#!/usr/bin/env perl #************************************************************************* # Copyright (c) 2012 UChicago Argonne LLC, as Operator of Argonne # National Laboratory. # SPDX-License-Identifier: EPICS # EPICS BASE is distributed subject to a Software License Agreement found # in file LICENSE that is included with this distribution. #************************************************************************* use strict; use FindBin qw($Bin); use lib ("$Bin/../../lib/perl"); use DBD; use DBD::Parser; use EPICS::Getopts; use EPICS::macLib; use EPICS::PodMD; use EPICS::Readfile; use Pod::Usage; =head1 NAME dbdToMD.pl - Convert DBD file with POD to Markdown =head1 SYNOPSIS B [B<-h>] [B<-D>] [B<-I> dir] [B<-o> file] file.dbd.pod =head1 DESCRIPTION Generates MArkdown documentation from a B<.dbd.pod> file. =head1 OPTIONS B understands the following options: =over 4 =item B<-h> Help, display usage information. =item B<-H> Conversion help, display information about converting reference documentation from the EPICS Wiki into a B<.dbd.pod> file for use with this tool. =item B<-I> Path to look for include files. =item B<-o> file Name of the output file to be created. =back If no output filename is set, the file created will be named after the input file, removing any directory components in the path and replacing any B<.dbd.pod> file extension with B<.md>. =cut our ($opt_h, $opt_H, @opt_I, $opt_o); my $tool = 'dbdToMD.pl'; getopts('hHI@o:') or pod2usage(2); pod2usage(-verbose => 2) if $opt_H; pod2usage(1) if $opt_h; pod2usage("$tool: No input file given.\n") if @ARGV != 1; my $dbd = DBD->new(); my $infile = shift @ARGV; $infile =~ m/\.dbd.pod$/ or pod2usage("$tool: Input file '$infile' must have '.dbd.pod' extension.\n"); ParseDBD($dbd, Readfile($infile, 0, \@opt_I)); if (!$opt_o) { ($opt_o = $infile) =~ s/\.dbd\.pod$/.md/; $opt_o =~ s/^.*\///; $opt_o =~ s/dbCommonRecord/dbCommon/; } (my $title = $opt_o) =~ s/\.md$//; open my $out, '>', $opt_o or die "Can't create $opt_o: $!\n"; $SIG{__DIE__} = sub { die @_ if $^S; # Ignore eval deaths close $out; unlink $opt_o; }; sub make_fragment { my $fragment = $_[1]; $fragment =~ s/\W+/-/g; $fragment = lc($fragment); $_[1] = $fragment; } my $podRst = EPICS::PodMD->new( perldoc_url_prefix => '', perldoc_fragment_format => make_fragment, markdown_fragment_format => make_fragment, ); # Parse the Pod text from the root DBD object my $pod = join "\n", map { # Handle a 'recordtype' Pod directive if (m/^ =recordtype \s+ (\w+) /x) { my $rn = $1; my $rtyp = $dbd->recordtype($rn); die "Unknown recordtype '$rn' in $infile POD directive\n" unless $rtyp; rtypeToMD($rtyp, $dbd); } # Handle a 'menu' Pod directive elsif (m/^ =menu \s+ (\w+) /x) { my $mn = $1; my $menu = $dbd->menu($mn); die "Unknown menu '$mn' in $infile POD directive\n" unless $menu; menuToMD($menu); } elsif (m/^ =title \s+ (.*)/x) { $title = $1; "=head1 $title"; } else { $_; } } $dbd->pod; $podRst->output_fh($out); $podRst->parse_string_document($pod); close $out; sub menuToMD { my ($menu) = @_; my $index = 0; return "| Index | Identifier | Choice String |", "| ----- | -------------------------------------- | ---------------------- |", map({choiceTableRow($_, $index++)} $menu->choices); } sub choiceTableRow { my ($ch, $index) = @_; my ($id, $name) = @{$ch}; return sprintf("| %5d | %-*s | %-22s |", $index, 38-( () = $id =~ /_/g), $id, $name); } sub rtypeToMD { my ($rtyp, $dbd) = @_; return map { # Handle a 'fields' Pod directive if (m/^ =fields \s+ (\w+ (?:\s* , \s* \w+ )* )/x) { my @names = split /\s*,\s*/, $1; # Look up the named fields my @fields = map { my $field = $rtyp->field($_); die "Unknown field name '$_' in $infile POD\n" unless $field; $field; } @names; # Generate Pod for the table "| Field | Summary | Type | DCT | Default | Read | Write | CA PP |", "| ----- | -------------------------- | ------------- | --- | ------- | ---- | ----- | ----- |", map({fieldTableRow($_, $dbd)} @fields); } # Handle a 'menu' Pod directive elsif (m/^ =menu \s+ (\w+) /x) { my $mn = $1; my $menu = $dbd->menu($mn); die "Unknown menu '$mn' in $infile POD directive\n" unless $menu; menuToMD($menu); } else { # Raw text line $_; } } $rtyp->pod; } sub fieldTableRow { my ($fld, $dbd) = @_; my @md; push @md, sprintf("%-5s", $fld->name), sprintf("%-26s", $fld->attribute('prompt')); my $type = $fld->public_type; if ($type eq 'STRING') { $type .