#!/bin/bash

abs_path=$(cd `dirname $0`; pwd)
hdfs_url="hdfs://192.168.141.208:9000"
hdfs_path="/dw/recommend/model/831_recsys_rov_train_data"
hdfs_feature_path="/dw/recommend/model/831_recsys_analysis_data"
data_path="${abs_path}/../data"
model_path="${abs_path}/../model"
monitor_py="${abs_path}/../src/tools/rec_monitor_push.py"
model_name="推荐模型str+_v1"

run_mode=""
if(($#==1))
then
    run_mode=$1
else
    exit -1
fi

if [[ "$run_mode" != "run" ]]
then
    exit -1
fi

# 0. check data
try_times=10
wait_time=300s
data_size_threshold=25000
feature_size_threshold=10000
data_date=$(date +%Y%m%d -d "-1 $days day")
before_data_date=$(date +%Y%m%d -d "-2 $days day")
for((i=0; i<=$try_times; i++))
do
  hdfs_file=$(printf "%s/%s/part-00063.gz" $hdfs_path $data_date)
  hdfs dfs -fs $hdfs_url -test -e $hdfs_file
  if [ $? -ne 0 ]
  then
    if [ $i -ge $try_times ]
    then
      echo `date` "$hdfs_file is not exist"
      exit -1
    else
      echo `date` "check $i, $hdfs_file is not exist, sleep $wait_time"
      sleep $wait_time
    fi
  else
    echo `date` "$hdfs_file is exist"
    data_size=$(hdfs dfs -fs $hdfs_url -text $hdfs_file | wc -l | awk '{print $1}')
    if [ $data_size -le $data_size_threshold ]
    then
      level=error
      msg=" ${model_name}更新失败, $hdfs_file $data_size <= $data_size_threshold"
      #python3 $monitor_py --level "$level" --model "$model_name" --msg "$msg"
      echo `date` "$msg"
      exit -1
    fi
    break
  fi
done

# 0.1 download feature
hdfs_feature_file=$(printf "%s/%s/part-00000.gz" $hdfs_feature_path $data_date)
hdfs dfs -fs $hdfs_url -test -e $hdfs_feature_file
if [ $? -ne 0 ]
then
  echo `date` "$hdfs_feature_file is not exist"
  exit -1
else
  local_feature_file="${data_path}/feature_${data_date}.csv"
  hdfs dfs -fs $hdfs_url -text $hdfs_feature_file | grep -v -E '@wh@|@unknown|e1@|b8_6h|b8_12h|b9_6h|b9_12h|c9_.*cate|c9_lr1s' > $local_feature_file &
  wait
  sleep 30s
  feature_size=$(wc -l $local_feature_file | awk '{print $1}')
  if [ $feature_size -le $feature_size_threshold ]
  then
    exho `date` "$local_feature_file, $feature_size <= $feature_size_threshold"
    exit -1
  fi
fi

# 1. download data
down_sh="${abs_path}/download_data.sh"
echo `date` "sh +x $down_sh $before_data_date $data_date"
sh +x $down_sh $before_data_date $data_date &
wait
sleep 30s

# 2. train model
train_sh="${abs_path}/train_model_data.sh"
sub_model_path="${model_path}/${data_date}"
if [ ! -d $sub_model_path ]
then
  mkdir $sub_model_path
fi
train_start_date=$(date -d "$data_date -13 day" +"%Y%m%d")
train_end_date=$data_date
save_model_file="${sub_model_path}/model_fm_for_recsys_v1_str.txt"
echo `date` "sh +x $train_sh $train_start_date $train_end_date $save_model_file"
sh +x $train_sh $train_start_date $train_end_date $save_model_file $local_feature_file &
wait
sleep 30s

# 3. update model
update_sh="${abs_path}/update_model.sh"
echo `date` "sh +x $update_sh $save_model_file $model_name"
sh +x $update_sh $save_model_file $model_name &
wait
sleep 30s

# 4. remove data
remove_sh="${abs_path}/remove_data.sh"
sh +x $remove_sh
wait