Imported Upstream version 1.1~hg20130304upstream/1.1_hg20130304

1 files changed, 188 insertions, 69 deletions

diff --git a/src/cmd/yacc/yacc.go b/src/cmd/yacc/yacc.go
index e94228152..0c18f93b6 100644
--- a/src/cmd/yacc/yacc.go
+++ b/src/cmd/yacc/yacc.go
@@ -51,6 +51,8 @@ import (
 	"fmt"
 	"os"
 	"strings"
+	"unicode"
+	"unicode/utf8"
 )
 
 // the following are adjustable
@@ -153,6 +155,8 @@ var ftable *bufio.Writer    // y.go file
 var fcode = &bytes.Buffer{} // saved code
 var foutput *bufio.Writer   // y.output file
 
+var fmtImported bool // output file has recorded an import of "fmt"
+
 var oflag string  // -o [y.go]		- y.go file
 var vflag string  // -v [y.output]	- y.output file
 var lflag bool    // -l			- disable line directives
@@ -323,7 +327,6 @@ var resrv = []Resrv{
 var zznewstate = 0
 
 const EOF = -1
-const UTFmax = 0x3f
 
 func main() {
 
@@ -716,8 +719,8 @@ func moreprod() {
 }
 
 //
-// define s to be a terminal if t=0
-// or a nonterminal if t=1
+// define s to be a terminal if nt==0
+// or a nonterminal if nt==1
 //
 func defin(nt int, s string) int {
 	val := 0
@@ -750,56 +753,66 @@ func defin(nt int, s string) int {
 
 	// establish value for token
 	// single character literal
-	if s[0] == ' ' && len(s) == 1+1 {
-		val = int(s[1])
-	} else if s[0] == ' ' && s[1] == '\\' { // escape sequence
-		if len(s) == 2+1 {
-			// single character escape sequence
-			switch s[2] {
-			case '\'':
-				val = '\''
-			case '"':
-				val = '"'
-			case '\\':
-				val = '\\'
-			case 'a':
-				val = '\a'
-			case 'b':
-				val = '\b'
-			case 'n':
-				val = '\n'
-			case 'r':
-				val = '\r'
-			case 't':
-				val = '\t'
-			case 'v':
-				val = '\v'
-			default:
-				errorf("invalid escape %v", s[1:3])
-			}
-		} else if s[2] == 'u' && len(s) == 2+1+4 { // \unnnn sequence
-			val = 0
-			s = s[3:]
-			for s != "" {
-				c := int(s[0])
-				switch {
-				case c >= '0' && c <= '9':
-					c -= '0'
-				case c >= 'a' && c <= 'f':
-					c -= 'a' - 10
-				case c >= 'A' && c <= 'F':
-					c -= 'A' - 10
+	if s[0] == ' ' {
+		s = s[1:]
+		r, size := utf8.DecodeRuneInString(s)
+		if r == utf8.RuneError && size == 1 {
+			errorf("invalid UTF-8 sequence %q", s)
+		}
+		val = int(r)
+		if val == '\\' { // escape sequence
+			switch {
+			case len(s) == 2:
+				// single character escape sequence
+				switch s[1] {
+				case '\'':
+					val = '\''
+				case '"':
+					val = '"'
+				case '\\':
+					val = '\\'
+				case 'a':
+					val = '\a'
+				case 'b':
+					val = '\b'
+				case 'f':
+					val = '\f'
+				case 'n':
+					val = '\n'
+				case 'r':
+					val = '\r'
+				case 't':
+					val = '\t'
+				case 'v':
+					val = '\v'
 				default:
-					errorf("illegal \\unnnn construction")
+					errorf("invalid escape %s", s)
 				}
-				val = val*16 + c
-				s = s[1:]
-			}
-			if val == 0 {
-				errorf("'\\u0000' is illegal")
+			case s[1] == 'u' && len(s) == 2+4, // \unnnn sequence
+				s[1] == 'U' && len(s) == 2+8: // \Unnnnnnnn sequence
+				val = 0
+				s = s[2:]
+				for s != "" {
+					c := int(s[0])
+					switch {
+					case c >= '0' && c <= '9':
+						c -= '0'
+					case c >= 'a' && c <= 'f':
+						c -= 'a' - 10
+					case c >= 'A' && c <= 'F':
+						c -= 'A' - 10
+					default:
+						errorf(`illegal \u or \U construction`)
+					}
+					val = val*16 + c
+					s = s[1:]
+				}
+			default:
+				errorf("invalid escape %s", s)
 			}
-		} else {
-			errorf("unknown escape")
+		}
+		if val == 0 {
+			errorf("token value 0 is illegal")
 		}
 	} else {
 		val = extval
@@ -985,7 +998,7 @@ func gettok() int {
 
 func getword(c rune) {
 	tokname = ""
-	for isword(c) || isdigit(c) || c == '_' || c == '.' || c == '$' {
+	for isword(c) || isdigit(c) || c == '.' || c == '$' {
 		tokname += string(c)
 		c = getrune(finput)
 	}
@@ -1073,6 +1086,7 @@ out:
 
