improved tokenizer, added tests

config/config.go

@@ -1,9 +1,9 @@
 package config
 
 import (
+	"alrm/alarm"
 	"fmt"
 	"time"
-	"alrm/alarm"
 )
 
 type Config struct {

config/parser.go

@@ -8,13 +8,13 @@ import (
 )
 
 const (
-	TK_NONE = iota
-	TK_SET
-	TK_MONITOR
-	TK_GROUP
-	TK_HOST
-	TK_CHECK
-	TK_ALARM
+	PR_NONE = iota
+	PR_SET
+	PR_MONITOR
+	PR_GROUP
+	PR_HOST
+	PR_CHECK
+	PR_ALARM
 )
 
 type Parser struct {
@@ -39,20 +39,20 @@ func (p *Parser) Parse(fn string) (*Config, error) {
 		tk := tok.Text()
 	stateswitch:
 		switch p.state() {
-		case TK_NONE:
+		case PR_NONE:
 			switch strings.ToLower(tk) {
 			case "monitor":
-				p.setState(TK_MONITOR)
+				p.setState(PR_MONITOR)
 			case "set":
-				p.setState(TK_SET)
+				p.setState(PR_SET)
 			case "alarm":
-				p.setState(TK_ALARM)
+				p.setState(PR_ALARM)
 			default:
 				return nil, fmt.Errorf("invalid token in %s, line %d: \"%s\"",
 					fn, tok.Line(), tk)
 			}
 
-		case TK_SET:
+		case PR_SET:
 			key := strings.ToLower(tk)
 			if !tok.Scan() {
 				return nil, fmt.Errorf("empty value name for set in %s, line %d",
@@ -76,20 +76,20 @@ func (p *Parser) Parse(fn string) (*Config, error) {
 			}
 			p.prevState()
 
-		case TK_MONITOR:
+		case PR_MONITOR:
 			switch strings.ToLower(tk) {
 			case "host":
-				p.setState(TK_HOST)
+				p.setState(PR_HOST)
 
 			case "group":
-				p.setState(TK_GROUP)
+				p.setState(PR_GROUP)
 
 			default:
 				p.prevState()
 				goto stateswitch
 			}
 
-		case TK_GROUP:
+		case PR_GROUP:
 			if p.lastGroup == nil {
 				p.lastGroup, err = config.NewGroup(tk)
 				if err != nil {
@@ -102,14 +102,14 @@ func (p *Parser) Parse(fn string) (*Config, error) {
 
 			switch strings.ToLower(tk) {
 			case "host":
-				p.setState(TK_HOST)
+				p.setState(PR_HOST)
 
 			default:
 				p.prevState()
 				goto stateswitch
 			}
 
-		case TK_HOST:
+		case PR_HOST:
 			// If a host has no group, inherit the host name
 			if p.lastGroup == nil {
 				p.lastGroup, err = config.NewGroup(tk)
@@ -139,14 +139,14 @@ func (p *Parser) Parse(fn string) (*Config, error) {
 				p.lastHost.Address = tok.Text()
 
 			case "check":
-				p.setState(TK_CHECK)
+				p.setState(PR_CHECK)
 
 			default:
 				p.prevState()
 				goto stateswitch
 			}
 
-		case TK_CHECK:
+		case PR_CHECK:
 			if p.lastCheck == nil {
 				p.lastCheck, err = p.lastHost.NewCheck(tk)
 				if err != nil {
@@ -166,7 +166,7 @@ func (p *Parser) Parse(fn string) (*Config, error) {
 				goto stateswitch
 			}
 
-		case TK_ALARM:
+		case PR_ALARM:
 			if p.lastAlarm == nil {
 				if p.lastAlarmName == "" {
 					p.lastAlarmName = tk
@@ -204,19 +204,19 @@ func (p *Parser) Parse(fn string) (*Config, error) {
 
 func (p *Parser) state() int {
 	if len(p.states) < 1 {
-		return TK_NONE
+		return PR_NONE
 	}
 	return p.states[len(p.states)-1]
 }
 
 func (p *Parser) setState(state int) {
 	switch state {
-	case TK_SET, TK_MONITOR:
+	case PR_SET, PR_MONITOR:
 		fallthrough
-	case TK_GROUP:
+	case PR_GROUP:
 		p.lastGroup = nil
 		fallthrough
-	case TK_HOST:
+	case PR_HOST:
 		p.lastHost = nil
 		p.lastCheck = nil
 	}
@@ -239,20 +239,20 @@ func (p *Parser) prevState() int {
 
