Cannot close object, library is destroyed. This may cause a memory leak!

[aceliuchanghong]

def run_surya_table_detection(
    PDF_PATH,
    det_model,
    det_processor,
    layout_model,
    layout_processor,
    table_rec_model,
    table_rec_processor,
    rec_model,
    rec_processor,
):
    """
    table
    """

    new_det_model = det_model, det_processor, layout_model, layout_processor
    new_rec_models = table_rec_model, table_rec_processor, rec_model, rec_processor

    start_time = time.time()

    images, highres_images, names, text_lines = load_pdfs_images(PDF_PATH)
    table_detection = extract_tables(
        images, highres_images, text_lines, new_det_model, new_rec_models
    )

    end_time = time.time()
    elapsed_time = end_time - start_time
    logger.info(f"surya_table_detection耗时: {elapsed_time:.2f}秒")
    return table_detection



if __name__ == "__main__":    
    PDF_PATH = "no_git_oic/页面提取自－NPD2317设计开发记录.pdf"

    det_model_path = os.getenv("SURYA_DET3_MODEL_PATH")
    rec_model_path = os.getenv("SURYA_REC2_MODEL_PATH")
    layout_model_path = os.getenv("SURYA_LAYOUT4_MODEL_PATH")
    order_model_path = os.getenv("SURYA_ORDER_MODEL_PATH")
    table_rec_model_path = os.getenv("SURYA_TABLEREC_MODEL_PATH")

    start_time = time.time()

    rec_processor = load_rec_processor()
    det_model = load_det_model(det_model_path)
    det_processor = load_det_processor(det_model_path)
    rec_model = load_rec_model(rec_model_path)

    layout_model = load_det_model(layout_model_path)
    layout_processor = load_det_processor(layout_model_path)

    order_model = load_order_processor(order_model_path)
    order_processor = load_order_model(order_model_path)

    table_rec_model = load_table_rec_model(table_rec_model_path)
    table_rec_processor = load_table_rec_processor()

    end_time = time.time()
    elapsed_time = end_time - start_time
    logger.info(f"surya模型加载耗时: {elapsed_time:.2f}秒")


    table_detection = run_surya_table_detection(
        PDF_PATH,
        det_model,
        det_processor,
        layout_model,
        layout_processor,
        table_rec_model,
        table_rec_processor,
        rec_model,
        rec_processor,
    )
    logger.info(f"table_detection:\n{table_detection}")

