data_files.proto

syntax = "proto3";

// articles

message StatisticRow {
    float statval = 1;
    repeated int32 ordinal_by_universe = 2;
    repeated int32 percentile_by_population_by_universe = 4;
}

message FirstOrLast {
    // index into article.rows
    int32 article_row_idx = 1;
    // index into article.universes
    int32 article_universes_idx = 2;
    bool is_first = 3;
}

message RelatedButton {
    string longname = 1;
    string shortname = 2;
    string row_type = 3;
}

message RelatedButtons {
    string relationship_type = 1;
    repeated RelatedButton buttons = 2;
}

message Histogram {
    float bin_min = 1;
    float bin_size = 2;
    // the counts are normalized to sum to 2**16
    repeated int32 counts = 3;
}

message TimeSeries {
    repeated float values = 1;
}

message ExtraStatistic {
    optional Histogram histogram = 1;
    optional TimeSeries timeseries = 2;
}

message Article {
    string shortname = 1;
    string longname = 2;
    string source = 3;
    string article_type = 4;
    bytes statistic_indices_packed = 9;
    repeated StatisticRow rows = 5;
    repeated FirstOrLast overall_first_or_last = 10;
    repeated RelatedButtons related = 6;
    repeated string universes = 7;
    repeated ExtraStatistic extra_stats = 8;
}


// geojson

message Coordinate {
    float lon = 1;
    float lat = 2;
}

message Ring {
    repeated Coordinate coords = 1;
}

message Polygon {
    repeated Ring rings = 1;
}

message MultiPolygon {
    repeated Polygon polygons = 1;
}

message Feature {
    oneof geometry {
        Polygon polygon = 1;
        MultiPolygon multipolygon = 2;
    }
    repeated int32 zones = 3;
    float center_lon = 4;
}

// string lists

message StringList {
    repeated string elements = 1;
}

message SearchIndex {
    repeated string elements = 1;
    repeated uint32 priorities = 2;
}

message OrderList {
    repeated int32 order_idxs = 1;
}

message DataList {
    repeated float value = 1;
    repeated int32 population_percentile = 2;
}

message OrderLists {
    repeated string statnames = 1;
    repeated OrderList order_lists = 2;
}

message DataLists {
    repeated string statnames = 1;
    repeated DataList data_lists = 2;
}

// unified result

message AllStats {
    repeated float stats = 1;
}

message ConsolidatedShapes {
    repeated string longnames = 1;
    repeated Feature shapes = 2;
}

message ConsolidatedStatistics {
    repeated string longnames = 1;
    repeated string shortnames = 2;
    repeated AllStats stats = 3;
}

// quiz sampling data

message QuizDataForStat {
    repeated float stats = 1;
}

message QuizFullData {
    repeated QuizDataForStat stats = 1;
}

message QuizQuestionTronche {
    repeated int32 geography_a = 1;
    repeated int32 geography_b = 2;
    repeated int32 stat = 3;
    // probability within tronche
    int32 neg_log_prob_x10_basis = 4;
    repeated int32 neg_log_prob_x10_minus_basis = 5;
}