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;
25*b1663d7eSMauro Carvalho Chehabmy $warn = 0;
26*b1663d7eSMauro Carvalho Chehab
27*b1663d7eSMauro Carvalho Chehabif (! -d ".git") {
28*b1663d7eSMauro Carvalho Chehab	printf "Warning: can't check if file exists, as this is not a git tree";
29*b1663d7eSMauro Carvalho Chehab	exit 0;
30*b1663d7eSMauro Carvalho Chehab}
31d2656095SMauro Carvalho Chehab
32d2656095SMauro Carvalho ChehabGetOptions(
33d2656095SMauro Carvalho Chehab	'fix' => \$fix,
34*b1663d7eSMauro Carvalho Chehab	'warn' => \$warn,
35d2656095SMauro Carvalho Chehab	'h|help|usage' => \$help,
36d2656095SMauro Carvalho Chehab);
37d2656095SMauro Carvalho Chehab
38d2656095SMauro Carvalho Chehabif ($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\:\`([^\`]+)\`,);
54894ee5ffSMauro Carvalho Chehab
55894ee5ffSMauro Carvalho Chehab	my $d = $1;
56894ee5ffSMauro Carvalho Chehab	my $doc_ref = $2;
57894ee5ffSMauro Carvalho Chehab
58894ee5ffSMauro Carvalho Chehab	my $f = $doc_ref;
59894ee5ffSMauro Carvalho Chehab
60894ee5ffSMauro Carvalho Chehab	$d =~ s,(.*/).*,$1,;
61894ee5ffSMauro Carvalho Chehab	$f =~ s,.*\<([^\>]+)\>,$1,;
62894ee5ffSMauro Carvalho Chehab
63894ee5ffSMauro Carvalho Chehab	$f ="$d$f.rst";
64894ee5ffSMauro Carvalho Chehab
65894ee5ffSMauro Carvalho Chehab	next if (grep -e, glob("$f"));
66894ee5ffSMauro Carvalho Chehab
67894ee5ffSMauro Carvalho Chehab	if ($fix && !$doc_fix) {
68894ee5ffSMauro Carvalho Chehab		print STDERR "\nWARNING: Currently, can't fix broken :doc:`` fields\n";
69894ee5ffSMauro Carvalho Chehab	}
70894ee5ffSMauro Carvalho Chehab	$doc_fix++;
71894ee5ffSMauro Carvalho Chehab
72894ee5ffSMauro Carvalho Chehab	print STDERR "$f: :doc:`$doc_ref`\n";
73894ee5ffSMauro Carvalho Chehab}
74894ee5ffSMauro Carvalho Chehabclose IN;
75894ee5ffSMauro Carvalho Chehab
76d2656095SMauro Carvalho Chehabopen IN, "git grep 'Documentation/'|"
77d2656095SMauro Carvalho Chehab     or die "Failed to run git grep";
78d2656095SMauro Carvalho Chehabwhile (<IN>) {
79d2656095SMauro Carvalho Chehab	next if (!m/^([^:]+):(.*)/);
80d2656095SMauro Carvalho Chehab
81d2656095SMauro Carvalho Chehab	my $f = $1;
82d2656095SMauro Carvalho Chehab	my $ln = $2;
83d2656095SMauro Carvalho Chehab
84fe3e4b9cSMauro Carvalho Chehab	# On linux-next, discard the Next/ directory
85fe3e4b9cSMauro Carvalho Chehab	next if ($f =~ m,^Next/,);
86fe3e4b9cSMauro Carvalho Chehab
872d69708fSMauro Carvalho Chehab	# Makefiles and scripts contain nasty expressions to parse docs
882d69708fSMauro Carvalho Chehab	next if ($f =~ m/Makefile/ || $f =~ m/\.sh$/);
892d69708fSMauro Carvalho Chehab
90d2656095SMauro Carvalho Chehab	# Skip this script
91d2656095SMauro Carvalho Chehab	next if ($f eq $scriptname);
92d2656095SMauro Carvalho Chehab
932d69708fSMauro Carvalho Chehab	if ($ln =~ m,\b(\S*)(Documentation/[A-Za-z0-9\_\.\,\~/\*\[\]\?+-]*)(.*),) {
94d2656095SMauro Carvalho Chehab		my $prefix = $1;
95d2656095SMauro Carvalho Chehab		my $ref = $2;
96d2656095SMauro Carvalho Chehab		my $base = $2;
972d69708fSMauro Carvalho Chehab		my $extra = $3;
98d2656095SMauro Carvalho Chehab
992d69708fSMauro Carvalho Chehab		# some file references are like:
1002d69708fSMauro Carvalho Chehab		# /usr/src/linux/Documentation/DMA-{API,mapping}.txt
1012d69708fSMauro Carvalho Chehab		# For now, ignore them
1022d69708fSMauro Carvalho Chehab		next if ($extra =~ m/^{/);
1032d69708fSMauro Carvalho Chehab
1042d69708fSMauro Carvalho Chehab		# Remove footnotes at the end like:
1052d69708fSMauro Carvalho Chehab		# Documentation/devicetree/dt-object-internal.txt[1]
1062d69708fSMauro Carvalho Chehab		$ref =~ s/(txt|rst)\[\d+]$/$1/;
