# Copyright 2019 Google LLC
#
# 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
#
#     https://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 functools
import os

import numpy as np
import tensorflow as tf
from absl import app
from absl import flags

from fixmatch import FixMatch
from libml import data, utils, augment, ctaugment

FLAGS = flags.FLAGS


class AugmentPoolCTACutOutStyle(augment.AugmentPoolCTA):
    @staticmethod
    def numpy_apply_policies(arglist):
        x, cta, probe, anchoring = arglist
        if x.ndim == 3:
            assert probe
            policy = cta.policy(probe=True)
            return dict(policy=policy,
                        probe=ctaugment.apply(x, policy),
                        image=x)
        assert not probe
        cutout_policy = lambda: cta.policy(probe=False) + [ctaugment.OP('cutout', (1,))]
        aug0 = [x[0]] if anchoring[0] == 'weak' else [ctaugment.apply(x[0], cutout_policy())]
        aug1 = [y for y in x[1:]] if anchoring[1] == 'weak' else [ctaugment.apply(y, cutout_policy()) for y in x[1:]]
        return dict(image=np.stack(aug0 + aug1).astype('f'))

    def queue_images(self):
        batch = self.get_samples()
        args = [(x, batch['cta'], batch['probe'], batch['anchoring']) for x in batch['image']]
        self.queue.append(augment.PoolEntry(payload=augment.POOL.imap(self.numpy_apply_policies, args), batch=batch))


class AB_FixMatch_Anchoring(FixMatch):
    AUGMENT_POOL_CLASS = AugmentPoolCTACutOutStyle

    def gen_labeled_fn(self, data_iterator):
        def wrap():
            batch = self.session.run(data_iterator)
            batch['cta'] = self.augmenter
            batch['probe'] = True
            batch['anchoring'] = 'weak.strong'.split('.')
            return batch

        return self.AUGMENT_POOL_CLASS(wrap)

    def gen_unlabeled_fn(self, data_iterator):
        def wrap():
            batch = self.session.run(data_iterator)
            batch['cta'] = self.augmenter
            batch['probe'] = False
            batch['anchoring'] = FLAGS.anchoring.split('.')
            return batch

        return self.AUGMENT_POOL_CLASS(wrap)

    def model(self, batch, lr, wd, wu, confidence, uratio, ema=0.999, **kwargs):
        hwc = [self.dataset.height, self.dataset.width, self.dataset.colors]
        xt_in = tf.placeholder(tf.float32, [batch] + hwc, 'xt')  # Training labeled
        x_in = tf.placeholder(tf.float32, [None] + hwc, 'x')  # Eval images
        y_in = tf.placeholder(tf.float32, [batch * uratio, 2] + hwc, 'y')  # Training unlabeled (weak, strong)
        l_in = tf.placeholder(tf.int32, [batch], 'labels')  # Labels

        tf.summary.image('weak', y_in[:, 0])
        tf.summary.image('strong', y_in[:, 1])

        lrate = tf.clip_by_value(tf.to_float(self.step) / (FLAGS.train_kimg << 10), 0, 1)
        lr *= tf.cos(lrate * (7 * np.pi) / (2 * 8))
        tf.summary.scalar('monitors/lr', lr)

        # Compute logits for xt_in and y_in
        classifier = lambda x, **kw: self.classifier(x, **kw, **kwargs).logits
        skip_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
        x = utils.interleave(tf.concat([xt_in, y_in[:, 0], y_in[:, 1]], 0), 2 * uratio + 1)
        logits = utils.para_cat(lambda x: classifier(x, training=True), x)
        logits = utils.de_interleave(logits, 2 * uratio+1)
        post_ops = [v for v in tf.get_collection(tf.GraphKeys.UPDATE_OPS) if v not in skip_ops]
        logits_x = logits[:batch]
        logits_weak, logits_strong = tf.split(logits[batch:], 2)
        del logits, skip_ops

