factor out cleanup method

[pve-docs.git] / scan-adoc-refs

#!/usr/bin/perl

use strict;
use warnings;
use IO::File;
use JSON;

use Data::Dumper;

my $environments = {
    default => 1,
    wiki => 1,
    manvolnum => 1,
    pvelogo => 0, # ignore
};

my $fileinfo = {};

my $start_env = [];
foreach my $e (keys %$environments) {
    push @$start_env, $e if $environments->{$e};
}

my $env_stack = [$start_env];
my $env_name_stack = [];

sub reset_environment_stack {
    $env_stack = [$start_env];
    $env_name_stack = [];
}

sub push_environment {
    my ($env, $not) = @_;

    die "undefined environment '$env'\n" if !defined($environments->{$env});

    # FIXME: this seems wrong  (nested env?)?
    return if !$environments->{$env}; # do not track

    if ($not) {
	my $new_env = [];
	foreach my $e (@{$env_stack->[-1]}) {
	    if ($e ne $env) {
		push @$new_env, $e;
	    }
	}
	die "empty environment" if !scalar($new_env);
	push @$env_stack, $new_env;
    } else {
	push @$env_stack, [$env];
    }

    push @$env_name_stack, $env;
}

sub pop_environment {
    my ($env) = @_;

    die "undefined environment '$env'\n" if !defined($environments->{$env});

    return if !$environments->{$env}; # do not track

    pop @$env_stack;
    my $res = pop @$env_name_stack;

    die "environment missmatch ($res != $env)\n" if $res ne $env;
}

sub register_include {
    my ($filename, $include_filename, $env_list) = @_;

    return if $include_filename !~ m/\.adoc$/; # skip attributes.txt

    foreach my $e (@$env_list) {
	$fileinfo->{include}->{$e}->{$filename}->{$include_filename} = 1;
    }
}

sub register_blockid {
    my ($filename, $blockid, $reftext, $env_list) = @_;

    foreach my $e (@$env_list) {
	my $fn = $fileinfo->{blockid}->{$e}->{$blockid};
	die "blockid '$blockid' already defined in $fn"
	    if defined($fn);
	$fileinfo->{blockid}->{$e}->{$blockid} = $filename;
	$fileinfo->{reftext}->{$e}->{$blockid} = $reftext
	    if defined($reftext);
    }
}

sub register_title {
    my ($filename, $env, $doctype, $title) = @_;

    # fixme: what about other macros?
    $title =~ s/\{pve\}/Proxmox VE/g;
    $title =~ s!http://\S+\[(.*?)\]!$1!g;

    $fileinfo->{titles}->{$env}->{$filename} = $title;

    if (defined($doctype)) {
	$fileinfo->{doctype}->{$env}->{$filename} = $doctype;
    } else {
	die "unable to change title (no doctype)"
	    if !defined($fileinfo->{doctype}->{$env}->{$filename});
    }
}

sub scan_adoc_file {
    my ($filename) = @_;

    reset_environment_stack();

    # print "SCAN $filename\n";

    my $fh = IO::File->new("$filename", "r") or
	die "unable to open file '$filename' - $!\n";

    my $env_last_line = {};

    while (defined (my $line = <$fh>)) {
	if ($line =~ m/^if(n?)def::(\S+)\[(.*)\]\s*$/) {
	    my ($not, $env, $text) = ($1, $2, $3);
	    die "unsuported ifdef usage - implement me" if $text;
	    push_environment($env, $not);
	    next;
	} elsif ($line =~ m/^endif::(\S+)\[(.*)\]\s*$/) {
	    my ($env, $text) = ($1, $2);
	    die "unsuported ifdef usage - implement me" if $text;
	    pop_environment($env);
	    next;
	} elsif ($line =~ m/^include::(\S+)\[.*\]\s*$/) {
	    register_include($filename, $1, $env_stack->[-1]);
	    next;
	}

	# try to detect titles
	foreach my $e (@{$env_stack->[-1]}) {
	    my $title = $fileinfo->{titles}->{$e}->{$filename};
	    next if defined($title);

	    if ($line =~ m/^=====+$/) {
		register_title($filename, $e, 0, $env_last_line->{$e});
	    } elsif ($line =~ m/^-----+$/) {
		register_title($filename, $e, 1, $env_last_line->{$e});
	    } elsif ($line =~ m/^= +(\S.*?)( +=)?$/) {
		register_title($filename, $e, 0, $1);
	    } elsif ($line =~ m/^== +(\S.*?)( +==)?$/) {
		register_title($filename, $e, 1, $1);
	    }

	    $env_last_line->{$e} = $line;
	    chomp $env_last_line->{$e};
	}

	if ($line =~ m/^:(\S+?):\s*(.*\S)?\s*$/) {
	    my ($key, $value) = ($1, $2);
	    if ($key eq 'pve-toplevel') {
		
		foreach my $e (@{$env_stack->[-1]}) {
		    my $title = $fileinfo->{titles}->{$e}->{$filename};
		    die "not title for toplevel file '$filename' (env=$e)\n"
			if !defined($title);
		    $fileinfo->{toplevel}->{$e}->{$filename} = 1;
		}
	    } elsif ($key eq 'title') {
		foreach my $e (@{$env_stack->[-1]}) {
		    register_title($filename, $e, undef, $value);
		}
	    }
	}

	if ($line =~ m/^\[\[(.*)\]\]\s*$/) {
	    my $blockid = $1;
	    die "implement me" if $blockid =~m/,/;
	    my $reftext = '';
	    register_blockid($filename, $blockid, $reftext, $env_stack->[-1]);
	}
	# fixme: "anchor:"
	# bibliography anchors
	if ($line =~ m/\[\[\[([^\]]*)\]\]\]/) {
	    my $blockid = $1;
	    die "implement me" if $blockid =~m/,/;
	    register_blockid($filename, $blockid, "&#91;$blockid&#93;", $env_stack->[-1]);
	}
    }
}

my $scanned_files = {};
while (my $filename = shift) {
    next if $filename !~ m/\.adoc$/; # skip attributes.txt
    next if $scanned_files->{$filename};

    scan_adoc_file($filename);
    $scanned_files->{$filename} = 1;
}

sub resolve_link_target {
    my ($env, $filename) = @_;

    my $include_hash = $fileinfo->{include}->{$env};

    my $repeat = 1;

    while ($repeat) {
	$repeat = 0;
	foreach my $fn (keys %$include_hash) {
	    if ($include_hash->{$fn}->{$filename}) {
		$filename = $fn;
		$repeat = 1;
		last;
	    }
	}
    }

    return $filename;
}

# now resolve blockids
foreach my $e (@$start_env) {
    my $blockid_hash = $fileinfo->{blockid}->{$e};
    foreach my $blockid (keys %$blockid_hash) {
	my $fn = resolve_link_target($e, $blockid_hash->{$blockid});
	if ($e eq 'wiki') {
	    my $title = $fileinfo->{titles}->{$e}->{$fn};
	    $title =~ s/\s/_/g;
	    die "found not title for '$fn' in env '$e'" if !$title;
	    $fileinfo->{blockid_target}->{$e}->{$blockid} = "link:/wiki/$title#$blockid";
	} else {
	    $fileinfo->{blockid_target}->{$e}->{$blockid} = $fn;
	}
    }
}


print to_json($fileinfo, { pretty => 1 } );