1072d69708fSMauro Carvalho Chehab
1082d69708fSMauro Carvalho Chehab		# Remove ending ']' without any '['
1092d69708fSMauro Carvalho Chehab		$ref =~ s/\].*// if (!($ref =~ m/\[/));
1102d69708fSMauro Carvalho Chehab
1112d69708fSMauro Carvalho Chehab		# Remove puntuation marks at the end
112d2656095SMauro Carvalho Chehab		$ref =~ s/[\,\.]+$//;
113d2656095SMauro Carvalho Chehab
114d2656095SMauro Carvalho Chehab		my $fulref = "$prefix$ref";
115d2656095SMauro Carvalho Chehab
116d2656095SMauro Carvalho Chehab		$fulref =~ s/^(\<file|ref)://;
117d2656095SMauro Carvalho Chehab		$fulref =~ s/^[\'\`]+//;
118d2656095SMauro Carvalho Chehab		$fulref =~ s,^\$\(.*\)/,,;
119d2656095SMauro Carvalho Chehab		$base =~ s,.*/,,;
120d2656095SMauro Carvalho Chehab
121d2656095SMauro Carvalho Chehab		# Remove URL false-positives
122d2656095SMauro Carvalho Chehab		next if ($fulref =~ m/^http/);
123d2656095SMauro Carvalho Chehab
1245d395fa6SMauro Carvalho Chehab		# Remove sched-pelt false-positive
1255d395fa6SMauro Carvalho Chehab		next if ($fulref =~ m,^Documentation/scheduler/sched-pelt$,);
1265d395fa6SMauro Carvalho Chehab
127d25c0634SMauro Carvalho Chehab		# Discard some build examples from Documentation/target/tcm_mod_builder.txt
128d25c0634SMauro Carvalho Chehab		next if ($fulref =~ m,mnt/sdb/lio-core-2.6.git/Documentation/target,);
129d25c0634SMauro Carvalho Chehab
130d2656095SMauro Carvalho Chehab		# Check if exists, evaluating wildcards
131d2656095SMauro Carvalho Chehab		next if (grep -e, glob("$ref $fulref"));
132d2656095SMauro Carvalho Chehab
133a78513c6SMauro Carvalho Chehab		# Accept relative Documentation patches for tools/
134a78513c6SMauro Carvalho Chehab		if ($f =~ m/tools/) {
135a78513c6SMauro Carvalho Chehab			my $path = $f;
136a78513c6SMauro Carvalho Chehab			$path =~ s,(.*)/.*,$1,;
1374904aeedSMauro Carvalho Chehab			next if (grep -e, glob("$path/$ref $path/../$ref $path/$fulref"));
138a78513c6SMauro Carvalho Chehab		}
139a78513c6SMauro Carvalho Chehab
140aeaacbfeSMauro Carvalho Chehab		# Discard known false-positives
141aeaacbfeSMauro Carvalho Chehab		if (defined($false_positives{$f})) {
142aeaacbfeSMauro Carvalho Chehab			next if ($false_positives{$f} eq $fulref);
143aeaacbfeSMauro Carvalho Chehab		}
144aeaacbfeSMauro Carvalho Chehab
145d2656095SMauro Carvalho Chehab		if ($fix) {
146be600e5aSMauro Carvalho Chehab			if (!($ref =~ m/(scripts|Kconfig|Kbuild)/)) {
147d2656095SMauro Carvalho Chehab				$broken_ref{$ref}++;
148d2656095SMauro Carvalho Chehab			}
149*b1663d7eSMauro Carvalho Chehab		} elsif ($warn) {
150*b1663d7eSMauro Carvalho Chehab			print STDERR "Warning: $f references a file that doesn't exist: $fulref\n";
151d2656095SMauro Carvalho Chehab		} else {
152d2656095SMauro Carvalho Chehab			print STDERR "$f: $fulref\n";
153d2656095SMauro Carvalho Chehab		}
154d2656095SMauro Carvalho Chehab	}
155d2656095SMauro Carvalho Chehab}
156894ee5ffSMauro Carvalho Chehabclose IN;
157d2656095SMauro Carvalho Chehab
158d2656095SMauro Carvalho Chehabexit 0 if (!$fix);
159d2656095SMauro Carvalho Chehab
160d2656095SMauro Carvalho Chehab# Step 2: Seek for file name alternatives
161d2656095SMauro Carvalho Chehabprint "Auto-fixing broken references. Please double-check the results\n";
162d2656095SMauro Carvalho Chehab
163d2656095SMauro Carvalho Chehabforeach my $ref (keys %broken_ref) {
164d2656095SMauro Carvalho Chehab	my $new =$ref;
165d2656095SMauro Carvalho Chehab
1669e78e7fcSMauro Carvalho Chehab	my $basedir = ".";
1679e78e7fcSMauro Carvalho Chehab	# On translations, only seek inside the translations directory
