1d2656095SMauro Carvalho Chehab#!/usr/bin/env perl
2d2656095SMauro Carvalho Chehab# SPDX-License-Identifier: GPL-2.0
3d2656095SMauro Carvalho Chehab#
4e8939222SJani Nikula# Treewide grep for references to files under Documentation, and report
5e8939222SJani Nikula# non-existing files in stderr.
6e8939222SJani Nikula
7d2656095SMauro Carvalho Chehabuse warnings;
8d2656095SMauro Carvalho Chehabuse strict;
9d2656095SMauro Carvalho Chehabuse Getopt::Long qw(:config no_auto_abbrev);
10e8939222SJani Nikula
11aeaacbfeSMauro Carvalho Chehab# NOTE: only add things here when the file was gone, but the text wants
12aeaacbfeSMauro Carvalho Chehab# to mention a past documentation file, for example, to give credits for
13aeaacbfeSMauro Carvalho Chehab# the original work.
14aeaacbfeSMauro Carvalho Chehabmy %false_positives = (
15ce5c5d65SMauro Carvalho Chehab	"Documentation/scsi/scsi_mid_low_api.rst" => "Documentation/Configure.help",
16aeaacbfeSMauro Carvalho Chehab	"drivers/vhost/vhost.c" => "Documentation/virtual/lguest/lguest.c",
17aeaacbfeSMauro Carvalho Chehab);
18aeaacbfeSMauro Carvalho Chehab
19d2656095SMauro Carvalho Chehabmy $scriptname = $0;
20d2656095SMauro Carvalho Chehab$scriptname =~ s,.*/([^/]+/),$1,;
21d2656095SMauro Carvalho Chehab
22d2656095SMauro Carvalho Chehab# Parse arguments
23d2656095SMauro Carvalho Chehabmy $help = 0;
24d2656095SMauro Carvalho Chehabmy $fix = 0;
25b1663d7eSMauro Carvalho Chehabmy $warn = 0;
26b1663d7eSMauro Carvalho Chehab
27*709dedfdSRob Herringif (! -e ".git") {
28d98dbbe0STiezhu Yang	printf "Warning: can't check if file exists, as this is not a git tree\n";
29b1663d7eSMauro Carvalho Chehab	exit 0;
30b1663d7eSMauro Carvalho Chehab}
31d2656095SMauro Carvalho Chehab
32d2656095SMauro Carvalho ChehabGetOptions(
33d2656095SMauro Carvalho Chehab	'fix' => \$fix,
34b1663d7eSMauro Carvalho Chehab	'warn' => \$warn,
35d2656095SMauro Carvalho Chehab	'h|help|usage' => \$help,
36d2656095SMauro Carvalho Chehab);
37d2656095SMauro Carvalho Chehab
38d2656095SMauro Carvalho Chehabif ($help != 0) {
3940fc3eb0SMauro Carvalho Chehab    print "$scriptname [--help] [--fix]\n";
40d2656095SMauro Carvalho Chehab    exit -1;
41d2656095SMauro Carvalho Chehab}
42d2656095SMauro Carvalho Chehab
43d2656095SMauro Carvalho Chehab# Step 1: find broken references
44d2656095SMauro Carvalho Chehabprint "Finding broken references. This may take a while...  " if ($fix);
45d2656095SMauro Carvalho Chehab
46d2656095SMauro Carvalho Chehabmy %broken_ref;
47d2656095SMauro Carvalho Chehab
48894ee5ffSMauro Carvalho Chehabmy $doc_fix = 0;
49894ee5ffSMauro Carvalho Chehab
50894ee5ffSMauro Carvalho Chehabopen IN, "git grep ':doc:\`' Documentation/|"
51894ee5ffSMauro Carvalho Chehab     or die "Failed to run git grep";
52894ee5ffSMauro Carvalho Chehabwhile (<IN>) {
53894ee5ffSMauro Carvalho Chehab	next if (!m,^([^:]+):.*\:doc\:\`([^\`]+)\`,);
54290d5388SMauro Carvalho Chehab	next if (m,sphinx/,);
55894ee5ffSMauro Carvalho Chehab
56290d5388SMauro Carvalho Chehab	my $file = $1;
57894ee5ffSMauro Carvalho Chehab	my $d = $1;
58894ee5ffSMauro Carvalho Chehab	my $doc_ref = $2;
59894ee5ffSMauro Carvalho Chehab
60894ee5ffSMauro Carvalho Chehab	my $f = $doc_ref;
61894ee5ffSMauro Carvalho Chehab
62894ee5ffSMauro Carvalho Chehab	$d =~ s,(.*/).*,$1,;
63894ee5ffSMauro Carvalho Chehab	$f =~ s,.*\<([^\>]+)\>,$1,;
64894ee5ffSMauro Carvalho Chehab
