Fix parsing of <= and >= in templated lambda

Skipping between matching delimiters is now done at the token level
rather than the character level.

Partly addresses #2630

CHANGES.current

@@ -7,6 +7,12 @@ the issue number to the end of the URL: https://github.com/swig/swig/issues/
 Version 4.2.0 (in progress)
 ===========================
 
+2023-06-07: olly
+	    #2630 Fix parsing of <= and >= in templated lambda.
+
+	    Skipping between matching delimiters is now done at the token level
+	    rather than the character level.
+
 2023-06-02: mmomtchev
 	    [Javascript] #2622 Fix support for %typemap(default) and improve
 	    support for default arguments in general.

Examples/test-suite/cpp20_lambda_template.i

@@ -9,4 +9,8 @@
 %inline %{
 #include <vector>
 auto templated_lambda = []<typename T>(std::vector<T> t){};
+
+// Regression test for parsing bug fixed in SWIG 4.2.0 (#2630)
+auto y = []<bool X = 1>=2> { return 1; };
+auto z = []<bool X = 1<=2> { return 1; };
 %}

Source/Swig/scanner.c

@@ -1554,126 +1554,55 @@ void Scanner_skip_line(Scanner *s) {
  * ----------------------------------------------------------------------------- */
 
 int Scanner_skip_balanced(Scanner *s, int startchar, int endchar) {
-  char c;
+  int old_line = s->line;
-  int num_levels = 1;
+  long position = Tell(s->str);
-  int state = 0;
-  String *locator = 0;
-  Clear(s->text);
-  Setfile(s->text, Getfile(s->str));
-  Setline(s->text, s->line);
 
-  Putc((char)startchar, s->text);
+  int num_levels = 1;
-  while (num_levels > 0) {
+  int starttok, endtok;
-    if ((c = nextchar(s)) == 0) {
+  switch (endchar) {
-      Delete(locator);
+    case '}':
-      return -1;
+      starttok = SWIG_TOKEN_LBRACE;
-    }
+      endtok = SWIG_TOKEN_RBRACE;
-    switch (state) {
-    case 0:
-      if (c == startchar)
-	num_levels++;
-      else if (c == endchar)
-	num_levels--;
-      else if (c == '/')
-	state = 10;
-      else if (c == '\"')
-	state = 20;
-      else if (c == '\'')
-	state = 30;
       break;
-    case 10:
+    case ')':
-      if (c == '/')
+      starttok = SWIG_TOKEN_LPAREN;
-	state = 11;
+      endtok = SWIG_TOKEN_RPAREN;
-      else if (c == '*')
-	state = 12;
-      else if (c == startchar) {
-	state = 0;
-	num_levels++;
-      }
-      else
-	state = 0;
       break;
-    case 11:
+    case ']':
-      if (c == '\n')
+      starttok = SWIG_TOKEN_LBRACKET;
-	state = 0;
+      endtok = SWIG_TOKEN_RBRACKET;
-      else
-	state = 11;
       break;
-    case 12: /* first character inside C comment */
+    case '>':
-      if (c == '*')
+      starttok = SWIG_TOKEN_LESSTHAN;
-	state = 14;
+      endtok = SWIG_TOKEN_GREATERTHAN;
-      else if (c == '@')
-	state = 40;
-      else
-	state = 13;
-      break;
-    case 13:
-      if (c == '*')
-	state = 14;
-      break;
-    case 14: /* possible end of C comment */
-      if (c == '*')
-	state = 14;
-      else if (c == '/')
-	state = 0;
-      else
-	state = 13;
-      break;
-    case 20:
-      if (c == '\"')
-	state = 0;
-      else if (c == '\\')
-	state = 21;
-      break;
-    case 21:
-      state = 20;
-      break;
-    case 30:
-      if (c == '\'')
-	state = 0;
-      else if (c == '\\')
-	state = 31;
-      break;
-    case 31:
-      state = 30;
-      break;
-    /* 40-45 SWIG locator checks - a C comment with contents starting: @SWIG */
-    case 40:
-      state = (c == 'S') ? 41 : (c == '*') ? 14 : 13;
-      break;
-    case 41:
-      state = (c == 'W') ? 42 : (c == '*') ? 14 : 13;
-      break;
-    case 42:
-      state = (c == 'I') ? 43 : (c == '*') ? 14 : 13;
-      break;
-    case 43:
-      state = (c == 'G') ? 44 : (c == '*') ? 14 : 13;
-      if (c == 'G') {
-	Delete(locator);
-	locator = NewString("/*@SWIG");
-      }
-      break;
-    case 44:
-      if (c == '*')
-	state = 45;
-      Putc(c, locator);
-      break;
-    case 45: /* end of SWIG locator in C comment */
-      if (c == '/') {
-	state = 0;
-	Putc(c, locator);
-	Scanner_locator(s, locator);
-      } else {
-	/* malformed locator */
-	state = (c == '*') ? 14 : 13;
-      }
       break;
     default:
-      break;
+      assert(0);
+  }
+
+  while (1) {
+    int tok = Scanner_token(s);
+    if (tok == starttok) {
+      num_levels++;
+    } else if (tok == endtok) {
+      if (--num_levels == 0) break;
+    } else if (tok == SWIG_TOKEN_COMMENT) {
+      char *loc = Char(s->text);
+      if (strncmp(loc, "/*@SWIG", 7) == 0 && loc[Len(s->text)-3] == '@') {
+	Scanner_locator(s, s->text);
+      }
+    } else if (tok == 0) {
+      return -1;
     }
   }
-  Delete(locator);
+
+  Delete(s->text);
+  s->text = NewStringWithSize(Char(s->str) + position - 1,
+			      Tell(s->str) - position + 1);
+  Char(s->text)[0] = startchar;
+  Setfile(s->text, Getfile(s->str));
+  Setline(s->text, old_line);
+
   return 0;
 }
 