 //
 // saves code between %{ and %}
+// adds an import for __fmt__ the first time
 //
 func cpycode() {
 	lno := lineno
@@ -1085,15 +1099,18 @@ func cpycode() {
 	if !lflag {
 		fmt.Fprintf(ftable, "\n//line %v:%v\n", infile, lineno)
 	}
+	// accumulate until %}
+	code := make([]rune, 0, 1024)
 	for c != EOF {
 		if c == '%' {
 			c = getrune(finput)
 			if c == '}' {
+				emitcode(code, lno+1)
 				return
 			}
-			ftable.WriteRune('%')
+			code = append(code, '%')
 		}
-		ftable.WriteRune(c)
+		code = append(code, c)
 		if c == '\n' {
 			lineno++
 		}
@@ -1104,6 +1121,107 @@ func cpycode() {
 }
 
 //
+// emits code saved up from between %{ and %}
+// called by cpycode
+// adds an import for __yyfmt__ after the package clause
+//
+func emitcode(code []rune, lineno int) {
+	for i, line := range lines(code) {
+		writecode(line)
+		if !fmtImported && isPackageClause(line) {
+			fmt.Fprintln(ftable, `import __yyfmt__ "fmt"`)
+			fmt.Fprintf(ftable, "//line %v:%v\n\t\t", infile, lineno+i)
+			fmtImported = true
+		}
+	}
+}
+
+//
+// does this line look like a package clause?  not perfect: might be confused by early comments.
+//
+func isPackageClause(line []rune) bool {
+	line = skipspace(line)
+
+	// must be big enough.
+	if len(line) < len("package X\n") {
+		return false
+	}
+
+	// must start with "package"
+	for i, r := range []rune("package") {
+		if line[i] != r {
+			return false
+		}
+	}
+	line = skipspace(line[len("package"):])
+
+	// must have another identifier.
+	if len(line) == 0 || (!unicode.IsLetter(line[0]) && line[0] != '_') {
+		return false
+	}
+	for len(line) > 0 {
+		if !unicode.IsLetter(line[0]) && !unicode.IsDigit(line[0]) && line[0] != '_' {
+			break
+		}
+		line = line[1:]
+	}
+	line = skipspace(line)
+
+	// eol, newline, or comment must follow
+	if len(line) == 0 {
+		return true
+	}
+	if line[0] == '\r' || line[0] == '\n' {
+		return true
+	}
+	if len(line) >= 2 {
+		return line[0] == '/' && (line[1] == '/' || line[1] == '*')
+	}
+	return false
+}
+
+//
+// skip initial spaces
+//
+func skipspace(line []rune) []rune {
+	for len(line) > 0 {
+		if line[0] != ' ' && line[0] != '\t' {
+			break
+		}
+		line = line[1:]
+	}
+	return line
+}
+
+//
+// break code into lines
+//
+func lines(code []rune) [][]rune {
+	l := make([][]rune, 0, 100)
+	for len(code) > 0 {
+		// one line per loop
+		var i int
+		for i = range code {
+			if code[i] == '\n' {
+				break
+			}
+		}
+		l = append(l, code[:i+1])
+		code = code[i+1:]
+	}
+	return l
+}
+
+//
+// writes code to ftable
+//
+func writecode(code []rune) {
+	for _, r := range code {
+		ftable.WriteRune(r)
+	}
+}
+
+//
 // skip over comments
 // skipcom is called after reading a '/'
 //
@@ -1230,7 +1348,7 @@ loop:
 				if j >= max {
 					errorf("Illegal use of $%v", j)
 				}
-			} else if isword(c) || c == '_' || c == '.' {
+			} else if isword(c) || c == '.' {
 				// look for $name
 				ungetrune(finput, c)
 				if gettok() != IDENTIFIER {
@@ -2775,7 +2893,7 @@ func others() {
 		j = tokset[i].value
 		if j >= 0 && j < 256 {
 			if temp1[j] != 0 {
-				fmt.Print("yacc bug -- cant have 2 different Ts with same value\n")
+				fmt.Print("yacc bug -- cannot have 2 different Ts with same value\n")
 				fmt.Printf("	%s and %s\n", tokset[i].name, tokset[temp1[j]].name)
 				nerrors++
 			}
@@ -2799,7 +2917,7 @@ func others() {
 		j = tokset[i].value - PRIVATE
 		if j >= 0 && j < 256 {
 			if temp1[j] != 0 {
-				fmt.Print("yacc bug -- cant have 2 different Ts with same value\n")
+				fmt.Print("yacc bug -- cannot have 2 different Ts with same value\n")
 				fmt.Printf("	%s and %s\n", tokset[i].name, tokset[temp1[j]].name)
 				nerrors++
 			}
@@ -2982,7 +3100,7 @@ var peekrune rune
 func isdigit(c rune) bool { return c >= '0' && c <= '9' }
 