65290d5388SMauro Carvalho Chehab	if ($f =~ m,^/,) {
66290d5388SMauro Carvalho Chehab		$f = "$f.rst";
67290d5388SMauro Carvalho Chehab		$f =~ s,^/,Documentation/,;
68290d5388SMauro Carvalho Chehab	} else {
69894ee5ffSMauro Carvalho Chehab		$f = "$d$f.rst";
70290d5388SMauro Carvalho Chehab	}
71894ee5ffSMauro Carvalho Chehab
72894ee5ffSMauro Carvalho Chehab	next if (grep -e, glob("$f"));
73894ee5ffSMauro Carvalho Chehab
74894ee5ffSMauro Carvalho Chehab	if ($fix && !$doc_fix) {
75894ee5ffSMauro Carvalho Chehab		print STDERR "\nWARNING: Currently, can't fix broken :doc:`` fields\n";
76894ee5ffSMauro Carvalho Chehab	}
77894ee5ffSMauro Carvalho Chehab	$doc_fix++;
78894ee5ffSMauro Carvalho Chehab
79290d5388SMauro Carvalho Chehab	print STDERR "$file: :doc:`$doc_ref`\n";
80894ee5ffSMauro Carvalho Chehab}
81894ee5ffSMauro Carvalho Chehabclose IN;
82894ee5ffSMauro Carvalho Chehab
83d2656095SMauro Carvalho Chehabopen IN, "git grep 'Documentation/'|"
84d2656095SMauro Carvalho Chehab     or die "Failed to run git grep";
85d2656095SMauro Carvalho Chehabwhile (<IN>) {
86d2656095SMauro Carvalho Chehab	next if (!m/^([^:]+):(.*)/);
87d2656095SMauro Carvalho Chehab
88d2656095SMauro Carvalho Chehab	my $f = $1;
89d2656095SMauro Carvalho Chehab	my $ln = $2;
90d2656095SMauro Carvalho Chehab
91fe3e4b9cSMauro Carvalho Chehab	# On linux-next, discard the Next/ directory
92fe3e4b9cSMauro Carvalho Chehab	next if ($f =~ m,^Next/,);
93fe3e4b9cSMauro Carvalho Chehab
942d69708fSMauro Carvalho Chehab	# Makefiles and scripts contain nasty expressions to parse docs
952d69708fSMauro Carvalho Chehab	next if ($f =~ m/Makefile/ || $f =~ m/\.sh$/);
962d69708fSMauro Carvalho Chehab
97d2656095SMauro Carvalho Chehab	# Skip this script
98d2656095SMauro Carvalho Chehab	next if ($f eq $scriptname);
99d2656095SMauro Carvalho Chehab
100407b584dSMauro Carvalho Chehab	# Ignore the dir where documentation will be built
101407b584dSMauro Carvalho Chehab	next if ($ln =~ m,\b(\S*)Documentation/output,);
102407b584dSMauro Carvalho Chehab
1032d69708fSMauro Carvalho Chehab	if ($ln =~ m,\b(\S*)(Documentation/[A-Za-z0-9\_\.\,\~/\*\[\]\?+-]*)(.*),) {
104d2656095SMauro Carvalho Chehab		my $prefix = $1;
105d2656095SMauro Carvalho Chehab		my $ref = $2;
106d2656095SMauro Carvalho Chehab		my $base = $2;
1072d69708fSMauro Carvalho Chehab		my $extra = $3;
108d2656095SMauro Carvalho Chehab
1092d69708fSMauro Carvalho Chehab		# some file references are like:
1102d69708fSMauro Carvalho Chehab		# /usr/src/linux/Documentation/DMA-{API,mapping}.txt
1112d69708fSMauro Carvalho Chehab		# For now, ignore them
1122d69708fSMauro Carvalho Chehab		next if ($extra =~ m/^{/);
1132d69708fSMauro Carvalho Chehab
1142d69708fSMauro Carvalho Chehab		# Remove footnotes at the end like:
1152d69708fSMauro Carvalho Chehab		# Documentation/devicetree/dt-object-internal.txt[1]
1162d69708fSMauro Carvalho Chehab		$ref =~ s/(txt|rst)\[\d+]$/$1/;
1172d69708fSMauro Carvalho Chehab
1182d69708fSMauro Carvalho Chehab		# Remove ending ']' without any '['
1192d69708fSMauro Carvalho Chehab		$ref =~ s/\].*// if (!($ref =~ m/\[/));
1202d69708fSMauro Carvalho Chehab
1212d69708fSMauro Carvalho Chehab		# Remove puntuation marks at the end
122d2656095SMauro Carvalho Chehab		$ref =~ s/[\,\.]+$//;
123d2656095SMauro Carvalho Chehab
124d2656095SMauro Carvalho Chehab		my $fulref = "$prefix$ref";
125d2656095SMauro Carvalho Chehab
