tools/tpch-raw-oo/18.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(duckdb$rapi_load_rfuns(drv@database_ref))
invisible(DBI::dbExecute(con, 'CREATE MACRO "=="(x, y) AS (x == y)'))
invisible(DBI::dbExecute(con, 'CREATE MACRO "___coalesce"(x, y) AS COALESCE(x, y)'))
df1 <- lineitem
"summarise"
rel1 <- duckdb$rel_from_df(con, df1, experimental = experimental)
"summarise"
rel2 <- duckdb$rel_project(
  rel1,
  list(
    {
      tmp_expr <- duckdb$expr_reference("l_orderkey")
      duckdb$expr_set_alias(tmp_expr, "l_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_partkey")
      duckdb$expr_set_alias(tmp_expr, "l_partkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_suppkey")
      duckdb$expr_set_alias(tmp_expr, "l_suppkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_linenumber")
      duckdb$expr_set_alias(tmp_expr, "l_linenumber")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_quantity")
      duckdb$expr_set_alias(tmp_expr, "l_quantity")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$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_tax")
      duckdb$expr_set_alias(tmp_expr, "l_tax")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_returnflag")
      duckdb$expr_set_alias(tmp_expr, "l_returnflag")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_linestatus")
      duckdb$expr_set_alias(tmp_expr, "l_linestatus")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_shipdate")
      duckdb$expr_set_alias(tmp_expr, "l_shipdate")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_commitdate")
      duckdb$expr_set_alias(tmp_expr, "l_commitdate")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_receiptdate")
      duckdb$expr_set_alias(tmp_expr, "l_receiptdate")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_shipinstruct")
      duckdb$expr_set_alias(tmp_expr, "l_shipinstruct")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_shipmode")
      duckdb$expr_set_alias(tmp_expr, "l_shipmode")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_comment")
      duckdb$expr_set_alias(tmp_expr, "l_comment")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_window(duckdb$expr_function("row_number", list()), list(), list(), offset_expr = NULL, default_expr = NULL)
      duckdb$expr_set_alias(tmp_expr, "___row_number")
      tmp_expr
    }
  )
)
"summarise"
rel3 <- duckdb$rel_aggregate(
  rel2,
  groups = list(duckdb$expr_reference("l_orderkey")),
  aggregates = list(
    {
      tmp_expr <- duckdb$expr_function("min", list(duckdb$expr_reference("___row_number")))
      duckdb$expr_set_alias(tmp_expr, "___row_number")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_function("sum", list(duckdb$expr_reference("l_quantity")))
      duckdb$expr_set_alias(tmp_expr, "sum")
      tmp_expr
    }
  )
)
"summarise"
rel4 <- duckdb$rel_order(rel3, list(duckdb$expr_reference("___row_number")))
"summarise"
rel5 <- duckdb$rel_project(
  rel4,
  list(
    {
      tmp_expr <- duckdb$expr_reference("l_orderkey")
      duckdb$expr_set_alias(tmp_expr, "l_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("sum")
      duckdb$expr_set_alias(tmp_expr, "sum")
      tmp_expr
    }
  )
)
"filter"
rel6 <- duckdb$rel_project(
  rel5,
  list(
    {
      tmp_expr <- duckdb$expr_reference("l_orderkey")
      duckdb$expr_set_alias(tmp_expr, "l_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("sum")
      duckdb$expr_set_alias(tmp_expr, "sum")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_window(duckdb$expr_function("row_number", list()), list(), list(), offset_expr = NULL, default_expr = NULL)
      duckdb$expr_set_alias(tmp_expr, "___row_number")
      tmp_expr
    }
  )
)
"filter"
rel7 <- duckdb$rel_filter(
  rel6,
  list(
    duckdb$expr_function(
      "r_base::>",
      list(
        duckdb$expr_reference("sum"),
        if ("experimental" %in% names(formals(duckdb$expr_constant))) {
          duckdb$expr_constant(300, experimental = experimental)
        } else {
          duckdb$expr_constant(300)