(.venv) (base) root@v100gpu-0002:/mnt/data/llch/my_lm_log# python test/ocr/test_surya.py 
Loaded detection model /mnt/data/llch/surya/surya_det3 on device cuda with dtype torch.float16
Loaded recognition model /mnt/data/llch/surya/surya_rec2 on device cuda with dtype torch.float16
Loaded detection model /mnt/data/llch/surya/surya_layout4 on device cuda with dtype torch.float16
Loaded reading order model /mnt/data/llch/surya/surya_order on device cuda with dtype torch.float16
Loaded recognition model /mnt/data/llch/surya/surya_tablerec on device cuda with dtype torch.float16
INFO:__main__:surya模型加载耗时: 18.78秒
Detecting bboxes: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00,  2.33it/s]
Detecting bboxes: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00,  1.54it/s]
Detecting bboxes: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00,  5.48it/s]
Recognizing Text: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00,  1.26it/s]
Recognizing tables: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00,  6.09it/s]
INFO:__main__:surya_table_detection耗时: 2.47秒
INFO:__main__:table_detection:
[ExtractPageResult(cells=[[SpanTableCell(bbox=[24.0, 43.0, 163.0, 69.0], text='项目名称', row_ids=[0], col_ids=[0]), SpanTableCell(bbox=[230.0, 44.0, 1097.0, 73.0], text='(G)CT41G-2220-X7R-50V-22µF-K(N)电容器', row_ids=[0], col_ids=[1]), SpanTableCell(bbox=[23.0, 104.0, 161.0, 130.0], text='预计周期', row_ids=[1], col_ids=[0]), SpanTableCell(bbox=[441.0, 104.0, 730.0, 132.0], text='2023 年 02月 01日', row_ids=[1], col_ids=[1]), SpanTableCell(bbox=[229.0, 105.0, 362.0, 131.0], text='起始时间', row_ids=[1], col_ids=[1]), SpanTableCell(bbox=[804.0, 105.0, 937.0, 132.0], text='终止时间', row_ids=[1], col_ids=[1]), SpanTableCell(bbox=[1016.0, 105.0, 1300.0, 132.0], text='2023年12月 30日', row_ids=[1], col_ids=[4]), SpanTableCell(bbox=[228.0, 156.0, 441.0, 181.0], text='□换先研究项目', row_ids=[2], col_ids=[1]), SpanTableCell(bbox=[424.0, 201.0, 584.0, 228.0], text='四非系列化', row_ids=[3], col_ids=[2]), SpanTableCell(bbox=[228.0, 212.0, 407.0, 236.0], text='△A. 新产品', row_ids=[2], col_ids=[1]), SpanTableCell(bbox=[624.0, 212.0, 1361.0, 237.0], text='(□A1 全新产品 □A2 改进型新产品 ☑A3 扩展型新产', row_ids=[3], col_ids=[1]), SpanTableCell(bbox=[622.0, 252.0, 1272.0, 281.0], text='品 □A4 降低成本型新产品 □A5 仿制型新产品)', row_ids=[4], col_ids=[1]), SpanTableCell(bbox=[228.0, 255.0, 363.0, 278.0], text='、新材料', row_ids=[2], col_ids=[1]), SpanTableCell(bbox=[428.0, 256.0, 583.0, 283.0], text='□系列型谱', row_ids=[4], col_ids=[2]), SpanTableCell(bbox=[24.0, 304.0, 160.0, 331.0], text='项目类型', row_ids=[2], col_ids=[0]), SpanTableCell(bbox=[227.0, 319.0, 1064.0, 345.0], text='cB. 工艺改进cC. 基础预研aD. 设备开发及工装夹具开发', row_ids=[5], col_ids=[1]), SpanTableCell(bbox=[621.0, 373.0, 927.0, 396.0], text='□政府项目 归口单位:', row_ids=[6], col_ids=[1]), SpanTableCell(bbox=[227.0, 414.0, 379.0, 440.0], text='國内部项目', row_ids=[6], col_ids=[1]), SpanTableCell(bbox=[426.0, 414.0, 577.0, 440.0], text='ロ外部项目', row_ids=[6], col_ids=[2]), SpanTableCell(bbox=[619.0, 415.0, 928.0, 440.0], text='□纵向项目 归口单位:', row_ids=[7], col_ids=[1]), SpanTableCell(bbox=[621.0, 458.0, 928.0, 484.0], text='□横向项目 合作单位:', row_ids=[8], col_ids=[1]), SpanTableCell(bbox=[20.0, 510.0, 164.0, 537.0], text='成果形式', row_ids=[9], col_ids=[0]), SpanTableCell(bbox=[227.0, 515.0, 1247.0, 544.0], text='A. 样品( )B. 产品(1) C. 专利( )D. 论文( )E. 