path: root/tools/mpfrlint

#!/usr/bin/env zsh

# Check possible problems in the MPFR source.

set -e
setopt EXTENDED_GLOB

# mpfrlint can be run from the tools directory
oldpwd=$PWD
[[ -d src ]] || [[ $oldpwd:t != tools ]] || cd ..
err=0

if [[ -t 1 ]] then
  term=1
  pfx=""
  sfx=""
fi

err-if-output()
{
  local dir msg checkoutput=1 checkstatus=1 output st h line
  while [[ $1 == -* ]]
  do
    case $1 in
      --dir=*)
        dir=${1#--dir=} ;;
      --msg=*)
        msg=${1#--msg=} ;;
      -o)
        # The command generates output even in case of success;
        # do not regard this as an error.
        checkoutput="" ;;
      -t)
        # The command normally returns with a non-zero exit status;
        # do not regard this as an error.
        checkstatus="" ;;
      *)
        echo "unrecognized option '$1' for $0" >&2
        exit 1 ;;
    esac
    shift
  done
  [[ -n $dir ]] && pushd $dir
  set +e
  # Note: the double quotes for $@[2,-1] are important to pass empty args.
  output=(${(f)"$("$@[2,-1]" 2>&1)"})
  st=$?
  if [[ ( -n "$checkstatus" && $st -ne 0 ) ||
        ( -n "$checkoutput" && -n "$output" ) ]] then
    if [[ -n "$1" && -t 1 ]] then
      [[ -n "$pfx" ]] || pfx=$(tput 2>/dev/null bold)
      [[ -n "$sfx" ]] || sfx=$(tput 2>/dev/null sgr0)
    fi
    h=${1:+$pfx$1:$sfx }
    for line in ${output:-"exit status = $st"}
    do
      printf "%s%s\n" $h $line
    done
    [[ -n "$msg" ]] && printf "%s %s\n" "$pfx-->$sfx" "$msg"
    err=1
  fi
  set -e
  [[ -n $dir ]] && popd
  # Do not fail.
  true
}

if [[ $1 == test ]] then
  export LC_ALL=C
  err-if-output Err-e exit 17
  err-if-output Err-f false
  err-if-output Err-t true
  err-if-output "" echo "Empty first argument"
  err-if-output "1 line" echo foo
  err-if-output "2 lines" printf "foo\nbar\n"
  err-if-output -o Err-f false
  err-if-output -o Err-t true
  err-if-output -o "cp test" cp
  err-if-output -o "1 line" echo foo
  err-if-output -t Err-f false
  err-if-output -t Err-t true
  err-if-output -t error echo output
  echo "Test done."
  exit
: <<EOF
The output should be:

Err-e: exit status = 17
Err-f: exit status = 1
Empty first argument
1 line: foo
2 lines: foo
2 lines: bar
Err-f: exit status = 1
cp test: cp: missing file operand
cp test: Try 'cp --help' for more information.
error: output
Test done.

EOF
fi

############################################################################

# Note: In the source, ignore everything related to mini-gmp.
srctests=({src,tests}/**/*.[ch]~*mini-gmp.*)
#srctests=(`find src tests -name '*.[ch]' ! -name '*mini-gmp.*'`)

# Detect the definition of reserved macro names.
#
# See ISO C 7.1.3 (Reserved identifiers) and 7.31 (Future library directions).
# The definition of a reserved identifier is undefined behavior
# (this includes the future library directions).
#
# Note: The following check may have false positives in some cases,
# but this may also correspond to bad coding, in particular because
# code moves. We assume that if a header is used somewhere, it may
# be used everywhere in the MPFR code. This particularly concerns
# macros defined via mpfr-impl.h or mpfr-test.h, which are included
# in almost every tests.
#
# The case of EXP in src/mpfr-gmp.h is not really fixable due to the
# possible use of gmp-impl.h, but we should make sure that <errno.h>
# is never included in this case (see comment in the code).
names='E[0-9A-Z]|FE_[A-Z]|LC_[A-Z]|(PRI|SCN)[Xa-z]|SIG_?[A-Z]|TIME_[A-Z]'
msg='ISO C 7.1.3 (Reserved identifiers) and 7.31 (Future library directions).'
grep -E "# *define ($names)" $srctests | \
  grep -v 'src/mpfr-gmp.h:#define EXP(' | \
  err-if-output --msg="$msg" "reserved identifiers (macro names)" cat

