# Copyright (c) 2022 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.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import os
import sys

__dir__ = os.path.dirname(os.path.abspath(__file__))
sys.path.append(__dir__)
sys.path.insert(0, os.path.abspath(os.path.join(__dir__, "../..")))

os.environ["FLAGS_allocator_strategy"] = "auto_growth"

import cv2
import json
import numpy as np
import time

import paddleocr.tools.infer.utility as utility
from paddleocr.tools.infer_kie_token_ser_re import make_input
from paddleocr.ppocr.postprocess import build_post_process
from paddleocr.ppocr.utils.logging import get_logger
from paddleocr.ppocr.utils.visual import draw_ser_results, draw_re_results
from paddleocr.ppocr.utils.utility import get_image_file_list, check_and_read
from paddleocr.ppstructure.utility import parse_args
from paddleocr.ppstructure.kie.predict_kie_token_ser import SerPredictor

logger = get_logger()


class SerRePredictor(object):
    def __init__(self, args):
        self.use_visual_backbone = args.use_visual_backbone
        self.ser_engine = SerPredictor(args)
        if args.re_model_dir is not None:
            postprocess_params = {"name": "VQAReTokenLayoutLMPostProcess"}
            self.postprocess_op = build_post_process(postprocess_params)
            (
                self.predictor,
                self.input_tensor,
                self.output_tensors,
                self.config,
            ) = utility.create_predictor(args, "re", logger)
        else:
            self.predictor = None

    def __call__(self, img):
        starttime = time.time()
        ser_results, ser_inputs, ser_elapse = self.ser_engine(img)
        if self.predictor is None:
            return ser_results, ser_elapse

        re_input, entity_idx_dict_batch = make_input(ser_inputs, ser_results)
        if self.use_visual_backbone == False:
            re_input.pop(4)
        for idx in range(len(self.input_tensor)):
            self.input_tensor[idx].copy_from_cpu(re_input[idx])

        self.predictor.run()
        outputs = []
        for output_tensor in self.output_tensors:
            output = output_tensor.copy_to_cpu()
            outputs.append(output)
        preds = dict(
            loss=outputs[1],
            pred_relations=outputs[2],
            hidden_states=outputs[0],
        )

        post_result = self.postprocess_op(
            preds, ser_results=ser_results, entity_idx_dict_batch=entity_idx_dict_batch
        )

        elapse = time.time() - starttime
        return post_result, elapse


def main(args):
    image_file_list = get_image_file_list(args.image_dir)
    ser_re_predictor = SerRePredictor(args)
    count = 0
    total_time = 0

    os.makedirs(args.output, exist_ok=True)
    with open(
            os.path.join(args.output, "infer.txt"), mode="w", encoding="utf-8"
    ) as f_w:
        for image_file in image_file_list:
            img, flag, _ = check_and_read(image_file)
            if not flag:
                img = cv2.imread(image_file)
                img = img[:, :, ::-1]
            if img is None:
                logger.info("error in loading image:{}".format(image_file))
                continue
            re_res, elapse = ser_re_predictor(img)
            re_res = re_res[0]

            res_str = "{}\t{}\n".format(
                image_file,
                json.dumps(
                    {
                        "ocr_info": re_res,
                    },
                    ensure_ascii=False,
                ),
            )
            f_w.write(res_str)
            if ser_re_predictor.predictor is not None:
                img_res = draw_re_results(
                    image_file, re_res, font_path=args.vis_font_path
                )
                img_save_path = os.path.join(
                    args.output,
                    os.path.splitext(os.path.basename(image_file))[0] + "_ser_re.jpg",
                )
            else:
                img_res = draw_ser_results(
                    image_file, re_res, font_path=args.vis_font_path
                )
                img_save_path = os.path.join(
                    args.output,
                    os.path.splitext(os.path.basename(image_file))[0] + "_ser.jpg",
                )

            cv2.imwrite(img_save_path, img_res)
            logger.info("save vis result to {}".format(img_save_path))
            if count > 0:
                total_time += elapse
            count += 1
            logger.info("Predict time of {}: {}".format(image_file, elapse))


if __name__ == "__main__":
    args = parse_args()
    args.mode = 'kie'
    args.use_visual_backbone = False
    args.kie_algorithm = 'LayoutXLM'
    args.re_model_dir = './models/re_vi_layoutxlm_xfund_infer'
    args.ser_model_dir = './models/ser_vi_layoutxlm_xfund_infer'
    args.ser_dict_path = './models/xfund_class_list.txt'
    args.vis_font_path = './models/simfang.ttf'
    args.ocr_order_method = "tb-yx"
    args.det_model_dir = './models/ch_PP-OCRv3_det_infer'
    args.rec_model_dir = './models/ch_PP-OCRv4_rec_infer'
    args.image_dir = './1.png'
    main(args)

import cv2
import time
from paddleocr.ppstructure.kie.predict_kie_token_ser_re import SerRePredictor
from paddleocr.ppstructure.utility import parse_args
from paddleocr.tools.infer_kie_token_ser_re import make_input


