hdf5/bin/checkposix

#!/usr/bin/env perl
require 5.003;
use warnings;

#
# Copyright by The HDF Group.
# Copyright by the Board of Trustees of the University of Illinois.
# All rights reserved.
#
# This file is part of HDF5.  The full HDF5 copyright notice, including
# terms governing use, modification, and redistribution, is contained in
# the COPYING file, which can be found at the root of the source code
# distribution tree, or in https://www.hdfgroup.org/licenses.
# If you do not have access to either file, you may request a copy from
# help@hdfgroup.org.
#
# Dana Robinson
# Spring 2019
# (Original by Robb Matzke)
#
# Purpose:  Given the names of C source files this script will print the
#           file name, line number, and function name of any function that
#           doesn't begin with the letter 'h' or 'H' as stipulated by the
#           HDF5 programming style guide.
#
#           Emacs users can run this script as the compile command and
#           use 'next-error' (usually bound to M-`) to find each name
#           violation.

use File::Basename;

# Loop over all files passed to the function
foreach $arg (@ARGV) {

    # Get the filename from the path
    $filename = fileparse($arg);

    # Skip files that don't include H5private.h
    # H5system. has to be inspected by hand since it wraps POSIX files
    #
    # H5detect and H5make_libsettings are created before the library exists
    # so calls that link to function replacements won't work. We'll ignore
    # it here.
    #
    # If a user specifies one file, process it no matter what so people
    # can inspect files we normally skip (like H5system.c).

    $ignore = 0;

    # Ignored files in src/
    if($#ARGV gt 0 and $filename =~ /H5FDmulti|H5FDstdio|H5VLpassthru|H5system|H5detect|H5make_libsettings/) {
        $ignore = 1;
    }
    # Ignored atomic test files in test/
    if($#ARGV gt 0 and $filename =~ /atomic_reader|atomic_writer/) {
        $ignore = 1;
    }
    # Ignored filter plugins in test/
    if($#ARGV gt 0 and $filename =~ /^filter_plugin\d_/) {
        $ignore = 1;
    }
    # Ignored generators in test/
    if($#ARGV gt 0 and $filename =~ /^gen_/) {
        $ignore = 1;
    }

    if($ignore) {
        print "$filename is exempt from using Standard library macro wrappers\n";
        next;
    }

    # Open the file
    open(my $fh, "<", $arg) or do {
        warn "NOTE: Unable to open $arg: !$\n";
        next;
    };

    # Loop over all lines in the file to find undecorated functions
    while (<$fh>) {

        # Get rid of comments by removing the inside part.
        s|/\*.*?\*/||g;
        if ($in_comment) {
            if (/\*\//) {
                s|.*?\*/||;
                $in_comment = 0;
            } else {
                $_="\n";
            }
        } elsif (m|/\*|) {
            s|/\*.*||;
            $in_comment = 1;
        }

        # Get rid of string constants if they begin and end on this line.
        s/([\'\"])([^\1]|\\\1)*?\1/$1$1/g;

        # Get rid of preprocessor directives
        s/^\#.*//;

        # Skip callbacks invoked as methods in a struct
        next if $_ =~ /\b(\)?]?->|\.)\(?([a-z_A-Z]\w*)\s*\(/;

        # Now find all function calls on this line which don't start with 'H'
        while (($name)=/\b([a-z_A-GI-Z]\w*)\s*\(/) {
            $_ = $';
              
            # Ignore C statements that look sort of like function
            # calls.
            next if $name =~ /^(if|for|offsetof|return|sizeof|switch|while|void)$/;

            # Ignore things that get misdetected because of the simplified
            # parsing that takes place here.
            next if $name =~ /^(int|herr_t|_term_interface|_term_package)$/;

            # These are really HDF5 functions/macros even though they don't
            # start with `h' or `H'.
            next if $name =~ /^FUNC_(ENTER|LEAVE)(_(NO)?API|_PACKAGE|_STATIC)?(_NAMECHECK_ONLY|_NOFS|_NOCLEAR|_NOINIT|_NOPUSH)?(_NOFUNC|_TAG)?$/;
            next if $name =~ /^(BEGIN|END)_FUNC$/;
            next if $name =~ /^U?INT(8|16|32|64)(ENCODE|DECODE)(_VAR)?$/;
            next if $name =~ /^CI_(PRINT_STATS|INC_SRC|INC_DST)$/;
            next if $name =~ /^(ABS|ADDR_OVERFLOW|ALL_MEMBERS|BOUND|CONSTR|DETECT_[I|F|M]|DOWN)$/;
            next if $name =~ /^(MIN3?|MAX3?|NELMTS|POWER_OF_TWO|REGION_OVERFLOW)$/;
            next if $name =~ /^(SIZE_OVERFLOW|UNIQUE_MEMBERS|S_ISDIR)$/;
            next if $name =~ /^addr_defined$/;
            next if $name =~ /^TERMINATOR$/;

            # Ignore callback invocation
            next if $name =~ /^(op|cb|OP|iter_op|func)$/;

            # Ignore main
            next if $name =~ /^main$/;

            # This often appears in preprocessor lines that span multiple lines
            next if $name =~ /^(defined)$/;

            # These are Windows system calls. Ignore them.
            next if $name =~ /^(_get_osfhandle|GetFileInformationByHandle|SetFilePointer|GetLastError|SetEndOfFile)$/;
            next if $name =~ /^(FindNextFile|FindClose|_tzset|Wgettimeofday|GetSystemTimeAsFileTime|Wgetlogin|GetUserName)$/;
            next if $name =~ /^(DeleteCriticalSection|TlsFree|TlsGetValue|CreateThread)$/;
            next if $name =~ /^(ExpandEnvironmentStringsA|LockFileEx|UnlockFileEx)$/;
            next if $name =~ /^(DllMain|LocalAlloc|LocalFree)$/;
            next if $name =~ /^(FindFirstFileA|FindNextFileA)$/;
            next if $name =~ /^(_beginthread|(Initialize|Enter|Leave)CriticalSection|TlsAlloc)$/;

