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