其它( )', row_ids=[9], col_ids=[1]), SpanTableCell(bbox=[21.0, 568.0, 166.0, 596.0], text='完成形式', row_ids=[10], col_ids=[0]), SpanTableCell(bbox=[226.0, 568.0, 493.0, 596.0], text='产品通过设计定型', row_ids=[10], col_ids=[1])]], rows_cols=[TableResult(cells=[TableCell(bbox=[24.0, 43.0, 163.0, 69.0], text='项目名称'), TableCell(bbox=[230.0, 44.0, 1097.0, 73.0], text='(G)CT41G-2220-X7R-50V-22µF-K(N)片式多层瓷介固定电容器'), TableCell(bbox=[23.0, 104.0, 161.0, 130.0], text='预计周期'), TableCell(bbox=[441.0, 104.0, 730.0, 132.0], text='2023 年 02月 01日'), TableCell(bbox=[229.0, 105.0, 362.0, 131.0], text='起始时间'), TableCell(bbox=[804.0, 105.0, 937.0, 132.0], text='终止时间'), TableCell(bbox=[1016.0, 105.0, 1300.0, 132.0], text='2023年12月 30日'), TableCell(bbox=[228.0, 156.0, 441.0, 181.0], text='□换先研究项目'), TableCell(bbox=[424.0, 201.0, 584.0, 228.0], text='四非系列化'), TableCell(bbox=[228.0, 212.0, 407.0, 236.0], text='△A. 新产品'), TableCell(bbox=[624.0, 212.0, 1361.0, 237.0], text='(□A1 全新产品 □A2 改进型新产品 ☑A3 扩展型新产'), TableCell(bbox=[622.0, 252.0, 1272.0, 281.0], text='品 □A4 降低成本型新产品 □A5 仿制型新产品)'), TableCell(bbox=[228.0, 255.0, 363.0, 278.0], text='、新材料'), TableCell(bbox=[428.0, 256.0, 583.0, 283.0], text='□系列型谱'), TableCell(bbox=[24.0, 304.0, 160.0, 331.0], text='项目类型'), TableCell(bbox=[227.0, 319.0, 1064.0, 345.0], text='cB. 工艺改进cC. 基础预研aD. 设备开发及工装夹具开发'), TableCell(bbox=[621.0, 373.0, 927.0, 396.0], text='□政府项目 归口单位:'), TableCell(bbox=[227.0, 414.0, 379.0, 440.0], text='國内部项目'), TableCell(bbox=[426.0, 414.0, 577.0, 440.0], text='ロ外部项目'), TableCell(bbox=[619.0, 415.0, 928.0, 440.0], text='□纵向项目 归口单位:'), TableCell(bbox=[621.0, 458.0, 928.0, 484.0], text='□横向项目 合作单位:'), TableCell(bbox=[20.0, 510.0, 164.0, 537.0], text='成果形式'), TableCell(bbox=[227.0, 515.0, 1247.0, 544.0], text='A. 样品( )B. 产品(1) C. 专利( )D. 论文( )E. 其它( )'), TableCell(bbox=[21.0, 568.0, 166.0, 596.0], text='完成形式'), TableCell(bbox=[226.0, 568.0, 493.0, 596.0], text='产品通过设计定型')], rows=[TableRow(bbox=[22.91015625, 43.265625, 1101.03515625, 72.515625], row_id=0), TableRow(bbox=[21.5625, 104.203125, 1299.140625, 131.015625], row_id=1), TableRow(bbox=[58.623046875, 157.21875, 458.876953125, 330.28125], row_id=2), TableRow(bbox=[227.75390625, 203.2265625, 1357.08984375, 236.7421875], row_id=3), TableRow(bbox=[222.36328125, 252.28125, 1270.83984375, 281.53125], row_id=4), TableRow(bbox=[235.83984375, 315.9609375, 1052.51953125, 342.1640625], row_id=5), TableRow(bbox=[240.556640625, 369.890625, 918.427734375, 441.796875], row_id=6), TableRow(bbox=[603.75, 414.6796875, 927.1875, 439.6640625], row_id=7), TableRow(bbox=[619.921875, 457.3359375, 927.1875, 483.5390625], row_id=8), TableRow(bbox=[18.8671875, 507.9140625, 1245.234375, 540.2109375], row_id=9), TableRow(bbox=[20.21484375, 567.9375, 494.58984375, 595.96875], row_id=10)], cols=[TableCol(bbox=[20.21484375, 42.9609375, 165.76171875, 594.4453125], col_id=0), TableCol(bbox=[222.36328125, 42.9609375, 1257.36328125, 594.4453125], col_id=1), TableCol(bbox=[431.923828125, 106.640625, 586.904296875, 441.796875], col_id=2), TableCol(bbox=[991.201171875, 103.59375, 1011.416015625, 470.4375], col_id=3), TableCol(bbox=[1014.111328125, 111.515625, 1298.466796875, 138.328125], col_id=4)], image_bbox=[0.0, 0.0, 1380.0, 624.0])], table_imgs=[<PIL.Image.Image image mode=RGB size=1380x624 at 0x7FA541791FD0>])]
-> Cannot close object, library is destroyed. This may cause a memory leak!
-> Cannot close object, library is destroyed. This may cause a memory leak!

-> Cannot close object, library is destroyed. This may cause a memory leak!

how can i sovle it

[VikParuchuri]

This is an issue with pypdfium2 - it needs to be the first import in your code. It's just a warning though.