0001 #!/usr/bin/env perl
0002 # SPDX-License-Identifier: GPL-2.0
0003 #
0004 # Treewide grep for references to files under Documentation, and report
0005 # non-existing files in stderr.
0006
0007 use warnings;
0008 use strict;
0009 use Getopt::Long qw(:config no_auto_abbrev);
0010
0011 # NOTE: only add things here when the file was gone, but the text wants
0012 # to mention a past documentation file, for example, to give credits for
0013 # the original work.
0014 my %false_positives = (
0015 "Documentation/scsi/scsi_mid_low_api.rst" => "Documentation/Configure.help",
0016 "drivers/vhost/vhost.c" => "Documentation/virtual/lguest/lguest.c",
0017 );
0018
0019 my $scriptname = $0;
0020 $scriptname =~ s,.*/([^/]+/),$1,;
0021
0022 # Parse arguments
0023 my $help = 0;
0024 my $fix = 0;
0025 my $warn = 0;
0026
0027 if (! -e ".git") {
0028 printf "Warning: can't check if file exists, as this is not a git tree\n";
0029 exit 0;
0030 }
0031
0032 GetOptions(
0033 'fix' => \$fix,
0034 'warn' => \$warn,
0035 'h|help|usage' => \$help,
0036 );
0037
0038 if ($help != 0) {
0039 print "$scriptname [--help] [--fix]\n";
0040 exit -1;
0041 }
0042
0043 # Step 1: find broken references
0044 print "Finding broken references. This may take a while... " if ($fix);
0045
0046 my %broken_ref;
0047
0048 my $doc_fix = 0;
0049
0050 open IN, "git grep ':doc:\`' Documentation/|"
0051 or die "Failed to run git grep";
0052 while (<IN>) {
0053 next if (!m,^([^:]+):.*\:doc\:\`([^\`]+)\`,);
0054 next if (m,sphinx/,);
0055
0056 my $file = $1;
0057 my $d = $1;
0058 my $doc_ref = $2;
0059
0060 my $f = $doc_ref;
0061
0062 $d =~ s,(.*/).*,$1,;
0063 $f =~ s,.*\<([^\>]+)\>,$1,;
0064
0065 if ($f =~ m,^/,) {
0066 $f = "$f.rst";
0067 $f =~ s,^/,Documentation/,;
0068 } else {
0069 $f = "$d$f.rst";
0070 }
0071
0072 next if (grep -e, glob("$f"));
0073
0074 if ($fix && !$doc_fix) {
0075 print STDERR "\nWARNING: Currently, can't fix broken :doc:`` fields\n";
0076 }
0077 $doc_fix++;
0078
0079 print STDERR "$file: :doc:`$doc_ref`\n";
0080 }
0081 close IN;
0082
0083 open IN, "git grep 'Documentation/'|"
0084 or die "Failed to run git grep";
0085 while (<IN>) {
0086 next if (!m/^([^:]+):(.*)/);
0087
0088 my $f = $1;
0089 my $ln = $2;
0090
0091 # On linux-next, discard the Next/ directory
0092 next if ($f =~ m,^Next/,);
0093
0094 # Makefiles and scripts contain nasty expressions to parse docs
0095 next if ($f =~ m/Makefile/ || $f =~ m/\.sh$/);
0096
0097 # It doesn't make sense to parse hidden files
0098 next if ($f =~ m#/\.#);
0099
0100 # Skip this script
0101 next if ($f eq $scriptname);
0102
0103 # Ignore the dir where documentation will be built
0104 next if ($ln =~ m,\b(\S*)Documentation/output,);
0105
0106 if ($ln =~ m,\b(\S*)(Documentation/[A-Za-z0-9\_\.\,\~/\*\[\]\?+-]*)(.*),) {
0107 my $prefix = $1;
0108 my $ref = $2;
0109 my $base = $2;
0110 my $extra = $3;
0111
0112 # some file references are like:
0113 # /usr/src/linux/Documentation/DMA-{API,mapping}.txt
0114 # For now, ignore them
0115 next if ($extra =~ m/^{/);
0116
0117 # Remove footnotes at the end like:
0118 # Documentation/devicetree/dt-object-internal.txt[1]
0119 $ref =~ s/(txt|rst)\[\d+]$/$1/;
0120
0121 # Remove ending ']' without any '['
0122 $ref =~ s/\].*// if (!($ref =~ m/\[/));
0123
0124 # Remove puntuation marks at the end
0125 $ref =~ s/[\,\.]+$//;
0126
0127 my $fulref = "$prefix$ref";