        }
      )
    )
  )
)
"filter"
rel8 <- duckdb$rel_order(rel7, list(duckdb$expr_reference("___row_number")))
"filter"
rel9 <- duckdb$rel_project(
  rel8,
  list(
    {
      tmp_expr <- duckdb$expr_reference("l_orderkey")
      duckdb$expr_set_alias(tmp_expr, "l_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("sum")
      duckdb$expr_set_alias(tmp_expr, "sum")
      tmp_expr
    }
  )
)
df2 <- orders
"inner_join"
rel10 <- duckdb$rel_from_df(con, df2, experimental = experimental)
"inner_join"
rel11 <- duckdb$rel_set_alias(rel10, "lhs")
"inner_join"
rel12 <- duckdb$rel_set_alias(rel9, "rhs")
"inner_join"
rel13 <- duckdb$rel_project(
  rel11,
  list(
    {
      tmp_expr <- duckdb$expr_reference("o_orderkey")
      duckdb$expr_set_alias(tmp_expr, "o_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_custkey")
      duckdb$expr_set_alias(tmp_expr, "o_custkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderstatus")
      duckdb$expr_set_alias(tmp_expr, "o_orderstatus")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_totalprice")
      duckdb$expr_set_alias(tmp_expr, "o_totalprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderdate")
      duckdb$expr_set_alias(tmp_expr, "o_orderdate")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderpriority")
      duckdb$expr_set_alias(tmp_expr, "o_orderpriority")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_clerk")
      duckdb$expr_set_alias(tmp_expr, "o_clerk")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_shippriority")
      duckdb$expr_set_alias(tmp_expr, "o_shippriority")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_comment")
      duckdb$expr_set_alias(tmp_expr, "o_comment")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_window(duckdb$expr_function("row_number", list()), list(), list(), offset_expr = NULL, default_expr = NULL)
      duckdb$expr_set_alias(tmp_expr, "___row_number_x")
      tmp_expr
    }
  )
)
"inner_join"
rel14 <- duckdb$rel_project(
  rel12,
  list(
    {
      tmp_expr <- duckdb$expr_reference("l_orderkey")
      duckdb$expr_set_alias(tmp_expr, "l_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("sum")
      duckdb$expr_set_alias(tmp_expr, "sum")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_window(duckdb$expr_function("row_number", list()), list(), list(), offset_expr = NULL, default_expr = NULL)
      duckdb$expr_set_alias(tmp_expr, "___row_number_y")
      tmp_expr
    }
  )
)
"inner_join"
rel15 <- duckdb$rel_join(
  rel13,
  rel14,
  list(
    duckdb$expr_function(
      "==",
      list(duckdb$expr_reference("o_orderkey", rel13), duckdb$expr_reference("l_orderkey", rel14))
    )
  ),
  "inner"
)
"inner_join"
rel16 <- duckdb$rel_order(
  rel15,
  list(duckdb$expr_reference("___row_number_x", rel13), duckdb$expr_reference("___row_number_y", rel14))
)
"inner_join"
rel17 <- duckdb$rel_project(
  rel16,
  list(
    {
      tmp_expr <- duckdb$expr_function(
        "___coalesce",
        list(duckdb$expr_reference("o_orderkey", rel13), duckdb$expr_reference("l_orderkey", rel14))
      )
      duckdb$expr_set_alias(tmp_expr, "o_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_custkey")
      duckdb$expr_set_alias(tmp_expr, "o_custkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderstatus")
      duckdb$expr_set_alias(tmp_expr, "o_orderstatus")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_totalprice")
      duckdb$expr_set_alias(tmp_expr, "o_totalprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderdate")
      duckdb$expr_set_alias(tmp_expr, "o_orderdate")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderpriority")
      duckdb$expr_set_alias(tmp_expr, "o_orderpriority")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_clerk")
      duckdb$expr_set_alias(tmp_expr, "o_clerk")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_shippriority")
      duckdb$expr_set_alias(tmp_expr, "o_shippriority")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_comment")
