGDhIU,2132
+vlmparse-0.1.10.dist-info/licenses/LICENSE,sha256=3TKJHk8hPBR5dbLWZ3IpfCftl-_m-iyBwpYQGZYxj14,1080
+vlmparse-0.1.10.dist-info/METADATA,sha256=OIRlJUlRioNzrehJIK2dmBcTFHI7A6H5uedu-EzDTQA,6077
+vlmparse-0.1.10.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+vlmparse-0.1.10.dist-info/entry_points.txt,sha256=gD5berP6HwE2wNIkls-Lw5goiceA8uMgPEd7ifnFJXs,47
+vlmparse-0.1.10.dist-info/top_level.txt,sha256=k4ni-GNH_iAX7liQEsk_KY_c3xgZgt8k9fsSs9IXLXs,9
+vlmparse-0.1.10.dist-info/RECORD,,