# Detect the possible use of forbidden macros in mpfr.h, such as those
# starting with "HAVE_" or "WANT_". Public macros defined by MPFR must
# start with "MPFR_".
err-if-output -t "" perl -ne '
  /^#/ && ! /^# *error / or next; while (/\b([_A-Z]+)\b/g) {
    my $m = $1;
    $m =~ /^(_*MPFR_|_*GMP_|__(GNUC|ICC|STDC)(_|$)|_MSC_|U?INTMAX_C$)/
      and next;
    print "Forbidden macro in mpfr.h line $.: $m\n" }' src/mpfr.h

# Test before other ones about preprocessing directives.
err-if-output \
  --msg="Do not put spaces before a preprocessing directive" \
  -t "space+#" grep -E '^ +# *(define|include|if|el|endif)' $srctests

err-if-output -t "math.h" grep '^# *include *<math\.h>' src/*.c

flaglist="underflow|overflow|divby0|nanflag|inexflag|erangeflag"
grep -E "mpfr_($flaglist)_p" src/*.{c,h} | \
  grep -v -i 'mpfr_clear_' | \
  grep -v '^src/exceptions.c:' | \
  grep -v '^src/mpfr-impl.h:#define mpfr_.*_p()' | \
  grep -v '^src/mpfr.h:__MPFR_DECLSPEC ' | \
  err-if-output "flags" cat
grep -E "mpfr_(clear|set)_($flaglist) *\(" src/*.{c,h} | \
  grep -v '^src/exceptions.c:' | \
  grep -v '^src/mpfr.h:' | \
  err-if-output "flags" cat

grconf()
{
  grep -v '^dnl ' acinclude.m4 configure.ac | \
    err-if-output -t "grconf '$1'" grep -E "$1"
}

grconf '^(.*if +|[[:space:]]*)(test|\[).* == '
grconf '="`'
grconf '[^a-z][ef]?grep[^a-z]'
grconf '[^a-z]sed[^a-z]'

err-if-output --msg="Use GMP_NUMB_BITS instead." \
  -t "GMP_LIMB_BITS" grep GMP_LIMB_BITS $srctests

grep GMP_RND $srctests | err-if-output -t "GMP_RND*" grep -v '#define GMP_RND'

# __MPFR_DECLSPEC (based on the __MPFR_WITHIN_MPFR status) must be used
# instead of __GMP_DECLSPEC (based on the __GMP_WITHIN_GMP status, always
# undefined in MPFR); in particular, the __GMP_DECLSPEC occurrences from
# GMP's longlong.h file must be changed to __MPFR_DECLSPEC when porting
# this file to MPFR. See:
#   https://sympa.inria.fr/sympa/arc/mpfr/2018-08/msg00000.html
#   https://sympa.inria.fr/sympa/arc/mpfr/2018-08/msg00001.html
err-if-output --msg="Use __MPFR_DECLSPEC instead." -t "__GMP_DECLSPEC" \
  grep --exclude=src/mpfr.h __GMP_DECLSPEC $srctests

# Use mpfr_div_2ui/mpfr_mul_2ui instead of mpfr_div_2exp/mpfr_mul_2exp.
grep 'mpfr_\(div\|mul\)_2exp' {src,tests}/*.{c,h} |\
  grep -v '^src/\(\(div\|mul\)_2exp\.c:\|mpf2mpfr\.h:#define mpf_\)' | \
  grep -v '^src/mpfr\.h:\(__MPFR_DECLSPEC\|#define mpfr_..._2exp\)' | \
  grep -v '^tests/\(reuse\|taway\)\.c: *test2ui (mpfr_..._2exp,' | \
  err-if-output --msg="Use mpfr_div_2ui and mpfr_mul_2ui instead." \
    "mpfr_div_2exp and mpfr_mul_2exp" cat

