Do not allow unrecognized escapes in regex.

* parser.l (REGOP): New regex alias for matching all regex
special characters.
(grammar): Several rules for regex special characters merged
together. New rule introduced to match a special character
after a backslash, making it literal. The old rule which makes
literal any character after a backslash now throws an error,
unless version 105 comaptibility is selected.

* txr.1: Documented this behavior change.

3 files changed, 46 insertions, 16 deletions

diff --git a/ChangeLog b/ChangeLog
index 12d4babb..60472e7c 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,19 @@
 2015-04-19  Kaz Kylheku  <kaz@kylheku.com>
 
+	Do not allow unrecognized escapes in regex.
+
+	* parser.l (REGOP): New regex alias for matching all regex
+	special characters.
+	(grammar): Several rules for regex special characters merged
+	together. New rule introduced to match a special character
+	after a backslash, making it literal. The old rule which makes
+	literal any character after a backslash now throws an error,
+	unless version 105 comaptibility is selected.
+
+	* txr.1: Documented this behavior change.
+
+2015-04-19  Kaz Kylheku  <kaz@kylheku.com>
+
 	Improvement in error reporting.
 
 	This fix prevents certain situations when an inappropriate form
diff --git a/parser.l b/parser.l
index 70c21d79..9036278f 100644
--- a/parser.l
+++ b/parser.l
@@ -49,6 +49,7 @@
 #include "hash.h"
 #include "parser.h"
 #include "eval.h"
+#include "txr.h"
 #include "y.tab.h"
 
 #define YY_INPUT(buf, result, max_size)           \
@@ -198,6 +199,8 @@ WS      [\t ]*
 HEX     [0-9A-Fa-f]
 OCT     [0-7]
 
+REGOP   [/()|.*?+~&%\[\]\-]
+
 ASC     [\x00-\x7f]
 ASCN    [\x00-\t\v-\x7f]
 U       [\x80-\xbf]
@@ -741,24 +744,24 @@ UONLY   {U2}{U}|{U3}{U}{U}|{U4}{U}{U}{U}
   return ERRTOK;
 }
 
-<REGEX>[.*?+~&%] {
+<REGEX>{REGOP} {
   yylval->chr = yytext[0];
   return yytext[0];
 }
 
-<REGEX>[\[\]\-] {
-  yylval->chr = yytext[0];
-  return yytext[0];
-}
-
-<REGEX>[()|] {
-  yylval->chr = yytext[0];
-  return yytext[0];
+<REGEX>[\\]{REGOP} {
+  yylval->chr = yytext[1];
+  return REGCHAR;
 }
 
 <REGEX>[\\]. {
-  yylval->chr = yytext[1];
-  return REGCHAR;
+  if (opt_compat && opt_compat <= 105) {
+    yylval->chr = yytext[1];
+    return REGCHAR;
+  }
+    
+  yyerrprepf(yyg, lit("unrecognized escape in regex"), nao);
+  return ERRTOK;
 }
 
 <REGEX>[\\] {
diff --git a/txr.1 b/txr.1
index 81ccb76c..6884e0ed 100644
--- a/txr.1
+++ b/txr.1
@@ -2052,14 +2052,13 @@ and the set matched by
 This operator is called intersection, logical and, or conjunction.
 
 .PP
-Any escaped character which does not fall into the above escaping conventions,
-or any unescaped character which is not a regular expression operator, denotes
-one-position match of that character itself.
+Any character which is not a regular expression operator, a backslash escape,
+or the slash delimiter, denotes one-position match of that character itself.
 
 Any of the special characters, including the delimiting
 .codn / ,
-can be escaped with a backslash to suppress its meaning and denote the
-character itself.
+and the backslash, can be escaped with a backslash to suppress its
+meaning and denote the character itself.
 
 Furthermore, all of the same escapes as are described in the section Special
 Characters in Text above are supported - the difference is that in regular
@@ -2072,6 +2071,12 @@ rather than
 Octal and hex character escapes can be optionally
 terminated by a semicolon, which is useful if the following characters are
 octal or hex digits not intended to be part of the escape.
+
+Only the above escapes are supported. Unlike in some other regular expression
+implementations, if a backlash appears before a character which isn't a regex
+special character or one of the supported escape sequences, it is an error.
+This wasn't true of historic versions of \*(TX. See the Compatibility section.
+
 .IP "Precedence table, highest to lowest:"
 .TS
 tab(!);
@@ -27924,6 +27929,14 @@ function automatically marks a stream open on a TTY devices as a real-time strea
 .code isatty
 function).
 
+Also allows unrecognized backslash escape sequences in regular
+expression syntax to simply denote the escaped character literally,
+as was historically the case prior to \*(TX 106, so that
+.code \ez
+for instance denotes
+.codn z .
+As of \*(TX 106, these are diagnosed as errors.
+
 .IP 102
 Up to \*(TX 102, the
 .code get-string