= sprintf("%-5s", ' [' . $fld->attribute('size') . ']'); } elsif ($type eq 'MENU') { my $mn = $fld->attribute('menu'); my $menu = $dbd->menu($mn); my $mnl = lc($mn); my $url = $menu ? "/menu-$mnl" : "${mn}.md"; #just pass a L directive for the parser $type .= " L<$mn|$url>"; } else { $type = sprintf("%-13s", $type); } push @md, $type; push @md, sprintf("%-3s", $fld->attribute('promptgroup') ? 'Yes' : 'No'); push @md, sprintf("%7s", $fld->attribute('initial') || ' '); push @md, sprintf("%-4s",$fld->readable); push @md, sprintf("%-5s",$fld->writable); push @md, sprintf("%-5s",$fld->attribute('pp') eq 'TRUE' ? 'Yes' : 'No'); return '| ' . join(' | ', @md) . ' |'; } # Native type presented to dbAccess users sub DBD::Recfield::public_type { my $fld = shift; m/^ =type \s+ (.+) /x && return $1 for $fld->comments; my $type = $fld->dbf_type; $type =~ s/^DBF_//; return $type; } # Check if this field is readable sub DBD::Recfield::readable { my $fld = shift; m/^ =read \s+ (?i) (Yes|No) /x && return $1 for $fld->comments; return 'Probably' if $fld->attribute('special') eq "SPC_DBADDR"; return $fld->dbf_type eq 'DBF_NOACCESS' ? 'No' : 'Yes'; } # Check if this field is writable sub DBD::Recfield::writable { my $fld = shift; m/^ =write \s+ (?i) (Yes|No) /x && return $1 for $fld->comments; my $special = $fld->attribute('special'); return 'No' if $special eq "SPC_NOMOD"; return 'Maybe' if $special eq "SPC_DBADDR"; return $fld->dbf_type eq "DBF_NOACCESS" ? 'No' : 'Yes'; } =pod =head1 Converting Wiki Record Reference to POD If you open the src/std/rec/aiRecord.dbd.pod file in your favourite plain text editor you'll see what input was required to generate the aiRecord.html file. The text markup language we're using is a standard called POD (Plain Old Documentation) which is used by Perl developers, but you don't need to know Perl at all to be able to use it. When we add POD markup to a record type, we rename its *Record.dbd file to .dbd.pod in the src/std/rec directory; no other changes are needed for the build system to find it by its new name. The POD content is effectively just a new kind of comment that appears in .dbd.pod files, which the formatter knows how to convert into Markdown. The build also generates a plain *Record.dbd file from this same input file by stripping out all of the POD markup. Documentation for Perl's POD markup standard can be found online at L or you may be able to type 'perldoc perlpod' into a Linux command-line to see the same text. We added a few POD keywords of our own to handle the table generation, and I'll cover those briefly below. POD text can appear almost anywhere in a dbd.pod file. It always starts with a line "=[keyword] [additional text...]" where [keyword] is "title", "head1" through "head4" etc.. The POD text ends with a line "=cut". There must be a blank line above every POD line, and in many cases below it as well. The POD keywords we have added are "title", "recordtype", "menu", "fields", "type", "read" and "write". The last 3 are less common but are used in some of the other record types such as the waveform and aSub records. The most interesting of our new keywords is "fields", which takes a list of record field names on the same line after the keyword and generates an Markdown Table describing those fields based on the field description found in the DBD parts. In the ai documentation the first such table covers the DTYP and INP fields. Note that the "=fields" line must appear inside the DBD's declaration of the record type, i.e. after the line recordtype(ai) { The "type", "read" and "write" POD keywords are used inside an individual record field declaration and provide information for the "Type", "Read" and "Write" columns of the field's table output for fields where this information is normally supplied by the record support code. Usage examples for these keywords can be found in the aai and aSub record types. If you look at the L file you'll see that the POD there starts by documenting a record-specific menu definition. The "menu" keyword generates a table that lists all the choices found in the named menu. Any MENU fields in the field tables that refer to a locally-defined menu will generate a link to a document section which must be titled "Menu [menuName]". =cut