# Note: GMP internals must not be used unless WANT_GMP_INTERNALS is defined.
# The check below attempts to handle that, but this is a heuristic (it might
# fail in case of nested #if's, but this is currently OK).
for file in $srctests
do
  perl -e 'undef $/; $_ = <>;
           s:/\*.*?\*/::sg;
           s:# *(el)?if.*WANT_GMP_INTERNALS(.|\n)*?# *(else|endif)::g;
           print' $file |
    err-if-output -t "GMP internals in $file" grep '__gmp[nz]_'
done

err-if-output --msg="Use mpfr_limb_ptr and mpfr_limb_srcptr instead." \
  -t "mp_ptr and mp_srcptr" grep -E 'mp_(src)?ptr' $srctests

# Do not use __mpfr_struct structure members in .c files.
err-if-output -t "__mpfr_struct members" \
  grep -E '[^0-9a-z_]_mpfr_(prec|sign|exp|d)' {src,tests}/*.c

# Detect some uses of "x != x" and "x == x". If this occurs in source code,
# x is probably a native FP number (otherwise the test is useless), but in
# such a case, the DOUBLE_ISNAN macro should be used.
err-if-output -t "x != x and x == x tests" \
  grep '( *\([^[:space:]]*\) *[!=]= *\1 *)' $srctests

for i in exp prec rnd
do
  grep "[^a-z]mp_${i}_t" $srctests | \
    grep -v "\(# *define\|# *ifndef\|typedef\) *mp_${i}_t" | \
    grep -v "\[mp_${i}_t\]" | \
    err-if-output "mp_*_t" cat
done

for file in $srctests
do
  err-if-output "MPFR_LOG_MSG format" perl -e '
    my $f = do { local $/; <> };
    while ($f =~ /MPFR_LOG_MSG\s*\(\s*\(.*?\)\s*\)/gs) {
      my $s = $&; print "$ARGV: $s\n" if
        index($s,"\\n\"") < 0 || $s !~ /"\s*,/
    }' $file
done

# Macros of the form:
#   #define FOO { ... }
# may be unsafe and could yield obscure failures where writing "FOO;" as
# this is here a block followed by a null statement. The following form
# is preferred in most of the cases:
#   #define FOO do { ... } while (0)
# so that "FOO;" is a single statement.
# To avoid false positives, a comment can be inserted, e.g.:
#   #define FOO /* initializer */ { 0, 17 }
for file in $srctests
do
  err-if-output "Missing 'do ... while (0)'" perl -e '
    while (<>) {
      my $s = $_;
      while ($s =~ s/\\\n//) { $s .= <> }
      $s =~ /^#\s*define\s+\w+(\([^)]*\))?\s*{/
        and $s =~ tr/ \t/ /s, print "$ARGV: $s";
    }' $file
done

# Do not use snprintf as it is not available in ISO C90.
# Even on platforms where it is available, the prototype
# may not be included (e.g. with gcc -ansi), so that the
# code may be compiled incorrectly.
grep '[^a-z_]snprintf *([^)]' $srctests | \
  err-if-output -t "snprintf" grep -v '/\*.*[^a-z_]snprintf *([^)]'

