====== Creating alternative XMLTV station lists for Hungary and Romania ======

The PERL xmltv grabber for .hu and .ro packaged with the original xmltv distribution is rather slow and buggy so I've decided to write my own using AWK and of course bash.

<code bash>
#!/bin/bash

WGET="wget -q -O- "

get_channel_list() {

    list=`$WGET "http://www.port.hu/pls/tv/tv.channel?i_ch=1&i_days=1&i_xday=1&i_where=1"`

#    echo -e "$list" | egrep -zo '<select name="i_ch".*</select>' | awk '/value="[0-9]+">.*<\/option>/ { print $1 }'
#    echo -e "$list" | tr -d '\n' | egrep -o '<select name="i_ch".*</select>' | sed
    echo -e "$list" | awk '\
BEGIN{a=0} \
/<select name="i_ch"/{a=1} \
a==1 && /value/ { \
    split($0,b,/value="/); \
    chan=b[2] + 0; \
    split($0,b,/value="[0-9]+">/); \
    split(b[2],c,/<\/option>/); \
    name=c[1]; \
    print chan "\t" name; \
} \
/<\/select>/{a=0} \
' | sort -n | sed "s/&/&amp;/g"

}

channels=`get_channel_list`

# print header
cat <<EOF
<?xml version="1.0" encoding="ISO-8859-1"?>
<!DOCTYPE tv SYSTEM "xmltv.dtd">

<tv source-info-url="http://www.port.hu/" source-data-url="http://www.port.hu" generator-info-name="XMLTV_MCREE" generator-info-url="http://www.tricon.hu/~mcree">

EOF

# print channel list
echo -e "$channels" | while read channel; do
    id=`echo -e "$channel" | cut -f1`
    name=`echo -e "$channel" | cut -f2`

    echo "<channel id=\"$id\">"
    echo "      <display-name lang=\"hu\">$name</display-name>"
    echo "</channel>"
done

# get programmes
for day in -1 +0 +1 +2 +3; do
offset=$(( $day + 1 ))
date=`date +%Y%m%d -d"now $day days"`
nextdate=`date +%Y%m%d -d"now $day days"`
zone=`date +%z`
echo -e "$channels" | while read channel; do
    id=`echo -e "$channel" | cut -f1`
    name=`echo -e "$channel" | cut -f2`

    programmes=`$WGET "http://www.port.hu/pls/tv/tv.channel?i_ch=$id&i_days=$offset&i_xday=1&i_where=1"`

    echo -e "$programmes" | awk '\
/<tr/{tr=1}
tr==1 && /<td.*<span class="btxt"/ {\
match($0,/([0-9]+:[0-9]+)/,a);\
split(a[1],d,/:/);\
if (d[1]!=0) {date="'$date'"} else {date="'$nextdate'"}; \
print "<programme start=\"" date d[1] d[2] "00 '$zone'\" channel=\"'$id'\">";\
getline;\
res=split($0,a,/(<[^>]*>)/);\
c=0;\
for (b=0;b<res;b++) \
    if (a[b] ~ /\w/) \
        if (c==0) {\
            c=1; \
            print "<title lang=\"hu\">" a[b] "</title>"\
        } else {\
            if (c==1) { c=2; print "<desc lang=\"hu\">"}\
            print a[b]\
        };\
    if (c>1) print "</desc>"; \
print "</programme>"; \
}\
/<\/tr>{tr=0}/
' | sed "s/&/&amp;/g"
done
done

# print footer
cat <<EOF

</tv>
EOF
</code>

{{tag>xmltv util bash awk}}

~~LINKBACK~~