tradelog

#!/bin/sh

export POSIXLY_CORRECT=yes
export LC_NUMERIC=en_US.UTF-8
export LC_ALL=C

#napoveda
print_help(){
  echo "NAME: tradelog - stock market log analyzer"
  echo "USAGE: tradelog [-h | --help]"
  echo "       tradelog [FILTER...] [COMMAND] [LOG...]"
  echo ""
  echo "COMMAND: "
  echo "   list-tick   -   print a list of stock symbols, so-called \"Tickers\""
  echo "   profit      -   print a total profit from closed positions"
  echo "   pos         -   print a list of values of currently held positions sorted in descending order by value"
  echo "   last-price  -   print the last known price for every ticker"
  echo "   hist-ord    -   print  a histogram of the number of transactions according to the ticker"
  echo "   graph-pos   -   print a graph of values of held positions according to the ticker"
  echo ""
  echo "FILTERS:"
  echo "   -a DATETIME -   after: only logs after this date are processed (excluding this date). DATETIME format is YYYY-MM-DD HH:MM:SS"
  echo "   -b DATETIME -   before: only logs before this date are processed (excluding this date)"
  echo "   -t TICKER   -   only entries corresponding to a given ticker are considered.
                           With multiple occurrences of the switch, the set of all listed ticker is taken"
  echo "   -w WIDTH    -   for the graph printing set its width, that means adjust the length of the longest line to WIDTH.
                           WIDTH has to be a positive integer. Multiple occurrences of the switch is an invalid execution"
  echo "   -h | --help -   print this help"
  echo ""
}

#Error vystup
err(){
    echo "E: $*" >>/dev/stderr
    exit 1
}
#V pripade pracovanie s prazdnym logom je vystup rovno prazdny a ukoncuje sa
empty_log(){
  EMPTY=$(eval "$*")
      if [ -z "$EMPTY" ]; then
          exit 0
      fi
}

#premenne pre vstup
COMMAND=""
LOG_FILES=""
GZ_LOG_FILES=""
B_TIME="9999-99-99"
A_TIME=""
TICKERS=""
WIDTH="0"
READ_INPUT=""
TIME_FORMAT="[0-9][0-9][0-9][0-9]-[0-1][0-9]-[0-3][0-9] [0-2][0-9]:[0-6][0-9]:[0-6][0-9]"

#zpracovat argumenty
while [ "$#" -gt 0 ]; do
  case "$1" in
  list-tick | pos | profit | last-price | hist-ord | graph-pos)
    if [ -n "$COMMAND" ]; then
        err "There can be used only one COMMAND at a time"
    fi
    COMMAND="$1"
    shift
    ;;
  -h)
    print_help
    exit 0
    ;;
  -w)
    if [ "$2" -le 0 ] || [ $WIDTH != "0" ]; then
        err "WIDTH has to be a positive integer without multiple occurrences"
    elif [ -z "$2" ]; then
      err "Missing argument"
    fi
    WIDTH="$2"
    shift
    shift #zpracovali sa 2 parametry subezne
    ;;
  -t)
    CHECK=$( echo "$2" | grep ".*;.*" ) #nemoze obsahovat ';'
    CHECK_2=$( echo "$2" | grep ".*[[:space:]].*" ) #nemoze obsahovat ' '
    if [ -n "$CHECK" ] || [ -n "$CHECK_2" ]; then
        err "Invalid ticker, cannot use ';' or ' '"
    elif [ -z "$2" ]; then
      err "Missing argument"
    fi

    if [ -z "$TICKERS" ]; then
      TICKERS="$2"
    else
      TICKERS="$2\|$TICKERS"
      #pipujeme tickery, '\' kvoli pouziti v uvozovkach eval\awk
    fi
    shift
    shift
    ;;
  -b)
    CHECK=$( echo "$2" | grep "$TIME_FORMAT")
    if [ -z "$CHECK" ]; then
        err "Invalid format of time, use YYYY-MM-DD HH:MM:SS"
    elif [ -z "$2" ]; then
      err "Missing argument"
    fi

    B_TIME="$2"
    shift
    shift
    ;;
  -a)
    CHECK=$( echo "$2" | grep "$TIME_FORMAT")
    if [ -z "$CHECK" ]; then
        err "Invalid format of time, use YYYY-MM-DD HH:MM:SS"
    elif [ -z "$2" ]; then
      err "Missing argument"
    fi

    A_TIME="$2"
    shift
    shift
    ;;
  *.gz)
    if ! [ -r "$1" ]; then
        err "Cannot read from file"
    fi
    GZ_LOG_FILES="$GZ_LOG_FILES $1"
    shift
    ;;
  *)
    if ! [ -r "$1" ]; then
        err "Cannot read from file"
    fi
    LOG_FILES="$LOG_FILES $1"
    shift
    ;;
  esac
done

#nacitanie logov
if [ -n "$GZ_LOG_FILES" ] && [ -n "$LOG_FILES" ]; then
    READ_INPUT="gzip -d -c $GZ_LOG_FILES | cat $LOG_FILES -"
    #logy a zipy
elif [ -n "$GZ_LOG_FILES" ]; then
    READ_INPUT="gzip -d -c $GZ_LOG_FILES"
    #zipy
elif [ -n "$LOG_FILES" ]; then
    READ_INPUT="cat $LOG_FILES"
    #logy
else
    READ_INPUT="cat"
    #stdin
fi

if [ -z "$TICKERS" ]; then
    TICK_FILTER="cat"
    #bez tikerov
else
    TICK_FILTER="grep '^.*;\($TICKERS\);'"
fi

