paddlex 3.0.0rc0__py3-none-any.whl → 3.0.1__py3-none-any.whl

paddlex/inference/serving/basic_serving/_pipeline_apps/table_recognition_v2.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -14,18 +14,21 @@
 
 from typing import Any, Dict, List
 
-from fastapi import FastAPI
-
+from .....utils.deps import function_requires_deps, is_dep_available
 from ...infra import utils as serving_utils
 from ...infra.config import AppConfig
-from ...infra.models import ResultResponse
+from ...infra.models import AIStudioResultResponse
 from ...schemas.table_recognition_v2 import INFER_ENDPOINT, InferRequest, InferResult
 from .._app import create_app, primary_operation
 from ._common import common
 from ._common import ocr as ocr_common
 
+if is_dep_available("fastapi"):
+    from fastapi import FastAPI
+
 
-def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
+@function_requires_deps("fastapi")
+def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> "FastAPI":
     app, ctx = create_app(
         pipeline=pipeline, app_config=app_config, app_aiohttp_session=True
     )
@@ -37,7 +40,7 @@ def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
         INFER_ENDPOINT,
         "infer",
     )
-    async def _infer(request: InferRequest) -> ResultResponse[InferResult]:
+    async def _infer(request: InferRequest) -> AIStudioResultResponse[InferResult]:
         pipeline = ctx.pipeline
 
         log_id = serving_utils.generate_log_id()
@@ -50,19 +53,18 @@ def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
             use_doc_unwarping=request.useDocUnwarping,
             use_layout_detection=request.useLayoutDetection,
             use_ocr_model=request.useOcrModel,
-            layout_threshold=request.layoutThreshold,
-            layout_nms=request.layoutNms,
-            layout_unclip_ratio=request.layoutUnclipRatio,
-            layout_merge_bboxes_mode=request.layoutMergeBboxesMode,
             text_det_limit_side_len=request.textDetLimitSideLen,
             text_det_limit_type=request.textDetLimitType,
             text_det_thresh=request.textDetThresh,
             text_det_box_thresh=request.textDetBoxThresh,
             text_det_unclip_ratio=request.textDetUnclipRatio,
             text_rec_score_thresh=request.textRecScoreThresh,
-            use_table_cells_ocr_results=request.useTableCellsOcrResults,
             use_e2e_wired_table_rec_model=request.useE2eWiredTableRecModel,
             use_e2e_wireless_table_rec_model=request.useE2eWirelessTableRecModel,
+            use_wired_table_cells_trans_to_html=request.useWiredTableCellsTransToHtml,
+            use_wireless_table_cells_trans_to_html=request.useWirelessTableCellsTransToHtml,
+            use_table_orientation_classify=request.useTableOrientationClassify,
+            use_ocr_results_with_table_cells=request.useOcrResultsWithTableCells,
         )
 
         table_rec_results: List[Dict[str, Any]] = []
@@ -96,7 +98,7 @@ def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
                 )
             )
 
-        return ResultResponse[InferResult](
+        return AIStudioResultResponse[InferResult](
             logId=log_id,
             result=InferResult(
                 tableRecResults=table_rec_results,

paddlex/inference/serving/basic_serving/_pipeline_apps/ts_anomaly_detection.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -14,16 +14,19 @@
 
 from typing import Any
 
-from fastapi import FastAPI
-
+from .....utils.deps import function_requires_deps, is_dep_available
 from ...infra import utils as serving_utils
 from ...infra.config import AppConfig
-from ...infra.models import ResultResponse
+from ...infra.models import AIStudioResultResponse
 from ...schemas.ts_anomaly_detection import INFER_ENDPOINT, InferRequest, InferResult
 from .._app import create_app, primary_operation
 
+if is_dep_available("fastapi"):
+    from fastapi import FastAPI
+
 
-def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
+@function_requires_deps("fastapi")
+def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> "FastAPI":
     app, ctx = create_app(
         pipeline=pipeline, app_config=app_config, app_aiohttp_session=True
     )
@@ -33,7 +36,7 @@ def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
         INFER_ENDPOINT,
         "infer",
     )
-    async def _infer(request: InferRequest) -> ResultResponse[InferResult]:
+    async def _infer(request: InferRequest) -> AIStudioResultResponse[InferResult]:
         pipeline = ctx.pipeline
         aiohttp_session = ctx.aiohttp_session
 
@@ -54,7 +57,7 @@ def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
         else:
             output_image = None
 
