tools/tpch-raw/09.R

In duckdblabs/duckplyr: A 'DuckDB'-Backed Version of 'dplyr'

qloadm("tools/tpch/001.qs")
duckdb <- asNamespace("duckdb")
drv <- duckdb::duckdb()
con <- DBI::dbConnect(drv)
experimental <- FALSE
invisible(
  DBI::dbExecute(
    con,
    'CREATE MACRO "grepl"(pattern, x) AS (CASE WHEN x IS NULL THEN FALSE ELSE regexp_matches(x, pattern) END)'
  )
)
invisible(DBI::dbExecute(con, 'CREATE MACRO "=="(x, y) AS (x == y)'))
invisible(DBI::dbExecute(con, 'CREATE MACRO "___coalesce"(x, y) AS COALESCE(x, y)'))
invisible(duckdb$rapi_load_rfuns(drv@database_ref))
df1 <- part
"select"
rel1 <- duckdb$rel_from_df(con, df1, experimental = experimental)
"select"
rel2 <- duckdb$rel_project(
  rel1,
  list(
    {
      tmp_expr <- duckdb$expr_reference("p_name")
      duckdb$expr_set_alias(tmp_expr, "p_name")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_partkey")
      duckdb$expr_set_alias(tmp_expr, "p_partkey")
      tmp_expr
    }
  )
)
"filter"
rel3 <- duckdb$rel_filter(
  rel2,
  list(
    duckdb$expr_function(
      "grepl",
      list(
        if ("experimental" %in% names(formals(duckdb$expr_constant))) {
          duckdb$expr_constant("green", experimental = experimental)
        } else {
          duckdb$expr_constant("green")
        },
        duckdb$expr_reference("p_name")
      )
    )
  )
)
"select"
rel4 <- duckdb$rel_project(
  rel3,
  list(
    {
      tmp_expr <- duckdb$expr_reference("p_partkey")
      duckdb$expr_set_alias(tmp_expr, "p_partkey")
      tmp_expr
    }
  )
)
df2 <- partsupp
"select"
rel5 <- duckdb$rel_from_df(con, df2, experimental = experimental)
"select"
rel6 <- duckdb$rel_project(
  rel5,
  list(
    {
      tmp_expr <- duckdb$expr_reference("ps_suppkey")
      duckdb$expr_set_alias(tmp_expr, "ps_suppkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("ps_partkey")
      duckdb$expr_set_alias(tmp_expr, "ps_partkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("ps_supplycost")
      duckdb$expr_set_alias(tmp_expr, "ps_supplycost")
      tmp_expr
    }
  )
)
"inner_join"
rel7 <- duckdb$rel_set_alias(rel6, "lhs")
"inner_join"
rel8 <- duckdb$rel_set_alias(rel4, "rhs")
"inner_join"
rel9 <- duckdb$rel_join(
  rel7,
  rel8,
  list(
    duckdb$expr_function(
      "==",
      list(duckdb$expr_reference("ps_partkey", rel7), duckdb$expr_reference("p_partkey", rel8))
    )
  ),
  "inner"
)
"inner_join"
rel10 <- duckdb$rel_project(
  rel9,
  list(
    {
      tmp_expr <- duckdb$expr_reference("ps_suppkey")
      duckdb$expr_set_alias(tmp_expr, "ps_suppkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_function(
        "___coalesce",
        list(duckdb$expr_reference("ps_partkey", rel7), duckdb$expr_reference("p_partkey", rel8))
      )
      duckdb$expr_set_alias(tmp_expr, "ps_partkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("ps_supplycost")
      duckdb$expr_set_alias(tmp_expr, "ps_supplycost")
      tmp_expr
    }
  )
)
df3 <- supplier
"select"
rel11 <- duckdb$rel_from_df(con, df3, experimental = experimental)
"select"
rel12 <- duckdb$rel_project(
  rel11,
  list(
    {
      tmp_expr <- duckdb$expr_reference("s_suppkey")
      duckdb$expr_set_alias(tmp_expr, "s_suppkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("s_nationkey")
      duckdb$expr_set_alias(tmp_expr, "s_nationkey")
      tmp_expr
    }
  )
)
df4 <- nation
"select"
rel13 <- duckdb$rel_from_df(con, df4, experimental = experimental)
"select"
rel14 <- duckdb$rel_project(
  rel13,
  list(
    {
      tmp_expr <- duckdb$expr_reference("n_nationkey")
      duckdb$expr_set_alias(tmp_expr, "n_nationkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("n_name")
      duckdb$expr_set_alias(tmp_expr, "n_name")
      tmp_expr
    }
  )
)
"inner_join"
rel15 <- duckdb$rel_set_alias(rel12, "lhs")
"inner_join"
rel16 <- duckdb$rel_set_alias(rel14, "rhs")
"inner_join"
rel17 <- duckdb$rel_join(
  rel15,
  rel16,
  list(
    duckdb$expr_function(
      "==",
      list(duckdb$expr_reference("s_nationkey", rel15), duckdb$expr_reference("n_nationkey", rel16))
    )
  ),
  "inner"
)
"inner_join"
rel18 <- duckdb$rel_project(
  rel17,
  list(
    {
      tmp_expr <- duckdb$expr_reference("s_suppkey")