      duckdb$expr_set_alias(tmp_expr, "o_comment")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("sum")
      duckdb$expr_set_alias(tmp_expr, "sum")
      tmp_expr
    }
  )
)
"inner_join"
rel18 <- duckdb$rel_set_alias(rel17, "lhs")
df3 <- customer
"inner_join"
rel19 <- duckdb$rel_from_df(con, df3, experimental = experimental)
"inner_join"
rel20 <- duckdb$rel_set_alias(rel19, "rhs")
"inner_join"
rel21 <- duckdb$rel_project(
  rel18,
  list(
    {
      tmp_expr <- duckdb$expr_reference("o_orderkey")
      duckdb$expr_set_alias(tmp_expr, "o_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_custkey")
      duckdb$expr_set_alias(tmp_expr, "o_custkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderstatus")
      duckdb$expr_set_alias(tmp_expr, "o_orderstatus")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_totalprice")
      duckdb$expr_set_alias(tmp_expr, "o_totalprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderdate")
      duckdb$expr_set_alias(tmp_expr, "o_orderdate")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderpriority")
      duckdb$expr_set_alias(tmp_expr, "o_orderpriority")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_clerk")
      duckdb$expr_set_alias(tmp_expr, "o_clerk")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_shippriority")
      duckdb$expr_set_alias(tmp_expr, "o_shippriority")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_comment")
      duckdb$expr_set_alias(tmp_expr, "o_comment")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("sum")
      duckdb$expr_set_alias(tmp_expr, "sum")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_window(duckdb$expr_function("row_number", list()), list(), list(), offset_expr = NULL, default_expr = NULL)
      duckdb$expr_set_alias(tmp_expr, "___row_number_x")
      tmp_expr
    }
  )
)
"inner_join"
rel22 <- duckdb$rel_project(
  rel20,
  list(
    {
      tmp_expr <- duckdb$expr_reference("c_custkey")
      duckdb$expr_set_alias(tmp_expr, "c_custkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("c_name")
      duckdb$expr_set_alias(tmp_expr, "c_name")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("c_address")
      duckdb$expr_set_alias(tmp_expr, "c_address")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("c_nationkey")
      duckdb$expr_set_alias(tmp_expr, "c_nationkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("c_phone")
      duckdb$expr_set_alias(tmp_expr, "c_phone")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("c_acctbal")
      duckdb$expr_set_alias(tmp_expr, "c_acctbal")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("c_mktsegment")
      duckdb$expr_set_alias(tmp_expr, "c_mktsegment")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("c_comment")
      duckdb$expr_set_alias(tmp_expr, "c_comment")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_window(duckdb$expr_function("row_number", list()), list(), list(), offset_expr = NULL, default_expr = NULL)
      duckdb$expr_set_alias(tmp_expr, "___row_number_y")
      tmp_expr
    }
  )
)
"inner_join"
rel23 <- duckdb$rel_join(
  rel21,
  rel22,
  list(
    duckdb$expr_function(
      "==",
      list(duckdb$expr_reference("o_custkey", rel21), duckdb$expr_reference("c_custkey", rel22))
    )
  ),
  "inner"
)
"inner_join"
rel24 <- duckdb$rel_order(
  rel23,
  list(duckdb$expr_reference("___row_number_x", rel21), duckdb$expr_reference("___row_number_y", rel22))
)
"inner_join"
rel25 <- duckdb$rel_project(
  rel24,
  list(
    {
      tmp_expr <- duckdb$expr_reference("o_orderkey")
      duckdb$expr_set_alias(tmp_expr, "o_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_function(
        "___coalesce",
        list(duckdb$expr_reference("o_custkey", rel21), duckdb$expr_reference("c_custkey", rel22))
      )
      duckdb$expr_set_alias(tmp_expr, "o_custkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderstatus")
      duckdb$expr_set_alias(tmp_expr, "o_orderstatus")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_totalprice")
