From 9616984b81840d5954baa5fc1dc53a097fdb7f97 Mon Sep 17 00:00:00 2001
From: cet001 <chadtakahashi@gmail.com>
Date: Thu, 20 Jun 2019 00:26:29 -0700
Subject: [PATCH] - Added `Field.Bytes()` method, which returns the field's
 backing []byte slice. - Exported `ParseUint32([]byte) uint32` function. -
 Fixed bug in `Field.Uint32()` whereby fields containing a number that
 overflows uint32 by more than a factor of 10 would result in a panic.

---
 hastycsv.go      | 62 +++++++++++++++++++++++++++++++++---------------
 hastycsv_test.go | 11 +++++++--
 2 files changed, 52 insertions(+), 21 deletions(-)

diff --git a/hastycsv.go b/hastycsv.go
index bd0e325..2b96a48 100644
--- a/hastycsv.go
+++ b/hastycsv.go
@@ -9,13 +9,27 @@ import (
 	"bytes"
 	"fmt"
 	"io"
+	"math"
 	"os"
 	"strconv"
 	"unsafe"
 )
 
-// Needed by Field.Uint32() parser.
-var base10exp = []uint64{1, 10, 100, 1000, 10000, 100000, 1000000, 10000000, 100000000, 1000000000}
+// Needed by ParseUint32() for better performance.
+var base10exp = []uint64{
+	1, 10, 100, 1000, 10000, 100000, 1000000, 10000000, 100000000,
+	1000000000,
+	10000000000,
+	100000000000,
+	1000000000000,
+	10000000000000,
+	100000000000000,
+	1000000000000000,
+	10000000000000000,
+	100000000000000000,
+	100000000000000000,
+	1000000000000000000,
+}
 
 // Definition of a callback function that serves as a sequential record iterator.
 // Read() and ReadFile() will stop reading the input records if this function
@@ -116,6 +130,11 @@ func (me Field) IsEmpty() bool {
 	return len(me.data) == 0
 }
 
+// Returns the backing byte slice of this field.
+func (me Field) Bytes() []byte {
+	return me.data
+}
+
 // Returns this field as a string.
 func (me Field) String() string {
 	return string(me.data)
@@ -135,27 +154,14 @@ func (me Field) ToLower() Field {
 
 // Parses this field as a Uint32.
 func (me Field) Uint32() uint32 {
-	v := uint64(0)
-	d := len(me.data)
-	for _, b := range me.data {
-		if b < '0' || b > '9' {
-			if me.reader.err == nil {
-				me.reader.err = fmt.Errorf("Field \"%v\" contains non-numeric character '%v'", string(me.data), string(b))
-			}
-			return 0
-		}
-		d--
-		v += uint64(b-'0') * base10exp[d]
-	}
-
-	if v > 4294967295 {
+	i, err := ParseUint32(me.data)
+	if err != nil {
 		if me.reader.err == nil {
-			me.reader.err = fmt.Errorf("%v overflows uint32", string(me.data))
+			me.reader.err = fmt.Errorf(`Can't parse field as uint32: %v`, err)
 		}
-		return 0
 	}
 
-	return uint32(v)
+	return i
 }
 
 // Parses this field as a float32.
@@ -170,6 +176,24 @@ func (me Field) Float32() float32 {
 	return float32(f)
 }
 
+func ParseUint32(data []byte) (uint32, error) {
+	v := uint64(0)
+	d := len(data)
+	for _, ch := range data {
+		if ch < '0' || ch > '9' {
+			return 0, fmt.Errorf(`"%v" contains non-numeric character '%v'`, string(data), string(ch))
+		}
+		d--
+		v += uint64(ch-'0') * base10exp[d]
+	}
+
+	if v > math.MaxUint32 {
+		return 0, fmt.Errorf(`"%v" overflows uint32`, string(data))
+	}
+
+	return uint32(v), nil
+}
+
 // Returns the string representation of this Field without creating a memory allocation.
 //
 // WARNING! The returned string points to this Field object, which is a mutable
diff --git a/hastycsv_test.go b/hastycsv_test.go
index 39c0fed..d4ad07d 100644
--- a/hastycsv_test.go
+++ b/hastycsv_test.go
@@ -45,6 +45,12 @@ func TestField_ToLower(t *testing.T) {
 	}
 }
 
+func TestField_Bytes(t *testing.T) {
+	assert.Equal(t, []byte{}, makeField("").Bytes())
+	assert.Equal(t, []byte{65, 66, 67}, makeField("ABC").Bytes())
+
+}
+
 func TestField_String(t *testing.T) {
 	values := []string{
 		"",
@@ -88,7 +94,8 @@ func TestField_Uint32_parseError(t *testing.T) {
 		"x",
 		"abc",
 		" ",
-		"4294967296", //uint32 overflow
+		"4294967296",      //uint32 overflow (by 1)
+		"999999999999999", // uint32 overflow (by a lot)
 	}
 
 	for _, badlyFormattedInt := range badlyFormattedInts {
@@ -256,7 +263,7 @@ Mary|25|130.5`)
 		return nil
 	})
 
-	assert.EqualError(t, err, "Line 1: Field \"123xyz\" contains non-numeric character 'x'")
+	assert.EqualError(t, err, "Line 1: Can't parse field as uint32: \"123xyz\" contains non-numeric character 'x'")
 }
 
 func TestReadFile(t *testing.T) {