9 files changed, 92 insertions(+), 30 deletions(-)

M cppp.1
M cppp.c
M ppproc.c
M ppproc.h
A tests/basic.c
M tests/good.c
M tests/phases.c
A tests/qqs.c
M tests/testall

@@ 52,6 52,11 @@ Treat
 .I SYMBOL
 as undefined during partial preprocessing.
 .TP
+.B \-t, \--trigraphs
+Enable trigraph handling. By default
+.B cppp
+ignores trigraph sequences in the input files.
+.TP
 .B \--help
 Display help and exit.
 .TP

@@ 23,6 23,7 @@ static char const *const yowzitch1 =
 static char const *const yowzitch2 =
     "      -D SYMBOL[=NUMBER]  Preprocess SYMBOL as defined [to NUMBER].\n"
     "      -U SYMBOL           Preprocess SYMBOL as undefined.\n"
+    "      -t, --trigraphs     Enable trigraph handling.\n"
     "      --help              Display this help and exit.\n"
     "      --version           Display version information and exit.\n\n";
 static char const *const yowzitch3 =


@@ 33,7 34,7 @@ static char const *const yowzitch3 =
 /* Version identifier.
  */
 static char const *const vourzhon =
-    "cppp: version 2.8\n"
+    "cppp: version 2.9\n"
     "Copyright (C) 2011-2022 by Brian Raiter <breadbox@muppetlabs.com>\n"
     "License GPLv2+: GNU GPL version 2 or later.\n"
     "This is free software; you are free to change and redistribute it.\n"


@@ 131,6 132,8 @@ static int readcmdline(int argc, char *argv[], symset *defs, symset *undefs)
             else if (removesymbolfromset(undefs, arg))
                 warn("undefining already-undefined symbol %s", arg);
             addsymboltoset(undefs, arg, 0L);
+        } else if (!strcmp(argv[i], "-t") || !strcmp(argv[i], "--trigraphs")) {
+            enabletrigraphs(TRUE);
         } else {
             fail("invalid option: %s", argv[i]);
         }

@@ 48,6 48,10 @@ struct ppproc {
     int         stack[STACK_SIZE];      /* state flags for each level */
 };
 
+/* This global flag controls trigraph handling.
+ */
+static int trigraphsenabled = FALSE;
+
 /* Allocates a partial preprocessor object.
  */
 ppproc *initppproc(symset const *defs, symset const *undefs)


@@ 71,6 75,13 @@ void freeppproc(ppproc *ppp)
     deallocate(ppp);
 }
 
+/* Enable and disable trigraph handling.
+ */
+void enabletrigraphs(int flag)
+{
+    trigraphsenabled = flag;
+}
+
 /* Set the state appropriate for the beginning of a file.
  */
 static void beginfile(ppproc *ppp)


@@ 457,7 468,7 @@ static int readline(ppproc *ppp, FILE *infile)
     erasemstr(ppp->line);
     while (ch != EOF) {
         appendmstr(ppp->line, ch);
-        if (back2 == '?' && back1 == '?') {
+        if (trigraphsenabled && back2 == '?' && back1 == '?') {
             switch (ch) {
               case '=':         replacement = '#';      break;
               case '(':         replacement = '[';      break;

@@ 23,6 23,11 @@ extern ppproc *initppproc(symset const *defs, symset const *undefs);
  */
 extern void freeppproc(ppproc *ppp);
 
+/* Enable and disable trigraph handling in the preprocessor. Trigraph
+ * support is disabled by default.
+ */
+extern void enabletrigraphs(int flag);
+
 /* Partially preprocesses infile's contents to outfile.
  */
 extern void partialpreprocess(ppproc *ppp, FILE *infile, FILE *outfile);

@@ 0,0 1,7 @@
+#ifdef foo
+-Dfoo
+#else
+#if !defined(foo)
+-Ufoo
+#endif
+#endif

@@ 1,11 1,3 @@
-#ifdef foo
--Dfoo
-#else
-#if !defined(foo)
--Ufoo
-#endif
-#endif
-$
 /*1*/#/*2*/ifdef/*3*/foo\
 /*4*/
 -Dfoo

@@ 16,9 16,9 @@ l\
 s\
 e  /* backslash-newline pair within digraph */
 -Ufoo
-??=??/
-e??/
-n??/
-d??/
-i??/
-f  /* trigraphs */
+#\
+e\
+n\
+d\
+i\
+f  /* backslash-newline pair with MS-DOS newline sequence */

@@ 0,0 1,13 @@
+??=if defined foo /* trigraph for # */
+-Dfoo
+??=elif ??-foo == ??-0 /* more trigraphs */
+-Ufoo
+??=??/
+else /* backslash-newline pair with trigraph */
+failed -Ufoo test
+??=??/
+e??/
+n??/
+d??/
+i??/
+f /* backslash-newline pair with trigraph and MS-DOS newline */

