File:SARS-CoV-2 infections in and around Poland en.svg

#!/bin/bash

# pl_sarscov2 - Polish neighbourhood SARS-CoV-2 pandemic data

# (C) Wikimedia Commons User:Boud, 2020 CC-BY-SA-4.0
# https://creativecommons.org/licenses/by-sa/4.0/
# https://creativecommons.org/2015/10/08/cc-by-sa-4-0-now-one-way-compatible-with-gplv3/

# This component downloads and tidies the data into files, and runs
# pl_sarscov2.m through octave.

# TODO: 2020-03-17: This script is hardwired to handle March. Minor
# generalisations to handle April, May or later will be needed.

## START in March 2020
START=4
## TODAY in March 2020
TODAY=26

DAILY=true
#DAILY=false

## exponential for comparison: value on first date
if (${DAILY}); then
    EXP_START=20
else
    EXP_START=100
fi
## exponential for comparison: +${PERCENT} per day
PERCENT=25

URL_PL="https://en.wikipedia.org/w/index.php?title=Template:2019%E2%80%9320_coronavirus_pandemic_data/Poland_medical_cases_chart&action=edit"

## EU terrestrial neighbours of PL
URL_DE="https://en.wikipedia.org/w/index.php?title=Template:2019%E2%80%9320_coronavirus_pandemic_data/Germany_medical_cases_chart&action=edit"
URL_LT="https://en.wikipedia.org/w/index.php?title=Template:2019%E2%80%9320_coronavirus_pandemic_data/Lithuania_medical_cases_chart&action=edit"
#CZ bar stacked
URL_CZ="https://en.wikipedia.org/w/index.php?title=Template:2019%E2%80%9320_coronavirus_pandemic_data/Czech_Republic_medical_cases_chart&action=edit"
#SK
URL_SK="https://en.wikipedia.org/w/index.php?title=Template:2019%E2%80%9320_coronavirus_pandemic_data/Slovakia_medical_cases_chart&action=edit"

## non-EU terrestrial neighbours of PL
# https://en.wikipedia.org/wiki/2020_coronavirus_pandemic_in_Belarus
URL_BY="https://en.wikipedia.org/w/index.php?title=Template:2019%E2%80%9320_coronavirus_pandemic_data/Belarus_medical_cases_chart&action=edit"
URL_RU="https://en.wikipedia.org/w/index.php?title=Template:2019%E2%80%9320_coronavirus_pandemic_data/Russia_medical_cases_chart&action=edit"
URL_UA="https://en.wikipedia.org/w/index.php?title=Template:2019%E2%80%9320_coronavirus_pandemic_data/Ukraine_medical_cases_chart&action=edit"

## set to YES to enable
RECOV=no

TMP_PL=/tmp/tmpsemilogpl.data

TMP_DE=/tmp/tmpsemilogde.data
TMP_LT=/tmp/tmpsemiloglt.data
TMP_CZ=/tmp/tmpsemilogcz.data
TMP_SK=/tmp/tmpsemilogsk.data

TMP_BY=/tmp/tmpsemilogby.data
TMP_RU=/tmp/tmpsemilogru.data
TMP_UA=/tmp/tmpsemilogua.data

TMP_OCT=/tmp/tmpsemilog_oct.data
TMP_WMEDIA=/tmp/tmpsemilog_wmedia.data

TMP1=$(mktemp /tmp/tmpsemilog_pl.XXXXXXX)
TMP2=$(mktemp /tmp/tmpsemilog_pl.XXXXXXX)
TMP3=$(mktemp /tmp/tmpsemilog_pl.XXXXXXX)
TMP4=$(mktemp /tmp/tmpsemilog_pl.XXXXXXX)
TMP5=$(mktemp /tmp/tmpsemilog_pl.XXXXXXX)
TMP6=$(mktemp /tmp/tmpsemilog_pl.XXXXXXX)
TMP7=$(mktemp /tmp/tmpsemilog_pl.XXXXXXX)
TMP8=$(mktemp /tmp/tmpsemilog_pl.XXXXXXX)
TMP_X=$(mktemp /tmp/tmpsemilog_pl.XXXXXXX)

function sleep_a_bit {
    ## We do not want to overload Wikimedia Foundation servers.
    sleep $(shuf -i 5-10|tail -n1) # wait 5 to 10 seconds
}

