Rewrite tokenizer to clean up and fix a bug with spaces before delimited block.

Fixes #501.

R=rsc
CC=golang-dev
https://golang.org/cl/181183

2 files changed, 73 insertions, 45 deletions

diff --git a/src/pkg/template/template.go b/src/pkg/template/template.go
index ef694b24b0..f1257b0915 100644
--- a/src/pkg/template/template.go
+++ b/src/pkg/template/template.go
@@ -224,63 +224,69 @@ func equal(s []byte, n int, t []byte) bool {
 // Action tokens on a line by themselves drop the white space on
 // either side, up to and including the newline.
 func (t *Template) nextItem() []byte {
-	sawLeft := false // are we waiting for an opening delimiter?
 	special := false // is this a {.foo} directive, which means trim white space?
 	// Delete surrounding white space if this {.foo} is the only thing on the line.
-	trim_white := t.p == 0 || t.buf[t.p-1] == '\n'
-	only_white := true // we have seen only white space so far
-	var i int
+	trimSpace := t.p == 0 || t.buf[t.p-1] == '\n'
 	start := t.p
-Loop:
-	for i = t.p; i < len(t.buf); i++ {
-		switch {
-		case t.buf[i] == '\n':
-			t.linenum++
-			i++
-			break Loop
-		case white(t.buf[i]):
-			// white space, do nothing
-		case !sawLeft && equal(t.buf, i, t.ldelim): // sawLeft checked because delims may be equal
-			// anything interesting already on the line?
-			if !only_white {
-				break Loop
-			}
-			// is it a directive or comment?
-			j := i + len(t.ldelim) // position after delimiter
-			if j+1 < len(t.buf) && (t.buf[j] == '.' || t.buf[j] == '#') {
-				special = true
-				if trim_white && only_white {
-					start = i
-				}
-			} else if i > t.p { // have some text accumulated so stop before delimiter
-				break Loop
+	var i int
+	newline := func() {
+		t.linenum++
+		i++
+	}
+	// Leading white space up to but not including newline
+	for i = start; i < len(t.buf); i++ {
+		if t.buf[i] == '\n' || !white(t.buf[i]) {
+			break
+		}
+	}
+	if trimSpace {
+		start = i
+	} else if i > start {
+		// white space is valid text
+		t.p = i
+		return t.buf[start:i]
+	}
+	// What's left is nothing, newline, delimited string, or plain text
+Switch:
+	switch {
+	case i == len(t.buf):
+		// EOF; nothing to do
+	case t.buf[i] == '\n':
+		newline()
+	case equal(t.buf, i, t.ldelim):
+		i += len(t.ldelim) // position after delimiter
+		if i+1 < len(t.buf) && (t.buf[i] == '.' || t.buf[i] == '#') {
+			special = true
+		}
+		for ; i < len(t.buf); i++ {
+			if t.buf[i] == '\n' {
+				break
 			}
-			sawLeft = true
-			i = j - 1
-		case equal(t.buf, i, t.rdelim):
-			if !sawLeft {
-				t.parseError("unmatched closing delimiter")
-				return nil
+			if equal(t.buf, i, t.rdelim) {
+				i += len(t.rdelim)
+				break Switch
 			}
-			sawLeft = false
-			i += len(t.rdelim)
-			break Loop
-		default:
-			only_white = false
 		}
-	}
-	if sawLeft {
 		t.parseError("unmatched opening delimiter")
 		return nil
+	default:
+		for ; i < len(t.buf); i++ {
+			if t.buf[i] == '\n' {
+				newline()
+				break
+			}
+			if equal(t.buf, i, t.ldelim) {
+				break
+			}
+		}
 	}
 	item := t.buf[start:i]
-	if special && trim_white {
+	if special && trimSpace {
 		// consume trailing white space
 		for ; i < len(t.buf) && white(t.buf[i]); i++ {
 			if t.buf[i] == '\n' {
-				t.linenum++
-				i++
-				break // stop after newline
+				newline()
+				break // stop before newline
 			}
 		}
 	}
diff --git a/src/pkg/template/template_test.go b/src/pkg/template/template_test.go
index 65dae3a490..0ae581c593 100644
--- a/src/pkg/template/template_test.go
+++ b/src/pkg/template/template_test.go
@@ -86,6 +86,7 @@ var formatters = FormatterMap{
 var tests = []*Test{
 	// Simple
 	&Test{"", "", ""},
+	&Test{"abc", "abc", ""},
 	&Test{"abc\ndef\n", "abc\ndef\n", ""},
 	&Test{" {.meta-left}   \n", "{", ""},
 	&Test{" {.meta-right}   \n", "}", ""},
@@ -173,6 +174,7 @@ var tests = []*Test{
 		out: "Header=77\n" +
 			"Header=77\n",
 	},
+
 	&Test{
 		in: "{.section data}{.end} {header}\n",
 
@@ -225,6 +227,17 @@ var tests = []*Test{
 			"ItemNumber2\n",
 	},
 	&Test{
+		in: "{.repeated section pdata }\n" +
+			"{item}\n" +
+			"{.alternates with}\n" +
+			"is\nover\nmultiple\nlines\n" +
+			" {.end}\n",
+
+		out: "ItemNumber1\n" +
+			"is\nover\nmultiple\nlines\n" +
+			"ItemNumber2\n",
+	},
+	&Test{
 		in: "{.section pdata }\n" +
 			"{.repeated section @ }\n" +
 			"{item}={value}\n" +
@@ -246,6 +259,13 @@ var tests = []*Test{
 		out: "elt1\n" +
 			"elt2\n",
 	},
+	// Same but with a space before {.end}: was a bug.
+	&Test{
+		in: "{.repeated section vec }\n" +
+			"{@} {.end}\n",
+
+		out: "elt1 elt2 \n",
+	},
 	&Test{
 		in: "{.repeated section integer}{.end}",
 
@@ -374,7 +394,9 @@ func TestAll(t *testing.T) {
 				t.Error("unexpected execute error:", err)
 			}
 		} else {
-			if err == nil || err.String() != test.err {
+			if err == nil {
+				t.Errorf("expected execute error %q, got nil", test.err)
+			} else if err.String() != test.err {
 				t.Errorf("expected execute error %q, got %q", test.err, err.String())
 			}
 		}