      duckdb$expr_set_alias(tmp_expr, "s_suppkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_function(
        "___coalesce",
        list(duckdb$expr_reference("s_nationkey", rel15), duckdb$expr_reference("n_nationkey", rel16))
      )
      duckdb$expr_set_alias(tmp_expr, "s_nationkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("n_name")
      duckdb$expr_set_alias(tmp_expr, "n_name")
      tmp_expr
    }
  )
)
"select"
rel19 <- duckdb$rel_project(
  rel18,
  list(
    {
      tmp_expr <- duckdb$expr_reference("s_suppkey")
      duckdb$expr_set_alias(tmp_expr, "s_suppkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("n_name")
      duckdb$expr_set_alias(tmp_expr, "n_name")
      tmp_expr
    }
  )
)
"inner_join"
rel20 <- duckdb$rel_set_alias(rel10, "lhs")
"inner_join"
rel21 <- duckdb$rel_set_alias(rel19, "rhs")
"inner_join"
rel22 <- duckdb$rel_join(
  rel20,
  rel21,
  list(
    duckdb$expr_function(
      "==",
      list(duckdb$expr_reference("ps_suppkey", rel20), duckdb$expr_reference("s_suppkey", rel21))
    )
  ),
  "inner"
)
"inner_join"
rel23 <- duckdb$rel_project(
  rel22,
  list(
    {
      tmp_expr <- duckdb$expr_function(
        "___coalesce",
        list(duckdb$expr_reference("ps_suppkey", rel20), duckdb$expr_reference("s_suppkey", rel21))
      )
      duckdb$expr_set_alias(tmp_expr, "ps_suppkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("ps_partkey")
      duckdb$expr_set_alias(tmp_expr, "ps_partkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("ps_supplycost")
      duckdb$expr_set_alias(tmp_expr, "ps_supplycost")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("n_name")
      duckdb$expr_set_alias(tmp_expr, "n_name")
      tmp_expr
    }
  )
)
df5 <- lineitem
"select"
rel24 <- duckdb$rel_from_df(con, df5, experimental = experimental)
"select"
rel25 <- duckdb$rel_project(
  rel24,
  list(
    {
      tmp_expr <- duckdb$expr_reference("l_suppkey")
      duckdb$expr_set_alias(tmp_expr, "l_suppkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_partkey")
      duckdb$expr_set_alias(tmp_expr, "l_partkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_orderkey")
      duckdb$expr_set_alias(tmp_expr, "l_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_extendedprice")
      duckdb$expr_set_alias(tmp_expr, "l_extendedprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_discount")
      duckdb$expr_set_alias(tmp_expr, "l_discount")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_quantity")
      duckdb$expr_set_alias(tmp_expr, "l_quantity")
      tmp_expr
    }
  )
)
"inner_join"
rel26 <- duckdb$rel_set_alias(rel25, "lhs")
"inner_join"
rel27 <- duckdb$rel_set_alias(rel23, "rhs")
"inner_join"
rel28 <- duckdb$rel_join(
  rel26,
  rel27,
  list(
    duckdb$expr_function(
      "==",
      list(duckdb$expr_reference("l_suppkey", rel26), duckdb$expr_reference("ps_suppkey", rel27))
    ),
    duckdb$expr_function(
      "==",
      list(duckdb$expr_reference("l_partkey", rel26), duckdb$expr_reference("ps_partkey", rel27))
    )
  ),
  "inner"
)
"inner_join"
rel29 <- duckdb$rel_project(
  rel28,
  list(
    {
      tmp_expr <- duckdb$expr_function(
        "___coalesce",
        list(duckdb$expr_reference("l_suppkey", rel26), duckdb$expr_reference("ps_suppkey", rel27))
      )
      duckdb$expr_set_alias(tmp_expr, "l_suppkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_function(
        "___coalesce",
        list(duckdb$expr_reference("l_partkey", rel26), duckdb$expr_reference("ps_partkey", rel27))
      )
      duckdb$expr_set_alias(tmp_expr, "l_partkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_orderkey")
      duckdb$expr_set_alias(tmp_expr, "l_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_extendedprice")
      duckdb$expr_set_alias(tmp_expr, "l_extendedprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_discount")
      duckdb$expr_set_alias(tmp_expr, "l_discount")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_quantity")
      duckdb$expr_set_alias(tmp_expr, "l_quantity")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("ps_supplycost")
      duckdb$expr_set_alias(tmp_expr, "ps_supplycost")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("n_name")
      duckdb$expr_set_alias(tmp_expr, "n_name")
