From 4546f1d936d42133dee61878520a172f1dccc9c6 Mon Sep 17 00:00:00 2001
From: rafapereirabr <rafa.pereira.br@gmail.com>
Date: Mon, 16 Dec 2024 23:27:03 -0300
Subject: [PATCH] draft query_aggregate_and_match with precision

---
 R/geocode.R                              | 14 ++--------
 tests/tests_rafa/tests_arrow_vs_duckdb.R | 35 +++++++++++++++---------
 2 files changed, 24 insertions(+), 25 deletions(-)

diff --git a/R/geocode.R b/R/geocode.R
index 824eb0b..5d21c29 100644
--- a/R/geocode.R
+++ b/R/geocode.R
@@ -58,8 +58,7 @@ geocode <- function(input_table,
   checkmate::assert_logical(cache)
   checkmate::assert_names(
     names(input_table),
-    must.include = "ID",
-    .var.name = "input_table"
+    must.include = "ID"
   )
 
 
@@ -572,16 +571,7 @@ geocode <- function(input_table,
   # DBI::dbReadTable(con, 'output_caso_01')
 
   # DBI::dbRemoveTable(con, 'output_caso_01')
-  # DBI::dbRemoveTable(con, 'output_caso_02')
-  # DBI::dbRemoveTable(con, 'output_caso_03')
-  # DBI::dbRemoveTable(con, 'output_caso_04')
-  # DBI::dbRemoveTable(con, 'output_caso_05')
-  # DBI::dbRemoveTable(con, 'output_caso_06')
-  # DBI::dbRemoveTable(con, 'output_caso_07')
-  # DBI::dbRemoveTable(con, 'output_caso_08')
-  # DBI::dbRemoveTable(con, 'output_caso_09')
-  # DBI::dbRemoveTable(con, 'output_caso_10')
-  # DBI::dbRemoveTable(con, 'output_caso_11')
+
 
 
 
diff --git a/tests/tests_rafa/tests_arrow_vs_duckdb.R b/tests/tests_rafa/tests_arrow_vs_duckdb.R
index 2f223e6..db3f02c 100644
--- a/tests/tests_rafa/tests_arrow_vs_duckdb.R
+++ b/tests/tests_rafa/tests_arrow_vs_duckdb.R
@@ -113,19 +113,28 @@ range_lat <- sd(df$lat) *2
 
 
 
-query_aggregate_and_match <- glue::glue(
-  "CREATE TABLE {output_tb} AS
-  WITH pre_aggregated_cnefe AS (
-    SELECT {cols_select}, AVG(lon) AS lon, AVG(lat) AS lat,
-    MAX(lon) - MIN(lon) as range_lon, MAX(lat) - MIN(lat) as range_lat,
-    FROM {y}
-    GROUP BY {cols_group}
-  )
-  SELECT {x}.ID, pre_aggregated_cnefe.lon, pre_aggregated_cnefe.lat
-  FROM {x} AS {x}
-  LEFT JOIN pre_aggregated_cnefe
-  ON {join_condition}
-  WHERE pre_aggregated_cnefe.lon IS NOT NULL;"
+query_aggregate_and_match <- sprintf(
+  "CREATE TABLE %s AS
+    WITH pre_aggregated_cnefe AS (
+      SELECT %s AVG(lon) AS lon, AVG(lat) AS lat,
+      2 * STDDEV_SAMP(lon) as range_lon, 2 * STDDEV_SAMP(lat) as range_lat
+      FROM %s
+      GROUP BY %s
+    )
+    SELECT %s.ID, pre_aggregated_cnefe.lon, pre_aggregated_cnefe.lat,
+    pre_aggregated_cnefe.range_lon, pre_aggregated_cnefe.range_lat
+    FROM %s AS %s
+    LEFT JOIN pre_aggregated_cnefe
+    ON %s
+    WHERE pre_aggregated_cnefe.lon IS NOT NULL;",
+  output_tb,     # new table
+  cols_select,   # select
+  y,             # from
+  cols_group,    # group
+  x,             # select
+  x, x,          # from
+  join_condition # on
 )
 
 
+