covid

#!/bin/sh -u
################################################################################
#-Get infecter of COVID-19 info in Japan.
#-
# [VERSION]        0.0.3
#
#+[USAGE]
#+  $0 [-hV]
#+  $0 [-ctu][-m 'cond']
#+  $0 [-tu][-f field][-m 'cond'] [field1 field2...]
#+
#-[OPTIONS]
#-  -c             print count of rows the match the condition.
#-
#-  -f field       print count of rows the match the condition each field items.
#-                 for can be used filed-name, see field-name.
#-
#-  -h             print script help.
#-
#-  -m cond        print only recode matched the conditions.
#-                 expression of awk can be used.
#-                 exp.) $0 -m 'age>10 && date~"2020/05"'
#-
#-  -t             print total count of rows in japan each of date
#-                 for can be used filed-name, see field-name.
#-
#-  -u             update database.
#-
#-  -V             print script version.
#-
#-  field          print target field. names can be used.
#-                   - id
#-                   - date
#-                   - area1
#-                   - area2
#-                   - area3
#-                   - age
#-                   - sex
#-                   - job
#-                   - note
#-                   - sts
#-
#-                 if use '-t'
#-                   - date
#-                   - inf
#-                   - infd
#-                   - die
#-                   - leav
#-                   - leavd
#-                   - pcr
#-                   - pcrd
#-
#-                 if not given field, print all fields.
#-
#-[EXIT-STATUS]
#-  0              no error
#-  1              parameter error
#-  2              process error
#-
#-[DEPENDENCY]
#-  - curl
#-  - https://dl.dropboxusercontent.com/s/6mztoeb6xf78g5w/COVID-19.csv
#-
#-[REFERENCE]
#-  - https://jag-japan.com/covid19map-readme/
#-
################################################################################
CMD_NAME=$(basename $0)

###################################################
# Configs
###################################################
# database
GET_CMD="curl -sSf"
DB_URL=https://dl.dropboxusercontent.com/s/6mztoeb6xf78g5w/COVID-19.csv
DB_DIR="${XDG_DATA_HOME:-${HOME}/.local/share/}/${CMD_NAME}"
[ ! -d "$DB_DIR" ] && mkdir -p "$DB_DIR"
RAW_DB_FILE="${DB_DIR}/COVID-19.csv"
ITEM_DB_FILE="${DB_DIR}/ITEM_DB_COVID-19.csv"
TOTAL_DB_FILE="${DB_DIR}/TOTAL_DB_COVID-19.csv"

###################################################
# Utilities
###################################################
usage_exit() {
  echo "$1" 1>&2
  sed -n '/^#+/s/^#+//p' "$0" | sed "s/\$0/${CMD_NAME}/g" 1>&2
  exit 1
}
help_exit() {
  sed -n '/^#[-+]/s/^#[-+]//p' "$0" | sed "s/\$0/${CMD_NAME}/g" 1>&2
  exit 0
}
err_exit() {
  echo "$1" 1>&2
  exit 2
}
version_exit() {
  sed -n "/^# \[VERSION\]/s/^# \[VERSION\] *//p" "$1" 1>&2
  exit 0
}

###################################################
# Analyze options & parameters
###################################################
# options
UPDATE_DB=
PICKUP_COND=
CNT_IN_FIELD=
CNT_OF_RECODE=
TOTAL_COUNT=
while getopts cf:hm:tuV OPT
do
  case $OPT in
    c ) CNT_OF_RECODE=1      ;;
    f ) CNT_IN_FIELD=$OPTARG ;;
    h ) help_exit "$0"       ;;
    m ) PICKUP_COND=$OPTARG  ;;
    t ) TOTAL_COUNT=1        ;;
    u ) UPDATE_DB=1          ;;
    V ) version_exit "$0"    ;;
    \? ) usage_exit "$0"     ;;
  esac
done
shift $(expr $OPTIND - 1)

# params
ERR_MSG=
if [ -n "$TOTAL_COUNT" ]; then
  for field in "$@";do
    case $field in
    date | inf | infd | die | leav | leavd | pcr | pcrd ) ;;
    * ) ERR_MSG="invalid field-name '$field'" ;;
    esac
  done
else
  for field in "$@";do
    case $field in
    id | date | area1 | area2 | area3 | age | sex | job | note | sts ) ;;
    * ) ERR_MSG="invalid field-name '$field'" ;;
    esac
  done
fi
[ -n "$ERR_MSG" ] && usage_exit "$ERR_MSG"

# make output format for awk
# "id date job age" --> "$1,$2,$8,$6"
item_name_to_pos(){
  sed -e 's/id/$1/g'    \
    -e 's/date/$2/g'    \
    -e 's/area1/$3/g'   \
    -e 's/area2/$4/g'   \
    -e 's/area3/$5/g'   \
    -e 's/age/$6/g'     \
    -e 's/sex/$7/g'     \
    -e 's/job/$8/g'     \
    -e 's/note/$9/g'    \
    -e 's/sts/$10/g'
}
total_name_to_pos(){
  sed -e 's/date/$1/g' \
    -e 's/infd/$3/g'   \
    -e 's/inf/$2/g'    \
    -e 's/die/$4/g'    \
    -e 's/leavd/$6/g'  \
    -e 's/leav/$5/g'   \
    -e 's/pcrd/$8/g'   \
    -e 's/pcr/$7/g'
}
NAME_TO_POS=item_name_to_pos
DB_FILE="$ITEM_DB_FILE"
if [ -n "$TOTAL_COUNT" ]; then
  NAME_TO_POS=total_name_to_pos
  DB_FILE="$TOTAL_DB_FILE"
fi