function do_download {
    wget -O ${TMP_PL} "${URL_PL}"
    sleep_a_bit
    wget -O ${TMP_DE} "${URL_DE}"
    sleep_a_bit
    wget -O ${TMP_LT} "${URL_LT}"
    sleep_a_bit
    wget -O ${TMP_CZ} "${URL_CZ}"
    sleep_a_bit
    wget -O ${TMP_SK} "${URL_SK}"
    sleep_a_bit
    wget -O ${TMP_BY} "${URL_BY}"
    sleep_a_bit
    wget -O ${TMP_RU} "${URL_RU}"
    sleep_a_bit
    wget -O ${TMP_UA} "${URL_UA}"
}

function russia_hack {
    ## Russia hack
    ## This relates to the greps and regular expressions used
    ## below for parsing the source files.
    mv -v ${TMP_RU} ${TMP_X}
    sed -e 's/Row|||2|3|/Row|2020-03-04||2|3|/' ${TMP_X} | \
        sed -e 's/Row|||3|10|/Row|2020-03-10||3|10|/' ${TMP_X} > \
            ${TMP_RU}
}

function grep_medical_cases_octave {
    TMP_XX=$1
    TMP_OUT=$2
    ## day-of-month; total lab-confirmed cases; deaths; recoveries
    grep "^{{[mM]edical cases chart/Row" ${TMP_XX} | \
        grep "2020-03" | \
        tr -d ' ' | \
        awk -F '|' '{print $2,$5,"0"$3,"0"$4}' | sed -e 's/2020-03-//' > ${TMP_OUT}
    printf "\n" >> ${TMP_OUT}
}

function grep_bar_chart_octave {
    TMP_XX=$1
    TMP_OUT=$2
    ## day-of-month; total lab-confirmed cases; deaths; recoveries
    grep "^{{[Bb]ar stacked" ${TMP_XX} | \
        grep "2020-03" | \
        sed -e "s;.*[Bb]ar stacked.*2020-03-\([0-9]*\)[^0-9][^:]*expr: *\([0-9][0-9]*\)/[^:]*expr: *\([0-9][0-9]*\)/[^:]*expr: *\([0-9][0-9]*\)/.*;\1 \2 \3 \4;" | \
        awk '{print $1,$2+$3+$4,$2,$3}' > ${TMP_OUT}
    printf "\n" >> ${TMP_OUT}
}


function plot_mediawiki_top {
    if ( [ -e ${TMP_PL} ] &&  [ -e ${TMP_DE} ] &&  [ -e ${TMP_LT} ]  &&  [ -e ${TMP_BY} ] &&  [ -e ${TMP_RU} ] &&  [ -e ${TMP_CZ} ] &&  [ -e ${TMP_SK} ]  &&  [ -e ${TMP_UA} ] ); then
        cat <<EOF > ${TMP_WMEDIA}
{{Graph:Chart
 | width = 500
 | height = 300
 | type = line
 | yGrid =
 | xGrid =
 | xAxisAngle = -90
 | yScaleType = log
 | yAxisMin = 1
 | colors = #e00000, #005000, #007000, #00a000, #70a080, #000090, #5000b0, #1050ff, #000000
EOF
        printf " | x = " >> ${TMP_WMEDIA}
        for ((d=${START}; d<=${TODAY}; d++)); do
            if [ ${d} -le 31 ]; then 
                printf "${d}/3" >> ${TMP_WMEDIA}
            elif [ "${d}" -le 61 ]; then
                e=$((d-31))
                printf "${e}/4" >> ${TMP_WMEDIA}
            elif [ "${d}" -le 92 ]; then
                e=$((d-61))
                printf "${e}/5" >> ${TMP_WMEDIA}
            else
                printf "It is now June\! Please update the script\!"
            fi
            if [ ${d} -lt ${TODAY} ]; then
                printf ", " >> ${TMP_WMEDIA}
            fi
        done
        printf "\n" >> ${TMP_WMEDIA}
    else
        printf "${TMP_PL} or ${TMP_DE} or ... does not exist.\n"
        printf "Try the 'w' parameter to download.\n"
    fi
}

function plot_mediawiki_xx {
    printf " | $1= " >> ${TMP_WMEDIA}
    TMP_XX=$2
    TMP_OUT=$3    
    ## day-of-month; total lab-confirmed cases; deaths; recoveries
    ## TODO: update for April, May, ...
    grep "^{{[mM]edical cases chart/Row" ${TMP_XX} | \
        grep "2020-03" | \
        tr -d ' ' | \
        awk -F '|' '{print $2,$5}' | sed -e 's/2020-03-//' > ${TMP_OUT}
    n_old=0
    for ((d=${START}; d<=${TODAY}; d++)); do
        if (${DAILY}); then
            n_new=$(cat ${TMP_OUT} | \
                           awk -v d=${d} '{if ($1 == d) printf $2}')

            if [ "x${n_new}" = "x" ]; then
                # use the old value if the new value is invalid
                printf ""
                n_new=${n_old}
            fi

            delta_n=$((${n_new}-${n_old}))            
            printf "${delta_n}" >> ${TMP_WMEDIA}

            if [ "x${n_new}" != "x" ]; then
                # copy a valid new value to the old one
                n_old=${n_new}
            fi
            