-        return ResultResponse[InferResult](
+        return AIStudioResultResponse[InferResult](
             logId=serving_utils.generate_log_id(),
             result=InferResult(csv=output_csv, image=output_image),
         )

paddlex/inference/serving/basic_serving/_pipeline_apps/ts_classification.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -14,16 +14,19 @@
 
 from typing import Any
 
-from fastapi import FastAPI
-
+from .....utils.deps import function_requires_deps, is_dep_available
 from ...infra import utils as serving_utils
 from ...infra.config import AppConfig
-from ...infra.models import ResultResponse
+from ...infra.models import AIStudioResultResponse
 from ...schemas.ts_classification import INFER_ENDPOINT, InferRequest, InferResult
 from .._app import create_app, primary_operation
 
+if is_dep_available("fastapi"):
+    from fastapi import FastAPI
+
 
-def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
+@function_requires_deps("fastapi")
+def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> "FastAPI":
     app, ctx = create_app(
         pipeline=pipeline, app_config=app_config, app_aiohttp_session=True
     )
@@ -33,7 +36,7 @@ def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
         INFER_ENDPOINT,
         "infer",
     )
-    async def _infer(request: InferRequest) -> ResultResponse[InferResult]:
+    async def _infer(request: InferRequest) -> AIStudioResultResponse[InferResult]:
         pipeline = ctx.pipeline
         aiohttp_session = ctx.aiohttp_session
 
@@ -53,7 +56,7 @@ def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
         else:
             output_image = None
 
-        return ResultResponse[InferResult](
+        return AIStudioResultResponse[InferResult](
             logId=serving_utils.generate_log_id(),
             result=InferResult(label=label, score=score, image=output_image),
         )

paddlex/inference/serving/basic_serving/_pipeline_apps/ts_forecast.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -14,16 +14,19 @@
 
 from typing import Any
 
-from fastapi import FastAPI
-
+from .....utils.deps import function_requires_deps, is_dep_available
 from ...infra import utils as serving_utils
 from ...infra.config import AppConfig
-from ...infra.models import ResultResponse
+from ...infra.models import AIStudioResultResponse
 from ...schemas.ts_forecast import INFER_ENDPOINT, InferRequest, InferResult
 from .._app import create_app, primary_operation
 
+if is_dep_available("fastapi"):
+    from fastapi import FastAPI
+
 
-def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
+@function_requires_deps("fastapi")
+def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> "FastAPI":
     app, ctx = create_app(
         pipeline=pipeline, app_config=app_config, app_aiohttp_session=True
     )
@@ -33,7 +36,7 @@ def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
         INFER_ENDPOINT,
         "infer",
     )
-    async def _infer(request: InferRequest) -> ResultResponse[InferResult]:
+    async def _infer(request: InferRequest) -> AIStudioResultResponse[InferResult]:
         pipeline = ctx.pipeline
         aiohttp_session = ctx.aiohttp_session
 
@@ -54,7 +57,7 @@ def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
         else:
             output_image = None
 
-        return ResultResponse[InferResult](
+        return AIStudioResultResponse[InferResult](
             logId=serving_utils.generate_log_id(),
             result=InferResult(csv=output_csv, image=output_image),
         )

paddlex/inference/serving/basic_serving/_pipeline_apps/vehicle_attribute_recognition.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -14,11 +14,10 @@
 
 from typing import Any, Dict, List
 
-from fastapi import FastAPI
-
+from .....utils.deps import function_requires_deps, is_dep_available
 from ...infra import utils as serving_utils
 from ...infra.config import AppConfig
-from ...infra.models import ResultResponse
+from ...infra.models import AIStudioResultResponse
 from ...schemas.vehicle_attribute_recognition import (
     INFER_ENDPOINT,
     InferRequest,
@@ -26,8 +25,12 @@ from ...schemas.vehicle_attribute_recognition import (
 )
 from .._app import create_app, primary_operation
 
+if is_dep_available("fastapi"):
+    from fastapi import FastAPI
+
 
-def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
+@function_requires_deps("fastapi")
+def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> "FastAPI":
     app, ctx = create_app(
         pipeline=pipeline, app_config=app_config, app_aiohttp_session=True
     )
@@ -37,7 +40,7 @@ def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
         INFER_ENDPOINT,
         "infer",
     )
-    async def _infer(request: InferRequest) -> ResultResponse[InferResult]:
+    async def _infer(request: InferRequest) -> AIStudioResultResponse[InferResult]:
         pipeline = ctx.pipeline
         aiohttp_session = ctx.aiohttp_session
 