#filtrace(bud pomocou casu alebo tickerov, medzerou oddelene premenne)
READ_FILTERED="$READ_INPUT | awk -F ';' '{ if ( \$1 > \"$A_TIME\" ) {print \$0} }'"
READ_FILTERED="$READ_FILTERED | awk -F ';' '{ if ( \$1 < \"$B_TIME\" ) {print \$0} }'"
READ_FILTERED="$READ_FILTERED | $TICK_FILTER"

#vypisanie dlzky retazca v druhom stlpci a zoradenie podla danej dlzky
LENGTH="awk -F ' ' '{ printf \"%s %d\n\", \$0, length(\$2) }' | sort -r -n -t ' ' -k 3,3"

#formatovanie vystupu vo funkciach pos a last-price, nastavenie : zarovnanie dolava a ' ' pred nejdlhsi retazec
SPACES="awk -F ' ' 'BEGIN { start = 1; max_len = 0 }
                    { if (start) { start = 0; max_len = \$3 };
                      printf \"%-10s: %*s\n\",\$1, max_len, \$2 }' "

#vypocet hodnot aktuálně držených pozic, vystup: TICKER HODNOTA(zaokruhlena na 2 des. miesta)
POS="awk -F ';' 'BEGIN { amount = 0; tick = 0; value = 0 }
                       { if ( tick != \$2 && tick )
                                  { printf \"%s %.2f\n\", tick, amount * value;
                                    amount = 0 };
                         if ( \$3 == \"sell\" ) { amount = amount - \$6 }
                            else { amount = amount + \$6 }
                         tick = \$2;
                         value = \$4 }
                 END { printf \"%s %.2f\n\",tick,amount * value }' "

#vykreslenie grafu pre hist-ord a graph-pos, prijima: TICKER POCET_ZNAKOV
GRAPH="awk -F ' ' '{ if (\$2 > 0) { sign = \"#\";
                                    repeat = \$2 }
                              else { sign = \"!\"
                                     repeat = -\$2 }
                     printf \"%-10s:\", \$1;
                     if ( repeat != 0 ) {printf \" \"}
                     for ( i = 0; i < repeat; i++) { printf\"%s\", sign}
                     printf \"\n\" }'"

if [ "$COMMAND" = "list-tick" ]; then
      eval "$READ_FILTERED | awk -F ';' '{ print \$2 }' | sort -u"

elif [ "$COMMAND" = "profit" ]; then
      PROFIT="awk -F ';' 'BEGIN { profit = 0 }
                   { value = \$4 * \$6;
                     if ( \$3 == \"sell\" ) { profit = profit + value }
                        else { profit = profit - value } }
                   END { printf \"%.2f\n\", profit } '"

      eval "$READ_FILTERED | $PROFIT"

elif [ "$COMMAND" = "pos" ]; then
      if [ "$READ_INPUT" != "cat" ]; then
          empty_log "$READ_FILTERED"
      fi

      eval "$READ_FILTERED | sort | sort -s -t ';' -k 2,2 | $POS | $LENGTH | $SPACES | sort -r -n -t ':' -k 2,2"

elif [ "$COMMAND" = "last-price" ]; then
      if [ "$READ_INPUT" != "cat" ]; then
        empty_log "$READ_FILTERED"
      fi

      LAST_PRICE="awk -F ';' 'BEGIN { tick = 0; value = 0 }
                              { if ( tick != \$2 && tick ) { printf \"%s %.2f\n\", tick, value};
                                tick = \$2;
                                value = \$4 }
                              END { printf \"%s %.2f\n\", tick, value }' "

      eval "$READ_FILTERED | sort | sort -s -t ';' -k 2,2 | $LAST_PRICE | $LENGTH | $SPACES | sort"

elif [ "$COMMAND" = "hist-ord" ]; then
      if [ "$READ_INPUT" != "cat" ]; then
        empty_log "$READ_FILTERED"
      fi

      HIST="awk -F ';' 'BEGIN { tick = 0; number = 0 }
                        { if ( tick != \$2 && tick ) { print tick, number; number = 0};
                          tick = \$2;
                          number = number + 1 }
                        END { print tick, number }' "

      if [ "$WIDTH" != 0 ]; then
          RATIO="sort -n -r -t ' ' -k 2,2 |
                 awk -F ' ' 'BEGIN { start = 1; ratio = 0 }
                             { if (start) { start = 0; ratio = $WIDTH / \$2 };
                               printf \"%s %d\n\", \$1, \$2*ratio }' |
                 sort"
      else
        RATIO="cat"
      fi

      eval "$READ_FILTERED | sort -t ';' -k 2,2 | $HIST | $RATIO | $GRAPH"

elif [ "$COMMAND" = "graph-pos" ]; then
      if [ "$READ_INPUT" != "cat" ]; then
        empty_log "$READ_FILTERED"
      fi

      if [ "$WIDTH" != 0 ]; then
        RATIO=" awk -F ' ' '{ if (\$2 < 0) { abs = -\$2 }
                                  else { abs = \$2 }
                              printf \"%s %.2f\n\", \$0, abs}' |
                sort -n -r -t ' ' -k 3,3 |
                awk -F ' ' 'BEGIN { start = 1; ratio = 0 }
                            { if (start) { start = 0;
                                           ratio = $WIDTH / \$3};
                              printf \"%s %d\n\", \$1, \$2 * ratio }' |
                sort"
      else
        #bez zadanej sirky je implicitne nastaveny 1 znak = 1000 + zaokruhlenie dole
        WIDTH=1000;
        RATIO="awk -F ' ' '{ printf \"%s %d\n\",\$1, \$2 / $WIDTH }' "
      fi

      eval "$READ_FILTERED | sort | sort -s -t ';' -k 2,2 | $POS | $RATIO | $GRAPH "

else
  eval "$READ_FILTERED"
fi