      duckdb$expr_set_alias(tmp_expr, "o_totalprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderdate")
      duckdb$expr_set_alias(tmp_expr, "o_orderdate")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderpriority")
      duckdb$expr_set_alias(tmp_expr, "o_orderpriority")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_clerk")
      duckdb$expr_set_alias(tmp_expr, "o_clerk")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_shippriority")
      duckdb$expr_set_alias(tmp_expr, "o_shippriority")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_comment")
      duckdb$expr_set_alias(tmp_expr, "o_comment")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("sum")
      duckdb$expr_set_alias(tmp_expr, "sum")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("c_name")
      duckdb$expr_set_alias(tmp_expr, "c_name")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("c_address")
      duckdb$expr_set_alias(tmp_expr, "c_address")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("c_nationkey")
      duckdb$expr_set_alias(tmp_expr, "c_nationkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("c_phone")
      duckdb$expr_set_alias(tmp_expr, "c_phone")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("c_acctbal")
      duckdb$expr_set_alias(tmp_expr, "c_acctbal")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("c_mktsegment")
      duckdb$expr_set_alias(tmp_expr, "c_mktsegment")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("c_comment")
      duckdb$expr_set_alias(tmp_expr, "c_comment")
      tmp_expr
    }
  )
)
"select"
rel26 <- duckdb$rel_project(
  rel25,
  list(
    {
      tmp_expr <- duckdb$expr_reference("c_name")
      duckdb$expr_set_alias(tmp_expr, "c_name")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_custkey")
      duckdb$expr_set_alias(tmp_expr, "c_custkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderkey")
      duckdb$expr_set_alias(tmp_expr, "o_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_orderdate")
      duckdb$expr_set_alias(tmp_expr, "o_orderdate")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("o_totalprice")
      duckdb$expr_set_alias(tmp_expr, "o_totalprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("sum")
      duckdb$expr_set_alias(tmp_expr, "sum")
      tmp_expr
    }
  )
)
"arrange"
rel27 <- duckdb$rel_project(
  rel26,
  list(
    {
      tmp_expr <- duckdb$expr_reference("c_name")
      duckdb$expr_set_alias(tmp_expr, "c_name")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("c_custkey")
      duckdb$expr_set_alias(tmp_expr, "c_custkey")
      tmp_expr
    },
    {
      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
    },
    {
      tmp_expr <- duckdb$expr_reference("o_totalprice")
      duckdb$expr_set_alias(tmp_expr, "o_totalprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("sum")
      duckdb$expr_set_alias(tmp_expr, "sum")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_window(duckdb$expr_function("row_number", list()), list(), list(), offset_expr = NULL, default_expr = NULL)
      duckdb$expr_set_alias(tmp_expr, "___row_number")
      tmp_expr
    }
  )
)
"arrange"
rel28 <- duckdb$rel_order(
  rel27,
  list(duckdb$expr_reference("o_totalprice"), duckdb$expr_reference("o_orderdate"), duckdb$expr_reference("___row_number"))
)
"arrange"
rel29 <- duckdb$rel_project(
  rel28,
  list(
    {
      tmp_expr <- duckdb$expr_reference("c_name")
      duckdb$expr_set_alias(tmp_expr, "c_name")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("c_custkey")
      duckdb$expr_set_alias(tmp_expr, "c_custkey")
      tmp_expr
    },
    {
      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
    },
    {
      tmp_expr <- duckdb$expr_reference("o_totalprice")
      duckdb$expr_set_alias(tmp_expr, "o_totalprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("sum")
      duckdb$expr_set_alias(tmp_expr, "sum")
      tmp_expr
    }
  )
)
"head"
rel30 <- duckdb$rel_limit(rel29, 100)
rel30
duckdb$rel_to_altrep(rel30)