# 继承SerRePredictor类，重写__call__方法，实现同时输出Ser模型结果
class KIE(SerRePredictor):
    def __init__(self, para):
        super().__init__(para)

    def __call__(self, im):
        start_time = time.time()
        ser_results, ser_inputs, ser_elapse = self.ser_engine(im)
        if self.predictor is None:
            return ser_results, ser_elapse

        re_input, entity_idx_dict_batch = make_input(ser_inputs, ser_results)
        if not self.use_visual_backbone:
            re_input.pop(4)
        for idx in range(len(self.input_tensor)):
            self.input_tensor[idx].copy_from_cpu(re_input[idx])

        self.predictor.run()
        outputs = []
        for output_tensor in self.output_tensors:
            output = output_tensor.copy_to_cpu()
            outputs.append(output)
        preds = dict(
            loss=outputs[1],
            pred_relations=outputs[2],
            hidden_states=outputs[0],
        )

        post_result = self.postprocess_op(
            preds, ser_results=ser_results, entity_idx_dict_batch=entity_idx_dict_batch
        )

        elapse_time = time.time() - start_time
        return ser_results, post_result, elapse_time


if __name__ == "__main__":
    args = parse_args()
    args.mode = 'kie'
    args.show_log = True
    args.use_visual_backbone = False
    args.kie_algorithm = 'LayoutXLM'
    args.re_model_dir = './models/re_vi_layoutxlm_xfund_infer'
    args.ser_model_dir = './models/ser_vi_layoutxlm_xfund_infer'
    args.ser_dict_path = './models/xfund_class_list.txt'
    args.vis_font_path = './models/simfang.ttf'
    args.ocr_order_method = "tb-yx"
    args.det_model_dir = './models/ch_PP-OCRv3_det_infer'
    args.rec_model_dir = './models/ch_PP-OCRv4_rec_infer'
    args.det_limit_side_len = 1600,

    kie = KIE(args)
    img = cv2.imread('./1.png')
    ser_res, re_res, elapse = kie(img)
    print(ser_res, '\n', re_res, '\n', elapse)

import json


def transfer_xfun_data(json_path=None, output_file=None):
    with open(json_path, "r", encoding="utf-8") as fin:
        lines = fin.readlines()

    json_info = json.loads(lines[0])
    documents = json_info["documents"]
    with open(output_file, "w", encoding="utf-8") as fout:
        for idx, document in enumerate(documents):
            label_info = []
            img_info = document["img"]
            document = document["document"]
            image_path = img_info["fname"]

            for doc in document:
                x1, y1, x2, y2 = doc["box"]
                points = [[x1, y1], [x2, y1], [x2, y2], [x1, y2]]
                label_info.append(
                    {
                        "transcription": doc["text"],
                        "label": doc["label"],
                        "points": points,
                        "id": doc["id"],
                        "linking": doc["linking"],
                    }
                )

            fout.write(
                image_path + "\t" + json.dumps(label_info, ensure_ascii=False) + "\n"
            )

    print("===ok====")

ori_gt_path = r'D:\pycharmproject_2\kie\demo\dataset\zh.val.json'
output_path = r'D:\pycharmproject_2\kie\demo\dataset\xfun_val.json'

transfer_xfun_data(ori_gt_path, output_path)

# Copyright (c) 2020 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.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.


from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import os
import sys

__dir__ = os.path.dirname(os.path.abspath(__file__))
sys.path.append(__dir__)
sys.path.insert(0, os.path.abspath(os.path.join(__dir__, "..")))