 func (p *Parser) stateName() string {
 	switch p.state() {
-	case TK_NONE:
-		return "TK_NONE"
-	case TK_SET:
-		return "TK_SET"
-	case TK_MONITOR:
-		return "TK_MONITOR"
-	case TK_GROUP:
-		return "TK_GROUP"
-	case TK_HOST:
-		return "TK_HOST"
-	case TK_CHECK:
-		return "TK_CHECK"
-	case TK_ALARM:
-		return "TK_ALARM"
+	case PR_NONE:
+		return "PR_NONE"
+	case PR_SET:
+		return "PR_SET"
+	case PR_MONITOR:
+		return "PR_MONITOR"
+	case PR_GROUP:
+		return "PR_GROUP"
+	case PR_HOST:
+		return "PR_HOST"
+	case PR_CHECK:
+		return "PR_CHECK"
+	case PR_ALARM:
+		return "PR_ALARM"
 	default:
 		return "UNKNOWN"
 	}

config/testdata/comments-inline.tok

@@ -0,0 +1,5 @@
+one #one
+"two#three"
+# "three"
+four
+# EOF

config/testdata/comments.tok

@@ -1,5 +1,6 @@
-# one two three
-one two three
-#four five six
-four five six #seven eight nine
-# EOF
+# one
+one
+#two
+two
+   # three
+   three

config/testdata/quotes-multiline.tok

@@ -0,0 +1,6 @@
+"one
+two"   'three
+four'
+
+`five
+  six`

config/testdata/quotes.tok

@@ -1,8 +1,3 @@
-"one" "two three"
+"one" 'two' `three`
 
-"four five"
-
-" #six" "" "seven" "ei" "ght"
-
-"multi
-line"
+`four` 'five' "six"

config/testdata/simple-spaces.tok

@@ -0,0 +1 @@
+one two three    four five six

config/testdata/simple.tok

@@ -1 +0,0 @@
-one two three four five six

config/tokenizer.go

@@ -2,26 +2,37 @@ package config
 
 import (
 	"bufio"
-	"fmt"
+	"io"
 	"os"
+	"strings"
+	"unicode"
+)
+
+const (
+	TK_NONE = iota
+	TK_VAL
+	TK_QUOTE
+	TK_COMMENT
 )
 
 type Tokenizer struct {
-	line    int
+	curline int
+	repline int
 	file    *os.File
-	scanner *bufio.Scanner
+	reader  *bufio.Reader
+	text    string
+	err     error
 }
 
 func NewTokenizer(fn string) (*Tokenizer, error) {
 	var err error
-	tk := &Tokenizer{line: 1}
+	tk := &Tokenizer{curline: 1}
 	tk.file, err = os.Open(fn)
 	if err != nil {
 		return nil, err
 	}
 
-	tk.scanner = bufio.NewScanner(tk.file)
-	tk.scanner.Split(tk.Split)
+	tk.reader = bufio.NewReader(tk.file)
 	return tk, nil
 }
 
@@ -30,93 +41,114 @@ func (t *Tokenizer) Close() error {
 }
 
 func (t *Tokenizer) Scan() bool {
-	return t.scanner.Scan()
+	t.repline = t.curline
+	state := TK_NONE
+	t.text = ""
+
+	var b strings.Builder
+	var quo rune
+	for {
+		var r rune
+		r, _, t.err = t.reader.ReadRune()
+		if t.err != nil {
+			break
+		}
+
+		switch state {
+		case TK_NONE:
+			// When between values, increment both the reported line
+			// and the current line, since there's not yet anything
+			// to report
+			if r == '\n' {
+				t.repline++
+				t.curline++
+			}
+
+			// If we're between values and we encounter a space
+			// or a control character, ignore it
+			if unicode.IsSpace(r) || unicode.IsControl(r) {
+				continue
+			}
+
+			// If we're between values and we encounter a #, it's
+			// the beginning of a comment
+			if r == '#' {
+				state = TK_COMMENT
+				continue
+			}
+
+			// If we're between values and we get a quote character
+			// treat it as the beginning of a string literal
+			if r == '"' || r == '\'' || r == '`' {
+				state = TK_QUOTE
+				quo = r
+				continue
+			}
+
+			b.WriteRune(r)
+			state = TK_VAL
+
+		case TK_VAL:
+			// In values, only increment the current line, so
+			// if an error is reported, it reports the line
+			// the value starts on
+			if r == '\n' {
+				t.curline++
+			}
+
+			// If we're in a normal value and we encounter a space
+			// or a control character, end value
+			if unicode.IsSpace(r) || unicode.IsControl(r) {
+				goto end
+			}
+			b.WriteRune(r)
+
+		case TK_QUOTE:
+			// In quotes, only increment the current line, so
+			// if an error is reported, it reports the line
+			// the quoted value starts on
+			if r == '\n' {
+				t.curline++
+			}
+
+			// End this quote if it's another quote of the same rune
+			if r == quo {
+				goto end
+			}
+			b.WriteRune(r)
+
+		case TK_COMMENT:
+			// Comments are ignored, until a new line is encounter
+			// at which point, increment the current and reported line
+			if r == '\n' {
+				t.curline++
+				t.repline++
+				state = TK_NONE
+			}
+			continue
+		}
+	}
+
+end:
+	if t.err == nil || t.err == io.EOF {
+		if b.Len() > 0 {
+			t.text = b.String()
+		}
+	}
+	return t.err == nil
 }
 