# Constant checking should use either MPFR_STAT_STATIC_ASSERT
# or MPFR_ASSERTN(0) for not yet implemented corner cases.
# This test is a heuristic.
# Note: as long as the support of _MPFR_EXP_FORMAT = 4 is not complete,
# run-time assertions involving MPFR_EMAX_MAX, LONG_MAX, etc. should be
# used instead of static assertions to allow testing and correction of
# the code (then the removal of the assertions).
grep 'MPFR_ASSERT[DN][^a-z]*;' src/*.c | grep -v 'MPFR_ASSERTN *(0)' | \
  grep -v '\(MPFR_EMAX_MAX\|MPFR_EXP_MAX\).*LONG_MAX' | \
  grep -v '\(MPFR_EMIN_MIN\|MPFR_EXP_MIN\).*LONG_MIN' | \
  grep -v MPFR_BLOCK_EXCEP | \
  err-if-output --msg="Use MPFR_STAT_STATIC_ASSERT in general." \
    -t "Constant checking" cat

# ASSERT and ASSERT_ALWAYS must not be used for assertion checking.
# Use MPFR_STAT_STATIC_ASSERT for static assertions, otherwise either
# MPFR_ASSERTD (debug mode / hint for the compiler) or MPFR_ASSERTN.
err-if-output -t "ASSERT / ASSERT_ALWAYS" \
  grep -E '[^_]ASSERT(_ALWAYS)? *(\(|$)' {src,tests}/*.c

# Use MPFR_TMP_LIMBS_ALLOC.
err-if-output -t "Use MPFR_TMP_LIMBS_ALLOC" \
  grep 'MPFR_TMP_ALLOC.*\(BYTES_PER_MP_LIMB\|sizeof.*mp_limb_t\)' src/*.c

# Use simple mp_limb_t constants: MPFR_LIMB_ZERO, MPFR_LIMB_ONE, etc.
grep '(mp_limb_t) *-\?[01][^0-9]' $srctests | grep -v '#define MPFR_LIMB_' | \
  grep -v 'MPFR_ASSERTN *(MPFR_MANT.*== *(mp_limb_t)' | \
  err-if-output -t "Use simple mp_limb_t constants" cat

# "~0;" has already been seen in the code. Check also a bit more.
# Such code is either an error or poorly written, and in this case, quite
# fragile. In practice, this may work because most (or all) platforms use
# two's complement, and 0 being of type int (signed), this will give -1;
# then when converted to a larger unsigned type, this will give the maximum
# value as expected. However the reader may not know this explanation, and
# a small change of the code can easily make it incorrect. The correct way
# to obtain this result is to cast the constant to the expected type, then
# do the bitwise complement; alternatively, cast -1 to the expected type.
# For limbs, the macro MPFR_LIMB_MAX should be used.
err-if-output \
  --msg="Possibly incorrect type for ~ (use MPFR_LIMB_MAX for limbs)" \
  -t "Suspicious code" \
  grep '~[0-9] *[-+*&|;]' $srctests

# Code possibly wrong with some C/GMP implementations. In short, if the
# right operand of a shift depends on GMP_NUMB_BITS, then the left operand
# should be of type mp_limb_t. There might be false positives.
err-if-output \
  --msg="Use a constant of type mp_limb_t instead of unsigned long?" \
  -t "Suspicious code" \
  grep -E '[^A_Za-z_][0-9]+[UL]* *<<.*GMP_NUMB_BITS' src/*.c

for file in $srctests */Makefile.am acinclude.m4 configure.ac
do
  # Note: this is one less that the POSIX minimum limit in case
  # implementations are buggy like POSIX examples. :)
  err-if-output "" perl -ne "/.{2047,}/ and print \
    \"Line \$. of file $file has more than 2046 bytes.\n\"" "$file"
done

# Code style: a sign decimal constant for mpfr_set_inf and mpfr_set_zero
# should be either 1 or -1 (except for the tests in tset.c).
grep -E 'mpfr_set_(inf|zero) *\([^,]*, *[-+]?([02-9]|1[^)])' $srctests | \
  err-if-output -t "mpfr_set_(inf|zero) second argument" \
    grep -v tests/tset\\.c:

