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
33d2656095SMauro Carvalho Chehabopen IN, "git grep 'Documentation/'|"
34d2656095SMauro Carvalho Chehab     or die "Failed to run git grep";
35d2656095SMauro Carvalho Chehabwhile (<IN>) {
36d2656095SMauro Carvalho Chehab	next if (!m/^([^:]+):(.*)/);
37d2656095SMauro Carvalho Chehab
38d2656095SMauro Carvalho Chehab	my $f = $1;
39d2656095SMauro Carvalho Chehab	my $ln = $2;
40d2656095SMauro Carvalho Chehab
41*2d69708fSMauro Carvalho Chehab	# Makefiles and scripts contain nasty expressions to parse docs
42*2d69708fSMauro Carvalho Chehab	next if ($f =~ m/Makefile/ || $f =~ m/\.sh$/);
43*2d69708fSMauro Carvalho Chehab
44d2656095SMauro Carvalho Chehab	# Skip this script
45d2656095SMauro Carvalho Chehab	next if ($f eq $scriptname);
46d2656095SMauro Carvalho Chehab
47*2d69708fSMauro Carvalho Chehab	if ($ln =~ m,\b(\S*)(Documentation/[A-Za-z0-9\_\.\,\~/\*\[\]\?+-]*)(.*),) {
48d2656095SMauro Carvalho Chehab		my $prefix = $1;
49d2656095SMauro Carvalho Chehab		my $ref = $2;
50d2656095SMauro Carvalho Chehab		my $base = $2;
51*2d69708fSMauro Carvalho Chehab		my $extra = $3;
52d2656095SMauro Carvalho Chehab
53*2d69708fSMauro Carvalho Chehab		# some file references are like:
54*2d69708fSMauro Carvalho Chehab		# /usr/src/linux/Documentation/DMA-{API,mapping}.txt
55*2d69708fSMauro Carvalho Chehab		# For now, ignore them
56*2d69708fSMauro Carvalho Chehab		next if ($extra =~ m/^{/);
57*2d69708fSMauro Carvalho Chehab
58*2d69708fSMauro Carvalho Chehab		# Remove footnotes at the end like:
59*2d69708fSMauro Carvalho Chehab		# Documentation/devicetree/dt-object-internal.txt[1]
60*2d69708fSMauro Carvalho Chehab		$ref =~ s/(txt|rst)\[\d+]$/$1/;
61*2d69708fSMauro Carvalho Chehab
62*2d69708fSMauro Carvalho Chehab		# Remove ending ']' without any '['
63*2d69708fSMauro Carvalho Chehab		$ref =~ s/\].*// if (!($ref =~ m/\[/));
64*2d69708fSMauro Carvalho Chehab
65*2d69708fSMauro Carvalho Chehab		# Remove puntuation marks at the end
66d2656095SMauro Carvalho Chehab		$ref =~ s/[\,\.]+$//;
67d2656095SMauro Carvalho Chehab
68d2656095SMauro Carvalho Chehab		my $fulref = "$prefix$ref";
69d2656095SMauro Carvalho Chehab
70d2656095SMauro Carvalho Chehab		$fulref =~ s/^(\<file|ref)://;
71d2656095SMauro Carvalho Chehab		$fulref =~ s/^[\'\`]+//;
72d2656095SMauro Carvalho Chehab		$fulref =~ s,^\$\(.*\)/,,;
73d2656095SMauro Carvalho Chehab		$base =~ s,.*/,,;
74d2656095SMauro Carvalho Chehab
75d2656095SMauro Carvalho Chehab		# Remove URL false-positives
76d2656095SMauro Carvalho Chehab		next if ($fulref =~ m/^http/);
77d2656095SMauro Carvalho Chehab
78d2656095SMauro Carvalho Chehab		# Check if exists, evaluating wildcards
79d2656095SMauro Carvalho Chehab		next if (grep -e, glob("$ref $fulref"));
80d2656095SMauro Carvalho Chehab
81d2656095SMauro Carvalho Chehab		if ($fix) {
82be600e5aSMauro Carvalho Chehab			if (!($ref =~ m/(scripts|Kconfig|Kbuild)/)) {
83d2656095SMauro Carvalho Chehab				$broken_ref{$ref}++;
84d2656095SMauro Carvalho Chehab			}
85d2656095SMauro Carvalho Chehab		} else {
86d2656095SMauro Carvalho Chehab			print STDERR "$f: $fulref\n";
87d2656095SMauro Carvalho Chehab		}
88d2656095SMauro Carvalho Chehab	}
89d2656095SMauro Carvalho Chehab}
90d2656095SMauro Carvalho Chehab
91d2656095SMauro Carvalho Chehabexit 0 if (!$fix);
92d2656095SMauro Carvalho Chehab