126d2656095SMauro Carvalho Chehab		$fulref =~ s/^(\<file|ref)://;
127d2656095SMauro Carvalho Chehab		$fulref =~ s/^[\'\`]+//;
128d2656095SMauro Carvalho Chehab		$fulref =~ s,^\$\(.*\)/,,;
129d2656095SMauro Carvalho Chehab		$base =~ s,.*/,,;
130d2656095SMauro Carvalho Chehab
131d2656095SMauro Carvalho Chehab		# Remove URL false-positives
132d2656095SMauro Carvalho Chehab		next if ($fulref =~ m/^http/);
133d2656095SMauro Carvalho Chehab
1345d395fa6SMauro Carvalho Chehab		# Remove sched-pelt false-positive
1355d395fa6SMauro Carvalho Chehab		next if ($fulref =~ m,^Documentation/scheduler/sched-pelt$,);
1365d395fa6SMauro Carvalho Chehab
1374ca9bc22SMauro Carvalho Chehab		# Discard some build examples from Documentation/target/tcm_mod_builder.rst
138d25c0634SMauro Carvalho Chehab		next if ($fulref =~ m,mnt/sdb/lio-core-2.6.git/Documentation/target,);
139d25c0634SMauro Carvalho Chehab
140d2656095SMauro Carvalho Chehab		# Check if exists, evaluating wildcards
141d2656095SMauro Carvalho Chehab		next if (grep -e, glob("$ref $fulref"));
142d2656095SMauro Carvalho Chehab
143a78513c6SMauro Carvalho Chehab		# Accept relative Documentation patches for tools/
144a78513c6SMauro Carvalho Chehab		if ($f =~ m/tools/) {
145a78513c6SMauro Carvalho Chehab			my $path = $f;
146a78513c6SMauro Carvalho Chehab			$path =~ s,(.*)/.*,$1,;
1474904aeedSMauro Carvalho Chehab			next if (grep -e, glob("$path/$ref $path/../$ref $path/$fulref"));
148a78513c6SMauro Carvalho Chehab		}
149a78513c6SMauro Carvalho Chehab
150aeaacbfeSMauro Carvalho Chehab		# Discard known false-positives
151aeaacbfeSMauro Carvalho Chehab		if (defined($false_positives{$f})) {
152aeaacbfeSMauro Carvalho Chehab			next if ($false_positives{$f} eq $fulref);
153aeaacbfeSMauro Carvalho Chehab		}
154aeaacbfeSMauro Carvalho Chehab
155d2656095SMauro Carvalho Chehab		if ($fix) {
156be600e5aSMauro Carvalho Chehab			if (!($ref =~ m/(scripts|Kconfig|Kbuild)/)) {
157d2656095SMauro Carvalho Chehab				$broken_ref{$ref}++;
158d2656095SMauro Carvalho Chehab			}
159b1663d7eSMauro Carvalho Chehab		} elsif ($warn) {
160b1663d7eSMauro Carvalho Chehab			print STDERR "Warning: $f references a file that doesn't exist: $fulref\n";
161d2656095SMauro Carvalho Chehab		} else {
162d2656095SMauro Carvalho Chehab			print STDERR "$f: $fulref\n";
163d2656095SMauro Carvalho Chehab		}
164d2656095SMauro Carvalho Chehab	}
165d2656095SMauro Carvalho Chehab}
166894ee5ffSMauro Carvalho Chehabclose IN;
167d2656095SMauro Carvalho Chehab
168d2656095SMauro Carvalho Chehabexit 0 if (!$fix);
169d2656095SMauro Carvalho Chehab
170d2656095SMauro Carvalho Chehab# Step 2: Seek for file name alternatives
171d2656095SMauro Carvalho Chehabprint "Auto-fixing broken references. Please double-check the results\n";
172d2656095SMauro Carvalho Chehab
173d2656095SMauro Carvalho Chehabforeach my $ref (keys %broken_ref) {
174d2656095SMauro Carvalho Chehab	my $new =$ref;
175d2656095SMauro Carvalho Chehab
1769e78e7fcSMauro Carvalho Chehab	my $basedir = ".";
1779e78e7fcSMauro Carvalho Chehab	# On translations, only seek inside the translations directory
1789e78e7fcSMauro Carvalho Chehab	$basedir  = $1 if ($ref =~ m,(Documentation/translations/[^/]+),);
1799e78e7fcSMauro Carvalho Chehab
180d2656095SMauro Carvalho Chehab	# get just the basename
181d2656095SMauro Carvalho Chehab	$new =~ s,.*/,,;
182d2656095SMauro Carvalho Chehab
