#!/bin/sh
set -eu
base="${TMPDIR-/tmp}/mcpu-cpp-command-line-macros-$$"
mkdir -p "$base"
trap 'rm -rf "$base"' EXIT HUP INT TERM

cat > "$base/main.c" <<'EOT'
OBJECT
DEFAULT
EMPTY marker
СУММА(10, 7)
АНДРЕЙ
résumé
ПОРЯДОК
СТРОКА(АНДРЕЙ)
СКЛЕИТЬ(АНД,РЕЙ)
#ifdef __MCPU_CPP_VERSION__
BUILTIN_PRESENT
#else
BUILTIN_REMOVED
#endif
#if defined(КОМАНДА) && КОМАНДА == 31
CONDITIONAL_OK
#else
CONDITIONAL_BAD
#endif
EOT

"$MCPU_CPP" --no-config \
  -D OBJECT=17 \
  -DDEFAULT \
  -DEMPTY= \
  '-DСУММА(лево,право)=лево + право' \
  -DАНДРЕЙ=23 \
  -Drésumé=29 \
  -DПОРЯДОК=1 -U ПОРЯДОК -DПОРЯДОК=2 \
  '-DСТРОКА(x)=#x' \
  '-DСКЛЕИТЬ(a,b)=a ## b' \
  -DАНДРЕЙ=23 \
  -U__MCPU_CPP_VERSION__ \
  -DКОМАНДА=31 \
  "$base/main.c" -o "$base/out"

grep '^17$' "$base/out" >/dev/null
grep '^1$' "$base/out" >/dev/null
grep '^ marker$' "$base/out" >/dev/null
grep '^10 + 7$' "$base/out" >/dev/null
test "`grep -c '^23$' "$base/out"`" -eq 2
grep '^29$' "$base/out" >/dev/null
grep '^2$' "$base/out" >/dev/null
grep '^"АНДРЕЙ"$' "$base/out" >/dev/null
grep '^BUILTIN_REMOVED$' "$base/out" >/dev/null
if grep '^BUILTIN_PRESENT$' "$base/out" >/dev/null; then exit 1; fi
grep '^CONDITIONAL_OK$' "$base/out" >/dev/null
if grep '^CONDITIONAL_BAD$' "$base/out" >/dev/null; then exit 1; fi

# A function-like -D without '=' has the GNU meaning "replacement is 1".
printf '%s\n' 'F(99)' > "$base/default-func.c"
"$MCPU_CPP" --no-config '-DF(x)' "$base/default-func.c" -o "$base/default-func.out"
grep '^1$' "$base/default-func.out" >/dev/null

# Unicode XID continuation works on the command line as it does in source.
combining=$(printf 'e\314\201')
printf '%s\n' "$combining" > "$base/combining.c"
"$MCPU_CPP" --no-config "-D$combining=37" "$base/combining.c" -o "$base/combining.out"
grep '^37$' "$base/combining.out" >/dev/null

arabic_three=$(printf '\331\243')
name="ИМЯ$arabic_three"
printf '%s\n' "$name" > "$base/digit.c"
"$MCPU_CPP" --no-config "-D$name=43" "$base/digit.c" -o "$base/digit.out"
grep '^43$' "$base/digit.out" >/dev/null

# Command-line definitions pass through the same phase-3 comment cleanup as
# source #define directives.
printf '%s\n' 'COMMENTED' > "$base/comment.c"
"$MCPU_CPP" --no-config '-DCOMMENTED=left/* comment */right' \
  "$base/comment.c" -o "$base/comment.out"
grep '^left right$' "$base/comment.out" >/dev/null

# -dM observes the final command-line macro state.
printf '%s\n' '' > "$base/empty.c"
"$MCPU_CPP" --no-config -dM -DПОРЯДОК=1 -UПОРЯДОК -DПОРЯДОК=2 \
  "$base/empty.c" -o "$base/macros"
grep '^#define ПОРЯДОК 2$' "$base/macros" >/dev/null

# '$' is accepted after the first identifier character.  Single quotes protect
# it from shell parameter expansion; without quotes a backslash can do the same.
"$MCPU_CPP" --no-config -dD '-DАНДРЕЙ$_Y=62' -DОЛЬГА\$OLD=37 \
  "$base/empty.c" -o "$base/dollar-command-line"
grep '^#define АНДРЕЙ$_Y 62$' "$base/dollar-command-line" >/dev/null
grep '^#define ОЛЬГА$OLD 37$' "$base/dollar-command-line" >/dev/null

