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 = (
15aeaacbfeSMauro Carvalho Chehab	"Documentation/scsi/scsi_mid_low_api.txt" => "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 $fix = 0;
25d2656095SMauro Carvalho Chehab
26d2656095SMauro Carvalho ChehabGetOptions(
27d2656095SMauro Carvalho Chehab	'fix' => \$fix,
28d2656095SMauro Carvalho Chehab	'h|help|usage' => \$help,
29d2656095SMauro Carvalho Chehab);
30d2656095SMauro Carvalho Chehab
31d2656095SMauro Carvalho Chehabif ($help != 0) {
3240fc3eb0SMauro Carvalho Chehab    print "$scriptname [--help] [--fix]\n";
33d2656095SMauro Carvalho Chehab    exit -1;
34d2656095SMauro Carvalho Chehab}
35d2656095SMauro Carvalho Chehab
36d2656095SMauro Carvalho Chehab# Step 1: find broken references
37d2656095SMauro Carvalho Chehabprint "Finding broken references. This may take a while...  " if ($fix);
38d2656095SMauro Carvalho Chehab
39d2656095SMauro Carvalho Chehabmy %broken_ref;
40d2656095SMauro Carvalho Chehab
41894ee5ffSMauro Carvalho Chehabmy $doc_fix = 0;
42894ee5ffSMauro Carvalho Chehab
43894ee5ffSMauro Carvalho Chehabopen IN, "git grep ':doc:\`' Documentation/|"
44894ee5ffSMauro Carvalho Chehab     or die "Failed to run git grep";
45894ee5ffSMauro Carvalho Chehabwhile (<IN>) {
46894ee5ffSMauro Carvalho Chehab	next if (!m,^([^:]+):.*\:doc\:\`([^\`]+)\`,);
47894ee5ffSMauro Carvalho Chehab
48894ee5ffSMauro Carvalho Chehab	my $d = $1;
49894ee5ffSMauro Carvalho Chehab	my $doc_ref = $2;
50894ee5ffSMauro Carvalho Chehab
51894ee5ffSMauro Carvalho Chehab	my $f = $doc_ref;
52894ee5ffSMauro Carvalho Chehab
53894ee5ffSMauro Carvalho Chehab	$d =~ s,(.*/).*,$1,;
54894ee5ffSMauro Carvalho Chehab	$f =~ s,.*\<([^\>]+)\>,$1,;
55894ee5ffSMauro Carvalho Chehab
56894ee5ffSMauro Carvalho Chehab	$f ="$d$f.rst";
57894ee5ffSMauro Carvalho Chehab
58894ee5ffSMauro Carvalho Chehab	next if (grep -e, glob("$f"));
59894ee5ffSMauro Carvalho Chehab
60894ee5ffSMauro Carvalho Chehab	if ($fix && !$doc_fix) {
61894ee5ffSMauro Carvalho Chehab		print STDERR "\nWARNING: Currently, can't fix broken :doc:`` fields\n";
62894ee5ffSMauro Carvalho Chehab	}
63894ee5ffSMauro Carvalho Chehab	$doc_fix++;
64894ee5ffSMauro Carvalho Chehab
65894ee5ffSMauro Carvalho Chehab	print STDERR "$f: :doc:`$doc_ref`\n";
66894ee5ffSMauro Carvalho Chehab}
67894ee5ffSMauro Carvalho Chehabclose IN;
68894ee5ffSMauro Carvalho Chehab
69d2656095SMauro Carvalho Chehabopen IN, "git grep 'Documentation/'|"
70d2656095SMauro Carvalho Chehab     or die "Failed to run git grep";
71d2656095SMauro Carvalho Chehabwhile (<IN>) {
72d2656095SMauro Carvalho Chehab	next if (!m/^([^:]+):(.*)/);
73d2656095SMauro Carvalho Chehab
74d2656095SMauro Carvalho Chehab	my $f = $1;
75d2656095SMauro Carvalho Chehab	my $ln = $2;
76d2656095SMauro Carvalho Chehab