# In general, one needs to include mpfr-impl.h (note that some platforms
# such as MS Windows use a config.h, which is included by mpfr-impl.h).
for file in src/*.c
do
  [[ "$file" == src/(jyn_asympt|mini-gmp|round_raw_generic).c ]] || \
    grep -q '^# *include *"\(mpfr-impl\|fits.*\|gen_inverse\|random_deviate\)\.h"' $file || \
    { echo "Missing '#include \"mpfr-impl.h\"' in $file?" && err=1 }
done

# Detect mpfr_mul with identical 2nd and 3rd arguments, which can be
# replaced by mpfr_sqr. We exclude mul.c and sqr.c (which implement
# these functions) because such a mpfr_mul can be used on purpose.
err-if-output --msg='Use mpfr_sqr instead of mpfr_mul.' "mul/sqr" perl -ne \
  '/__MPFR_DECLSPEC/ or ($u,$v) = /mpfr_mul *\([^,]*,([^,]*),([^,]*),/ and
     do { $u =~ tr/ //d; $v =~ tr/ //d; $u eq $v and print "$ARGV: $_" }' \
  src/*.[ch]~src/(mul|sqr).c

# Re-seeding mpfr_rands in the tests is bad practice, as it would affect
# later tests, defeating the purpose of GMP_CHECK_RANDOMIZE.
err-if-output \
  --msg='Do not re-seed mpfr_rands; use another gmp_randstate_t variable.' \
  -t "mpfr_rands" grep --exclude=tests.c \
    'gmp_randseed.*mpfr_rands' tests/*.{c,h}

# "mpfr-impl.h" must not be included directly by the test programs.
# Otherwise __MPFR_WITHIN_MPFR will be defined, yielding failures
# under MS Windows with DLL.
err-if-output \
  --msg='Include "mpfr-test.h" instead of "mpfr-impl.h" to prevent __MPFR_WITHIN_MPFR from being defined.' \
  -t "mpfr-impl.h inclusion" grep --exclude=mpfr-test.h \
    '^ *# *include *"mpfr-impl.h"' tests/*.{c,h}

# Check that the usual test programs call tests_start_mpfr and tests_end_mpfr.
tprg=($(sed -n '/^check_PROGRAMS/,/[^\\]$/ {
  s/.*=//
  s/\\//
  p
  }' tests/Makefile.am))
[[ -n $tprg ]]
for t in $tprg
do
  [[ $t != mpf*_compat ]] || continue
  tc=tests/$t.c
  if grep -q "main *(" $tc; then
    for fn in tests_start_mpfr tests_end_mpfr
    do
      err-if-output "missing call to $fn in $tc" grep -q "$fn *();" $tc
    done
  fi
done

# mpfr_printf-like functions shouldn't be used in the tests,
# as they need <stdarg.h> (HAVE_STDARG defined).
for file in tests/*.c
do
  sed '/#if\(def\| *defined *(\)\? *HAVE_STDARG/,/#\(endif\|else\) .*HAVE_STDARG/d
       /\/\*.*\*\//d
       /\/\*/,/\*\//d' $file | \
    err-if-output -t "unprotected mpfr_printf-like function call in $file" \
      grep "mpfr_[a-z]*printf"
done

grep __gmp_ tests/*.c | \
  err-if-output -t "__gmp_" grep -v '^tests/tabort_defalloc'

grep -E '[^a-z_](m|re)alloc *\(' tests/*.c | \
  err-if-output -t "alloc" grep -Ev '^tests/(memory|talloc-cache).c:'

err-if-output --dir=doc "check-typography" ./check-typography

fdlv1="`sed -n '/Version / {
  s/.*Version //
  s/,.*//
  p
  q
  }' doc/fdl.texi`"
fdlv2="`sed -n '/GNU Free Documentation License/ {
  s/.*Version //
  s/ or.*//
  p
  q
  }' doc/mpfr.texi`"
[[ $fdlv1 == $fdlv2 ]] || { cat <<EOF && err=1 }
GFDL versions differ:
   fdl.texi: $fdlv1
  mpfr.texi: $fdlv2
EOF

err-if-output "ck-copyright-notice" tools/ck-copyright-notice

if which gcc > /dev/null 2> /dev/null; then
  err-if-output "ck-mparam" tools/ck-mparam
else
  echo "Warning! gcc is not installed. Cannot run ck-mparam." >&2
fi