1689e78e7fcSMauro Carvalho Chehab	$basedir  = $1 if ($ref =~ m,(Documentation/translations/[^/]+),);
1699e78e7fcSMauro Carvalho Chehab
170d2656095SMauro Carvalho Chehab	# get just the basename
171d2656095SMauro Carvalho Chehab	$new =~ s,.*/,,;
172d2656095SMauro Carvalho Chehab
173d2656095SMauro Carvalho Chehab	my $f="";
174d2656095SMauro Carvalho Chehab
175be600e5aSMauro Carvalho Chehab	# usual reason for breakage: DT file moved around
176be600e5aSMauro Carvalho Chehab	if ($ref =~ /devicetree/) {
1770ca862e6SMauro Carvalho Chehab		# usual reason for breakage: DT file renamed to .yaml
1780ca862e6SMauro Carvalho Chehab		if (!$f) {
1790ca862e6SMauro Carvalho Chehab			my $new_ref = $ref;
1800ca862e6SMauro Carvalho Chehab			$new_ref =~ s/\.txt$/.yaml/;
1810ca862e6SMauro Carvalho Chehab			$f=$new_ref if (-f $new_ref);
1820ca862e6SMauro Carvalho Chehab		}
1830ca862e6SMauro Carvalho Chehab
1840ca862e6SMauro Carvalho Chehab		if (!$f) {
185be600e5aSMauro Carvalho Chehab			my $search = $new;
186be600e5aSMauro Carvalho Chehab			$search =~ s,^.*/,,;
187be600e5aSMauro Carvalho Chehab			$f = qx(find Documentation/devicetree/ -iname "*$search*") if ($search);
188be600e5aSMauro Carvalho Chehab			if (!$f) {
189be600e5aSMauro Carvalho Chehab				# Manufacturer name may have changed
190be600e5aSMauro Carvalho Chehab				$search =~ s/^.*,//;
191be600e5aSMauro Carvalho Chehab				$f = qx(find Documentation/devicetree/ -iname "*$search*") if ($search);
192be600e5aSMauro Carvalho Chehab			}
193be600e5aSMauro Carvalho Chehab		}
1940ca862e6SMauro Carvalho Chehab	}
195d2656095SMauro Carvalho Chehab
196d2656095SMauro Carvalho Chehab	# usual reason for breakage: file renamed to .rst
197d2656095SMauro Carvalho Chehab	if (!$f) {
198d2656095SMauro Carvalho Chehab		$new =~ s/\.txt$/.rst/;
1999e78e7fcSMauro Carvalho Chehab		$f=qx(find $basedir -iname $new) if ($new);
200d2656095SMauro Carvalho Chehab	}
201d2656095SMauro Carvalho Chehab
202e1f319feSMauro Carvalho Chehab	# usual reason for breakage: use dash or underline
203e1f319feSMauro Carvalho Chehab	if (!$f) {
204e1f319feSMauro Carvalho Chehab		$new =~ s/[-_]/[-_]/g;
2059e78e7fcSMauro Carvalho Chehab		$f=qx(find $basedir -iname $new) if ($new);
206e1f319feSMauro Carvalho Chehab	}
207e1f319feSMauro Carvalho Chehab
208be600e5aSMauro Carvalho Chehab	# Wild guess: seek for the same name on another place
209be600e5aSMauro Carvalho Chehab	if (!$f) {
2109e78e7fcSMauro Carvalho Chehab		$f = qx(find $basedir -iname $new) if ($new);
211be600e5aSMauro Carvalho Chehab	}
212be600e5aSMauro Carvalho Chehab
213d2656095SMauro Carvalho Chehab	my @find = split /\s+/, $f;
214d2656095SMauro Carvalho Chehab
215d2656095SMauro Carvalho Chehab	if (!$f) {
216d2656095SMauro Carvalho Chehab		print STDERR "ERROR: Didn't find a replacement for $ref\n";
217d2656095SMauro Carvalho Chehab	} elsif (scalar(@find) > 1) {
218d2656095SMauro Carvalho Chehab		print STDERR "WARNING: Won't auto-replace, as found multiple files close to $ref:\n";
219d2656095SMauro Carvalho Chehab		foreach my $j (@find) {
220d2656095SMauro Carvalho Chehab			$j =~ s,^./,,;
221d2656095SMauro Carvalho Chehab			print STDERR "    $j\n";
222d2656095SMauro Carvalho Chehab		}
223d2656095SMauro Carvalho Chehab	} else {
224d2656095SMauro Carvalho Chehab		$f = $find[0];
225d2656095SMauro Carvalho Chehab		$f =~ s,^./,,;
226d2656095SMauro Carvalho Chehab		print "INFO: Replacing $ref to $f\n";
227d2656095SMauro Carvalho Chehab		foreach my $j (qx(git grep -l $ref)) {
228d2656095SMauro Carvalho Chehab			qx(sed "s\@$ref\@$f\@g" -i $j);
229d2656095SMauro Carvalho Chehab		}
230d2656095SMauro Carvalho Chehab	}
231d2656095SMauro Carvalho Chehab}
232