 func (t *Tokenizer) Text() string {
-	return t.scanner.Text()
+	return t.text
 }
 
 func (t *Tokenizer) Line() int {
-	return t.line
+	return t.repline
 }
 
 func (t *Tokenizer) Err() error {
-	return t.scanner.Err()
-}
-
-func (t *Tokenizer) Split(data []byte, atEOF bool) (int, []byte, error) {
-	if atEOF && len(data) == 0 {
-		return 0, nil, nil
-	}
-
-	var ignoreline bool
-	var started bool
-	var startidx int
-	var quote byte
-
-	for i := 0; i < len(data); i++ {
-		c := data[i]
-		//fmt.Printf("%c (%t) (%t)\n", c, started, ignoreline)
-		switch c {
-		case '\f', '\n', '\r':
-			if started {
-				return i, data[startidx:i], nil
-			}
-
-			t.line++
-			if ignoreline {
-				ignoreline = false
-				continue
-			}
-			fallthrough
-
-		case ' ', '\t', '\v':
-			if started && quote == 0 {
-				return i + 1, data[startidx:i], nil
-			}
-
-		case '\'', '"', '`':
-			// When the quote ends
-			if quote == c {
-				// if we've gotten data, return it
-				if started {
-					return i + 1, data[startidx:i], nil
-				}
-				// if we haven't return nothing
-				return i + 1, []byte{}, nil
-			}
-
-			// start a quoted string
-			if !ignoreline && quote == 0 {
-				quote = c
-			}
-
-		case '#':
-			if !started {
-				ignoreline = true
-			}
-
-		default:
-			if !ignoreline && !started {
-				started = true
-				startidx = i
-			}
-		}
-	}
-
-	if atEOF {
-		if quote != 0 {
-			return 0, nil, fmt.Errorf("unterminated quote")
-		}
-
-		if ignoreline {
-			return len(data), nil, nil
-		}
-		if started {
-			return len(data), data[startidx:], nil
-		}
-	}
-
-	return 0, nil, nil
+	if t.err == io.EOF {
+		return nil
+	}
+	return t.err
 }

config/tokenizer_test.go

@@ -1,26 +1,48 @@
 package config
 
 import (
-	"testing"
 	"encoding/json"
+	"testing"
 )
 
-func TestTokenizer(t *testing.T) {
-	runTest(t, "simple",
+func TestSimpleSpaces(t *testing.T) {
+	runTest(t, "simple-spaces",
 		`[["one","two","three","four","five","six"]]`,
 	)
-	runTest(t, "simple-broken",
+}
+
+func TestSimpleMultiline(t *testing.T) {
+	runTest(t, "simple-multiline",
 		`[["one","two","three"],["four","five"],[],[],["six"]]`,
 	)
-	runTest(t, "comments",
-	 `[[],["one","two","three"],[],["four","five","six"]]`,
-	)
+}
+
+func TestQuotes(t *testing.T) {
 	runTest(t, "quotes",
-	 `[["one","two three",[],["four five"],[],[" #six","","seven","ei","ght"],[],["multi\nline"]]`,
+		`[["one","two","three"],[],["four","five","six"]]`,
+	)
+}
+
+func TestQuotesMultiline(t *testing.T) {
+	runTest(t, "quotes-multiline",
+		`[["one\ntwo"],["three\nfour"],[],[],["five\n  six"]]`,
+	)
+}
+
+func TestComments(t *testing.T) {
+	runTest(t, "comments",
+		`[[],["one"],[],["two"],[],["three"]]`,
+	)
+}
+
+func TestCommentsInline(t *testing.T) {
+	runTest(t, "comments-inline",
+		`[["one"],["two#three"],[],["four"]]`,
 	)
 }
 
 func runTest(t *testing.T, bn string, exp string) {
+	t.Logf("Running testdata/%s.tok.. ", bn)
 	tok, err := NewTokenizer("testdata/" + bn + ".tok")
 	if err != nil {
 		t.Fatalf("%s", err.Error())
@@ -49,7 +71,7 @@ func runTest(t *testing.T, bn string, exp string) {
 
 	if exp != string(out) {
 		t.Logf("Expected: %s", exp)
-		t.Logf("Got: %s", out)
-		t.Fail()
+		t.Logf("Got:      %s", out)
+		t.FailNow()
 	}
 }