]> git.proxmox.com Git - pve-docs.git/blob - scan-adoc-refs
scan-adoc-refs: improve title parser, store doctype
[pve-docs.git] / scan-adoc-refs
1 #!/usr/bin/perl
2
3 use strict;
4 use warnings;
5 use IO::File;
6 use JSON;
7
8 use Data::Dumper;
9
10 my $environments = {
11 default => 1,
12 wiki => 1,
13 manvolnum => 1,
14 pvelogo => 0, # ignore
15 };
16
17 my $fileinfo = {};
18
19 my $start_env = [];
20 foreach my $e (keys %$environments) {
21 push @$start_env, $e if $environments->{$e};
22 }
23
24 my $env_stack = [$start_env];
25 my $env_name_stack = [];
26
27 sub reset_environment_stack {
28 $env_stack = [$start_env];
29 $env_name_stack = [];
30 }
31
32 sub push_environment {
33 my ($env, $not) = @_;
34
35 die "undefined environment '$env'\n" if !defined($environments->{$env});
36
37 # FIXME: this seems wrong (nested env?)?
38 return if !$environments->{$env}; # do not track
39
40 if ($not) {
41 my $new_env = [];
42 foreach my $e (@{$env_stack->[-1]}) {
43 if ($e ne $env) {
44 push @$new_env, $e;
45 }
46 }
47 die "empty environment" if !scalar($new_env);
48 push @$env_stack, $new_env;
49 } else {
50 push @$env_stack, [$env];
51 }
52
53 push @$env_name_stack, $env;
54 }
55
56 sub pop_environment {
57 my ($env) = @_;
58
59 die "undefined environment '$env'\n" if !defined($environments->{$env});
60
61 return if !$environments->{$env}; # do not track
62
63 pop @$env_stack;
64 my $res = pop @$env_name_stack;
65
66 die "environment missmatch ($res != $env)\n" if $res ne $env;
67 }
68
69 sub register_include {
70 my ($filename, $include_filename, $env_list) = @_;
71
72 return if $include_filename !~ m/\.adoc$/; # skip attributes.txt
73
74 foreach my $e (@$env_list) {
75 $fileinfo->{include}->{$e}->{$filename}->{$include_filename} = 1;
76 }
77 }
78
79 sub register_blockid {
80 my ($filename, $blockid, $reftext, $env_list) = @_;
81
82 foreach my $e (@$env_list) {
83 my $fn = $fileinfo->{blockid}->{$e}->{$blockid};
84 die "blockid '$blockid' already defined in $fn"
85 if defined($fn);
86 $fileinfo->{blockid}->{$e}->{$blockid} = $filename;
87 $fileinfo->{reftext}->{$e}->{$blockid} = $reftext
88 if defined($reftext);
89 }
90 }
91
92 sub scan_adoc_file {
93 my ($filename) = @_;
94
95 reset_environment_stack();
96
97 # print "SCAN $filename\n";
98
99 my $fh = IO::File->new("$filename", "r") or
100 die "unable to open file '$filename' - $!\n";
101
102 my $env_last_line = {};
103
104 while (defined (my $line = <$fh>)) {
105 if ($line =~ m/^if(n?)def::(\S+)\[(.*)\]\s*$/) {
106 my ($not, $env, $text) = ($1, $2, $3);
107 die "unsuported ifdef usage - implement me" if $text;
108 push_environment($env, $not);
109 next;
110 } elsif ($line =~ m/^endif::(\S+)\[(.*)\]\s*$/) {
111 my ($env, $text) = ($1, $2);
112 die "unsuported ifdef usage - implement me" if $text;
113 pop_environment($env);
114 next;
115 } elsif ($line =~ m/^include::(\S+)\[.*\]\s*$/) {
116 register_include($filename, $1, $env_stack->[-1]);
117 next;
118 }
