From 0ad0f3a85145f2b64cf5cbffa57cdb779180b724 Mon Sep 17 00:00:00 2001
From: Silvan Jegen <s.jegen@gmail.com>
Date: Sun, 25 Dec 2016 16:41:56 +0100
Subject: Implement our own scanner

This allows us to record byte offset and line number in the token.
---
 conf/parser.go  |  8 +++---
 conf/scanner.go | 82 ++++++++++++++++++++++++++++++++++++++++++---------------
 2 files changed, 65 insertions(+), 25 deletions(-)

(limited to 'conf')

diff --git a/conf/parser.go b/conf/parser.go
index 21bc604..85d9b11 100644
--- a/conf/parser.go
+++ b/conf/parser.go
@@ -17,12 +17,12 @@ type parser struct {
 
 func NewConfig(r io.Reader) *Config {
 	s := newScanner(r)
-	tt, tok, err := s.Next()
+	tok, err := s.Scan()
 	for err == nil {
-		fmt.Fprintf(os.Stderr, "tokentype: %d, token: %q, err: %v\n", tt, tok, err)
-		tt, tok, err = s.Next()
+		fmt.Fprintf(os.Stderr, "tokentype: %d, token: %q offset: %d, line: %d\n", tok.Type, tok.Lit, tok.Offset, tok.LineNr)
+		tok, err = s.Scan()
 	}
-	fmt.Fprintf(os.Stderr, "Error: tokentype: %d, token: %q, err: %v\n", tt, tok, err)
+	fmt.Fprintf(os.Stderr, "Error: tokentype: %d, token: %q, err: %v\n", tok.Type, tok.Lit, err)
 
 	return &Config{}
 }
diff --git a/conf/scanner.go b/conf/scanner.go
index 172103f..3d83534 100644
--- a/conf/scanner.go
+++ b/conf/scanner.go
@@ -1,8 +1,12 @@
 package conf
 
 import (
-	"bufio"
+	"fmt"
 	"io"
+	"io/ioutil"
+	"os"
+	"unicode"
+	"unicode/utf8"
 )
 
 type tokentype int
@@ -16,45 +20,81 @@ const (
 )
 
 type token struct {
-	Type tokentype
+	Type   tokentype
+	Offset int
+	LineNr int
+	Lit    string
 }
 
 type scanner struct {
-	r    io.Reader
-	bs   bufio.Scanner
-	prev token
-	cur  token
+	r       io.Reader
+	data    []byte
+	offset  int
+	curline int
 }
 
 func newScanner(r io.Reader) *scanner {
-	ns := bufio.NewScanner(r)
-	ns.Split(bufio.ScanWords)
+	// TODO: don't be lazy
+	data, err := ioutil.ReadAll(r)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "could not read data from Reader: %v\n", err)
+		os.Exit(1)
+	}
 
 	sc := scanner{
-		r:  r,
-		bs: *ns,
+		data:    data,
+		curline: 1,
 	}
 
 	return &sc
 }
 
-func getTokenType(s string) tokentype {
+func getTokenType(s []byte) tokentype {
 	return Nothing
 }
 
-func (s *scanner) Next() (tokentype, string, error) {
-	more := s.bs.Scan()
-	if !more {
-		err := s.bs.Err()
-		if err != nil {
-			return Nothing, "", err
+func (s *scanner) Scan() (token, error) {
+	processed := 0
+	tokenstarted := false
+	oldline := s.curline
+	for {
+		r, rlen := utf8.DecodeRune(s.data[s.offset+processed:])
+		if r == utf8.RuneError {
+			if rlen == 1 {
+				return token{}, fmt.Errorf("found invalid UTF8 at offset %d, before: %s", s.offset, s.data[s.offset])
+			} else if rlen == 0 {
+				return token{}, io.EOF
+			}
+		}
+
+		processed += rlen
+
+		if unicode.IsSpace(r) {
+			if r == '\n' {
+				s.curline++
+			}
+			if tokenstarted {
+				break
+			}
+			s.offset += rlen
+			processed = 0
+			continue
 		}
-		return Nothing, "", io.EOF
+		tokenstarted = true
 	}
-	tokstr := s.bs.Text()
-	token := getTokenType(tokstr)
 
-	return token, tokstr, nil
+	tokbytes := s.data[s.offset : s.offset+processed-1]
+	tokent := getTokenType(tokbytes)
+
+	s.offset += processed
+
+	ret := token{
+		Type:   tokent,
+		Offset: s.offset,
+		LineNr: oldline,
+		Lit:    string(tokbytes),
+	}
+	return ret, nil
 }
 
 func (s *scanner) peek() (tokentype, string, error) {
-- 
cgit v1.2.1-18-gbd029