      tmp_expr
    }
  )
)
"select"
rel30 <- duckdb$rel_project(
  rel29,
  list(
    {
      tmp_expr <- duckdb$expr_reference("l_orderkey")
      duckdb$expr_set_alias(tmp_expr, "l_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_extendedprice")
      duckdb$expr_set_alias(tmp_expr, "l_extendedprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_discount")
      duckdb$expr_set_alias(tmp_expr, "l_discount")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_quantity")
      duckdb$expr_set_alias(tmp_expr, "l_quantity")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("ps_supplycost")
      duckdb$expr_set_alias(tmp_expr, "ps_supplycost")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("n_name")
      duckdb$expr_set_alias(tmp_expr, "n_name")
      tmp_expr
    }
  )
)
df6 <- orders
"select"
rel31 <- duckdb$rel_from_df(con, df6, experimental = experimental)
"select"
rel32 <- duckdb$rel_project(
  rel31,
  list(
    {
      tmp_expr <- duckdb$expr_reference("o_orderkey")
      duckdb$expr_set_alias(tmp_expr, "o_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderdate")
      duckdb$expr_set_alias(tmp_expr, "o_orderdate")
      tmp_expr
    }
  )
)
"inner_join"
rel33 <- duckdb$rel_set_alias(rel32, "lhs")
"inner_join"
rel34 <- duckdb$rel_set_alias(rel30, "rhs")
"inner_join"
rel35 <- duckdb$rel_join(
  rel33,
  rel34,
  list(
    duckdb$expr_function(
      "==",
      list(duckdb$expr_reference("o_orderkey", rel33), duckdb$expr_reference("l_orderkey", rel34))
    )
  ),
  "inner"
)
"inner_join"
rel36 <- duckdb$rel_project(
  rel35,
  list(
    {
      tmp_expr <- duckdb$expr_function(
        "___coalesce",
        list(duckdb$expr_reference("o_orderkey", rel33), duckdb$expr_reference("l_orderkey", rel34))
      )
      duckdb$expr_set_alias(tmp_expr, "o_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderdate")
      duckdb$expr_set_alias(tmp_expr, "o_orderdate")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_extendedprice")
      duckdb$expr_set_alias(tmp_expr, "l_extendedprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_discount")
      duckdb$expr_set_alias(tmp_expr, "l_discount")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_quantity")
      duckdb$expr_set_alias(tmp_expr, "l_quantity")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("ps_supplycost")
      duckdb$expr_set_alias(tmp_expr, "ps_supplycost")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("n_name")
      duckdb$expr_set_alias(tmp_expr, "n_name")
      tmp_expr
    }
  )
)
"select"
rel37 <- duckdb$rel_project(
  rel36,
  list(
    {
      tmp_expr <- duckdb$expr_reference("l_extendedprice")
      duckdb$expr_set_alias(tmp_expr, "l_extendedprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_discount")
      duckdb$expr_set_alias(tmp_expr, "l_discount")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_quantity")
      duckdb$expr_set_alias(tmp_expr, "l_quantity")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("ps_supplycost")
      duckdb$expr_set_alias(tmp_expr, "ps_supplycost")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("n_name")
      duckdb$expr_set_alias(tmp_expr, "n_name")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderdate")
      duckdb$expr_set_alias(tmp_expr, "o_orderdate")
      tmp_expr
    }
  )
)
"mutate"
rel38 <- duckdb$rel_project(
  rel37,
  list(
    {
      tmp_expr <- duckdb$expr_reference("l_extendedprice")
      duckdb$expr_set_alias(tmp_expr, "l_extendedprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_discount")
      duckdb$expr_set_alias(tmp_expr, "l_discount")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_quantity")
      duckdb$expr_set_alias(tmp_expr, "l_quantity")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("ps_supplycost")
      duckdb$expr_set_alias(tmp_expr, "ps_supplycost")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("n_name")
      duckdb$expr_set_alias(tmp_expr, "n_name")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderdate")
      duckdb$expr_set_alias(tmp_expr, "o_orderdate")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("n_name")
      duckdb$expr_set_alias(tmp_expr, "nation")
      tmp_expr