77fe3e4b9cSMauro Carvalho Chehab	# On linux-next, discard the Next/ directory
78fe3e4b9cSMauro Carvalho Chehab	next if ($f =~ m,^Next/,);
79fe3e4b9cSMauro Carvalho Chehab
802d69708fSMauro Carvalho Chehab	# Makefiles and scripts contain nasty expressions to parse docs
812d69708fSMauro Carvalho Chehab	next if ($f =~ m/Makefile/ || $f =~ m/\.sh$/);
822d69708fSMauro Carvalho Chehab
83d2656095SMauro Carvalho Chehab	# Skip this script
84d2656095SMauro Carvalho Chehab	next if ($f eq $scriptname);
85d2656095SMauro Carvalho Chehab
862d69708fSMauro Carvalho Chehab	if ($ln =~ m,\b(\S*)(Documentation/[A-Za-z0-9\_\.\,\~/\*\[\]\?+-]*)(.*),) {
87d2656095SMauro Carvalho Chehab		my $prefix = $1;
88d2656095SMauro Carvalho Chehab		my $ref = $2;
89d2656095SMauro Carvalho Chehab		my $base = $2;
902d69708fSMauro Carvalho Chehab		my $extra = $3;
91d2656095SMauro Carvalho Chehab
922d69708fSMauro Carvalho Chehab		# some file references are like:
932d69708fSMauro Carvalho Chehab		# /usr/src/linux/Documentation/DMA-{API,mapping}.txt
942d69708fSMauro Carvalho Chehab		# For now, ignore them
952d69708fSMauro Carvalho Chehab		next if ($extra =~ m/^{/);
962d69708fSMauro Carvalho Chehab
972d69708fSMauro Carvalho Chehab		# Remove footnotes at the end like:
982d69708fSMauro Carvalho Chehab		# Documentation/devicetree/dt-object-internal.txt[1]
992d69708fSMauro Carvalho Chehab		$ref =~ s/(txt|rst)\[\d+]$/$1/;
1002d69708fSMauro Carvalho Chehab
1012d69708fSMauro Carvalho Chehab		# Remove ending ']' without any '['
1022d69708fSMauro Carvalho Chehab		$ref =~ s/\].*// if (!($ref =~ m/\[/));
1032d69708fSMauro Carvalho Chehab
1042d69708fSMauro Carvalho Chehab		# Remove puntuation marks at the end
105d2656095SMauro Carvalho Chehab		$ref =~ s/[\,\.]+$//;
106d2656095SMauro Carvalho Chehab
107d2656095SMauro Carvalho Chehab		my $fulref = "$prefix$ref";
108d2656095SMauro Carvalho Chehab
109d2656095SMauro Carvalho Chehab		$fulref =~ s/^(\<file|ref)://;
110d2656095SMauro Carvalho Chehab		$fulref =~ s/^[\'\`]+//;
111d2656095SMauro Carvalho Chehab		$fulref =~ s,^\$\(.*\)/,,;
112d2656095SMauro Carvalho Chehab		$base =~ s,.*/,,;
113d2656095SMauro Carvalho Chehab
114d2656095SMauro Carvalho Chehab		# Remove URL false-positives
115d2656095SMauro Carvalho Chehab		next if ($fulref =~ m/^http/);
116d2656095SMauro Carvalho Chehab
1175d395fa6SMauro Carvalho Chehab		# Remove sched-pelt false-positive
1185d395fa6SMauro Carvalho Chehab		next if ($fulref =~ m,^Documentation/scheduler/sched-pelt$,);
1195d395fa6SMauro Carvalho Chehab
120d25c0634SMauro Carvalho Chehab		# Discard some build examples from Documentation/target/tcm_mod_builder.txt
121d25c0634SMauro Carvalho Chehab		next if ($fulref =~ m,mnt/sdb/lio-core-2.6.git/Documentation/target,);
122d25c0634SMauro Carvalho Chehab
123d2656095SMauro Carvalho Chehab		# Check if exists, evaluating wildcards