119
120 # try to detect titles
121 foreach my $e (@{$env_stack->[-1]}) {
122 my $title = $fileinfo->{titles}->{$e}->{$filename};
123 next if defined($title);
124
125 if ($line =~ m/^=====+$/) {
126 $fileinfo->{titles}->{$e}->{$filename} = $env_last_line->{$e};
127 $fileinfo->{doctype}->{$e}->{$filename} = 0;
128 } elsif ($line =~ m/^-----+$/) {
129 $fileinfo->{titles}->{$e}->{$filename} = $env_last_line->{$e};
130 $fileinfo->{doctype}->{$e}->{$filename} = 1;
131 } elsif ($line =~ m/^= +(\S.*?)( +=)?$/) {
132 $fileinfo->{titles}->{$e}->{$filename} = $1;
133 $fileinfo->{doctype}->{$e}->{$filename} = 0;
134 } elsif ($line =~ m/^== +(\S.*?)( +==)?$/) {
135 $fileinfo->{titles}->{$e}->{$filename} = $1;
136 $fileinfo->{doctype}->{$e}->{$filename} = 1;
137 }
138
139 $env_last_line->{$e} = $line;
140 chomp $env_last_line->{$e};
141 }
142
143 if ($line =~ m/^:(\S+?):\s*(.*\S)?\s*$/) {
144 my ($key, $value) = ($1, $2);
145 if ($key eq 'pve-toplevel') {
146
147 foreach my $e (@{$env_stack->[-1]}) {
148 my $title = $fileinfo->{titles}->{$e}->{$filename};
149 die "not title for toplevel file '$filename' (env=$e)\n"
150 if !defined($title);
151 $fileinfo->{toplevel}->{$e}->{$filename} = 1;
152 }
153 }
154 }
155
156 if ($line =~ m/^\[\[(.*)\]\]\s*$/) {
157 my $blockid = $1;
158 die "implement me" if $blockid =~m/,/;
159 my $reftext = '';
160 register_blockid($filename, $blockid, $reftext, $env_stack->[-1]);
161 }
162 # fixme: "anchor:"
163 # bibliography anchors
164 if ($line =~ m/\[\[\[([^\]]*)\]\]\]/) {
165 my $blockid = $1;
166 die "implement me" if $blockid =~m/,/;
167 register_blockid($filename, $blockid, "&#91;$blockid&#93;", $env_stack->[-1]);
168 }
169 }
170 }
171
172 my $scanned_files = {};
173 while (my $filename = shift) {
174 next if $filename !~ m/\.adoc$/; # skip attributes.txt
175 next if $scanned_files->{$filename};
176
177 scan_adoc_file($filename);
178 $scanned_files->{$filename} = 1;
179 }
180
181 sub resolve_link_target {
182 my ($env, $filename) = @_;
183
184 my $include_hash = $fileinfo->{include}->{$env};
185
186 my $repeat = 1;
187
188 while ($repeat) {
189 $repeat = 0;
190 foreach my $fn (keys %$include_hash) {
191 if ($include_hash->{$fn}->{$filename}) {
192 $filename = $fn;
193 $repeat = 1;
194 last;
195 }
196 }
197 }
198
199 return $filename;
200 }
201
202 # now resolve blockids
203 foreach my $e (@$start_env) {
204 my $blockid_hash = $fileinfo->{blockid}->{$e};
205 foreach my $blockid (keys %$blockid_hash) {
206 my $fn = resolve_link_target($e, $blockid_hash->{$blockid});
207 if ($e eq 'wiki') {
208 my $title = $fileinfo->{titles}->{$e}->{$fn};
209 $title =~ s/\{pve\}/Proxmox VE/g;
210 $title =~ s/\s/_/g;
211 die "found not title for '$fn' in env '$e'" if !$title;
212 $fileinfo->{blockid_target}->{$e}->{$blockid} = "link:/wiki/$title#$blockid";
213 } else {
214 $fileinfo->{blockid_target}->{$e}->{$blockid} = $fn;
215 }
216 }
217 }
218
219
220 print to_json($fileinfo, { pretty => 1 } );