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
11d2656095SMauro Carvalho Chehabmy $scriptname = $0;
12d2656095SMauro Carvalho Chehab$scriptname =~ s,.*/([^/]+/),$1,;
13d2656095SMauro Carvalho Chehab
14d2656095SMauro Carvalho Chehab# Parse arguments
15d2656095SMauro Carvalho Chehabmy $help = 0;
16d2656095SMauro Carvalho Chehabmy $fix = 0;
17d2656095SMauro Carvalho Chehab
18d2656095SMauro Carvalho ChehabGetOptions(
19d2656095SMauro Carvalho Chehab	'fix' => \$fix,
20d2656095SMauro Carvalho Chehab	'h|help|usage' => \$help,
21d2656095SMauro Carvalho Chehab);
22d2656095SMauro Carvalho Chehab
23d2656095SMauro Carvalho Chehabif ($help != 0) {
2440fc3eb0SMauro Carvalho Chehab    print "$scriptname [--help] [--fix]\n";
25d2656095SMauro Carvalho Chehab    exit -1;
26d2656095SMauro Carvalho Chehab}
27d2656095SMauro Carvalho Chehab
28d2656095SMauro Carvalho Chehab# Step 1: find broken references
29d2656095SMauro Carvalho Chehabprint "Finding broken references. This may take a while...  " if ($fix);
30d2656095SMauro Carvalho Chehab
31d2656095SMauro Carvalho Chehabmy %broken_ref;
32d2656095SMauro Carvalho Chehab
33*894ee5ffSMauro Carvalho Chehabmy $doc_fix = 0;
34*894ee5ffSMauro Carvalho Chehab
35*894ee5ffSMauro Carvalho Chehabopen IN, "git grep ':doc:\`' Documentation/|"
36*894ee5ffSMauro Carvalho Chehab     or die "Failed to run git grep";
37*894ee5ffSMauro Carvalho Chehabwhile (<IN>) {
38*894ee5ffSMauro Carvalho Chehab	next if (!m,^([^:]+):.*\:doc\:\`([^\`]+)\`,);
39*894ee5ffSMauro Carvalho Chehab
40*894ee5ffSMauro Carvalho Chehab	my $d = $1;
41*894ee5ffSMauro Carvalho Chehab	my $doc_ref = $2;
42*894ee5ffSMauro Carvalho Chehab
43*894ee5ffSMauro Carvalho Chehab	my $f = $doc_ref;
44*894ee5ffSMauro Carvalho Chehab
45*894ee5ffSMauro Carvalho Chehab	$d =~ s,(.*/).*,$1,;
46*894ee5ffSMauro Carvalho Chehab	$f =~ s,.*\<([^\>]+)\>,$1,;
47*894ee5ffSMauro Carvalho Chehab
48*894ee5ffSMauro Carvalho Chehab	$f ="$d$f.rst";
49*894ee5ffSMauro Carvalho Chehab
50*894ee5ffSMauro Carvalho Chehab	next if (grep -e, glob("$f"));
51*894ee5ffSMauro Carvalho Chehab
52*894ee5ffSMauro Carvalho Chehab	if ($fix && !$doc_fix) {
53*894ee5ffSMauro Carvalho Chehab		print STDERR "\nWARNING: Currently, can't fix broken :doc:`` fields\n";
54*894ee5ffSMauro Carvalho Chehab	}
55*894ee5ffSMauro Carvalho Chehab	$doc_fix++;
56*894ee5ffSMauro Carvalho Chehab
57*894ee5ffSMauro Carvalho Chehab	print STDERR "$f: :doc:`$doc_ref`\n";
58*894ee5ffSMauro Carvalho Chehab}
59*894ee5ffSMauro Carvalho Chehabclose IN;
60*894ee5ffSMauro Carvalho Chehab
61d2656095SMauro Carvalho Chehabopen IN, "git grep 'Documentation/'|"
62d2656095SMauro Carvalho Chehab     or die "Failed to run git grep";
63d2656095SMauro Carvalho Chehabwhile (<IN>) {
64d2656095SMauro Carvalho Chehab	next if (!m/^([^:]+):(.*)/);
65d2656095SMauro Carvalho Chehab
66d2656095SMauro Carvalho Chehab	my $f = $1;
67d2656095SMauro Carvalho Chehab	my $ln = $2;
68d2656095SMauro Carvalho Chehab
69fe3e4b9cSMauro Carvalho Chehab	# On linux-next, discard the Next/ directory
70fe3e4b9cSMauro Carvalho Chehab	next if ($f =~ m,^Next/,);
71fe3e4b9cSMauro Carvalho Chehab
722d69708fSMauro Carvalho Chehab	# Makefiles and scripts contain nasty expressions to parse docs
732d69708fSMauro Carvalho Chehab	next if ($f =~ m/Makefile/ || $f =~ m/\.sh$/);
742d69708fSMauro Carvalho Chehab
75d2656095SMauro Carvalho Chehab	# Skip this script
76d2656095SMauro Carvalho Chehab	next if ($f eq $scriptname);
