共通キーワードに基づいてデータを形式で解析します。

愚かな:

BEGIN{

    # split by , or .mp4
    FS=",|\\.mp4"

    # sort array by numeric value
    PROCINFO["sorted_in"] = "@val_num_asc"
}

# store count in a, store others as key, value pair in d[keyword]
{ a[$2]++; d[$2][$3]= $1 }

END{
    for( keyword in d ){
        printf( "a:%d:{", a[keyword] )
        for( i in d[keyword] )
            printf( "i:%d;s:4:\"%d\";", i-1, d[keyword][i] )
        printf( "}\n" )
    }
}

修正する

キーワードの順序を維持するには：

BEGIN {
    # split by , or .mp4
    FS=",|\\.mp4"

    # sort array by numeric value
    PROCINFO["sorted_in"] = "@val_num_asc"
}

{
    # insert next ordered number into loop_order upon new keyword
    if(!($2 in d))
        loop_order[k++] = $2

    # store count in a, store others as key, value pair in d[$2]
    a[$2]++; d[$2][$3]= $1
}

END{
    for (j = 0; j < length(loop_order); ++j) {
        keyword = loop_order[j]
        printf( "a:%d:{", a[keyword] )
        for( i in d[keyword] )
            printf( "i:%d;s:4:\"%d\";", i-1, d[keyword][i] )
        printf( "}\n" )
    }
}