Merge pull request #210 from syntheticpp/colon-escape

add colon escaping 

6 files changed, 115 insertions, 27 deletions

diff --git a/src/lexer.cc b/src/lexer.cc
index 5bc467b..9e4392c 100644
--- a/src/lexer.cc
+++ b/src/lexer.cc
@@ -90,6 +90,26 @@ const char* Lexer::TokenName(Token t) {
   return NULL;  // not reached
 }
 
+const char* Lexer::TokenErrorHint(Token t) {
+  switch (t) {
+  case ERROR:    return "";
+  case BUILD:    return "";
+  case COLON:    return " ($ also escapes ':')";
+  case DEFAULT:  return "";
+  case EQUALS:   return "";
+  case IDENT:    return "";
+  case INCLUDE:  return "";
+  case INDENT:   return "";
+  case NEWLINE:  return "";
+  case PIPE2:    return "";
+  case PIPE:     return "";
+  case RULE:     return "";
+  case SUBNINJA: return "";
+  case TEOF:     return "";
+  }
+  return "";
+}
+
 void Lexer::UnreadToken() {
   ofs_ = last_token_;
 }
@@ -606,7 +626,7 @@ bool Lexer::ReadEvalString(EvalString* eval, bool path, string* err) {
 	}
 	++p;
 	yych = *p;
-	goto yy118;
+	goto yy120;
 yy91:
 	{
       eval->AddText(StringPiece(start, p - start));
@@ -629,7 +649,7 @@ yy94:
 	++p;
 	if ((yych = *p) <= '/') {
 		if (yych <= ' ') {
-			if (yych == '\n') goto yy107;
+			if (yych == '\n') goto yy109;
 			if (yych <= 0x1F) goto yy98;
 			goto yy100;
 		} else {
@@ -643,17 +663,21 @@ yy94:
 		}
 	} else {
 		if (yych <= '^') {
-			if (yych <= '9') goto yy104;
-			if (yych <= '@') goto yy98;
-			if (yych <= 'Z') goto yy104;
-			goto yy98;
+			if (yych <= ':') {
+				if (yych <= '9') goto yy104;
+				goto yy106;
+			} else {
+				if (yych <= '@') goto yy98;
+				if (yych <= 'Z') goto yy104;
+				goto yy98;
+			}
 		} else {
 			if (yych <= '`') {
 				if (yych <= '_') goto yy104;
 				goto yy98;
 			} else {
 				if (yych <= 'z') goto yy104;
-				if (yych <= '{') goto yy106;
+				if (yych <= '{') goto yy108;
 				goto yy98;
 			}
 		}
@@ -690,56 +714,62 @@ yy102:
 yy104:
 	++p;
 	yych = *p;
-	goto yy116;
+	goto yy118;
 yy105:
 	{
       eval->AddSpecial(StringPiece(start + 1, p - start - 1));
       continue;
     }
 yy106:
+	++p;
+	{
+      eval->AddText(StringPiece(":", 1));
+      continue;
+    }
+yy108:
 	yych = *(q = ++p);
 	if (yybm[0+yych] & 32) {
-		goto yy110;
+		goto yy112;
 	}
 	goto yy99;
-yy107:
+yy109:
 	++p;
 	yych = *p;
 	if (yybm[0+yych] & 16) {
-		goto yy107;
+		goto yy109;
 	}
 	{
       continue;
     }
-yy110:
+yy112:
 	++p;
 	yych = *p;
 	if (yybm[0+yych] & 32) {
-		goto yy110;
+		goto yy112;
 	}
-	if (yych == '}') goto yy113;
+	if (yych == '}') goto yy115;
 	p = q;
 	goto yy99;
-yy113:
+yy115:
 	++p;
 	{
       eval->AddSpecial(StringPiece(start + 2, p - start - 3));
       continue;
     }
-yy115:
+yy117:
 	++p;
 	yych = *p;
-yy116:
+yy118:
 	if (yybm[0+yych] & 64) {
-		goto yy115;
+		goto yy117;
 	}
 	goto yy105;
-yy117:
+yy119:
 	++p;
 	yych = *p;
-yy118:
+yy120:
 	if (yybm[0+yych] & 128) {
-		goto yy117;
+		goto yy119;
 	}
 	goto yy91;
 }
diff --git a/src/lexer.h b/src/lexer.h
index c872b98..75c1b2f 100644
--- a/src/lexer.h
+++ b/src/lexer.h
@@ -46,6 +46,10 @@ struct Lexer {
   /// Return a human-readable form of a token, used in error messages.
   static const char* TokenName(Token t);
 
+
+  /// Return a human-readable token hint, used in error messages.
+  static const char* TokenErrorHint(Token t);
+
   /// Start parsing some input.
   void Start(StringPiece filename, StringPiece input);
 
diff --git a/src/lexer.in.cc b/src/lexer.in.cc
index 6c00dbd..28a5bdf 100644
--- a/src/lexer.in.cc
+++ b/src/lexer.in.cc
@@ -89,6 +89,26 @@ const char* Lexer::TokenName(Token t) {
   return NULL;  // not reached
 }
 
