tools/tpch-raw/17.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$rapi_load_rfuns(drv@database_ref))
invisible(DBI::dbExecute(con, 'CREATE MACRO "=="(x, y) AS (x == y)'))
invisible(DBI::dbExecute(con, 'CREATE MACRO "___coalesce"(x, y) AS COALESCE(x, y)'))
invisible(
  DBI::dbExecute(
    con,
    r"[CREATE MACRO "___divide"(x, y) AS CASE WHEN y = 0 THEN CASE WHEN x = 0 THEN CAST('NaN' AS double) WHEN x > 0 THEN CAST('+Infinity' AS double) ELSE CAST('-Infinity' AS double) END ELSE CAST(x AS double) / y END]"
  )
)
df1 <- part
"filter"
rel1 <- duckdb$rel_from_df(con, df1, experimental = experimental)
"filter"
rel2 <- duckdb$rel_filter(
  rel1,
  list(
    duckdb$expr_function(
      "r_base::==",
      list(
        duckdb$expr_reference("p_brand"),
        if ("experimental" %in% names(formals(duckdb$expr_constant))) {
          duckdb$expr_constant("Brand#23", experimental = experimental)
        } else {
          duckdb$expr_constant("Brand#23")
        }
      )
    ),
    duckdb$expr_function(
      "r_base::==",
      list(
        duckdb$expr_reference("p_container"),
        if ("experimental" %in% names(formals(duckdb$expr_constant))) {
          duckdb$expr_constant("MED BOX", experimental = experimental)
        } else {
          duckdb$expr_constant("MED BOX")
        }
      )
    )
  )
)
df2 <- lineitem
"inner_join"
rel3 <- duckdb$rel_from_df(con, df2, experimental = experimental)
"inner_join"
rel4 <- duckdb$rel_set_alias(rel3, "lhs")
"inner_join"
rel5 <- duckdb$rel_set_alias(rel2, "rhs")
"inner_join"
rel6 <- duckdb$rel_join(
  rel4,
  rel5,
  list(
    duckdb$expr_function(
      "==",
      list(duckdb$expr_reference("l_partkey", rel4), duckdb$expr_reference("p_partkey", rel5))
    )
  ),
  "inner"
)
"inner_join"
rel7 <- duckdb$rel_project(
  rel6,
  list(
    {
      tmp_expr <- duckdb$expr_reference("l_orderkey")
      duckdb$expr_set_alias(tmp_expr, "l_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_function(
        "___coalesce",
        list(duckdb$expr_reference("l_partkey", rel4), duckdb$expr_reference("p_partkey", rel5))
      )
      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_reference("p_name")
      duckdb$expr_set_alias(tmp_expr, "p_name")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_mfgr")
      duckdb$expr_set_alias(tmp_expr, "p_mfgr")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_brand")
      duckdb$expr_set_alias(tmp_expr, "p_brand")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_type")
      duckdb$expr_set_alias(tmp_expr, "p_type")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_size")
      duckdb$expr_set_alias(tmp_expr, "p_size")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_container")
      duckdb$expr_set_alias(tmp_expr, "p_container")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_retailprice")
      duckdb$expr_set_alias(tmp_expr, "p_retailprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_comment")
      duckdb$expr_set_alias(tmp_expr, "p_comment")
      tmp_expr
    }
  )
)
"summarise"
rel8 <- duckdb$rel_aggregate(
  rel7,
  groups = list(duckdb$expr_reference("l_partkey")),
  aggregates = list(
    {
      tmp_expr <- duckdb$expr_function(
        "*",
        list(
          if ("experimental" %in% names(formals(duckdb$expr_constant))) {
            duckdb$expr_constant(0.2, experimental = experimental)
          } else {
            duckdb$expr_constant(0.2)
          },
          duckdb$expr_function("mean", list(duckdb$expr_reference("l_quantity")))
        )
      )
      duckdb$expr_set_alias(tmp_expr, "quantity_threshold")
      tmp_expr
    }
  )
)
"inner_join"
rel9 <- duckdb$rel_set_alias(rel7, "lhs")
"inner_join"
rel10 <- duckdb$rel_set_alias(rel8, "rhs")
"inner_join"
rel11 <- duckdb$rel_project(
  rel9,
  list(
    {
      tmp_expr <- duckdb$expr_reference("l_orderkey")
      duckdb$expr_set_alias(tmp_expr, "l_orderkey_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_partkey")
      duckdb$expr_set_alias(tmp_expr, "l_partkey_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_suppkey")
      duckdb$expr_set_alias(tmp_expr, "l_suppkey_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_linenumber")
      duckdb$expr_set_alias(tmp_expr, "l_linenumber_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_quantity")
      duckdb$expr_set_alias(tmp_expr, "l_quantity_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_extendedprice")
      duckdb$expr_set_alias(tmp_expr, "l_extendedprice_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_discount")
      duckdb$expr_set_alias(tmp_expr, "l_discount_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_tax")
      duckdb$expr_set_alias(tmp_expr, "l_tax_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_returnflag")
      duckdb$expr_set_alias(tmp_expr, "l_returnflag_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_linestatus")
      duckdb$expr_set_alias(tmp_expr, "l_linestatus_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_shipdate")
      duckdb$expr_set_alias(tmp_expr, "l_shipdate_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_commitdate")
      duckdb$expr_set_alias(tmp_expr, "l_commitdate_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_receiptdate")
      duckdb$expr_set_alias(tmp_expr, "l_receiptdate_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_shipinstruct")
      duckdb$expr_set_alias(tmp_expr, "l_shipinstruct_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_shipmode")
      duckdb$expr_set_alias(tmp_expr, "l_shipmode_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_comment")