@@ -73,7 +76,7 @@ def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
         else:
             output_image_base64 = None
 
-        return ResultResponse[InferResult](
+        return AIStudioResultResponse[InferResult](
             logId=serving_utils.generate_log_id(),
             result=InferResult(vehicles=objs, image=output_image_base64),
         )

paddlex/inference/serving/basic_serving/_pipeline_apps/video_classification.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -15,16 +15,19 @@
 import os
 from typing import Any, Dict, List
 
-from fastapi import FastAPI, HTTPException
-
+from .....utils.deps import function_requires_deps, is_dep_available
 from ...infra import utils as serving_utils
 from ...infra.config import AppConfig
-from ...infra.models import ResultResponse
+from ...infra.models import AIStudioResultResponse
 from ...schemas.video_classification import INFER_ENDPOINT, InferRequest, InferResult
 from .._app import create_app, primary_operation
 
+if is_dep_available("fastapi"):
+    from fastapi import FastAPI, HTTPException
+
 
-def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
+@function_requires_deps("fastapi")
+def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> "FastAPI":
     app, ctx = create_app(
         pipeline=pipeline, app_config=app_config, app_aiohttp_session=True
     )
@@ -34,7 +37,7 @@ def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
         INFER_ENDPOINT,
         "infer",
     )
-    async def _infer(request: InferRequest) -> ResultResponse[InferResult]:
+    async def _infer(request: InferRequest) -> AIStudioResultResponse[InferResult]:
         pipeline = ctx.pipeline
         aiohttp_session = ctx.aiohttp_session
 
@@ -65,7 +68,7 @@ def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
         for id_, name, score in zip(result["class_ids"], cat_names, result["scores"]):
             categories.append(dict(id=id_, name=name, score=score))
 
-        return ResultResponse[InferResult](
+        return AIStudioResultResponse[InferResult](
             logId=serving_utils.generate_log_id(),
             result=InferResult(categories=categories),
         )

paddlex/inference/serving/basic_serving/_pipeline_apps/video_detection.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -15,16 +15,19 @@
 import os
 from typing import Any, Dict, List
 
-from fastapi import FastAPI, HTTPException
-
+from .....utils.deps import function_requires_deps, is_dep_available
 from ...infra import utils as serving_utils
 from ...infra.config import AppConfig
-from ...infra.models import ResultResponse
+from ...infra.models import AIStudioResultResponse
 from ...schemas.video_detection import INFER_ENDPOINT, InferRequest, InferResult
 from .._app import create_app, primary_operation
 
+if is_dep_available("fastapi"):
+    from fastapi import FastAPI, HTTPException
+
 
-def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
+@function_requires_deps("fastapi")
+def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> "FastAPI":
     app, ctx = create_app(
         pipeline=pipeline, app_config=app_config, app_aiohttp_session=True
     )
@@ -34,7 +37,7 @@ def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
         INFER_ENDPOINT,
         "infer",
     )
-    async def _infer(request: InferRequest) -> ResultResponse[InferResult]:
+    async def _infer(request: InferRequest) -> AIStudioResultResponse[InferResult]:
         pipeline = ctx.pipeline
         aiohttp_session = ctx.aiohttp_session
 
@@ -81,7 +84,7 @@ def create_pipeline_app(pipeline: Any, app_config: AppConfig) -> FastAPI:
                 )
             )
 
-        return ResultResponse[InferResult](
+        return AIStudioResultResponse[InferResult](
             logId=serving_utils.generate_log_id(),
             result=InferResult(frames=frames),
         )

paddlex/inference/serving/basic_serving/_server.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -14,11 +14,16 @@
 
 import logging
 
-import uvicorn
-from fastapi import FastAPI
+from ....utils.deps import function_requires_deps, is_dep_available
 
+if is_dep_available("uvicorn"):
+    import uvicorn
+if is_dep_available("fastapi"):
+    from fastapi import FastAPI
 
