Pig latin Limit运算符应用于每个组属性

Question

我试图仅根据每个州的人口数返回五个最大的地方。我也试图按州名对结果进行排序，每个州的地点按人口递减顺序列出。我目前只给了前五个州和不每个州五个最大的地方。

-- Groups places by state name.
group_by_state_name_populated_place_name =
    GROUP project_using_state_name
    BY (state::name, place::name);

-- Counts population for each place in every state.
count_population_for_each_place_in_every_state =
    FOREACH group_by_state_name_populated_place_name
    GENERATE group.state::name AS state_name,
             group.place::name AS name,
             COUNT(project_using_state_name.population) AS population;

-- Orders population in each group found above to enable the use of limit.
order_groups_of_states_and_population =
    ORDER count_population_for_each_place_in_every_state 
    BY state_name ASC, population DESC, name ASC;

-- Limit the top 5 population for each state BUT currently returning just the first 5 tuples of the previous one and not 5 of each state.
limit_population =
    LIMIT order_groups_of_states_and_population 5;

Answer 1

下面的代码段可能有帮助

inp_data = load 'input_data.csv' using PigStorage(',') AS (state:chararray,place:chararray,population:long);

req_stats = FOREACH(GROUP inp_data BY state) {
    ordered = ORDER inp_data BY population DESC;
    required = LIMIT ordered 5;
    GENERATE FLATTEN(required);
};

req_stats_ordered = ORDER req_stats BY state, population DESC;

DUMP req_stats_ordered;