OUTPUT_ITEMS='$0'
if [ $# -gt 0 ]; then
  OUTPUT_ITEMS=`echo "$*" | $NAME_TO_POS | sed -e 's/[ ]\{1,\}/,/g'`
fi
# echo $OUTPUT_ITEMS

###################################################
# Main
###################################################
# get csv & make database
#==================================================
PICKUP_DB_ITEMS='$1,$40,$10,$11,$14,$6,$7,$32,$19,$18 "(" $3 ")"'
#PICKUP_DB_HEADER='id,date,area1,area2,area3,age,sex,job,note,sts'
PICKUP_DB_HEADER='ID,確定日,受診都道府県,居住地,居住地(詳細),年代,性別,職業,備考,状態'

TOTAL_DB_ITEMS='$40,$24,$25,$27,$28,$29,$30,$31'
#TOTAL_DB_HEADER='date,inf,infd,die,leav,leavd,pcr,pcrd'
TOTAL_DB_HEADER='確定日,感染者数累計,感染者数前日比,死者合計,退院数累計,退院数前日比,PCR検査実施人数,PCR検査実施人数前日比'

get_body_recode(){
  tr -d '\r'        \
    | sed '1d'      \
    | grep -v '^,'
}
delete_space(){
  sed 's/[  ]\{1,\},/,/g'     \
    | sed 's/,[  ]\{1,\}/,/g' \
    | sed 's/[  ]\{1,\}$//'
}
format_date(){
  sed 's@,\([0-9]\{4\}\)/\([1-9]\)/\([0-9]\{1,2\}\),@,\1/0\2/\3,@'  \
    | sed 's@,\([0-9]\{4\}\)/\([0-9][0-9]\)/\([0-9]\),@,\1/\2/0\3,@'
}
make_database(){
  get_body_recode   \
    | awk -F","     \
    '   BEGIN{ OFS="," }
        {
          if($19~"100歳以上"){$6=100}
          if($6==""){$6="不明"}
          if($6=="0-10"){$6="0"}
          print '"$PICKUP_DB_ITEMS"'
        }' \
    | sed 's/()$//' \
    | delete_space  \
    | format_date
}
make_total_database(){
  get_body_recode   \
    | delete_space  \
    | format_date   \
    | sort -t ',' -k 1 -nr \
    | awk -F"," \
    '   BEGIN{ OFS="," }
        $24!="" && $24!="#REF!" && !uniq[$40]++ {
          print '"$TOTAL_DB_ITEMS"'
        }'      \
    | sort -t ',' -k 1
}

# get raw database from web
if [ ! -f "$RAW_DB_FILE" -o -n "$UPDATE_DB" ]; then
  $GET_CMD "$DB_URL" > "$RAW_DB_FILE"
  if [ $? -ne 0 ]; then
    err_exit "can't db update error from $DB_URL"
  fi
  rm "$ITEM_DB_FILE" "$TOTAL_DB_FILE" 1>&2 2>/dev/null
fi

# make database for this script
if [ ! -f "$ITEM_DB_FILE" ]; then
  echo "$PICKUP_DB_HEADER" > "$ITEM_DB_FILE"
  cat "$RAW_DB_FILE"    \
    | make_database >> "$ITEM_DB_FILE"
fi

# make database for total count
if [ ! -f "$TOTAL_DB_FILE" ]; then
  echo "$TOTAL_DB_HEADER" > "${TOTAL_DB_FILE}"
  cat "$RAW_DB_FILE"    \
    | make_total_database >> "${TOTAL_DB_FILE}"
fi

# pickup & output items
#==================================================
PICKUP_COND=`echo $PICKUP_COND | $NAME_TO_POS \
  | sed 's/\([^!~><=]\)=\([^=]\)/\1==\2/g'`
if [ -n "$CNT_IN_FIELD" ]; then
  CNT_IN_FIELD=`echo $CNT_IN_FIELD | $NAME_TO_POS`
  # output count each of field
  cat "$DB_FILE"    \
    | sed '1d'      \
    | awk -F"," '
      BEGIN{ OFS="," }
      '"$PICKUP_COND"'{
        count['$CNT_IN_FIELD']++
      }
      END{
        total=0
        for (field in count) {
          print field,count[field]
          total+=count[field]
        }
        print ":TOTAL",total
      }'
elif [ -n "$CNT_OF_RECODE" ]; then
  # output count of pickup recodes
  cat "$DB_FILE"    \
    | sed '1d'      \
    | awk -F"," '
      BEGIN{ count=0 }
      '"$PICKUP_COND"'{
        count++
      }
      END{ print count }'
else
  # output pickup headr
  head -n 1 "$DB_FILE"  \
    | awk -F"," '
      BEGIN{ OFS="," }
      {
        print '"${OUTPUT_ITEMS}"'
      }'

  # output pickup recodes
  cat "$DB_FILE"    \
    | sed '1d'      \
    | awk -F"," '
      BEGIN{ OFS="," }
      '"$PICKUP_COND"'{
        print '"${OUTPUT_ITEMS}"'
      }'
fi