-def run_server(app: FastAPI, *, host: str, port: int) -> None:
+
+@function_requires_deps("fastapi", "uvicorn")
+def run_server(app: "FastAPI", *, host: str, port: int) -> None:
     # HACK: Fix duplicate logs
     uvicorn_version = tuple(int(x) for x in uvicorn.__version__.split("."))
     if uvicorn_version < (0, 19, 0):

paddlex/inference/serving/infra/__init__.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

paddlex/inference/serving/infra/config.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

paddlex/inference/serving/infra/models.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -17,10 +17,15 @@ from typing import Dict, Generic, List, Tuple, TypeVar, Union
 from pydantic import BaseModel, Discriminator
 from typing_extensions import Annotated, Literal, TypeAlias
 
+from ....utils.deps import is_dep_available
+
+if is_dep_available("openai"):
+    from openai.types.chat import ChatCompletion
+
 __all__ = [
-    "NoResultResponse",
+    "AIStudioNoResultResponse",
     "ResultT",
-    "ResultResponse",
+    "AIStudioResultResponse",
     "Response",
     "ImageInfo",
     "PDFPageInfo",
@@ -30,7 +35,7 @@ __all__ = [
 ]
 
 
-class NoResultResponse(BaseModel):
+class AIStudioNoResultResponse(BaseModel):
     logId: str
     errorCode: int
     errorMsg: str
@@ -39,14 +44,16 @@ class NoResultResponse(BaseModel):
 ResultT = TypeVar("ResultT", bound=BaseModel)
 
 
-class ResultResponse(BaseModel, Generic[ResultT]):
+class AIStudioResultResponse(BaseModel, Generic[ResultT]):
     logId: str
     result: ResultT
     errorCode: Literal[0] = 0
     errorMsg: Literal["Success"] = "Success"
 
 
-Response: TypeAlias = Union[ResultResponse, NoResultResponse]
+Response: TypeAlias = Union[
+    AIStudioResultResponse, AIStudioNoResultResponse, "ChatCompletion"
+]
 
 
 class ImageInfo(BaseModel):

paddlex/inference/serving/infra/storage.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -17,12 +17,11 @@ from os import PathLike
 from pathlib import Path
 from typing import Any, Dict, Optional, Protocol, Union, runtime_checkable
 
-from baidubce.auth.bce_credentials import BceCredentials
-from baidubce.bce_client_configuration import BceClientConfiguration
-from baidubce.services.bos.bos_client import BosClient
 from pydantic import BaseModel, Discriminator, SecretStr, TypeAdapter
 from typing_extensions import Annotated, Literal, assert_never
 
+from ....utils.deps import class_requires_deps
+
 __all__ = [
     "InMemoryStorageConfig",
     "FileSystemStorageConfig",
@@ -123,9 +122,15 @@ class FileSystemStorage(Storage):
         return self._directory / key
 
 
+@class_requires_deps("bce-python-sdk")
 class BOS(Storage):
     def __init__(self, config: BOSConfig) -> None:
+        from baidubce.auth.bce_credentials import BceCredentials
+        from baidubce.bce_client_configuration import BceClientConfiguration
+        from baidubce.services.bos.bos_client import BosClient
+
         super().__init__()
+
         bos_cfg = BceClientConfiguration(
             credentials=BceCredentials(
                 config.ak.get_secret_value(), config.sk.get_secret_value()

paddlex/inference/serving/infra/utils.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -16,25 +16,33 @@ import asyncio
 import base64
 import io
 import mimetypes
+import re
 import tempfile
 import uuid
 from functools import partial
 from typing import Awaitable, Callable, List, Optional, Tuple, TypeVar, Union, overload
-from urllib.parse import urlparse
+from urllib.parse import parse_qs, urlparse
 
-import aiohttp
-import cv2
-import filetype
-import fitz
 import numpy as np
 import pandas as pd
 import requests
-import yarl
 from PIL import Image
 from typing_extensions import Literal, ParamSpec, TypeAlias, assert_never
 
+from ....utils.deps import function_requires_deps, is_dep_available
 from .models import ImageInfo, PDFInfo, PDFPageInfo
 