@@ -1684,105 +1613,66 @@ int Scanner_skip_balanced(Scanner *s, int startchar, int endchar) {
  * ----------------------------------------------------------------------------- */
 
 String *Scanner_get_raw_text_balanced(Scanner *s, int startchar, int endchar) {
-  String *result = 0;
+  String *result = NULL;
-  char c;
   int old_line = s->line;
   String *old_text = Copy(s->text);
   long position = Tell(s->str);
 
   int num_levels = 1;
-  int state = 0;
+  int starttok, endtok;
-  Clear(s->text);
+  switch (endchar) {
-  Setfile(s->text, Getfile(s->str));
+    case '}':
-  Setline(s->text, s->line);
+      starttok = SWIG_TOKEN_LBRACE;
-  Putc((char)startchar, s->text);
+      endtok = SWIG_TOKEN_RBRACE;
-  while (num_levels > 0) {
-    if ((c = nextchar(s)) == 0) {
-      Clear(s->text);
-      Append(s->text, old_text);
-      Delete(old_text);
-      s->line = old_line;
-      return 0;
-    }
-    switch (state) {
-    case 0:
-      if (c == startchar)
-	num_levels++;
-      else if (c == endchar)
-	num_levels--;
-      else if (c == '/')
-	state = 10;
-      else if (c == '\"')
-	state = 20;
-      else if (c == '\'')
-	state = 30;
       break;
-    case 10:
+    case ')':
-      if (c == '/')
+      starttok = SWIG_TOKEN_LPAREN;
-	state = 11;
+      endtok = SWIG_TOKEN_RPAREN;
-      else if (c == '*')
-	state = 12;
-      else if (c == startchar) {
-	state = 0;
-	num_levels++;
-      }
-      else
-	state = 0;
       break;
-    case 11:
+    case ']':
-      if (c == '\n')
+      starttok = SWIG_TOKEN_LBRACKET;
-	state = 0;
+      endtok = SWIG_TOKEN_RBRACKET;
-      else
-	state = 11;
       break;
-    case 12: /* first character inside C comment */
+    case '>':
-      if (c == '*')
+      starttok = SWIG_TOKEN_LESSTHAN;
-	state = 14;
+      endtok = SWIG_TOKEN_GREATERTHAN;
-      else
-	state = 13;
-      break;
-    case 13:
-      if (c == '*')
-	state = 14;
-      break;
-    case 14: /* possible end of C comment */
-      if (c == '*')
-	state = 14;
-      else if (c == '/')
-	state = 0;
-      else
-	state = 13;
-      break;
-    case 20:
-      if (c == '\"')
-	state = 0;
-      else if (c == '\\')
-	state = 21;
-      break;
-    case 21:
-      state = 20;
-      break;
-    case 30:
-      if (c == '\'')
-	state = 0;
-      else if (c == '\\')
-	state = 31;
-      break;
-    case 31:
-      state = 30;
       break;
     default:
+      assert(0);
+  }
+
+  while (1) {
+    int tok = Scanner_token(s);
+    if (tok == starttok) {
+      num_levels++;
+    } else if (tok == endtok) {
+      if (--num_levels == 0) {
+	result = NewStringWithSize(Char(s->str) + position - 1,
+				   Tell(s->str) - position + 1);
+	Char(result)[0] = startchar;
+	Setfile(result, Getfile(s->str));
+	Setline(result, old_line);
+	break;
+      }
+    } else if (tok == SWIG_TOKEN_COMMENT) {
+      char *loc = Char(s->text);
+      if (strncmp(loc, "/*@SWIG", 7) == 0 && loc[Len(s->text)-3] == '@') {
+	Scanner_locator(s, s->text);
+      }
+    } else if (tok == 0) {
       break;
     }
   }
+
+  /* Reset the scanner state. */
   Seek(s->str, position, SEEK_SET);
-  result = Copy(s->text);
+  Delete(s->text);
-  Clear(s->text);
+  s->text = old_text;
-  Append(s->text, old_text);
-  Delete(old_text);
   s->line = old_line;
+
   return result;
 }
+
 /* -----------------------------------------------------------------------------
  * Scanner_isoperator()
  *