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, $reftext, $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         $fileinfo->{reftext}->{$e}->{$blockid} = $reftext
  94             if defined($reftext);
  95     }
  96 }
  97
  98 sub scan_adoc_file {
  99     my ($filename) = @_;
 100
 101     reset_environment_stack();
 102
 103     # print "SCAN $filename\n";
 104
 105     my $fh = IO::File->new("$filename", "r") or
 106         die "unable to open file '$filename' - $!\n";
 107
 108     my $env_last_line = {};
 109
 110     while (defined (my $line = <$fh>)) {
 111         if ($line =~ m/^if(n?)def::(\S+)\[(.*)\]\s*$/) {
 112             my ($not, $env, $text) = ($1, $2, $3);
 113             die "unsuported ifdef usage - implement me" if $text;
 114             push_environment($env, $not);
 115             next;
 116         } elsif ($line =~ m/^endif::(\S+)\[(.*)\]\s*$/) {
 117             my ($env, $text) = ($1, $2);
 118             die "unsuported ifdef usage - implement me" if $text;
 119             pop_environment($env);
 120             next;
 121         } elsif ($line =~ m/^include::(\S+)\[.*\]\s*$/) {
 122             register_include($filename, $1, $env_stack->[-1]);
 123             next;
 124         }
 125
 126         # try to detect titles
 127         foreach my $e (@{$env_stack->[-1]}) {
 128             my $title = $fileinfo->{titles}->{$e}->{$filename};
 129             next if defined($title);
 130
 131             if (($line =~ m/^=====+/) || ($line =~ m/^-----+/)) {
 132                 $fileinfo->{titles}->{$e}->{$filename} = $env_last_line->{$e};
 133             }
 134             $env_last_line->{$e} = $line;
 135             chomp $env_last_line->{$e};
 136         }
 137
 138         # fixme: also scan <<>>
 139
 140         while ($line =~ m/xref:([^\s\[\]]+)\[([^\]]*)\]/g) {
 141             # print "$filename xref:$1 [$2]\n";
 142         }
 143
 144         if ($line =~ m/^\[\[(.*)\]\]\s*$/) {
 145             my $blockid = $1;
 146             die "implement me" if $blockid =~m/,/;
 147             my $reftext = '';
 148             register_blockid($filename, $blockid, $reftext, $env_stack->[-1]);
 149         }
 150         # fixme: "anchor:"
 151         # bibliography anchors
 152         if ($line =~ m/\[\[\[([^\]]*)\]\]\]/) {
 153             my $blockid = $1;
 154             die "implement me" if $blockid =~m/,/;
 155             register_blockid($filename, $blockid, "&#91;$blockid&#93;", $env_stack->[-1]);
 156         }
 157     }
 158 }
 159
 160 my $scanned_files = {};
 161 while (my $filename = shift) {
 162     next if $filename !~ m/\.adoc$/; # skip attributes.txt
 163     next if $scanned_files->{$filename};
 164
 165     scan_adoc_file($filename);
 166     $scanned_files->{$filename} = 1;
 167 }
 168
 169 sub resolve_link_target {
 170     my ($env, $filename) = @_;
 171
 172     my $include_hash = $fileinfo->{include}->{$env};
 173
 174     my $repeat = 1;
 175
 176     while ($repeat) {
 177         $repeat = 0;
 178         foreach my $fn (keys %$include_hash) {
 179             next if $resolve_skip_files->{$env}->{$fn};
 180             if ($include_hash->{$fn}->{$filename}) {
 181                 $filename = $fn;
 182                 $repeat = 1;
 183                 last;
 184             }
 185         }
 186     }
 187
 188     return $filename;
 189 }
 190
 191 # now resolve blockids
 192 foreach my $e (@$start_env) {
 193     my $blockid_hash = $fileinfo->{blockid}->{$e};
 194     foreach my $blockid (keys %$blockid_hash) {
 195         my $fn = resolve_link_target($e, $blockid_hash->{$blockid});
 196         if ($e eq 'wiki') {
 197             my $title = $fileinfo->{titles}->{$e}->{$fn};
 198             $title =~ s/\{pve\}/Proxmox VE/g;
 199             $title =~ s/\s/_/g;
 200             die "found not title for '$fn' in env '$e'" if !$title;
 201             $fileinfo->{blockid_target}->{$e}->{$blockid} = "link:/wiki/$title#$blockid";
 202         } else {
 203             $fileinfo->{blockid_target}->{$e}->{$blockid} = $fn;
 204         }
 205     }
 206 }
 207
 208
 209 print to_json($fileinfo, { pretty => 1 } );