tradelog

#!/bin/sh

# IOS project 1
# author: Patrik Korytar
# school: FIT VUT
# date: 28.3.2021

export POSIXLY_CORRECT=yes
export LC_ALL=C

### FUNCTIONS

# print help
print_help()
{
  echo "Usage:  ./tradelog [-h|--help]"
  echo "        ./tradelog [FILTER] [COMMAND] [LOG [LOG [...]]"
  echo "Analyze stock-exchange records."
  echo ""
  echo "Options:"
  echo "        -h / --help -- print help"
  echo ""
  echo "Filters:"
  echo "        -a DATETIME -- print just records after this date and time"
  echo "        -b DATETIME -- print just records before this date and time"
  echo "        -t TICKER   -- print just this ticker (multiple can be set)"
  echo "        -w WIDTH    -- set width of histogram / graph"
  echo ""
  echo "Commands (can be one of):"
  echo "        list-tick   -- list all tickers"
  echo "        profit      -- print overall profit from all positions"
  echo "        pos         -- list values of current tickers' positions"
  echo "        last-price  -- list last known tickers' prices"
  echo "        hist-ord    -- show histogram of number of tickers' transactions"
  echo "        graph-pos   -- show graph of values of current tickers' positions"
}

# check if file exists; if not, end program
file_exists()
{
  find "$1" 1>/dev/null 2>/dev/null
  if [ "$?" -eq 1 ] # if the above find failed, ie if there is no such a file
  then
    echo "tradelog: $1: no such file"
    exit 1
  fi
}

# compare two dates; return: 1 if first date is newer, 2 if second date is newer, 0 if same
compare_date()
{
  awk -v first="$1" -v second="$2" 'BEGIN{ if(first > second) print 1; else if(fist < second) print 2; else print 0 }'
}

### VARIABLES

DATETIME_AFTER="0000-00-00 00:00:00"
DATIME_BEFORE="9999-99-99 99:99:99"
TICKERS=""
WIDTH=""

CMD=""

LOG_LOADED=0 # 0 if not loaded, 1 if loaded
DATA=""

### PROCESS ARGUMENTS

while [ "$#" -gt 0 ] # while there are arguments left
do
  case "$1" in
  -h | --help)
    print_help
    exit 0
    ;;
  -a) # if multiple -a are given, intersection is set (ie the newest date)
    CMP=$(compare_date "$2" "$DATETIME_AFTER")
    if [ "$CMP" -eq 1 ] # if date in argument is newer than current date
    then
      DATETIME_AFTER="$2"
    fi

    shift # shift -a argument
    shift # shift DATETIME argument
    ;;
  -b) # if multiple -b are given, intersection is set (ie the oldest date)
    CMP=$(compare_date "$2" "$DATIME_BEFORE")
    if [ "$CMP" -eq 2 ] # if date in argument is older than current date
    then
      DATIME_BEFORE="$2"
    fi

    shift
    shift
    ;;
  -t)
    TICKERS="$TICKERS|$2"
    shift
    shift
    ;;
  -w)
    if [ "$WIDTH" ] # max one WIDTH allowed
    then
      echo "tradelog: WIDTH already set"
      exit 1
    fi

    if ! [ "$2" -eq "$2" ] 2>/dev/null # WIDTH not a number
    then
      echo "tradelog: WIDTH must be a number"
      exit 1
    fi

    if [ "$2" -lt 0 ] # WIDTH less than zero
    then
      echo "tradelog: WIDTH must be positive number"
      exit 1
    fi

    WIDTH="$2"
    shift
    shift
    ;;
  list-tick | profit | pos | last-price | hist-ord | graph-pos)
    if [ "$CMD" ] # max one COMMAND allowed
    then
      echo "tradelog: COMMAND already set"
      exit 1
    fi

    CMD="$1"
    shift
    ;;
  *.gz )
    file_exists "$1"
    DATA="${DATA}$(gzip -d -c "$1")" # load from decompressed .gz file and append to rest of loaded data
    LOG_LOADED=1
    shift
    ;;
  *)
    file_exists "$1"
    DATA=$(echo "$DATA" | cat - "$1") # load from file and append to rest of loaded data
    LOG_LOADED=1
    shift
    ;;
  esac
done

if [ "$LOG_LOADED" -eq 0 ] # if no file specified
then
  DATA=$(cat -) # load from stdin
fi