124d2656095SMauro Carvalho Chehab		next if (grep -e, glob("$ref $fulref"));
125d2656095SMauro Carvalho Chehab
126a78513c6SMauro Carvalho Chehab		# Accept relative Documentation patches for tools/
127a78513c6SMauro Carvalho Chehab		if ($f =~ m/tools/) {
128a78513c6SMauro Carvalho Chehab			my $path = $f;
129a78513c6SMauro Carvalho Chehab			$path =~ s,(.*)/.*,$1,;
1304904aeedSMauro Carvalho Chehab			next if (grep -e, glob("$path/$ref $path/../$ref $path/$fulref"));
131a78513c6SMauro Carvalho Chehab		}
132a78513c6SMauro Carvalho Chehab
133aeaacbfeSMauro Carvalho Chehab		# Discard known false-positives
134aeaacbfeSMauro Carvalho Chehab		if (defined($false_positives{$f})) {
135aeaacbfeSMauro Carvalho Chehab			next if ($false_positives{$f} eq $fulref);
136aeaacbfeSMauro Carvalho Chehab		}
137aeaacbfeSMauro Carvalho Chehab
138d2656095SMauro Carvalho Chehab		if ($fix) {
139be600e5aSMauro Carvalho Chehab			if (!($ref =~ m/(scripts|Kconfig|Kbuild)/)) {
140d2656095SMauro Carvalho Chehab				$broken_ref{$ref}++;
141d2656095SMauro Carvalho Chehab			}
142d2656095SMauro Carvalho Chehab		} else {
143d2656095SMauro Carvalho Chehab			print STDERR "$f: $fulref\n";
144d2656095SMauro Carvalho Chehab		}
145d2656095SMauro Carvalho Chehab	}
146d2656095SMauro Carvalho Chehab}
147894ee5ffSMauro Carvalho Chehabclose IN;
148d2656095SMauro Carvalho Chehab
149d2656095SMauro Carvalho Chehabexit 0 if (!$fix);
150d2656095SMauro Carvalho Chehab
151d2656095SMauro Carvalho Chehab# Step 2: Seek for file name alternatives
152d2656095SMauro Carvalho Chehabprint "Auto-fixing broken references. Please double-check the results\n";
153d2656095SMauro Carvalho Chehab
154d2656095SMauro Carvalho Chehabforeach my $ref (keys %broken_ref) {
155d2656095SMauro Carvalho Chehab	my $new =$ref;
156d2656095SMauro Carvalho Chehab
1579e78e7fcSMauro Carvalho Chehab	my $basedir = ".";
1589e78e7fcSMauro Carvalho Chehab	# On translations, only seek inside the translations directory
1599e78e7fcSMauro Carvalho Chehab	$basedir  = $1 if ($ref =~ m,(Documentation/translations/[^/]+),);
1609e78e7fcSMauro Carvalho Chehab
161d2656095SMauro Carvalho Chehab	# get just the basename
162d2656095SMauro Carvalho Chehab	$new =~ s,.*/,,;
163d2656095SMauro Carvalho Chehab
164d2656095SMauro Carvalho Chehab	my $f="";
165d2656095SMauro Carvalho Chehab
166be600e5aSMauro Carvalho Chehab	# usual reason for breakage: DT file moved around
167be600e5aSMauro Carvalho Chehab	if ($ref =~ /devicetree/) {
168*0ca862e6SMauro Carvalho Chehab		# usual reason for breakage: DT file renamed to .yaml
169*0ca862e6SMauro Carvalho Chehab		if (!$f) {
170*0ca862e6SMauro Carvalho Chehab			my $new_ref = $ref;
171*0ca862e6SMauro Carvalho Chehab			$new_ref =~ s/\.txt$/.yaml/;
172*0ca862e6SMauro Carvalho Chehab			$f=$new_ref if (-f $new_ref);
173*0ca862e6SMauro Carvalho Chehab		}