    }
  )
)
"mutate"
rel39 <- duckdb$rel_project(
  rel38,
  list(
    {
      tmp_expr <- duckdb$expr_reference("l_extendedprice")
      duckdb$expr_set_alias(tmp_expr, "l_extendedprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_discount")
      duckdb$expr_set_alias(tmp_expr, "l_discount")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_quantity")
      duckdb$expr_set_alias(tmp_expr, "l_quantity")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("ps_supplycost")
      duckdb$expr_set_alias(tmp_expr, "ps_supplycost")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("n_name")
      duckdb$expr_set_alias(tmp_expr, "n_name")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderdate")
      duckdb$expr_set_alias(tmp_expr, "o_orderdate")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("nation")
      duckdb$expr_set_alias(tmp_expr, "nation")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_function(
        "r_base::as.integer",
        list(
          duckdb$expr_function(
            "strftime",
            list(
              duckdb$expr_reference("o_orderdate"),
              if ("experimental" %in% names(formals(duckdb$expr_constant))) {
                duckdb$expr_constant("%Y", experimental = experimental)
              } else {
                duckdb$expr_constant("%Y")
              }
            )
          )
        )
      )
      duckdb$expr_set_alias(tmp_expr, "o_year")
      tmp_expr
    }
  )
)
"mutate"
rel40 <- duckdb$rel_project(
  rel39,
  list(
    {
      tmp_expr <- duckdb$expr_reference("l_extendedprice")
      duckdb$expr_set_alias(tmp_expr, "l_extendedprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_discount")
      duckdb$expr_set_alias(tmp_expr, "l_discount")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_quantity")
      duckdb$expr_set_alias(tmp_expr, "l_quantity")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("ps_supplycost")
      duckdb$expr_set_alias(tmp_expr, "ps_supplycost")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("n_name")
      duckdb$expr_set_alias(tmp_expr, "n_name")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderdate")
      duckdb$expr_set_alias(tmp_expr, "o_orderdate")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("nation")
      duckdb$expr_set_alias(tmp_expr, "nation")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_year")
      duckdb$expr_set_alias(tmp_expr, "o_year")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_function(
        "-",
        list(
          duckdb$expr_function(
            "*",
            list(
              duckdb$expr_reference("l_extendedprice"),
              duckdb$expr_function(
                "-",
                list(
                  if ("experimental" %in% names(formals(duckdb$expr_constant))) {
                    duckdb$expr_constant(1, experimental = experimental)
                  } else {
                    duckdb$expr_constant(1)
                  },
                  duckdb$expr_reference("l_discount")
                )
              )
            )
          ),
          duckdb$expr_function(
            "*",
            list(duckdb$expr_reference("ps_supplycost"), duckdb$expr_reference("l_quantity"))
          )
        )
      )
      duckdb$expr_set_alias(tmp_expr, "amount")
      tmp_expr
    }
  )
)
"select"
rel41 <- duckdb$rel_project(
  rel40,
  list(
    {
      tmp_expr <- duckdb$expr_reference("nation")
      duckdb$expr_set_alias(tmp_expr, "nation")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_year")
      duckdb$expr_set_alias(tmp_expr, "o_year")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("amount")
      duckdb$expr_set_alias(tmp_expr, "amount")
      tmp_expr
    }
  )
)
"summarise"
rel42 <- duckdb$rel_aggregate(
  rel41,
  groups = list(duckdb$expr_reference("nation"), duckdb$expr_reference("o_year")),
  aggregates = list(
    {
      tmp_expr <- duckdb$expr_function("sum", list(duckdb$expr_reference("amount")))
      duckdb$expr_set_alias(tmp_expr, "sum_profit")
      tmp_expr
    }
  )
)
"arrange"
rel43 <- duckdb$rel_order(rel42, list(duckdb$expr_reference("nation"), duckdb$expr_reference("o_year")))
rel43
duckdb$rel_to_altrep(rel43)