films_descriptions

#!/bin/bash

IN_FILE_LIST=$1
IN_FILE_DESCR=$2

if [ -z "$IN_FILE_LIST" -o -z "$IN_FILE_DESCR" -o -z "$3" ]
then
    echo "Format isn't correct!"
    echo "in_file_list in_file_descriptions out_file update(optional)"
    exit
fi

USER_AGENT="Mozilla/5.0 (Windows; U; Windows NT 5.1; AppleWebKit/534.1 (KHTML, like Gecko) Chrome/6.0.437.3 Safari/534.1"

#cat $1 | head -n 47 | tail -n 3 > /tmp/init_list_fd
DELIM=`grep -m 1 "2012" "$IN_FILE_DESCR" | cut -c 16`

FILMS_NUMBER=`grep -c -E "*" "$IN_FILE_LIST"`
FILMS_I=0
function print_progress() {
    FILMS_I=`expr $FILMS_I + 100`
    PROGRESS=`expr $FILMS_I / $FILMS_NUMBER`
    echo -en "\rProgress: $PROGRESS% "
}

function file_download() {
    if [[ -e $HOME/Dropbox/films_about/$NUMBER.html ]]
    then
        local SZOR=`grep -c "СЗоР" $HOME/Dropbox/films_about/$NUMBER.html`
    fi

    if [[ ! -e $HOME/Dropbox/films_about/$NUMBER.html || $SZOR > 0 ]]
    then
#         wget -q -O $HOME/Dropbox/films_about/$NUMBER.html -U "$USER_AGENT" \
#         http://m.kinopoisk.ru/movie/$NUMBER/
#         lynx
        elinks -source http://m.kinopoisk.ru/movie/$NUMBER/ \
        | iconv -f cp1251 -t utf8 > $HOME/Dropbox/films_about/$NUMBER.html
        echo "Film $FILM downloaded"
    fi

    if [ ! -e $HOME/Dropbox/films_images/$NUMBER.jpg ]
    then
        wget -q -O $HOME/Dropbox/films_images/$NUMBER.jpg -U "$USER_AGENT" \
        http://st.kinopoisk.ru/images/film/$NUMBER.jpg
    fi
}

function html_parsing() {
    head -n $1 "$FILE_ABOUT" | tail -n 1 | cut -d ">" -f2 | cut -d "<" -f1
}

echo "Название|ВЭБ|Дата|Длительность|Жанр|Страна|Оригинальный заголовок|Описание|Картинка" > /tmp/films_full.csv
echo "Обложка|Название|ВЭБ|Дата|Длительность|Жанр|Страна|Оригинальный заголовок|Описание" > /tmp/films_full_gd.csv

cat $IN_FILE_LIST | while read FILM
do
    print_progress
    STRING=`grep -m 1 "$FILM" "$IN_FILE_DESCR"`
    URL=`echo "$STRING" | cut -d "$DELIM" -f2`
    NUMBER=`echo $URL | cut -d "/" -f7`
    DATA_EXIST=`echo "$STRING" | cut -d "$DELIM" -f3`
    #echo "$FILM"
    #echo "$STRING"
    if [[ "$4" == "update" || "$DATA_EXIST" == "" ]]
    then
        file_download

        FILE_ABOUT="$HOME/Dropbox/films_about/$NUMBER.html"
        NAME_RU=`html_parsing 41`
        #echo $NAME_RU
        LINE_44=`html_parsing 42`

        GENRE=`html_parsing 46`
        COUNTRY=`html_parsing 47`
        DESCR=`grep "descr" "$FILE_ABOUT" | cut -d ">" -f2 | cut -d "<" -f1`
        PICT="$HOME/Dropbox/films_images/$NUMBER.jpg"
        #GENRE=`echo $GENRE | sed 's|,||g'`
        RUSSIA=`echo $COUNTRY | grep -c -E "Россия|СССР|Украина|Беларусь"`

        if [[ "$RUSSIA" -ne "0" ]]
        then
            NAME_EN="$NAME_RU"
            YEAR=`echo $LINE_44 | cut -d "," -f1`
            DURATION=`echo $LINE_44 | cut -d "," -f2`
        else
            NAME_EN=`echo $LINE_44 | cut -d "," -f1`
            YEAR=`echo $LINE_44 | cut -d "," -f2`
            if [[ "$YEAR" = " The" || "$YEAR" = " Le" \
                || "$YEAR" = " La" || "$YEAR" = " Man" || "$YEAR" = " Robot" ]]
            then
                NAME_EN="$NAME_EN,$YEAR"
                YEAR=`echo $LINE_44 | cut -d "," -f3`
                DURATION=`echo $LINE_44 | cut -d "," -f4`
            else
                DURATION=`echo $LINE_44 | cut -d "," -f3`
            fi
        fi
        IMG="=Image(\"http://st.kinopoisk.ru/images/film/$NUMBER.jpg\",1)"

        echo "$FILM|$URL|$YEAR|$DURATION|$GENRE|$COUNTRY|$NAME_EN|$DESCR|$PICT" >> /tmp/films_full.csv
        echo "$IMG|$FILM|$URL|$YEAR|$DURATION|$GENRE|$COUNTRY|$NAME_EN|$DESCR" >> /tmp/films_full_gd.csv
    else
        echo "$STRING" >> /tmp/films_full.csv
    fi
done
mv /tmp/films_full.csv $3
mv /tmp/films_full_gd.csv films_full_gd.csv
echo -e "\n"