import yaml
import paddle
import paddle.distributed as dist

from paddleocr.ppocr.data import build_dataloader, set_signal_handlers
from paddleocr.ppocr.modeling.architectures import build_model
from paddleocr.ppocr.losses import build_loss
from paddleocr.ppocr.optimizer import build_optimizer
from paddleocr.ppocr.postprocess import build_post_process
from paddleocr.ppocr.metrics import build_metric
from paddleocr.ppocr.utils.save_load import load_model
from paddleocr.ppocr.utils.utility import set_seed
from paddleocr.ppocr.modeling.architectures import apply_to_static
import paddleocr.tools.program as program
import paddleocr.tools.naive_sync_bn as naive_sync_bn

dist.get_world_size()


def main(config, device, logger, vdl_writer, seed):
    # init dist environment
    if config["Global"]["distributed"]:
        dist.init_parallel_env()

    global_config = config["Global"]

    # build dataloader
    set_signal_handlers()
    train_dataloader = build_dataloader(config, "Train", device, logger, seed)
    if len(train_dataloader) == 0:
        logger.error(
            "No Images in train dataset, please ensure\n"
            + "\t1. The images num in the train label_file_list should be larger than or equal with batch size.\n"
            + "\t2. The annotation file and path in the configuration file are provided normally."
        )
        return

    if config["Eval"]:
        valid_dataloader = build_dataloader(config, "Eval", device, logger, seed)
    else:
        valid_dataloader = None
    step_pre_epoch = len(train_dataloader)

    # build post process
    post_process_class = build_post_process(config["PostProcess"], global_config)

    # build model
    # for rec algorithm
    if hasattr(post_process_class, "character"):
        char_num = len(getattr(post_process_class, "character"))
        if config["Architecture"]["algorithm"] in [
            "Distillation",
        ]:  # distillation model
            for key in config["Architecture"]["Models"]:
                if (
                    config["Architecture"]["Models"][key]["Head"]["name"] == "MultiHead"
                ):  # for multi head
                    if config["PostProcess"]["name"] == "DistillationSARLabelDecode":
                        char_num = char_num - 2
                    if config["PostProcess"]["name"] == "DistillationNRTRLabelDecode":
                        char_num = char_num - 3
                    out_channels_list = {}
                    out_channels_list["CTCLabelDecode"] = char_num
                    # update SARLoss params
                    if (
                        list(config["Loss"]["loss_config_list"][-1].keys())[0]
                        == "DistillationSARLoss"
                    ):
                        config["Loss"]["loss_config_list"][-1]["DistillationSARLoss"][
                            "ignore_index"
                        ] = (char_num + 1)
                        out_channels_list["SARLabelDecode"] = char_num + 2
                    elif any(
                        "DistillationNRTRLoss" in d
                        for d in config["Loss"]["loss_config_list"]
                    ):
                        out_channels_list["NRTRLabelDecode"] = char_num + 3

                    config["Architecture"]["Models"][key]["Head"][
                        "out_channels_list"
                    ] = out_channels_list
                else:
                    config["Architecture"]["Models"][key]["Head"][
                        "out_channels"
                    ] = char_num
        elif config["Architecture"]["Head"]["name"] == "MultiHead":  # for multi head
            if config["PostProcess"]["name"] == "SARLabelDecode":
                char_num = char_num - 2
            if config["PostProcess"]["name"] == "NRTRLabelDecode":
                char_num = char_num - 3
            out_channels_list = {}
            out_channels_list["CTCLabelDecode"] = char_num
            # update SARLoss params
            if list(config["Loss"]["loss_config_list"][1].keys())[0] == "SARLoss":
                if config["Loss"]["loss_config_list"][1]["SARLoss"] is None:
                    config["Loss"]["loss_config_list"][1]["SARLoss"] = {
                        "ignore_index": char_num + 1
                    }
                else:
                    config["Loss"]["loss_config_list"][1]["SARLoss"]["ignore_index"] = (
                        char_num + 1
                    )
                out_channels_list["SARLabelDecode"] = char_num + 2
            elif list(config["Loss"]["loss_config_list"][1].keys())[0] == "NRTRLoss":
                out_channels_list["NRTRLabelDecode"] = char_num + 3
            config["Architecture"]["Head"]["out_channels_list"] = out_channels_list
        else:  # base rec model
            config["Architecture"]["Head"]["out_channels"] = char_num

        if config["PostProcess"]["name"] == "SARLabelDecode":  # for SAR model
            config["Loss"]["ignore_index"] = char_num - 1

    model = build_model(config["Architecture"])

    use_sync_bn = config["Global"].get("use_sync_bn", False)
    if use_sync_bn:
        if config["Global"].get("use_npu", False):
            naive_sync_bn.convert_syncbn(model)
        else:
            model = paddle.nn.SyncBatchNorm.convert_sync_batchnorm(model)
        logger.info("convert_sync_batchnorm")

    model = apply_to_static(model, config, logger)