93d2656095SMauro Carvalho Chehab# Step 2: Seek for file name alternatives
94d2656095SMauro Carvalho Chehabprint "Auto-fixing broken references. Please double-check the results\n";
95d2656095SMauro Carvalho Chehab
96d2656095SMauro Carvalho Chehabforeach my $ref (keys %broken_ref) {
97d2656095SMauro Carvalho Chehab	my $new =$ref;
98d2656095SMauro Carvalho Chehab
99d2656095SMauro Carvalho Chehab	# get just the basename
100d2656095SMauro Carvalho Chehab	$new =~ s,.*/,,;
101d2656095SMauro Carvalho Chehab
102d2656095SMauro Carvalho Chehab	my $f="";
103d2656095SMauro Carvalho Chehab
104be600e5aSMauro Carvalho Chehab	# usual reason for breakage: DT file moved around
105be600e5aSMauro Carvalho Chehab	if ($ref =~ /devicetree/) {
106be600e5aSMauro Carvalho Chehab		my $search = $new;
107be600e5aSMauro Carvalho Chehab		$search =~ s,^.*/,,;
108be600e5aSMauro Carvalho Chehab		$f = qx(find Documentation/devicetree/ -iname "*$search*") if ($search);
109be600e5aSMauro Carvalho Chehab		if (!$f) {
110be600e5aSMauro Carvalho Chehab			# Manufacturer name may have changed
111be600e5aSMauro Carvalho Chehab			$search =~ s/^.*,//;
112be600e5aSMauro Carvalho Chehab			$f = qx(find Documentation/devicetree/ -iname "*$search*") if ($search);
113be600e5aSMauro Carvalho Chehab		}
114be600e5aSMauro Carvalho Chehab	}
115d2656095SMauro Carvalho Chehab
116d2656095SMauro Carvalho Chehab	# usual reason for breakage: file renamed to .rst
117d2656095SMauro Carvalho Chehab	if (!$f) {
118d2656095SMauro Carvalho Chehab		$new =~ s/\.txt$/.rst/;
119d2656095SMauro Carvalho Chehab		$f=qx(find . -iname $new) if ($new);
120d2656095SMauro Carvalho Chehab	}
121d2656095SMauro Carvalho Chehab
122e1f319feSMauro Carvalho Chehab	# usual reason for breakage: use dash or underline
123e1f319feSMauro Carvalho Chehab	if (!$f) {
124e1f319feSMauro Carvalho Chehab		$new =~ s/[-_]/[-_]/g;
125e1f319feSMauro Carvalho Chehab		$f=qx(find . -iname $new) if ($new);
126e1f319feSMauro Carvalho Chehab	}
127e1f319feSMauro Carvalho Chehab
128be600e5aSMauro Carvalho Chehab	# Wild guess: seek for the same name on another place
129be600e5aSMauro Carvalho Chehab	if (!$f) {
130be600e5aSMauro Carvalho Chehab		$f = qx(find . -iname $new) if ($new);
131be600e5aSMauro Carvalho Chehab	}
132be600e5aSMauro Carvalho Chehab
133d2656095SMauro Carvalho Chehab	my @find = split /\s+/, $f;
134d2656095SMauro Carvalho Chehab
135d2656095SMauro Carvalho Chehab	if (!$f) {
136d2656095SMauro Carvalho Chehab		print STDERR "ERROR: Didn't find a replacement for $ref\n";
137d2656095SMauro Carvalho Chehab	} elsif (scalar(@find) > 1) {
138d2656095SMauro Carvalho Chehab		print STDERR "WARNING: Won't auto-replace, as found multiple files close to $ref:\n";
139d2656095SMauro Carvalho Chehab		foreach my $j (@find) {
140d2656095SMauro Carvalho Chehab			$j =~ s,^./,,;
141d2656095SMauro Carvalho Chehab			print STDERR "    $j\n";
142d2656095SMauro Carvalho Chehab		}
143d2656095SMauro Carvalho Chehab	} else {
144d2656095SMauro Carvalho Chehab		$f = $find[0];
145d2656095SMauro Carvalho Chehab		$f =~ s,^./,,;
146d2656095SMauro Carvalho Chehab		print "INFO: Replacing $ref to $f\n";
147d2656095SMauro Carvalho Chehab		foreach my $j (qx(git grep -l $ref)) {
148d2656095SMauro Carvalho Chehab			qx(sed "s\@$ref\@$f\@g" -i $j);
149d2656095SMauro Carvalho Chehab		}
150d2656095SMauro Carvalho Chehab	}
151d2656095SMauro Carvalho Chehab}
152