 func isword(c rune) bool {
-	return c >= 0xa0 || (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z')
+	return c >= 0xa0 || c == '_' || (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z')
 }
 
 func mktemp(t string) string { return t }
@@ -3110,12 +3228,13 @@ type $$Lexer interface {
 const $$Flag = -1000
 
 func $$Tokname(c int) string {
-	if c > 0 && c <= len($$Toknames) {
-		if $$Toknames[c-1] != "" {
-			return $$Toknames[c-1]
+	// 4 is TOKSTART above
+	if c >= 4 && c-4 < len($$Toknames) {
+		if $$Toknames[c-4] != "" {
+			return $$Toknames[c-4]
 		}
 	}
-	return fmt.Sprintf("tok-%v", c)
+	return __yyfmt__.Sprintf("tok-%v", c)
 }
 
 func $$Statname(s int) string {
@@ -3124,7 +3243,7 @@ func $$Statname(s int) string {
 			return $$Statenames[s]
 		}
 	}
-	return fmt.Sprintf("state-%v", s)
+	return __yyfmt__.Sprintf("state-%v", s)
 }
 
 func $$lex1(lex $$Lexer, lval *$$SymType) int {
@@ -3157,7 +3276,7 @@ out:
 		c = $$Tok2[1] /* unknown char */
 	}
 	if $$Debug >= 3 {
-		fmt.Printf("lex %U %s\n", uint(char), $$Tokname(c))
+		__yyfmt__.Printf("lex %U %s\n", uint(char), $$Tokname(c))
 	}
 	return c
 }
@@ -3184,7 +3303,7 @@ ret1:
 $$stack:
 	/* put a state and value onto the stack */
 	if $$Debug >= 4 {
-		fmt.Printf("char %v in %v\n", $$Tokname($$char), $$Statname($$state))
+		__yyfmt__.Printf("char %v in %v\n", $$Tokname($$char), $$Statname($$state))
 	}
 
 	$$p++
@@ -3253,8 +3372,8 @@ $$default:
 			$$lex.Error("syntax error")
 			Nerrs++
 			if $$Debug >= 1 {
-				fmt.Printf("%s", $$Statname($$state))
-				fmt.Printf("saw %s\n", $$Tokname($$char))
+				__yyfmt__.Printf("%s", $$Statname($$state))
+				__yyfmt__.Printf("saw %s\n", $$Tokname($$char))
 			}
 			fallthrough
 
@@ -3273,7 +3392,7 @@ $$default:
 
 				/* the current p has no shift on "error", pop stack */
 				if $$Debug >= 2 {
-					fmt.Printf("error recovery pops state %d\n", $$S[$$p].yys)
+					__yyfmt__.Printf("error recovery pops state %d\n", $$S[$$p].yys)
 				}
 				$$p--
 			}
@@ -3282,7 +3401,7 @@ $$default:
 
 		case 3: /* no shift yet; clobber input char */
 			if $$Debug >= 2 {
-				fmt.Printf("error recovery discards %s\n", $$Tokname($$char))
+				__yyfmt__.Printf("error recovery discards %s\n", $$Tokname($$char))
 			}
 			if $$char == $$EofCode {
 				goto ret1
@@ -3294,7 +3413,7 @@ $$default:
 
 	/* reduction by production $$n */
 	if $$Debug >= 2 {
-		fmt.Printf("reduce %v in:\n\t%v\n", $$n, $$Statname($$state))
+		__yyfmt__.Printf("reduce %v in:\n\t%v\n", $$n, $$Statname($$state))
 	}
 
 	$$nt := $$n