# '$' is not an identifier-start character.
if "$MCPU_CPP" --no-config '-D$BAD=1' "$base/empty.c" -o "$base/bad" \
     >"$base/dollar-start.stdout" 2>"$base/dollar-start.stderr"; then
  exit 1
fi
grep 'macro name expected after #define' "$base/dollar-start.stderr" >/dev/null

# For -D/-U an invalid tail after a valid identifier is discarded rather than
# becoming part of the replacement list.  The value after '=' is preserved.
"$MCPU_CPP" --no-config -dD '-DАНДРЕЙ@XYZ=62' '-DABC+TAIL=17' \
  '-UREMOVE@TAIL' -DREMOVE=99 "$base/empty.c" -o "$base/invalid-tail"
grep '^#define АНДРЕЙ 62$' "$base/invalid-tail" >/dev/null
grep '^#define ABC 17$' "$base/invalid-tail" >/dev/null
grep '^#define REMOVE 99$' "$base/invalid-tail" >/dev/null
if grep '^#define АНДРЕЙ @XYZ 62$' "$base/invalid-tail" >/dev/null; then
  exit 1
fi

# Explicit '=' in an -U tail is just another invalid suffix and is discarded.
"$MCPU_CPP" --no-config -DNAME=1 '-UNAME=bad' -dM "$base/empty.c" -o "$base/undef-tail"
if grep '^#define NAME ' "$base/undef-tail" >/dev/null; then
  exit 1
fi

# In -dD output command-line definitions have their own GNU-style origin
# marker and must not be reported as predefined/builtin macros.
"$MCPU_CPP" --no-config -dD -DАНДРЕЙ_К=62 -DОЛЬГА=37 \
  "$base/empty.c" -o "$base/dD-command-line"
awk '
  /^#define АНДРЕЙ_К 62$/ {
    if( previous != "# 0 \"<command-line>\"") exit 1
    andrey = 1
  }
  /^#define ОЛЬГА 37$/ {
    if( previous != "# 0 \"<command-line>\"") exit 1
    olga = 1
  }
  /^#define __MCPU_CPP_VERSION__ / {
    if( previous != "# 0 \"<built-in>\"") exit 1
    builtin = 1
  }
  { previous = $0 }
  END { if( !andrey || !olga || !builtin ) exit 1 }
' "$base/dD-command-line"

# An unterminated block comment in a command-line definition is diagnosed.
if "$MCPU_CPP" --no-config '-DBROKEN=left/*' "$base/empty.c" -o "$base/bad" \
     >"$base/comment-bad.stdout" 2>"$base/comment-bad.stderr"; then
  exit 1
fi
grep -- '-D: unterminated comment' "$base/comment-bad.stderr" >/dev/null

# Malformed macro names are rejected by the ordinary #define/#undef parsers.
if "$MCPU_CPP" --no-config -D1ABC=2 "$base/empty.c" -o "$base/bad" \
     >"$base/bad.stdout" 2>"$base/bad.stderr"; then
  exit 1
fi
grep 'macro name expected after #define' "$base/bad.stderr" >/dev/null


# Only -D/-U payloads are decoded.  Invalid UTF-8 and valid supplementary-plane
# UTF-8 are rejected because the internal text model is strict UCS-2.
bad_utf8=$(printf '\377')
if "$MCPU_CPP" --no-config "-D${bad_utf8}=1" "$base/empty.c" -o "$base/bad" \
     >"$base/utf8.stdout" 2>"$base/utf8.stderr"; then
  exit 1
fi
grep -- '-D: invalid UTF-8 or non-UCS-2 character' "$base/utf8.stderr" >/dev/null

non_ucs2=$(printf '\360\237\230\200')
if "$MCPU_CPP" --no-config "-D${non_ucs2}=1" "$base/empty.c" -o "$base/bad" \
     >"$base/nonucs2.stdout" 2>"$base/nonucs2.stderr"; then
  exit 1
fi
grep -- '-D: invalid UTF-8 or non-UCS-2 character' "$base/nonucs2.stderr" >/dev/null

if "$MCPU_CPP" --no-config "-U${bad_utf8}" "$base/empty.c" -o "$base/bad" \
     >"$base/utf8u.stdout" 2>"$base/utf8u.stderr"; then
  exit 1
fi
grep -- '-U: invalid UTF-8 or non-UCS-2 character' "$base/utf8u.stderr" >/dev/null