            # These are MPI function calls. Ignore them.
            next if $name =~ /^(MPI_|MPE_)/;

            # These are POSIX threads function calls. Ignore them.
            next if $name =~ /^pthread_/;

            # These are zlib & szlib function calls. Ignore them.
            next if $name =~ /^(inflate|SZ_)/;
            next if $name =~ /^compress2$/;

            # These is an H5Dfill function. Ignore it in this file.
            if($filename =~ /H5Dfill/) {
                next if $name =~ /^(alloc_func)$/;
            }

            # These are H5Zscaleoffset functions. Ignore them in this file.
            if($filename =~ /H5Zscaleoffset/) {
                next if $name =~ /^(pow_fun|round_fun|abs_fun|lround_fun|llround_fun)$/;
            }

            # This is a macro parameter in H5Rint.c. Ignore it in this file.
            if($filename =~ /H5Rint/) {
                next if $name =~ /^(func)$/;
            }

            # Internal calls in the HDFS VFD (H5FDhdfs.c). Ignore it in this file.
            if($filename =~ /H5FDhdfs/) {
                next if $name =~ /^(hdfs)/;
            }

            # Macros, etc. from the mirror VFD (H5FDmirror.c). Ignore in this file.
            if($filename =~ /H5FDmirror/) {
                next if $name =~ /^(LOG)/;
                next if $name =~ /^(BSWAP_64|is_host_little_endian)$/;
            }

            # These are things in H5FDs3comms.c and H5FDros3.c. Ignore them in these files.
            if($filename =~ /H5FDs3comms|H5FDros3/) {
                next if $name =~ /^(curl_|curlwritecallback|gmnow)/;
                next if $name =~ /^(ros3_|ROS3_|S3COMMS_)/;
                next if $name =~ /^(EVP_sha256|SHA256|ISO8601NOW)$/;
            }

            # TESTING (not comprehensive - just noise reduction)
            
            # Test macros and functions (testhdf5.h)
            next if $name =~ /^(AddTest|TestErrPrintf|TestSummary|TestCleanup|TestShutdown)$/;
            next if $name =~ /^(CHECK|CHECK_PTR|CHECK_PTR_NULL|CHECK_PTR_EQ|CHECK_I)$/;
            next if $name =~ /^(VERIFY|VERIFY_STR|VERIFY_TYPE|MESSAGE|ERROR)$/;

            # Test macros and functions (h5test.h)
            next if $name =~ /^(TESTING|PASSED|SKIPPED|PUTS_ERROR|FAIL_PUTS_ERROR|FAIL_STACK_ERROR|TEST_ERROR|AT)$/;
            next if $name =~ /^(GetTestExpress)$/;

            # Ignore functions that start with test_ or check_
            next if $name =~ /^test_/;
            next if $name =~ /^check_/;

            # Ignore functions that start with h5_
            next if $name =~ /^h5_/;

            # Ignore process completed status
            next if $name =~ /(WIFEXITED|WEXITSTATUS|WIFSIGNALED|WTERMSIG|WCOREDUMP|WIFSTOPPED|WSTOPSIG)/;

            # Ignore usage functions
            next if $name =~ /^usage$/;

            # Ignore callbacks
            next if $name =~ /(_cb\d?)$/;

            # Specific tests (not even remotely comprehensive)

            # accum test code
            if($filename =~ /accum/) {
                next if $name =~ /^(accum_)/;
            }

            # cache test code
            if($filename =~ /cache/) {
                next if $name =~ /(_entry|_entries|_cache|_check|_dependency|_status|_op)$/;
                next if $name =~ /^(verify_|smoke_check_|row_major_|col_major_)/;
                next if $name =~ /^(resize_configs_are_equal|CACHE_ERROR)$/
            }

            # Splitter VFD test code. Ignore in vfd.c.
            if($filename =~ /vfd/) {
                next if $name =~ /^(SPLITTER_|splitter_)/;
                next if $name =~ /(_splitter_)/;
                next if $name =~ /^(file_exists)$/;
            }

            # S3 VFD test code. Ignore in ros3.c and s3comms.c.
            # HDFS VFD test code. Ignore in hdfs.c.
            if($filename =~ /ros3|s3comms|hdfs/) {
                next if $name =~ /^(JSVERIFY|JSFAILED_|JSERR_|jserr_|FAIL_)/;
                next if $name =~ /^(curl_)/;
                next if $name =~ /^(S3COMMS_FORMAT_CREDENTIAL|ISO8601NOW|gmnow)$/;
            }

            # VDS test code. Ignore in vds.c.
            if($filename =~ /vds/) {
                next if $name =~ /^(vds_)/;
            }

            print "$filename:$.: $name\n";
        }

    }

    # Close the file
    close($fh);
}

if($#ARGV gt 0) {
    print "\n";
    print "NOTE:\n";
    print "If any files were skipped due to being exempt, you can inspect them manually\n";
    print "by using this script on them one at a time, which will always process the file.\n";
}