174*0ca862e6SMauro Carvalho Chehab
175*0ca862e6SMauro Carvalho Chehab		if (!$f) {
176be600e5aSMauro Carvalho Chehab			my $search = $new;
177be600e5aSMauro Carvalho Chehab			$search =~ s,^.*/,,;
178be600e5aSMauro Carvalho Chehab			$f = qx(find Documentation/devicetree/ -iname "*$search*") if ($search);
179be600e5aSMauro Carvalho Chehab			if (!$f) {
180be600e5aSMauro Carvalho Chehab				# Manufacturer name may have changed
181be600e5aSMauro Carvalho Chehab				$search =~ s/^.*,//;
182be600e5aSMauro Carvalho Chehab				$f = qx(find Documentation/devicetree/ -iname "*$search*") if ($search);
183be600e5aSMauro Carvalho Chehab			}
184be600e5aSMauro Carvalho Chehab		}
185*0ca862e6SMauro Carvalho Chehab	}
186d2656095SMauro Carvalho Chehab
187d2656095SMauro Carvalho Chehab	# usual reason for breakage: file renamed to .rst
188d2656095SMauro Carvalho Chehab	if (!$f) {
189d2656095SMauro Carvalho Chehab		$new =~ s/\.txt$/.rst/;
1909e78e7fcSMauro Carvalho Chehab		$f=qx(find $basedir -iname $new) if ($new);
191d2656095SMauro Carvalho Chehab	}
192d2656095SMauro Carvalho Chehab
193e1f319feSMauro Carvalho Chehab	# usual reason for breakage: use dash or underline
194e1f319feSMauro Carvalho Chehab	if (!$f) {
195e1f319feSMauro Carvalho Chehab		$new =~ s/[-_]/[-_]/g;
1969e78e7fcSMauro Carvalho Chehab		$f=qx(find $basedir -iname $new) if ($new);
197e1f319feSMauro Carvalho Chehab	}
198e1f319feSMauro Carvalho Chehab
199be600e5aSMauro Carvalho Chehab	# Wild guess: seek for the same name on another place
200be600e5aSMauro Carvalho Chehab	if (!$f) {
2019e78e7fcSMauro Carvalho Chehab		$f = qx(find $basedir -iname $new) if ($new);
202be600e5aSMauro Carvalho Chehab	}
203be600e5aSMauro Carvalho Chehab
204d2656095SMauro Carvalho Chehab	my @find = split /\s+/, $f;
205d2656095SMauro Carvalho Chehab
206d2656095SMauro Carvalho Chehab	if (!$f) {
207d2656095SMauro Carvalho Chehab		print STDERR "ERROR: Didn't find a replacement for $ref\n";
208d2656095SMauro Carvalho Chehab	} elsif (scalar(@find) > 1) {
209d2656095SMauro Carvalho Chehab		print STDERR "WARNING: Won't auto-replace, as found multiple files close to $ref:\n";
210d2656095SMauro Carvalho Chehab		foreach my $j (@find) {
211d2656095SMauro Carvalho Chehab			$j =~ s,^./,,;
212d2656095SMauro Carvalho Chehab			print STDERR "    $j\n";
213d2656095SMauro Carvalho Chehab		}
214d2656095SMauro Carvalho Chehab	} else {
215d2656095SMauro Carvalho Chehab		$f = $find[0];
216d2656095SMauro Carvalho Chehab		$f =~ s,^./,,;
217d2656095SMauro Carvalho Chehab		print "INFO: Replacing $ref to $f\n";
218d2656095SMauro Carvalho Chehab		foreach my $j (qx(git grep -l $ref)) {
219d2656095SMauro Carvalho Chehab			qx(sed "s\@$ref\@$f\@g" -i $j);
220d2656095SMauro Carvalho Chehab		}
221d2656095SMauro Carvalho Chehab	}
222d2656095SMauro Carvalho Chehab}
223