+if is_dep_available("aiohttp"):
+    import aiohttp
+if is_dep_available("opencv-contrib-python"):
+    import cv2
+if is_dep_available("filetype"):
+    import filetype
+if is_dep_available("pypdfium2"):
+    import pypdfium2 as pdfium
+if is_dep_available("yarl"):
+    import yarl
+
 __all__ = [
     "FileType",
     "generate_log_id",
@@ -89,7 +97,22 @@ def infer_file_type(url: str) -> Optional[FileType]:
     file_type = mimetypes.guess_type(filename)[0]
 
     if file_type is None:
-        return None
+        # HACK: The support for BOS URLs with query params is implementation-based,
+        # not interface-based.
+        is_bos_url = re.fullmatch(r"\w+\.bcebos\.com", url_parts.netloc) is not None
+        if is_bos_url and url_parts.query:
+            params = parse_qs(url_parts.query)
+            if (
+                "responseContentDisposition" in params
+                and len(params["responseContentDisposition"]) == 1
+            ):
+                match_ = re.match(
+                    r"attachment;filename=(.*)", params["responseContentDisposition"][0]
+                )
+                if match_:
+                    file_type = mimetypes.guess_type(match_.group(1))[0]
+        if file_type is None:
+            return None
 
     if file_type.startswith("image/"):
         return "IMAGE"
@@ -103,6 +126,7 @@ def infer_file_type(url: str) -> Optional[FileType]:
         return None
 
 
+@function_requires_deps("filetype")
 def infer_file_ext(file: str) -> Optional[str]:
     if is_url(file):
         url_parts = urlparse(file)
@@ -116,6 +140,7 @@ def infer_file_ext(file: str) -> Optional[str]:
         return "." + filetype.guess_extension(bytes_)
 
 
+@function_requires_deps("opencv-contrib-python")
 def image_bytes_to_array(data: bytes) -> np.ndarray:
     return cv2.imdecode(np.frombuffer(data, np.uint8), cv2.IMREAD_COLOR)
 
@@ -131,6 +156,7 @@ def image_to_bytes(image: Image.Image, format: str = "JPEG") -> bytes:
     return img_bytes
 
 
+@function_requires_deps("opencv-contrib-python")
 def image_array_to_bytes(image: np.ndarray, ext: str = ".jpg") -> bytes:
     image = cv2.imencode(ext, image)[1]
     return image.tobytes()
@@ -150,30 +176,29 @@ def base64_encode(data: bytes) -> str:
     return base64.b64encode(data).decode("ascii")
 
 
+@function_requires_deps("pypdfium2", "opencv-contrib-python")
 def read_pdf(
     bytes_: bytes, max_num_imgs: Optional[int] = None
 ) -> Tuple[List[np.ndarray], PDFInfo]:
     images: List[np.ndarray] = []
     page_info_list: List[PDFPageInfo] = []
-    with fitz.open("pdf", bytes_) as doc:
-        for page in doc:
-            if max_num_imgs is not None and len(images) >= max_num_imgs:
-                break
-            # TODO: Do not always use zoom=2.0
-            zoom = 2.0
-            deg = 0
-            mat = fitz.Matrix(zoom, zoom).prerotate(deg)
-            pixmap = page.get_pixmap(matrix=mat, alpha=False)
-            image = np.frombuffer(pixmap.samples, dtype=np.uint8).reshape(
-                pixmap.h, pixmap.w, pixmap.n
-            )
-            image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
-            images.append(image)
-            page_info = PDFPageInfo(
-                width=pixmap.w,
-                height=pixmap.h,
-            )
-            page_info_list.append(page_info)
+    doc = pdfium.PdfDocument(bytes_)
+    for page in doc:
+        if max_num_imgs is not None and len(images) >= max_num_imgs:
+            break
+        # TODO: Do not always use zoom=2.0
+        zoom = 2.0
+        deg = 0
+        image = page.render(scale=zoom, rotation=deg).to_pil()
+        image = image.convert("RGB")
+        image = np.array(image)
+        image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
+        images.append(image)
+        page_info = PDFPageInfo(
+            width=image.shape[1],
+            height=image.shape[0],
+        )
+        page_info_list.append(page_info)
     pdf_info = PDFInfo(
         numPages=len(page_info_list),
         pages=page_info_list,
@@ -243,7 +268,8 @@ def get_raw_bytes(file: str) -> bytes:
         return base64.b64decode(file)
 
 
-async def get_raw_bytes_async(file: str, session: aiohttp.ClientSession) -> bytes:
+@function_requires_deps("aiohttp", "yarl")
+async def get_raw_bytes_async(file: str, session: "aiohttp.ClientSession") -> bytes:
     if is_url(file):
         async with session.get(yarl.URL(file, encoded=True)) as resp:
             return await resp.read()

paddlex/inference/serving/schemas/__init__.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

paddlex/inference/serving/schemas/anomaly_detection.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

paddlex/inference/serving/schemas/doc_preprocessor.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.