        # Labeled cross-entropy
        loss_xe = tf.nn.sparse_softmax_cross_entropy_with_logits(labels=l_in, logits=logits_x)
        loss_xe = tf.reduce_mean(loss_xe)
        tf.summary.scalar('losses/xe', loss_xe)

        # Pseudo-label cross entropy for unlabeled data
        pseudo_labels = tf.stop_gradient(tf.nn.softmax(logits_weak))
        loss_xeu = tf.nn.sparse_softmax_cross_entropy_with_logits(labels=tf.argmax(pseudo_labels, axis=1),
                                                                  logits=logits_strong)
        pseudo_mask = tf.to_float(tf.reduce_max(pseudo_labels, axis=1) >= confidence)
        tf.summary.scalar('monitors/mask', tf.reduce_mean(pseudo_mask))
        loss_xeu = tf.reduce_mean(loss_xeu * pseudo_mask)
        tf.summary.scalar('losses/xeu', loss_xeu)

        # L2 regularization
        loss_wd = sum(tf.nn.l2_loss(v) for v in utils.model_vars('classify') if 'kernel' in v.name)
        tf.summary.scalar('losses/wd', loss_wd)

        ema = tf.train.ExponentialMovingAverage(decay=ema)
        ema_op = ema.apply(utils.model_vars())
        ema_getter = functools.partial(utils.getter_ema, ema)
        post_ops.append(ema_op)

        train_op = tf.train.MomentumOptimizer(lr, 0.9, use_nesterov=True).minimize(
            loss_xe + wu * loss_xeu + wd * loss_wd, colocate_gradients_with_ops=True)
        with tf.control_dependencies([train_op]):
            train_op = tf.group(*post_ops)

        return utils.EasyDict(
            xt=xt_in, x=x_in, y=y_in, label=l_in, train_op=train_op,
            classify_raw=tf.nn.softmax(classifier(x_in, training=False)),  # No EMA, for debugging.
            classify_op=tf.nn.softmax(classifier(x_in, getter=ema_getter, training=False)))


def main(argv):
    utils.setup_main()
    del argv  # Unused.
    dataset = data.PAIR_DATASETS()[FLAGS.dataset]()
    log_width = utils.ilog2(dataset.width)
    model = AB_FixMatch_Anchoring(
        os.path.join(FLAGS.train_dir, dataset.name, AB_FixMatch_Anchoring.cta_name(), FLAGS.anchoring),
        dataset,
        lr=FLAGS.lr,
        wd=FLAGS.wd,
        arch=FLAGS.arch,
        batch=FLAGS.batch,
        nclass=dataset.nclass,
        wu=FLAGS.wu,
        confidence=FLAGS.confidence,
        uratio=FLAGS.uratio,
        scales=FLAGS.scales or (log_width - 2),
        filters=FLAGS.filters,
        repeat=FLAGS.repeat)
    model.train(FLAGS.train_kimg << 10, FLAGS.report_kimg << 10)


if __name__ == '__main__':
    utils.setup_tf()
    flags.DEFINE_float('confidence', 0.95, 'Confidence threshold.')
    flags.DEFINE_float('wd', 0.0005, 'Weight decay.')
    flags.DEFINE_float('wu', 1, 'Pseudo label loss weight.')
    flags.DEFINE_integer('filters', 32, 'Filter size of convolutions.')
    flags.DEFINE_integer('repeat', 4, 'Number of residual layers per stage.')
    flags.DEFINE_integer('scales', 0, 'Number of 2x2 downscalings in the classifier.')
    flags.DEFINE_integer('uratio', 7, 'Unlabeled batch size ratio.')
    flags.DEFINE_string('anchoring', 'weak.strong', 'Augmentation anchoring strategy.')
    FLAGS.set_default('augment', 'd.d.d')
    FLAGS.set_default('dataset', 'cifar10.3@250-1')
    FLAGS.set_default('batch', 64)
    FLAGS.set_default('lr', 0.03)
    FLAGS.set_default('train_kimg', 1 << 16)
    app.run(main)