allows for table.name (no tidyselxn for table.name style yet)

docs/examples/UserGuide/key_differences.jl

@@ -35,6 +35,8 @@ copy_to(db, df, "df_mem"); # copying over the data frame to an in-memory databas
     @collect
  end
 
+# Notice the difference between the summarized tables above and below.
+
 @chain db_table(db, :df_mem) begin
     @group_by(groups)
     @mutate(max = maximum(percent), min = minimum(percent))

src/TBD_macros.jl

@@ -12,8 +12,24 @@ macro select(sqlquery, exprs...)
             columns_str = join(["SELECT ", join([string(column) for column in columns], ", ")])
             $(esc(sqlquery)).select = columns_str
             $(esc(sqlquery)).metadata.current_selxn .= 0
-            selected_indices = indexin(columns, $(esc(sqlquery)).metadata.name)
-            $(esc(sqlquery)).metadata.current_selxn[selected_indices[.!isnothing.(selected_indices)]] .= 1
+            for col in columns
+                if occursin(".", col)
+                    table_col_split = split(col, ".")
+                    table_name, col_name = table_col_split[1], table_col_split[2]
+
+                    # Iterate and update current_selxn based on matches
+                    for idx in eachindex($(esc(sqlquery)).metadata.current_selxn)
+                        if $(esc(sqlquery)).metadata.table_name[idx] == table_name && 
+                           $(esc(sqlquery)).metadata.name[idx] == col_name
+                            $(esc(sqlquery)).metadata.current_selxn[idx] = 1
+                        end
+                    end
+                else
+                    # Direct matching for columns without 'table.' prefix
+                    matching_indices = findall($(esc(sqlquery)).metadata.name .== col)
+                    $(esc(sqlquery)).metadata.current_selxn[matching_indices] .= 1
+                end
+            end
         end
 
         $(esc(sqlquery))
@@ -138,7 +154,7 @@ end
 
 
 
-function process_mutate_expression(expr, sq, select_expressions)
+function process_mutate_expression(expr, sq, select_expressions, cte_name)
     if isa(expr, Expr) && expr.head == :(=) && isa(expr.args[1], Symbol)
         col_name = string(expr.args[1])
         col_expr = expr_to_sql(expr.args[2], sq)  # Convert to SQL expression
@@ -152,7 +168,7 @@ function process_mutate_expression(expr, sq, select_expressions)
             # Append the mutation as a new column expression
             push!(select_expressions, string(col_expr, " AS ", col_name))
             # Update metadata to include this new column
-            push!(sq.metadata, Dict("name" => col_name, "type" => "UNKNOWN", "current_selxn" => 1))
+            push!(sq.metadata, Dict("name" => col_name, "type" => "UNKNOWN", "current_selxn" => 1, "table_name" => cte_name))
         end
     else
         throw("Unsupported expression format in @mutate: $(expr)")
@@ -220,10 +236,10 @@ macro mutate(sqlquery, mutations...)
                 end
                 if isa(expr, Expr) && expr.head == :tuple
                     for subexpr in expr.args
-                        process_mutate_expression(subexpr, sq, select_expressions)
+                        process_mutate_expression(subexpr, sq, select_expressions, cte_name)
                     end
                 else
-                    process_mutate_expression(expr, sq, select_expressions)
+                    process_mutate_expression(expr, sq, select_expressions, cte_name)
                 end
             end
             cte_sql = " " * join(select_expressions, ", ") * " FROM " * sq.from
@@ -350,7 +366,7 @@ function process_summary_expression(expr, sq, summary_str)
         summary_operation = string(summary_operation)
         summary_column = expr_to_sql(expr.args[1], sq, from_summarize = true)
         summary_column = string(summary_column)
-        push!(sq.metadata, Dict("name" => summary_column, "type" => "UNKNOWN", "current_selxn" => 1))
+        push!(sq.metadata, Dict("name" => summary_column, "type" => "UNKNOWN", "current_selxn" => 1, "table_name" => sq.from))
 
         push!(summary_str, summary_operation * " AS " * summary_column)
     else

src/TidierDB.jl

@@ -65,16 +65,6 @@ end
 
 
 
-
-function get_table_metadata(db::SQLite.DB, table_name::String)
-    query = "PRAGMA table_info($table_name);"
-    result = SQLite.DBInterface.execute(db, query) |> DataFrame
-    result[!, :current_selxn] .= 1
-    resize!(result.current_selxn, nrow(result))
-    return select(result, 2 => :name, 3 => :type, :current_selxn)
-end
-
-
 function finalize_ctes(ctes::Vector{CTE})
     if isempty(ctes)
         return ""
@@ -139,6 +129,16 @@ function finalize_query(sqlquery::SQLQuery)
 end
 