77d2656095SMauro Carvalho Chehab
782d69708fSMauro Carvalho Chehab	if ($ln =~ m,\b(\S*)(Documentation/[A-Za-z0-9\_\.\,\~/\*\[\]\?+-]*)(.*),) {
79d2656095SMauro Carvalho Chehab		my $prefix = $1;
80d2656095SMauro Carvalho Chehab		my $ref = $2;
81d2656095SMauro Carvalho Chehab		my $base = $2;
822d69708fSMauro Carvalho Chehab		my $extra = $3;
83d2656095SMauro Carvalho Chehab
842d69708fSMauro Carvalho Chehab		# some file references are like:
852d69708fSMauro Carvalho Chehab		# /usr/src/linux/Documentation/DMA-{API,mapping}.txt
862d69708fSMauro Carvalho Chehab		# For now, ignore them
872d69708fSMauro Carvalho Chehab		next if ($extra =~ m/^{/);
882d69708fSMauro Carvalho Chehab
892d69708fSMauro Carvalho Chehab		# Remove footnotes at the end like:
902d69708fSMauro Carvalho Chehab		# Documentation/devicetree/dt-object-internal.txt[1]
912d69708fSMauro Carvalho Chehab		$ref =~ s/(txt|rst)\[\d+]$/$1/;
922d69708fSMauro Carvalho Chehab
932d69708fSMauro Carvalho Chehab		# Remove ending ']' without any '['
942d69708fSMauro Carvalho Chehab		$ref =~ s/\].*// if (!($ref =~ m/\[/));
952d69708fSMauro Carvalho Chehab
962d69708fSMauro Carvalho Chehab		# Remove puntuation marks at the end
97d2656095SMauro Carvalho Chehab		$ref =~ s/[\,\.]+$//;
98d2656095SMauro Carvalho Chehab
99d2656095SMauro Carvalho Chehab		my $fulref = "$prefix$ref";
100d2656095SMauro Carvalho Chehab
101d2656095SMauro Carvalho Chehab		$fulref =~ s/^(\<file|ref)://;
102d2656095SMauro Carvalho Chehab		$fulref =~ s/^[\'\`]+//;
103d2656095SMauro Carvalho Chehab		$fulref =~ s,^\$\(.*\)/,,;
104d2656095SMauro Carvalho Chehab		$base =~ s,.*/,,;
105d2656095SMauro Carvalho Chehab
106d2656095SMauro Carvalho Chehab		# Remove URL false-positives
107d2656095SMauro Carvalho Chehab		next if ($fulref =~ m/^http/);
108d2656095SMauro Carvalho Chehab
1095d395fa6SMauro Carvalho Chehab		# Remove sched-pelt false-positive
1105d395fa6SMauro Carvalho Chehab		next if ($fulref =~ m,^Documentation/scheduler/sched-pelt$,);
1115d395fa6SMauro Carvalho Chehab
112d25c0634SMauro Carvalho Chehab		# Discard some build examples from Documentation/target/tcm_mod_builder.txt
113d25c0634SMauro Carvalho Chehab		next if ($fulref =~ m,mnt/sdb/lio-core-2.6.git/Documentation/target,);
114d25c0634SMauro Carvalho Chehab
115d2656095SMauro Carvalho Chehab		# Check if exists, evaluating wildcards
116d2656095SMauro Carvalho Chehab		next if (grep -e, glob("$ref $fulref"));
117d2656095SMauro Carvalho Chehab
118a78513c6SMauro Carvalho Chehab		# Accept relative Documentation patches for tools/
119a78513c6SMauro Carvalho Chehab		if ($f =~ m/tools/) {
120a78513c6SMauro Carvalho Chehab			my $path = $f;
121a78513c6SMauro Carvalho Chehab			$path =~ s,(.*)/.*,$1,;
122a78513c6SMauro Carvalho Chehab			next if (grep -e, glob("$path/$ref $path/$fulref"));
123a78513c6SMauro Carvalho Chehab		}
124a78513c6SMauro Carvalho Chehab
125d2656095SMauro Carvalho Chehab		if ($fix) {
126be600e5aSMauro Carvalho Chehab			if (!($ref =~ m/(scripts|Kconfig|Kbuild)/)) {
127d2656095SMauro Carvalho Chehab				$broken_ref{$ref}++;
128d2656095SMauro Carvalho Chehab			}
129d2656095SMauro Carvalho Chehab		} else {
130d2656095SMauro Carvalho Chehab			print STDERR "$f: $fulref\n";
131d2656095SMauro Carvalho Chehab		}
132d2656095SMauro Carvalho Chehab	}
133d2656095SMauro Carvalho Chehab}
134*894ee5ffSMauro Carvalho Chehabclose IN;
135d2656095SMauro Carvalho Chehab
136d2656095SMauro Carvalho Chehabexit 0 if (!$fix);
