C (MSYS2 + UCRT64) で duckdb を使って Content-Length を返さないサーバーの parquet にアクセスしてみた

main.c

#include <stdio.h>
#include <duckdb.h>

int main(int argc, char **argv)
{
    duckdb_database db;
    duckdb_connection con;

    if (duckdb_open(NULL, &db) == DuckDBError) {
        printf("duckdb_open failed\n");
        return 1;
    }
    if (duckdb_connect(db, &con) == DuckDBError) {
        printf("duckdb_connect failed\n");
        return 1;
    }

    // run queries...
    duckdb_state state;
    duckdb_result result;

    state = duckdb_query(con, "INSTALL httpfs", NULL);
    if (state == DuckDBError) {
        printf("failed: INSTALL httpfs\n");
        return 1;
    }
    state = duckdb_query(con, "LOAD httpfs", NULL);
    if (state == DuckDBError) {
        printf("failed: LOAD httpfs\n");
        return 1;
    }

#if 1
    state = duckdb_query(con, "SELECT avg(c_acctbal) FROM 'https://shell.duckdb.org/data/tpch/0_01/parquet/customer.parquet'", &result);
#else
    //state = duckdb_query(con, "SELECT avg(c_acctbal) FROM 'customer.parquet'", &result);
#endif
    if (state == DuckDBError) {
        printf("duckdb_query failed: %s\n", duckdb_result_error(&result));
    }
    while (true) {
        duckdb_data_chunk chunk = duckdb_fetch_chunk(result);
        if (!chunk) {
            break;
        }
        printf("column#0: name=%s type=%d\n",
                duckdb_column_name(&result, 0),
                duckdb_column_type(&result, 0));
        idx_t row_count = duckdb_data_chunk_get_size(chunk);
        duckdb_vector col0 = duckdb_data_chunk_get_vector(chunk, 0);
        double *col0_data = (double *)duckdb_vector_get_data(col0);
        uint64_t *col0_validity = duckdb_vector_get_validity(col0);
        printf("row_count=%d\n", row_count);

        for (idx_t row = 0; row < row_count; row++) {
            if (duckdb_validity_row_is_valid(col0_validity, row)) {
                printf("%e", col0_data[row]);
            } else {
                printf("NULL");
            }
            printf("\n");
        }
        duckdb_destroy_data_chunk(&chunk);
    }

    // cleanup
    duckdb_destroy_result(&result);
    duckdb_disconnect(&con);
    duckdb_close(&db);
    return 0;
}

Makefile

LDLIBS=-lduckdb

main: main.c

clean:
	rm -f *.o
	rm -f main