Merge pull request #8 from sonirico/feat/csv-streaming-response

enhacement: csv parser can handle quotes and separators

csvparser/columns.go

@@ -0,0 +1,76 @@
+package csvparser
+
+type (
+	Col[T any] interface {
+		Parse(data []byte, item *T) (int, error)
+		//Compile(x T, writer io.Writer) error
+	}
+
+	opts struct {
+		sep byte
+	}
+
+	ColFactory[T any] func(opts) Col[T]
+
+	StringColumn[T any] struct {
+		inner  StringType
+		setter func(x *T, v string)
+		getter func(x T) string
+	}
+
+	IntColumn[T any] struct {
+		inner  IntegerType
+		setter func(x *T, v int)
+		getter func(x T) int
+	}
+
+	BoolColumn[T any] struct {
+		inner  StringType
+		setter func(x T, v bool)
+		getter func(x T) bool
+	}
+)
+
+func (s StringColumn[T]) Parse(data []byte, item *T) (int, error) {
+	val, n, err := s.inner.Parse(data)
+	if err != nil {
+		return n, err
+	}
+	s.setter(item, val)
+	return n, nil
+}
+
+func (c IntColumn[T]) Parse(data []byte, item *T) (int, error) {
+	val, n, err := c.inner.Parse(data)
+	if err != nil {
+		return n, err
+	}
+	c.setter(item, val)
+	return n, nil
+}
+
+func StringCol[T any](
+	quote byte,
+	getter func(T) string,
+	setter func(*T, string),
+) ColFactory[T] {
+	return func(opts opts) Col[T] {
+		return StringColumn[T]{
+			inner:  StrType(quote, opts.sep),
+			getter: getter, setter: setter,
+		}
+	}
+}
+
+func IntCol[T any](
+	quote byte,
+	getter func(T) int,
+	setter func(*T, int),
+) ColFactory[T] {
+	return func(opts opts) Col[T] {
+		return IntColumn[T]{
+			inner:  IntType(quote, opts.sep),
+			getter: getter, setter: setter,
+		}
+	}
+}

csvparser/errors.go

@@ -4,4 +4,5 @@ import "github.com/pkg/errors"
 
 var (
 	ErrColumnMismatch = errors.New("column mismatch")
+	ErrQuoteExpected  = errors.New("quote was expected")
 )

csvparser/parser.go

@@ -0,0 +1,64 @@
+package csvparser
+
+import (
+	"bytes"
+)
+
+var (
+	quote = []byte{byte('"')}
+
+	QuoteDouble byte = '"'
+	QuoteSimple byte = '\''
+	QuoteNone   byte = 0
+
+	SeparatorComma     byte = ','
+	SeparatorSemicolon byte = ';'
+	SeparatorTab       byte = '\t'
+)
+
+type (
+	Parser[T any] struct {
+		separator byte
+		columns   []Col[T]
+	}
+)
+
+func (p Parser[T]) Parse(data []byte, item *T) (err error) {
+	data = bytes.TrimSpace(data) // cleanup phase
+	sepLen := 1                  // len(p.separator)
+
+	for i, col := range p.columns {
+		var read int
+		read, err = col.Parse(data, item)
+		if err != nil {
+			return
+		}
+
+		// TODO: handle read =0
+		_ = i
+
+		if read > len(data) {
+			break
+		}
+
+		// create a cursor to have better readability under the fact the column types will only parse
+		// its desired data, letting the parser have the liability to advance de cursor.
+		cursor := read
+		if read+sepLen <= len(data) {
+			cursor += sepLen
+		}
+
+		data = data[cursor:]
+	}
+	return nil
+}
+
+func New[T any](sep byte, cols ...ColFactory[T]) Parser[T] {
+	columns := make([]Col[T], len(cols))
+	opt := opts{sep: sep}
+
+	for i, c := range cols {
+		columns[i] = c(opt)
+	}
+	return Parser[T]{separator: sep, columns: columns}
+}