+const char* Lexer::TokenErrorHint(Token t) {
+  switch (t) {
+  case ERROR:    return "";
+  case BUILD:    return "";
+  case COLON:    return " ($ also escapes ':')";
+  case DEFAULT:  return "";
+  case EQUALS:   return "";
+  case IDENT:    return "";
+  case INCLUDE:  return "";
+  case INDENT:   return "";
+  case NEWLINE:  return "";
+  case PIPE2:    return "";
+  case PIPE:     return "";
+  case RULE:     return "";
+  case SUBNINJA: return "";
+  case TEOF:     return "";
+  }
+  return "";
+}
+
 void Lexer::UnreadToken() {
   ofs_ = last_token_;
 }
@@ -214,6 +234,10 @@ bool Lexer::ReadEvalString(EvalString* eval, bool path, string* err) {
       eval->AddSpecial(StringPiece(start + 1, p - start - 1));
       continue;
     }
+    "$:" {
+      eval->AddText(StringPiece(":", 1));
+      continue;
+    }
     "$". {
       last_token_ = start;
       return Error("bad $-escape (literal $ must be written as $$)", err);
diff --git a/src/lexer_test.cc b/src/lexer_test.cc
index 1c5894e..5795e5e 100644
--- a/src/lexer_test.cc
+++ b/src/lexer_test.cc
@@ -29,12 +29,12 @@ TEST(Lexer, ReadVarValue) {
 }
 
 TEST(Lexer, ReadEvalStringEscapes) {
-  Lexer lexer("$ $$ab $\ncde\n");
+  Lexer lexer("$ $$ab c$: $\ncde\n");
   EvalString eval;
   string err;
   EXPECT_TRUE(lexer.ReadVarValue(&eval, &err));
   EXPECT_EQ("", err);
-  EXPECT_EQ("[ $ab cde]",
+  EXPECT_EQ("[ $ab c: cde]",
             eval.Serialize());
 }
 
diff --git a/src/parsers.cc b/src/parsers.cc
index c1badd8..5d347b2 100644
--- a/src/parsers.cc
+++ b/src/parsers.cc
@@ -313,6 +313,7 @@ bool ManifestParser::ExpectToken(Lexer::Token expected, string* err) {
   if (token != expected) {
     string message = string("expected ") + Lexer::TokenName(expected);
     message += string(", got ") + Lexer::TokenName(token);
+    message += Lexer::TokenErrorHint(expected);
     return lexer_.Error(message, err);
   }
   return true;
diff --git a/src/parsers_test.cc b/src/parsers_test.cc
index f3ac517..2a30a83 100644
--- a/src/parsers_test.cc
+++ b/src/parsers_test.cc
@@ -377,14 +377,30 @@ TEST_F(ParserTest, Errors) {
     ManifestParser parser(&state, NULL);
     string err;
     EXPECT_FALSE(parser.ParseTest("rule cat\n"
-                                  "  command = cat\nbuild $: cat foo\n",
+                                  "  command = cat\n"
+                                  "build $.: cat foo\n",
                                   &err));
     EXPECT_EQ("input:3: bad $-escape (literal $ must be written as $$)\n"
-              "build $: cat foo\n"
+              "build $.: cat foo\n"
               "      ^ near here"
               , err);
   }
 
+
+  {
+    State state;
+    ManifestParser parser(&state, NULL);
+    string err;
+    EXPECT_FALSE(parser.ParseTest("rule cat\n"
+                                  "  command = cat\n"
+                                  "build $: cat foo\n",
+                                  &err));
+    EXPECT_EQ("input:3: expected ':', got newline ($ also escapes ':')\n"
+              "build $: cat foo\n"
+              "                ^ near here"
+              , err);
+  }
+
   {
     State state;
     ManifestParser parser(&state, NULL);
@@ -413,10 +429,10 @@ TEST_F(ParserTest, Errors) {
     ManifestParser parser(&state, NULL);
     string err;
     EXPECT_FALSE(parser.ParseTest("rule cc\n  command = foo\n"
-                                  "build $: cc bar.cc\n",
+                                  "build $.: cc bar.cc\n",
                                   &err));
     EXPECT_EQ("input:3: bad $-escape (literal $ must be written as $$)\n"
-              "build $: cc bar.cc\n"
+              "build $.: cc bar.cc\n"
               "      ^ near here"
               , err);
   }
@@ -425,6 +441,19 @@ TEST_F(ParserTest, Errors) {
     State state;
     ManifestParser parser(&state, NULL);
     string err;
+    EXPECT_FALSE(parser.ParseTest("rule cc\n  command = foo\n"
+                                  "build $: cc bar.cc\n",
+                                  &err));
+    EXPECT_EQ("input:3: expected ':', got newline ($ also escapes ':')\n"
+              "build $: cc bar.cc\n"
+              "                  ^ near here"
+              , err);
+  }
+
+  {
+    State state;
+    ManifestParser parser(&state, NULL);
+    string err;
     EXPECT_FALSE(parser.ParseTest("default\n",
                                   &err));
     EXPECT_EQ("input:1: expected target name\n"