183d2656095SMauro Carvalho Chehab	my $f="";
184d2656095SMauro Carvalho Chehab
185be600e5aSMauro Carvalho Chehab	# usual reason for breakage: DT file moved around
186be600e5aSMauro Carvalho Chehab	if ($ref =~ /devicetree/) {
1870ca862e6SMauro Carvalho Chehab		# usual reason for breakage: DT file renamed to .yaml
1880ca862e6SMauro Carvalho Chehab		if (!$f) {
1890ca862e6SMauro Carvalho Chehab			my $new_ref = $ref;
1900ca862e6SMauro Carvalho Chehab			$new_ref =~ s/\.txt$/.yaml/;
1910ca862e6SMauro Carvalho Chehab			$f=$new_ref if (-f $new_ref);
1920ca862e6SMauro Carvalho Chehab		}
1930ca862e6SMauro Carvalho Chehab
1940ca862e6SMauro Carvalho Chehab		if (!$f) {
195be600e5aSMauro Carvalho Chehab			my $search = $new;
196be600e5aSMauro Carvalho Chehab			$search =~ s,^.*/,,;
197be600e5aSMauro Carvalho Chehab			$f = qx(find Documentation/devicetree/ -iname "*$search*") if ($search);
198be600e5aSMauro Carvalho Chehab			if (!$f) {
199be600e5aSMauro Carvalho Chehab				# Manufacturer name may have changed
200be600e5aSMauro Carvalho Chehab				$search =~ s/^.*,//;
201be600e5aSMauro Carvalho Chehab				$f = qx(find Documentation/devicetree/ -iname "*$search*") if ($search);
202be600e5aSMauro Carvalho Chehab			}
203be600e5aSMauro Carvalho Chehab		}
2040ca862e6SMauro Carvalho Chehab	}
205d2656095SMauro Carvalho Chehab
206d2656095SMauro Carvalho Chehab	# usual reason for breakage: file renamed to .rst
207d2656095SMauro Carvalho Chehab	if (!$f) {
208d2656095SMauro Carvalho Chehab		$new =~ s/\.txt$/.rst/;
2099e78e7fcSMauro Carvalho Chehab		$f=qx(find $basedir -iname $new) if ($new);
210d2656095SMauro Carvalho Chehab	}
211d2656095SMauro Carvalho Chehab
212e1f319feSMauro Carvalho Chehab	# usual reason for breakage: use dash or underline
213e1f319feSMauro Carvalho Chehab	if (!$f) {
214e1f319feSMauro Carvalho Chehab		$new =~ s/[-_]/[-_]/g;
2159e78e7fcSMauro Carvalho Chehab		$f=qx(find $basedir -iname $new) if ($new);
216e1f319feSMauro Carvalho Chehab	}
217e1f319feSMauro Carvalho Chehab
218be600e5aSMauro Carvalho Chehab	# Wild guess: seek for the same name on another place
219be600e5aSMauro Carvalho Chehab	if (!$f) {
2209e78e7fcSMauro Carvalho Chehab		$f = qx(find $basedir -iname $new) if ($new);
221be600e5aSMauro Carvalho Chehab	}
222be600e5aSMauro Carvalho Chehab
223d2656095SMauro Carvalho Chehab	my @find = split /\s+/, $f;
224d2656095SMauro Carvalho Chehab
225d2656095SMauro Carvalho Chehab	if (!$f) {
226d2656095SMauro Carvalho Chehab		print STDERR "ERROR: Didn't find a replacement for $ref\n";
227d2656095SMauro Carvalho Chehab	} elsif (scalar(@find) > 1) {
228d2656095SMauro Carvalho Chehab		print STDERR "WARNING: Won't auto-replace, as found multiple files close to $ref:\n";
229d2656095SMauro Carvalho Chehab		foreach my $j (@find) {
230d2656095SMauro Carvalho Chehab			$j =~ s,^./,,;
231d2656095SMauro Carvalho Chehab			print STDERR "    $j\n";
232d2656095SMauro Carvalho Chehab		}
233d2656095SMauro Carvalho Chehab	} else {
234d2656095SMauro Carvalho Chehab		$f = $find[0];
235d2656095SMauro Carvalho Chehab		$f =~ s,^./,,;
236d2656095SMauro Carvalho Chehab		print "INFO: Replacing $ref to $f\n";
237d2656095SMauro Carvalho Chehab		foreach my $j (qx(git grep -l $ref)) {
238d2656095SMauro Carvalho Chehab			qx(sed "s\@$ref\@$f\@g" -i $j);
239d2656095SMauro Carvalho Chehab		}
240d2656095SMauro Carvalho Chehab	}
241d2656095SMauro Carvalho Chehab}
242