            ## debug only
            #printf "n_new=${n_new}; n_old=${n_old}; delta_n=${delta_n}\n"
            
            if [ ${d} -lt ${TODAY} ]; then
                printf ", " >> ${TMP_WMEDIA}
            fi
        else
            cat ${TMP_OUT} | \
                awk -v d=${d} '{if ($1 == d) printf $2}' >> ${TMP_WMEDIA}
            if [ ${d} -lt ${TODAY} ]; then
                printf ", " >> ${TMP_WMEDIA}
            fi
        fi
    done
    printf "\n" >> ${TMP_WMEDIA}
}

function plot_mediawiki_exponential {
    ## avoid FR/PL language environment with decimal comma!
    OLD_LANG=${LANG}
    LANG=C
    EXP_FINAL=$(echo "x" | awk -v n1=${EXP_START} -v power="1.${PERCENT}" -v days=$((${TODAY}-${START})) \
                               '{print int(n1*power^days+0.5)}')
    printf " | y9= ${EXP_START}, " >> ${TMP_WMEDIA}
    for ((d=$((${START}+1)); d<${TODAY}; d++)); do
        printf ", "  >> ${TMP_WMEDIA}
    done
    printf "${EXP_FINAL}\n" >> ${TMP_WMEDIA}

    LANG=${OLD_LANG}
}


function plot_mediawiki {
    plot_mediawiki_top

    plot_mediawiki_xx "y1" ${TMP_PL} ${TMP_OCT}.PL
    plot_mediawiki_xx "y2" ${TMP_DE} ${TMP_OCT}.DE
    plot_mediawiki_xx "y3" ${TMP_CZ} ${TMP_OCT}.CZ
    plot_mediawiki_xx "y4" ${TMP_SK} ${TMP_OCT}.SK
    plot_mediawiki_xx "y5" ${TMP_LT} ${TMP_OCT}.LT
    plot_mediawiki_xx "y6" ${TMP_RU} ${TMP_OCT}.RU
    plot_mediawiki_xx "y7" ${TMP_BY} ${TMP_OCT}.BY
    plot_mediawiki_xx "y8" ${TMP_UA} ${TMP_OCT}.UA

    plot_mediawiki_exponential

    plot_mediawiki_bottom
}

function plot_mediawiki_bottom {
    cat <<EOF >> ${TMP_WMEDIA}
 | legend =
 | showSymbols =
 | y1Title = PL
 | y2Title = DE
 | y3Title = CZ
 | y4Title = SK
 | y5Title = LT
 | y6Title = RU
 | y7Title = BY
 | y8Title = UA
 | y9Title = +${PERCENT}%/day
 | xAxisTitle=day/month in 2020
 | yAxisTitle=count
 }}
EOF
}

function plot_octave {
    if ( [ -e ${TMP_PL} ] &&  [ -e ${TMP_DE} ] &&  [ -e ${TMP_LT} ]  &&  [ -e ${TMP_BY} ] &&  [ -e ${TMP_RU} ] ); then
        grep_medical_cases_octave ${TMP_PL} ${TMP_OCT}.PL
        grep_medical_cases_octave ${TMP_DE} ${TMP_OCT}.DE
        grep_medical_cases_octave ${TMP_LT} ${TMP_OCT}.LT
        grep_medical_cases_octave ${TMP_CZ} ${TMP_OCT}.CZ # updated 2020-03-19
        grep_medical_cases_octave ${TMP_SK} ${TMP_OCT}.SK # updated 2020-03-19
        grep_medical_cases_octave ${TMP_BY} ${TMP_OCT}.BY
        grep_medical_cases_octave ${TMP_RU} ${TMP_OCT}.RU
        grep_medical_cases_octave ${TMP_UA} ${TMP_OCT}.UA

        #octave --persist --no-gui pl_sarscov2.m # debug
        octave --no-gui pl_sarscov2.m
    else
        printf "${TMP_PL} or ${TMP_DE} or ... does not exist.\n"
        printf "Try the 'w' parameter to download.\n"
    fi
}

## parameter 'w' to 'wget' a fresh copy
if [ "x$1" = "xw" ]; then
    do_download
    russia_hack
fi

#This is to generate .svg plots.
plot_octave

#This is to generate data for using with mediawiki/lua {{Graph:Chart}}.
#plot_mediawiki

rm -f /tmp/tmpsemilog_pl.*