@@ 7,28 7,26 @@ PROG=${1:-./cppp}
 fail()
 {
   echo "Unit test failure:" "$@" >&2
-  exit 1
+#  exit 1
 }
 
-# Run through an input file that responds to the preprocessor symbol
-# "foo", and verify that it behaves as expected.
+# Run through a single input file that responds to the preprocessor
+# symbol "foo", and verify that it behaves as expected.
 #
 goodparsingtest()
 {
   infile=$1
   "$PROG" "$infile" | cmp -s "$infile" || fail "altered $infile with no flags."
-  IFS='$' read -a inputs -d@ <$infile
-  for input in "${inputs[@]}" ; do
-    for flag in -Dfoo -Ufoo ; do
-      out=$(echo "$input" | "$PROG" $flag 2>&1)
-      test $? == 0 || fail "non-zero exit code for $infile with flag $flag."
-      test $out == $flag || fail "expected: $flag, got output: \"$out\""
-      test $out == $flag || fail "bad output for $infile with flag $flag."
-    done
+  for flag in -Dfoo -Ufoo ; do
+    out=$("$PROG" $flag "$infile" 2>&1)
+    test $? == 0 || fail "non-zero exit code for $infile with flag $flag."
+    test "$out" == $flag || fail "expected: $flag, got output: \"$out\""
+    test "$out" == $flag || fail "for $infile with flag $flag."
   done
 }
 
-# Run an invalid input file and verify that it produces an error message.
+# Run through an input file with multiple sections and verify that
+# each section produces an error message.
 #
 badparsingtest()
 {


@@ 41,6 39,25 @@ badparsingtest()
   done
 }
 
+# Run an input file containing trigraphs and verify that trigraphs are
+# only detected when enabled.
+#
+trigraphtest()
+{
+  infile=$1
+  "$PROG" "$infile" | cmp -s "$infile" || fail "altered $infile with no flags."
+  "$PROG" -Dfoo "$infile" | cmp -s "$infile" ||
+      fail "altered $infile with trigraphs disabled."
+  for tflag in -t --trigraphs ; do
+    for dflag in -Dfoo -Ufoo ; do
+      out=$("$PROG" $tflag $dflag "$infile" 2>&1)
+      test $? == 0 || fail "error for $infile with flags \"$tflag $dflag\"."
+      test "$out" == $dflag || fail "expected: $dflag, got output: \"$out\""
+      test "$out" == $dflag || fail "for $infile with flags \"$tflag $dflag\"."
+    done
+  done
+}
+
 # Run an input file through every combination of three preprocessor
 # symbols, with each one be defined, undefined, and unspecified.
 #


@@ 61,6 78,7 @@ fulltest()
     test $? == 0 || fail "non-zero exit code for $infile with flags $flags."
   done | diff - "$checkfile"
   test $? == 0 || fail "bad output for $infile."
+
 }
 
 # Run an input file containing two preprocessor symbols that can take


@@ 87,12 105,18 @@ misctests()
 {
   "$PROG" </dev/null >/dev/null
   test $? == 0 || fail "unable to run test program: $PROG."
-  out=$("$PROG" --help)
+  out=$("$PROG" --help </dev/null)
   test $? == 0 || fail "--help option not recognized."
   test -z "$out" && fail "no --help text."
-  out=$("$PROG" --version)
+  out=$("$PROG" --version </dev/null)
   test $? == 0 || fail "--version option not recoginzed."
   test -z "$out" && fail "no --version text."
+  out=$("$PROG" -D </dev/null 2>&1 >/dev/null)
+  test $? != 0 || fail "no warning for missing option argument."
+  test -z "$out" && fail "no error message for missing option argument."
+  out=$("$PROG" --nosuchoption </dev/null 2>&1 >/dev/null)
+  test $? != 0 || fail "no warning for invalid option."
+  test -z "$out" && fail "no error message for invalid option."
   out=$("$PROG" -Dfoo nosuchfile.c 2>&1 >/dev/null)
   test $? != 0 || fail "no warning for invalid filename."
   test -z "$out" && fail "no error message for invalid filename."


@@ 103,10 127,12 @@ misctests()
 #
 
 misctests
+goodparsingtest tests/basic.c
 goodparsingtest tests/good.c
 goodparsingtest tests/lines.c
 badparsingtest tests/bad.c
 goodparsingtest tests/phases.c
+trigraphtest tests/qqs.c
 for f in tests/full*.c ; do
   fulltest "$f" "${f%.c}.out"
 done