    # build loss
    loss_class = build_loss(config["Loss"])

    # build optim
    optimizer, lr_scheduler = build_optimizer(
        config["Optimizer"],
        epochs=config["Global"]["epoch_num"],
        step_each_epoch=len(train_dataloader),
        model=model,
    )

    # build metric
    eval_class = build_metric(config["Metric"])

    logger.info("train dataloader has {} iters".format(len(train_dataloader)))
    if valid_dataloader is not None:
        logger.info("valid dataloader has {} iters".format(len(valid_dataloader)))

    use_amp = config["Global"].get("use_amp", False)
    amp_level = config["Global"].get("amp_level", "O2")
    amp_dtype = config["Global"].get("amp_dtype", "float16")
    amp_custom_black_list = config["Global"].get("amp_custom_black_list", [])
    amp_custom_white_list = config["Global"].get("amp_custom_white_list", [])
    if os.path.exists(
        os.path.join(config["Global"]["save_model_dir"], "train_result.json")
    ):
        try:
            os.remove(
                os.path.join(config["Global"]["save_model_dir"], "train_result.json")
            )
        except:
            pass
    if use_amp:
        AMP_RELATED_FLAGS_SETTING = {
            "FLAGS_max_inplace_grad_add": 8,
        }
        if paddle.is_compiled_with_cuda():
            AMP_RELATED_FLAGS_SETTING.update(
                {
                    "FLAGS_cudnn_batchnorm_spatial_persistent": 1,
                    "FLAGS_gemm_use_half_precision_compute_type": 0,
                }
            )
        paddle.set_flags(AMP_RELATED_FLAGS_SETTING)
        scale_loss = config["Global"].get("scale_loss", 1.0)
        use_dynamic_loss_scaling = config["Global"].get(
            "use_dynamic_loss_scaling", False
        )
        scaler = paddle.amp.GradScaler(
            init_loss_scaling=scale_loss,
            use_dynamic_loss_scaling=use_dynamic_loss_scaling,
        )
        if amp_level == "O2":
            model, optimizer = paddle.amp.decorate(
                models=model,
                optimizers=optimizer,
                level=amp_level,
                master_weight=True,
                dtype=amp_dtype,
            )
    else:
        scaler = None

    # load pretrain model
    pre_best_model_dict = load_model(
        config, model, optimizer, config["Architecture"]["model_type"]
    )

    if config["Global"]["distributed"]:
        model = paddle.DataParallel(model)
    # start train
    program.train(
        config,
        train_dataloader,
        valid_dataloader,
        device,
        model,
        loss_class,
        optimizer,
        lr_scheduler,
        post_process_class,
        eval_class,
        pre_best_model_dict,
        logger,
        step_pre_epoch,
        vdl_writer,
        scaler,
        amp_level,
        amp_custom_black_list,
        amp_custom_white_list,
        amp_dtype,
    )


def test_reader(config, device, logger):
    loader = build_dataloader(config, "Train", device, logger)
    import time

    starttime = time.time()
    count = 0
    try:
        for data in loader():
            count += 1
            if count % 1 == 0:
                batch_time = time.time() - starttime
                starttime = time.time()
                logger.info(
                    "reader: {}, {}, {}".format(count, len(data[0]), batch_time)
                )
    except Exception as e:
        logger.info(e)
    logger.info("finish reader: {}, Success!".format(count))


if __name__ == "__main__":
    config, device, logger, vdl_writer = program.preprocess(is_train=True)
    seed = config["Global"]["seed"] if "seed" in config["Global"] else 1024
    set_seed(seed)
    main(config, device, logger, vdl_writer, seed)
    # test_reader(config, device, logger)

# Copyright (c) 2020 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.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import os
import sys

__dir__ = os.path.dirname(os.path.abspath(__file__))
sys.path.append(__dir__)
sys.path.insert(0, os.path.abspath(os.path.join(__dir__, "..")))

import argparse

from paddleocr.tools.program import load_config, merge_config, ArgsParser
from paddleocr.ppocr.utils.export_model import export


def main():
    FLAGS = ArgsParser().parse_args()
    config = load_config(FLAGS.config)
    config = merge_config(config, FLAGS.opt)
    # export model
    export(config)


if __name__ == "__main__":
    main()