137d2656095SMauro Carvalho Chehab
138d2656095SMauro Carvalho Chehab# Step 2: Seek for file name alternatives
139d2656095SMauro Carvalho Chehabprint "Auto-fixing broken references. Please double-check the results\n";
140d2656095SMauro Carvalho Chehab
141d2656095SMauro Carvalho Chehabforeach my $ref (keys %broken_ref) {
142d2656095SMauro Carvalho Chehab	my $new =$ref;
143d2656095SMauro Carvalho Chehab
144d2656095SMauro Carvalho Chehab	# get just the basename
145d2656095SMauro Carvalho Chehab	$new =~ s,.*/,,;
146d2656095SMauro Carvalho Chehab
147d2656095SMauro Carvalho Chehab	my $f="";
148d2656095SMauro Carvalho Chehab
149be600e5aSMauro Carvalho Chehab	# usual reason for breakage: DT file moved around
150be600e5aSMauro Carvalho Chehab	if ($ref =~ /devicetree/) {
151be600e5aSMauro Carvalho Chehab		my $search = $new;
152be600e5aSMauro Carvalho Chehab		$search =~ s,^.*/,,;
153be600e5aSMauro Carvalho Chehab		$f = qx(find Documentation/devicetree/ -iname "*$search*") if ($search);
154be600e5aSMauro Carvalho Chehab		if (!$f) {
155be600e5aSMauro Carvalho Chehab			# Manufacturer name may have changed
156be600e5aSMauro Carvalho Chehab			$search =~ s/^.*,//;
157be600e5aSMauro Carvalho Chehab			$f = qx(find Documentation/devicetree/ -iname "*$search*") if ($search);
158be600e5aSMauro Carvalho Chehab		}
159be600e5aSMauro Carvalho Chehab	}
160d2656095SMauro Carvalho Chehab
161d2656095SMauro Carvalho Chehab	# usual reason for breakage: file renamed to .rst
162d2656095SMauro Carvalho Chehab	if (!$f) {
163d2656095SMauro Carvalho Chehab		$new =~ s/\.txt$/.rst/;
164d2656095SMauro Carvalho Chehab		$f=qx(find . -iname $new) if ($new);
165d2656095SMauro Carvalho Chehab	}
166d2656095SMauro Carvalho Chehab
167e1f319feSMauro Carvalho Chehab	# usual reason for breakage: use dash or underline
168e1f319feSMauro Carvalho Chehab	if (!$f) {
169e1f319feSMauro Carvalho Chehab		$new =~ s/[-_]/[-_]/g;
170e1f319feSMauro Carvalho Chehab		$f=qx(find . -iname $new) if ($new);
171e1f319feSMauro Carvalho Chehab	}
172e1f319feSMauro Carvalho Chehab
173be600e5aSMauro Carvalho Chehab	# Wild guess: seek for the same name on another place
174be600e5aSMauro Carvalho Chehab	if (!$f) {
175be600e5aSMauro Carvalho Chehab		$f = qx(find . -iname $new) if ($new);
176be600e5aSMauro Carvalho Chehab	}
177be600e5aSMauro Carvalho Chehab
178d2656095SMauro Carvalho Chehab	my @find = split /\s+/, $f;
179d2656095SMauro Carvalho Chehab
180d2656095SMauro Carvalho Chehab	if (!$f) {
181d2656095SMauro Carvalho Chehab		print STDERR "ERROR: Didn't find a replacement for $ref\n";
182d2656095SMauro Carvalho Chehab	} elsif (scalar(@find) > 1) {
183d2656095SMauro Carvalho Chehab		print STDERR "WARNING: Won't auto-replace, as found multiple files close to $ref:\n";
184d2656095SMauro Carvalho Chehab		foreach my $j (@find) {
185d2656095SMauro Carvalho Chehab			$j =~ s,^./,,;
186d2656095SMauro Carvalho Chehab			print STDERR "    $j\n";
187d2656095SMauro Carvalho Chehab		}
188d2656095SMauro Carvalho Chehab	} else {
189d2656095SMauro Carvalho Chehab		$f = $find[0];
190d2656095SMauro Carvalho Chehab		$f =~ s,^./,,;
191d2656095SMauro Carvalho Chehab		print "INFO: Replacing $ref to $f\n";
192d2656095SMauro Carvalho Chehab		foreach my $j (qx(git grep -l $ref)) {
193d2656095SMauro Carvalho Chehab			qx(sed "s\@$ref\@$f\@g" -i $j);
194d2656095SMauro Carvalho Chehab		}
195d2656095SMauro Carvalho Chehab	}
196d2656095SMauro Carvalho Chehab}
197