      duckdb$expr_set_alias(tmp_expr, "l_comment_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_name")
      duckdb$expr_set_alias(tmp_expr, "p_name_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_mfgr")
      duckdb$expr_set_alias(tmp_expr, "p_mfgr_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_brand")
      duckdb$expr_set_alias(tmp_expr, "p_brand_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_type")
      duckdb$expr_set_alias(tmp_expr, "p_type_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_size")
      duckdb$expr_set_alias(tmp_expr, "p_size_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_container")
      duckdb$expr_set_alias(tmp_expr, "p_container_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_retailprice")
      duckdb$expr_set_alias(tmp_expr, "p_retailprice_x")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_comment")
      duckdb$expr_set_alias(tmp_expr, "p_comment_x")
      tmp_expr
    }
  )
)
"inner_join"
rel12 <- duckdb$rel_project(
  rel10,
  list(
    {
      tmp_expr <- duckdb$expr_reference("l_partkey")
      duckdb$expr_set_alias(tmp_expr, "l_partkey_y")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("quantity_threshold")
      duckdb$expr_set_alias(tmp_expr, "quantity_threshold_y")
      tmp_expr
    }
  )
)
"inner_join"
rel13 <- duckdb$rel_join(
  rel11,
  rel12,
  list(
    duckdb$expr_function(
      "==",
      list(duckdb$expr_reference("l_partkey_x", rel11), duckdb$expr_reference("l_partkey_y", rel12))
    )
  ),
  "inner"
)
"inner_join"
rel14 <- duckdb$rel_project(
  rel13,
  list(
    {
      tmp_expr <- duckdb$expr_reference("l_orderkey_x")
      duckdb$expr_set_alias(tmp_expr, "l_orderkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_function(
        "___coalesce",
        list(duckdb$expr_reference("l_partkey_x", rel11), duckdb$expr_reference("l_partkey_y", rel12))
      )
      duckdb$expr_set_alias(tmp_expr, "l_partkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_suppkey_x")
      duckdb$expr_set_alias(tmp_expr, "l_suppkey")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_linenumber_x")
      duckdb$expr_set_alias(tmp_expr, "l_linenumber")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_quantity_x")
      duckdb$expr_set_alias(tmp_expr, "l_quantity")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_extendedprice_x")
      duckdb$expr_set_alias(tmp_expr, "l_extendedprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_discount_x")
      duckdb$expr_set_alias(tmp_expr, "l_discount")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_tax_x")
      duckdb$expr_set_alias(tmp_expr, "l_tax")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_returnflag_x")
      duckdb$expr_set_alias(tmp_expr, "l_returnflag")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_linestatus_x")
      duckdb$expr_set_alias(tmp_expr, "l_linestatus")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_shipdate_x")
      duckdb$expr_set_alias(tmp_expr, "l_shipdate")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_commitdate_x")
      duckdb$expr_set_alias(tmp_expr, "l_commitdate")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_receiptdate_x")
      duckdb$expr_set_alias(tmp_expr, "l_receiptdate")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_shipinstruct_x")
      duckdb$expr_set_alias(tmp_expr, "l_shipinstruct")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_shipmode_x")
      duckdb$expr_set_alias(tmp_expr, "l_shipmode")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("l_comment_x")
      duckdb$expr_set_alias(tmp_expr, "l_comment")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_name_x")
      duckdb$expr_set_alias(tmp_expr, "p_name")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_mfgr_x")
      duckdb$expr_set_alias(tmp_expr, "p_mfgr")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_brand_x")
      duckdb$expr_set_alias(tmp_expr, "p_brand")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_type_x")
      duckdb$expr_set_alias(tmp_expr, "p_type")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_size_x")
      duckdb$expr_set_alias(tmp_expr, "p_size")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_container_x")
      duckdb$expr_set_alias(tmp_expr, "p_container")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_retailprice_x")
      duckdb$expr_set_alias(tmp_expr, "p_retailprice")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("p_comment_x")
      duckdb$expr_set_alias(tmp_expr, "p_comment")
      tmp_expr
    },
    {
      tmp_expr <- duckdb$expr_reference("quantity_threshold_y")
      duckdb$expr_set_alias(tmp_expr, "quantity_threshold")
      tmp_expr
    }
  )
)
"filter"
rel15 <- duckdb$rel_filter(
  rel14,
  list(
    duckdb$expr_function(
      "r_base::<",
      list(duckdb$expr_reference("l_quantity"), duckdb$expr_reference("quantity_threshold"))
    )
  )
)
"summarise"
rel16 <- duckdb$rel_aggregate(
  rel15,
  groups = list(),
  aggregates = list(
    {
      tmp_expr <- duckdb$expr_function(
        "___divide",
        list(
          duckdb$expr_function("sum", list(duckdb$expr_reference("l_extendedprice"))),
          if ("experimental" %in% names(formals(duckdb$expr_constant))) {
            duckdb$expr_constant(7, experimental = experimental)
          } else {
            duckdb$expr_constant(7)
          }
        )
      )
      duckdb$expr_set_alias(tmp_expr, "avg_yearly")
      tmp_expr
    }
  )
)
"summarise"
rel17 <- duckdb$rel_distinct(rel16)
rel17
duckdb$rel_to_altrep(rel17)