Convert the special "EBCDIC on an ASCII system" test.

src/pcre2_compile.c

@@ -304,10 +304,17 @@ static const short int escapes[] = {
 #else
 
 /* This is the "abnormal" table for EBCDIC systems without UTF-8 support.
-It runs from 'a' to '9'. */
+It runs from 'a' to '9'. For some minimal testing of EBCDIC features, the code 
+is sometimes compiled on an ASCII system. In this case, we must not use CHAR_a 
+because it is defined as 'a', which of course picks up the ASCII value. */
 
+#if 'a' == 0x81                 /* Check for a real EBCDIC environment */
 #define ESCAPES_FIRST  CHAR_a
 #define ESCAPES_LAST   CHAR_9
+#else                           /* Testing in an ASCII environment */
+#define ESCAPES_FIRST  ((unsigned char)'\x81')   /* EBCDIC 'a' */
+#define ESCAPES_LAST   ((unsigned char)'\xf9')   /* EBCDIC '9' */
+#endif
 
 static const short int escapes[] = {
 /*  80 */            7, -ESC_b,       0, -ESC_d, ESC_e,  ESC_f,      0,
@@ -328,7 +335,7 @@ static const short int escapes[] = {
 /*  F8 */     0,     0
 };
 
-#endif
+#endif   /* EBCDIC */
 
 
 /* Table of special "verbs" like (*PRUNE). This is a short table, so it is

testdata/testinputEBC

@@ -0,0 +1,121 @@
+# This is a specialized test for checking, when PCRE2 is compiled with the
+# EBCDIC option but in an ASCII environment, that newline and white space
+# functionality is working. It catches cases where explicit values such as 0x0a
+# have been used instead of names like CHAR_LF. Needless to say, it is not a
+# genuine EBCDIC test! In patterns, alphabetic characters that follow a
+# backslash must be in EBCDIC code. In data, NL, NEL, LF, ESC, and DEL must be
+# in EBCDIC, but can of course be specified as escapes.
+
+# Test default newline and variations
+
+/^A/m
+    ABC
+    12\x15ABC
+
+/^A/m,newline=any
+    12\x15ABC
+    12\x0dABC
+    12\x0d\x15ABC
+    12\x25ABC
+
+/^A/m,newline=anycrlf
+    12\x15ABC
+    12\x0dABC
+    12\x0d\x15ABC
+    ** Fail
+    12\x25ABC
+
+# Test \h
+
+/^A\ˆ/
+    A B
+
+# Test \H
+
+/^A\È/
+    AB
+    ** Fail
+    A B
+
+# Test \R
+
+/^A\Ù/
+    A\x15B
+    A\x0dB
+    A\x25B
+    A\x0bB
+    A\x0cB
+    ** Fail
+    A B
+
+# Test \v
+
+/^A\¥/
+    A\x15B
+    A\x0dB
+    A\x25B
+    A\x0bB
+    A\x0cB
+    ** Fail
+    A B
+
+# Test \V
+
+/^A\å/
+    A B
+    ** Fail
+    A\x15B
+    A\x0dB
+    A\x25B
+    A\x0bB
+    A\x0cB
+    
+# For repeated items, use an atomic group so that the output is the same
+# for DFA matching (otherwise it may show multiple matches).
+
+# Test \h+
+
+/^A(?>\ˆ+)/
+    A B
+
+# Test \H+
+
+/^A(?>\È+)/
+    AB
+    ** Fail
+    A B
+
+# Test \R+
+
+/^A(?>\Ù+)/
+    A\x15B
+    A\x0dB
+    A\x25B
+    A\x0bB
+    A\x0cB
+    ** Fail
+    A B
+
+# Test \v+
+
+/^A(?>\¥+)/
+    A\x15B
+    A\x0dB
+    A\x25B
+    A\x0bB
+    A\x0cB
+    ** Fail
+    A B
+
+# Test \V+
+
+/^A(?>\å+)/
+    A B
+    ** Fail
+    A\x15B
+    A\x0dB
+    A\x25B
+    A\x0bB
+    A\x0cB
+
+# End