 
+function get_table_metadata(db::SQLite.DB, table_name::String)
+    query = "PRAGMA table_info($table_name);"
+    result = SQLite.DBInterface.execute(db, query) |> DataFrame
+    result[!, :current_selxn] .= 1
+    resize!(result.current_selxn, nrow(result))
+    result[!, :table_name] .= table_name
+    # Adjust the select statement to include the new table_name column
+    return DataFrames.select(result, 2 => :name, 3 => :type, :current_selxn, :table_name)
+end
+
 function get_table_metadata(conn::LibPQ.Connection, table_name::String)
     query = """
     SELECT column_name, data_type
@@ -148,9 +148,12 @@ function get_table_metadata(conn::LibPQ.Connection, table_name::String)
     """
     result = LibPQ.execute(conn, query) |> DataFrame
     result[!, :current_selxn] .= 1
-    return select(result, 1 => :name, 2 => :type, :current_selxn)
+    result[!, :table_name] .= table_name
+    # Adjust the select statement to include the new table_name column
+    return select(result, 1 => :name, 2 => :type, :current_selxn, :table_name)
 end
 
+
 # DuckDB
 function get_table_metadata(conn::DuckDB.Connection, table_name::String)
     query = """
@@ -161,7 +164,9 @@ function get_table_metadata(conn::DuckDB.Connection, table_name::String)
     """
     result = DuckDB.execute(conn, query) |> DataFrame
     result[!, :current_selxn] .= 1
-    return select(result, 1 => :name, 2 => :type, :current_selxn)
+    result[!, :table_name] .= table_name
+    # Adjust the select statement to include the new table_name column
+    return select(result, 1 => :name, 2 => :type, :current_selxn, :table_name)
 end
 
 # MySQL
@@ -177,7 +182,9 @@ function get_table_metadata(conn::MySQL.Connection, table_name::String)
     result = DBInterface.execute(conn, query) |> DataFrame
     result[!, :DATA_TYPE] = map(x -> String(x), result.DATA_TYPE)
     result[!, :current_selxn] .= 1
-    return select(result, :COLUMN_NAME => :name, :DATA_TYPE => :type, :current_selxn)
+    result[!, :table_name] .= table_name
+    # Adjust the select statement to include the new table_name column
+    return select(result, :COLUMN_NAME => :name, 2 => :type, :current_selxn, :table_name)
 end
 
 # MSSQL
@@ -193,8 +200,10 @@ function get_table_metadata(conn::ODBC.Connection, table_name::String)
     result = DBInterface.execute(conn, query) |> DataFrame
     #result[!, :DATA_TYPE] = map(x -> String(x), result.DATA_TYPE)
     result[!, :current_selxn] .= 1
-    return select(result, :column_name => :name, :data_type => :type, :current_selxn)
-  end
+    result[!, :table_name] .= table_name
+    # Adjust the select statement to include the new table_name column
+    return select(result, :column_name => :name, :data_type => :type, :current_selxn, :table_name,)
+end
 
  # ClickHouse
 function get_table_metadata(conn::ClickHouse.ClickHouseSock, table_name::String)
@@ -209,8 +218,10 @@ function get_table_metadata(conn::ClickHouse.ClickHouseSock, table_name::String)
     result = ClickHouse.select_df(conn,query)
 
     result[!, :current_selxn] .= 1
-    return select(result, :column_name => :name, :data_type => :type, :current_selxn)
-  end 
+    result[!, :table_name] .= table_name
+    # Adjust the select statement to include the new table_name column
+    return select(result, 1 => :name, 2 => :type, :current_selxn, :table_name)
+end
 
 function db_table(db, table::Symbol)
     table_name = string(table)

src/db_parsing.jl

@@ -29,6 +29,10 @@ function parse_tidy_db(exprs, metadata::DataFrame)
     exprs_iterable = isa(exprs, Tuple) ? collect(exprs) : exprs
 
     for expr in exprs_iterable
+        if occursin(".", string(expr))
+            push!(included_columns, string(expr))
+            continue
+        end
         is_excluded = isa(expr, Expr) && expr.head == :call && expr.args[1] == :(!)
         actual_expr = is_excluded ? expr.args[2] : expr
 
@@ -66,6 +70,10 @@ function parse_tidy_db(exprs, metadata::DataFrame)
             end
         elseif isa(actual_expr, Symbol) || isa(actual_expr, String)
             # Handle single column name
+            if occursin(".", string(actual_expr))
+                push!(included_columns, string(actual_expr))
+                continue
+            end
             col_name = isa(actual_expr, Symbol) ? string(actual_expr) : actual_expr
             if is_excluded
                 push!(excluded_columns, col_name)
@@ -92,10 +100,11 @@ function parse_tidy_db(exprs, metadata::DataFrame)
     else
         included_columns = setdiff(included_columns, excluded_columns)
     end
-
+    
     return included_columns
 end
 
+
 function parse_if_else(expr)
     transformed_expr = MacroTools.postwalk(expr) do x
         # Ensure we're dealing with an Expr object and it's a call to if_else