### FILTERS

DATA=$(echo "$DATA" | awk -F \; -v datetime_after="$DATETIME_AFTER" '{ if($1 > datetime_after) print $0; }')
DATA=$(echo "$DATA" | awk -F \; -v datetime_before="$DATIME_BEFORE" '{ if($1 < datetime_before) print $0; }')

if [ -z "$TICKERS" ] # if no ticker specified
then
  TICKERS=".*" # regex for all tickers
fi
DATA=$(echo "$DATA" | awk -F \; -v filter="^($TICKERS)$" '{ if($2 ~ filter) print $0; }') # filter tickers using regex

### COMMANDS

case "$CMD" in
list-tick) # get all tickers, sort them and print just unique
  DATA=$(echo "$DATA" | awk -F \; '{ print $2; }' | sort | uniq)
  ;;
profit) # compute overall profit and print it
  DATA=$(echo "$DATA" | awk -F \; \
  ' {
      if($3 == "sell") sold += $4 * $6;
      if($3 == "buy") bought += $4 * $6;
    }

    END {
      printf("%.2f", sold - bought);
    } ')
  ;;
pos) # sort dates, compute values of current tickers' positions and print for each ticker sorted using values
  DATA=$(echo "$DATA" | sort -t \; -k1,1 | awk -F \; \
  ' {
      if($3 == "buy") count[$2] += $6;
      if($3 == "sell") count[$2] -= $6;
      price[$2] = $4; # dates are sorted, so last price is always last for each ticker
    }

    END {
      numlen = 0;
      for(i in count)
      {
        pos[i] = sprintf("%.2f", count[i] * price[i]);
        if(length(pos[i]) > numlen) numlen = length(pos[i]);
      }

      for(i in pos)
        printf("%-10s: %*.2f\n", i, numlen, pos[i]);
    } ' | sort -n -r -t : -k2,2)
  ;;
last-price) # sort dates, find last known tickers' prices and print for each ticker sorted
  DATA=$(echo "$DATA" | sort -t \; -k1,1 | awk -F \; \
  ' {
      price[$2] = sprintf("%.2f", $4); # dates are sorted, so last price is always last for each ticker
    }

    END {
      numlen = 0;
      for(i in price)
        if(length(price[i]) > numlen) numlen = length(price[i]);
      for(i in price)
        printf("%-10s: %*.2f\n", i, numlen, price[i]);
    } ' | sort)
  ;;
hist-ord) # compute number of tickers' transactions and show histogram for each ticker sorted
  DATA=$(echo "$DATA" | awk -F \; -v width="$WIDTH" \
  ' {
      count[$2]++;
    }

    END {
      if(length(width) != 0) # only needed when WIDTH set
      {
        max = 0;
        for(i in count)
          if(count[i] > max) max = count[i];
      }

      for(i in count)
      {
        printf("%-10s:", i);

        if(length(width) == 0) # WIDTH not set
          len = count[i];
        else # WIDTH set
          len = int((count[i] / max) * width);

        if(len != 0) # print space only if histogram follows
          printf(" ");
        for(j = 0; j < len; j++)
          printf("#");

        printf("\n");
      }
    } ' | sort)
  ;;
graph-pos) # sort dates, compute values of current tickers' positions and show graph for each ticker sorted
  DATA=$(echo "$DATA" | sort -t \; -k1,1 | awk -F \; -v width="$WIDTH" \
  ' function abs(x)
    {
      return x < 0 ? -x : x;
    }

    {
      if($3 == "buy") count[$2] += $6;
      if($3 == "sell") count[$2] -= $6;
      price[$2] = $4; # dates are sorted, so last price is always last for each ticker
    }

    END {
      max = 0;
      for(i in count)
      {
        pos[i] = count[i] * price[i];
        if(length(width) != 0) # only needed when WIDTH set
          if(abs(pos[i]) > max) max = abs(pos[i]);
      }

      for(i in count)
      {
        printf("%-10s:", i);

        if(length(width) == 0) # WIDTH not set
          len = int(abs(pos[i]) / 1000);
        else # WIDTH set
          len = int((abs(pos[i]) / max) * width);

        if(len != 0) # print space only if graph follows
          printf(" ");
        for(j = 0; j < len; j++)
        {
          if(pos[i] > 0)
            printf "#";
          else
            printf "!";
        }

        printf("\n");
      }
    } ' | sort)
  ;;
esac

### PRINT ALL DATA

echo "$DATA"

exit 0