testdata/testoutputEBC

@@ -0,0 +1,182 @@
+# This is a specialized test for checking, when PCRE2 is compiled with the
+# EBCDIC option but in an ASCII environment, that newline and white space
+# functionality is working. It catches cases where explicit values such as 0x0a
+# have been used instead of names like CHAR_LF. Needless to say, it is not a
+# genuine EBCDIC test! In patterns, alphabetic characters that follow a
+# backslash must be in EBCDIC code. In data, NL, NEL, LF, ESC, and DEL must be
+# in EBCDIC, but can of course be specified as escapes.
+
+# Test default newline and variations
+
+/^A/m
+    ABC
+ 0: A
+    12\x15ABC
+ 0: A
+
+/^A/m,newline=any
+    12\x15ABC
+ 0: A
+    12\x0dABC
+ 0: A
+    12\x0d\x15ABC
+ 0: A
+    12\x25ABC
+ 0: A
+
+/^A/m,newline=anycrlf
+    12\x15ABC
+ 0: A
+    12\x0dABC
+ 0: A
+    12\x0d\x15ABC
+ 0: A
+    ** Fail
+No match
+    12\x25ABC
+No match
+
+# Test \h
+
+/^A\ˆ/
+    A B
+ 0: A\x20
+
+# Test \H
+
+/^A\È/
+    AB
+ 0: AB
+    ** Fail
+No match
+    A B
+No match
+
+# Test \R
+
+/^A\Ù/
+    A\x15B
+ 0: A\x15
+    A\x0dB
+ 0: A\x0d
+    A\x25B
+ 0: A\x25
+    A\x0bB
+ 0: A\x0b
+    A\x0cB
+ 0: A\x0c
+    ** Fail
+No match
+    A B
+No match
+
+# Test \v
+
+/^A\¥/
+    A\x15B
+ 0: A\x15
+    A\x0dB
+ 0: A\x0d
+    A\x25B
+ 0: A\x25
+    A\x0bB
+ 0: A\x0b
+    A\x0cB
+ 0: A\x0c
+    ** Fail
+No match
+    A B
+No match
+
+# Test \V
+
+/^A\å/
+    A B
+ 0: A\x20
+    ** Fail
+No match
+    A\x15B
+No match
+    A\x0dB
+No match
+    A\x25B
+No match
+    A\x0bB
+No match
+    A\x0cB
+No match
+    
+# For repeated items, use an atomic group so that the output is the same
+# for DFA matching (otherwise it may show multiple matches).
+
+# Test \h+
+
+/^A(?>\ˆ+)/
+    A B
+ 0: A\x20
+
+# Test \H+
+
+/^A(?>\È+)/
+    AB
+ 0: AB
+    ** Fail
+No match
+    A B
+No match
+
+# Test \R+
+
+/^A(?>\Ù+)/
+    A\x15B
+ 0: A\x15
+    A\x0dB
+ 0: A\x0d
+    A\x25B
+ 0: A\x25
+    A\x0bB
+ 0: A\x0b
+    A\x0cB
+ 0: A\x0c
+    ** Fail
+No match
+    A B
+No match
+
+# Test \v+
+
+/^A(?>\¥+)/
+    A\x15B
+ 0: A\x15
+    A\x0dB
+ 0: A\x0d
+    A\x25B
+ 0: A\x25
+    A\x0bB
+ 0: A\x0b
+    A\x0cB
+ 0: A\x0c
+    ** Fail
+No match
+    A B
+No match
+
+# Test \V+
+
+/^A(?>\å+)/
+    A B
+ 0: A\x20B
+    ** Fail
+No match
+    A\x15B
+No match
+    A\x0dB
+No match
+    A\x25B
+No match
+    A\x0bB
+No match
+    A\x0cB
+No match
+
+# End