0128
0129 $fulref =~ s/^(\<file|ref)://;
0130 $fulref =~ s/^[\'\`]+//;
0131 $fulref =~ s,^\$\(.*\)/,,;
0132 $base =~ s,.*/,,;
0133
0134 # Remove URL false-positives
0135 next if ($fulref =~ m/^http/);
0136
0137 # Remove sched-pelt false-positive
0138 next if ($fulref =~ m,^Documentation/scheduler/sched-pelt$,);
0139
0140 # Discard some build examples from Documentation/target/tcm_mod_builder.rst
0141 next if ($fulref =~ m,mnt/sdb/lio-core-2.6.git/Documentation/target,);
0142
0143 # Check if exists, evaluating wildcards
0144 next if (grep -e, glob("$ref $fulref"));
0145
0146 # Accept relative Documentation patches for tools/
0147 if ($f =~ m/tools/) {
0148 my $path = $f;
0149 $path =~ s,(.*)/.*,$1,;
0150 $path =~ s,testing/selftests/bpf,bpf/bpftool,;
0151 next if (grep -e, glob("$path/$ref $path/../$ref $path/$fulref"));
0152 }
0153
0154 # Discard known false-positives
0155 if (defined($false_positives{$f})) {
0156 next if ($false_positives{$f} eq $fulref);
0157 }
0158
0159 if ($fix) {
0160 if (!($ref =~ m/(scripts|Kconfig|Kbuild)/)) {
0161 $broken_ref{$ref}++;
0162 }
0163 } elsif ($warn) {
0164 print STDERR "Warning: $f references a file that doesn't exist: $fulref\n";
0165 } else {
0166 print STDERR "$f: $fulref\n";
0167 }
0168 }
0169 }
0170 close IN;
0171
0172 exit 0 if (!$fix);
0173
0174 # Step 2: Seek for file name alternatives
0175 print "Auto-fixing broken references. Please double-check the results\n";
0176
0177 foreach my $ref (keys %broken_ref) {
0178 my $new =$ref;
0179
0180 my $basedir = ".";
0181 # On translations, only seek inside the translations directory
0182 $basedir = $1 if ($ref =~ m,(Documentation/translations/[^/]+),);
0183
0184 # get just the basename
0185 $new =~ s,.*/,,;
0186
0187 my $f="";
0188
0189 # usual reason for breakage: DT file moved around
0190 if ($ref =~ /devicetree/) {
0191 # usual reason for breakage: DT file renamed to .yaml
0192 if (!$f) {
0193 my $new_ref = $ref;
0194 $new_ref =~ s/\.txt$/.yaml/;
0195 $f=$new_ref if (-f $new_ref);
0196 }
0197
0198 if (!$f) {
0199 my $search = $new;
0200 $search =~ s,^.*/,,;
0201 $f = qx(find Documentation/devicetree/ -iname "*$search*") if ($search);
0202 if (!$f) {
0203 # Manufacturer name may have changed
0204 $search =~ s/^.*,//;
0205 $f = qx(find Documentation/devicetree/ -iname "*$search*") if ($search);
0206 }
0207 }
0208 }
0209
0210 # usual reason for breakage: file renamed to .rst
0211 if (!$f) {
0212 $new =~ s/\.txt$/.rst/;
0213 $f=qx(find $basedir -iname $new) if ($new);
0214 }
0215
0216 # usual reason for breakage: use dash or underline
0217 if (!$f) {
0218 $new =~ s/[-_]/[-_]/g;
0219 $f=qx(find $basedir -iname $new) if ($new);
0220 }
0221
0222 # Wild guess: seek for the same name on another place
0223 if (!$f) {
0224 $f = qx(find $basedir -iname $new) if ($new);
0225 }
0226
0227 my @find = split /\s+/, $f;
0228
0229 if (!$f) {
0230 print STDERR "ERROR: Didn't find a replacement for $ref\n";
0231 } elsif (scalar(@find) > 1) {
0232 print STDERR "WARNING: Won't auto-replace, as found multiple files close to $ref:\n";
0233 foreach my $j (@find) {
0234 $j =~ s,^./,,;
0235 print STDERR " $j\n";
0236 }
0237 } else {
0238 $f = $find[0];
0239 $f =~ s,^./,,;
0240 print "INFO: Replacing $ref to $f\n";
0241 foreach my $j (qx(git grep -l $ref)) {
0242 qx(sed "s\@$ref\@$f\@g" -i $j);
0243 }
0244 }
0245 }