# Note about the TZ value: GMT0 and UTC0 are both specified by POSIX,
# and UTC0 is the preferred value, but old systems only accept GMT0.
# The "0" is important ("GMT" alone does not work on Tru64 Unix).
texisvnd=`LC_ALL=C TZ=GMT0 svn info doc/mpfr.texi 2> /dev/null | sed -n 's/Last Changed Date:.*, [0-9]* \([A-Z][a-z][a-z] [0-9][0-9][0-9][0-9]\)).*/\1/p'`
if [[ $? -eq 0 ]] && [[ -n "$texisvnd" ]] then
  texidate=`sed -n 's/@set UPDATED-MONTH \([A-Z][a-z][a-z]\).*\( [0-9][0-9][0-9][0-9]\)/\1\2/p' doc/mpfr.texi`
  [[ $texidate == $texisvnd ]] || { cat <<EOF && err=1 }
mpfr.texi's UPDATED-MONTH seems to be incorrect:
  mpfr.texi's UPDATED-MONTH: $texidate
  Last Changed Date in WC:   $texisvnd
EOF
fi

acv1=`sed -n 's/.*autoconf \([0-9.]\+\) (at least).*/\1/p' doc/README.dev`
acv2=`sed -n 's/AC_PREREQ(\([0-9.]\+\).*/\1/p' acinclude.m4`
[[ $acv1 == $acv2 ]] || { cat <<EOF && err=1 }
autoconf minimal versions differ:
  README.dev:   $acv1
  acinclude.m4: $acv2
EOF

gmpv1=`sed -n 's/.*error "GMP \([0-9.]\+\) or .*/\1/p' configure.ac`
gmpv2=`sed -n 's/.*AC_MSG_ERROR.*GMP \([0-9.]\+\) or .*/\1/p' configure.ac`
gmpv3=`sed -n 's/.*MPFR requires GMP version \([0-9.]\+\) or .*/\1/p' INSTALL`
gmpv4=`sed -n 's/.*GNU MP (version \([0-9.]\+\) or .*/\1/p' doc/mpfr.texi`
[[ $gmpv1 == $gmpv2 && $gmpv1 == $gmpv3 && $gmpv1 == $gmpv4 ]] ||
  { cat <<EOF && err=1 }
Minimal GMP version mismatch:
  configure.ac 1: $gmpv1
  configure.ac 2: $gmpv2
  INSTALL:        $gmpv3
  doc/mpfr.texi:  $gmpv4
EOF

# In case of problems, one can skip files with -S or use "grep -v"...
# The codespell.exclude could be simplified after this is fixed:
#   https://github.com/codespell-project/codespell/issues/17
#   https://github.com/codespell-project/codespell/issues/535
if which codespell > /dev/null 2> /dev/null; then
  doc=(FAQ.html Makefile.am README.dev add-with-carry.c check-typography \
       faq.xsl fdl.texi mini-gmp mpfr.texi sum.txt update-faq)
  err-if-output "codespell" codespell ${term:+--enable-colors} \
    -q3 -I codespell.ignore -x codespell.exclude \
    AUTHORS BUGS INSTALL NEWS README TODO doc/${^doc} examples $srctests
else
  echo "Warning! codespell is not installed. Cannot check spelling." >&2
fi

err-if-output "ck-clz_tab" tools/ck-clz_tab
err-if-output "ck-inits-clears" tools/ck-inits-clears
err-if-output "ck-version-info" tools/ck-version-info

############################################################################

[[ $err -eq 0 ]] || echo "Terminated with errors." >&2
exit $err