Make scanning of C and C++ source even more consistent

Use the same code for scanning all input that gets copied to the output.
westes · Dec 3, 2018 · a6a3160 · a6a3160
1 parent 20c91b4
commit a6a3160
Showing 1 changed file with 27 additions and 20 deletions.
diff --git a/src/scan.l b/src/scan.l
@@ -43,13 +43,24 @@ extern bool tablesverify, tablesext;
 extern int trlcontxt; /* Set in  parse.y for each rule. */
 extern const char *escaped_qstart, *escaped_qend;
 
+
+static bool write_to_buf;
 #define M4QSTART "[""["
 #define M4QEND "]""]"
 
 #define ESCAPED_QSTART "[" M4QEND M4QSTART "[" M4QEND M4QSTART
 #define ESCAPED_QEND M4QEND "]" M4QSTART M4QEND "]" M4QSTART
 
-#define ACTION_ECHO do add_action(yytext); while(0)
+#define ACTION_ECHO \
+  do \
+    if (strlen(yytext) != yyleng) \
+      synerr(_( "NUL character in input file")); \
+    else if (write_to_buf) \
+      buf_strnappend(&top_buf, yytext, yyleng); \
+    else \
+      add_action(yytext); \
+  while(0)
+
 #define ACTION_IFDEF(def, should_define) \
 	do if (should_define) action_define(def, 1); while (0)
 
@@ -175,6 +186,7 @@ M4QEND      "]""]"
                 ++linenum;
                 buf_linedir( &top_buf, infilename?infilename:"<stdin>", linenum);
                 brace_depth = 1;
+                write_to_buf = true;
                 yy_push_state(CODEBLOCK_MATCH_BRACE);
             }
 
@@ -226,19 +238,13 @@ M4QEND      "]""]"
 <COMMENT,CODE_COMMENT>{ /* */
         [^\[\]\*\n]*  ACTION_ECHO;
         .           ACTION_ECHO;
-
-	{NL}	    ++linenum; ACTION_ECHO;
-}
-<COMMENT>{
-	"*/"	    add_action("*/]""]"); yy_pop_state();
-}
-<CODE_COMMENT>{
         "*"{INVALID_ESCAPED_NEWLINE}+"/" {
           synerr(_("Escaped newline in comment end sequence\n"));
 	}
-
-        "*/"        ACTION_ECHO; yy_pop_state();
+	{NL}	    ++linenum; ACTION_ECHO;
 }
+<COMMENT>"*/"           add_action("*/]""]"); yy_pop_state();
+<CODE_COMMENT>"*/"      ACTION_ECHO; yy_pop_state();
 
 <COMMENT_DISCARD>{
         /* This is the same as COMMENT, but is discarded rather than output. */
@@ -291,6 +297,7 @@ M4QEND      "]""]"
 <CODEBLOCK_MATCH_BRACE>{
     "}"     {
                 if( --brace_depth == 0){
+                    write_to_buf = false;
                     /* TODO: Matched. */
                     yy_pop_state();
                 }else
@@ -309,9 +316,6 @@ M4QEND      "]""]"
 
     {M4QSTART}  buf_strnappend(&top_buf, escaped_qstart, (int) strlen(escaped_qstart));
     {M4QEND}    buf_strnappend(&top_buf, escaped_qend, (int) strlen(escaped_qend));
-    ([^{}\r\n\[\]]+)|[^{}\r\n]  {
-       buf_strnappend(&top_buf, yytext, yyleng);
-    }
 
     <<EOF>>     {
                 linenum = brace_start_line;
@@ -909,13 +913,6 @@ nmstr[yyleng - 2 - end_is_ws] = '\0';  /* chop trailing brace */
 			}
 }
 <CODEBLOCK,ACTION,PERCENT_BRACE_ACTION>{
-	"/"{INVALID_ESCAPED_NEWLINE}+"*" {
-          synerr(_("Invalid escaped newline in comment-open-sequence"));
-	  ACTION_ECHO;
-	  yy_push_state(CODE_COMMENT);
-        }
-	"/*"		ACTION_ECHO; yy_push_state(CODE_COMMENT);
-
 	"reject" {
             ACTION_ECHO;
             CHECK_REJECT(yytext);
@@ -924,6 +921,16 @@ nmstr[yyleng - 2 - end_is_ws] = '\0';  /* chop trailing brace */
             ACTION_ECHO;
             CHECK_YYMORE(yytext);
         }
+}
+
+<CODEBLOCK,ACTION,PERCENT_BRACE_ACTION,CODEBLOCK_MATCH_BRACE>{
+	"/"{INVALID_ESCAPED_NEWLINE}+"*" {
+          synerr(_("Invalid escaped newline in comment-open-sequence"));
+	  ACTION_ECHO;
+	  yy_push_state(CODE_COMMENT);
+        }
+	"/*"		ACTION_ECHO; yy_push_state(CODE_COMMENT);
+
 	[^[:alpha:]_{}""''/\n\[\]]+	ACTION_ECHO;
         {NAME}		ACTION_ECHO;
         "'"([^''\\\n]|\\.)"'" ACTION_ECHO; /* character constant */