From 6c8c9a9c81abad7e55d2a7f75d776cd6684fe8b9 Mon Sep 17 00:00:00 2001
From: rokostik <rokostik@gmail.com>
Date: Fri, 15 Dec 2023 22:29:12 +0100
Subject: [PATCH] add spreadsheet autotype function

---
 evaldo/builtins_spreadsheet.go | 86 ++++++++++++++++++++++++++++++++--
 examples/spreadsheet/data.csv  | 22 ++++-----
 2 files changed, 92 insertions(+), 16 deletions(-)

diff --git a/evaldo/builtins_spreadsheet.go b/evaldo/builtins_spreadsheet.go
index 347fa8f1..78919eca 100644
--- a/evaldo/builtins_spreadsheet.go
+++ b/evaldo/builtins_spreadsheet.go
@@ -118,12 +118,14 @@ var Builtins_spreadsheet = map[string]*env.Builtin{
 					return MakeBuiltinError(ps, "Unable to parse file as CSV.", "load\\csv")
 				}
 				spr := env.NewSpreadsheet(rows[0])
-				for _, row := range rows {
-					anyRow := make([]any, len(row))
-					for i, v := range row {
-						anyRow[i] = v
+				if len(rows) > 1 {
+					for _, row := range rows[1:] {
+						anyRow := make([]any, len(row))
+						for i, v := range row {
+							anyRow[i] = *env.NewString(v)
+						}
+						spr.AddRow(*env.NewSpreadsheetRow(anyRow, spr))
 					}
-					spr.AddRow(*env.NewSpreadsheetRow(anyRow, spr))
 				}
 				return *spr
 			default:
@@ -337,6 +339,23 @@ var Builtins_spreadsheet = map[string]*env.Builtin{
 			}
 		},
 	},
+	"autotype": {
+		Argsn: 2,
+		Doc:   "",
+		Fn: func(ps *env.ProgramState, arg0 env.Object, arg1 env.Object, arg2 env.Object, arg3 env.Object, arg4 env.Object) (res env.Object) {
+			switch spr := arg0.(type) {
+			case env.Spreadsheet:
+				switch percent := arg1.(type) {
+				case env.Decimal:
+					return AutoType(ps, &spr, percent.Value)
+				default:
+					return MakeArgError(ps, 2, []env.Type{env.DecimalType}, "autotype")
+				}
+			default:
+				return MakeArgError(ps, 1, []env.Type{env.SpreadsheetType}, "autotype")
+			}
+		},
+	},
 }
 
 func GenerateColumn(ps *env.ProgramState, s env.Spreadsheet, name env.Word, extractCols env.Block, code env.Block) env.Object {
@@ -469,3 +488,60 @@ func Limit(ps *env.ProgramState, s env.Spreadsheet, n int) env.Object {
 	nspr.Rows = s.Rows[0:n]
 	return *nspr
 }
+
+func AutoType(ps *env.ProgramState, s *env.Spreadsheet, percent float64) env.Object {
+	colTypeCount := make(map[int]map[string]int)
+	for i := range s.Cols {
+		colTypeCount[i] = make(map[string]int)
+	}
+	for _, row := range s.Rows {
+		for i, val := range row.Values {
+			switch stringVal := val.(type) {
+			case env.String:
+				if _, err := strconv.Atoi(stringVal.Value); err == nil {
+					colTypeCount[i]["int"]++
+				} else if _, err = strconv.ParseFloat(stringVal.Value, 64); err == nil {
+					colTypeCount[i]["dec"]++
+				} else {
+					colTypeCount[i]["str"]++
+				}
+			default:
+				continue
+			}
+		}
+	}
+
+	lenRows := len(s.Rows)
+	newS := env.NewSpreadsheet(s.Cols)
+	for range s.Rows {
+		newRow := make([]any, len(s.Cols))
+		newS.AddRow(*env.NewSpreadsheetRow(newRow, newS))
+	}
+
+	for colNum, typeCount := range colTypeCount {
+		minRows := int(float64(lenRows) * percent)
+		var newType string
+		// if there's a mix of floats and ints, make it a float
+		if typeCount["dec"] > 0 && typeCount["dec"]+typeCount["int"] >= minRows {
+			newType = "dec"
+		} else if typeCount["int"] >= minRows {
+			newType = "int"
+		} else {
+			newType = "str"
+		}
+		for i, row := range s.Rows {
+			switch newType {
+			case "int":
+				intVal, _ := strconv.Atoi(row.Values[colNum].(env.String).Value)
+				newS.Rows[i].Values[colNum] = *env.NewInteger(int64(intVal))
+			case "dec":
+				floatVal, _ := strconv.ParseFloat(row.Values[colNum].(env.String).Value, 64)
+				newS.Rows[i].Values[colNum] = *env.NewDecimal(floatVal)
+			case "str":
+				newS.Rows[i].Values[colNum] = row.Values[colNum]
+			}
+		}
+	}
+
+	return *newS
+}
diff --git a/examples/spreadsheet/data.csv b/examples/spreadsheet/data.csv
index 03eef445..8963a230 100644
--- a/examples/spreadsheet/data.csv
+++ b/examples/spreadsheet/data.csv
@@ -1,11 +1,11 @@
-ID,Name,Department
-1,John Doe,Marketing
-2,Jane Smith,Engineering
-3,Robert Johnson,Human Resources
-4,Julia Davis,Sales
-5,James Brown,Engineering
-6,Mary Williams,Marketing
-7,Michael Miller,Human Resources
-8,Linda Wilson,Sales
-9,William Moore,Engineering
-10,Elizabeth Taylor,Marketing
\ No newline at end of file
+ID,Name,Department,Years at Company
+1,John Doe,Marketing,2
+2,Jane Smith,Engineering,5
+3,Robert Johnson,Human Resources,1
+4,Julia Davis,Sales,3
+5,James Brown,Engineering,2
+6,Mary Williams,Marketing,4
+7,Michael Miller,Human Resources,3
+8,Linda Wilson,Sales,1
+9,William Moore,Engineering,2
+10,Elizabeth Taylor,Marketing,3
\ No newline at end of file