initial implementation of -recorder (for latexml)

bin/latexml

@@ -26,7 +26,7 @@ use Encode;
 # Parse command line
 my ($verbosity, $strict, $comments, $noparse, $includestyles) = (0, 0, 1, 0, 0);
 my ($format, $destination, $help, $showversion)               = ('xml', '');
-my ($preamble, $postamble, $logfile)                          = (undef, undef, undef);
+my ($preamble, $postamble, $logfile, $recorder, $recordfile)  = (undef, undef, undef, undef, undef);
 my ($documentid);
 my $inputencoding;
 my $mode  = undef;
@@ -42,6 +42,8 @@ GetOptions("destination=s" => \$destination,
   "verbose"         => sub { $verbosity++; },
   "strict"          => \$strict,
   "log=s"           => \$logfile,
+  "recorder"        => \$recorder,
+  "recordfile=s"    => \$recordfile,
   "xml"             => sub { $format = 'xml'; },
   "tex"             => sub { $format = 'tex'; },
   "box"             => sub { $format = 'box'; },
@@ -77,6 +79,7 @@ my $latexml;
 my $digested;
 my $serialized;
 eval {    # Catch errors
+  UseRecordFile($recordfile);
   UseLog($logfile);
   Note("$LaTeXML::IDENTITY processing $source");
 
@@ -128,12 +131,15 @@ UseLog(undef);
 if    (!$serialized) { }
 elsif ($destination) {
   my $OUTPUT;
+  RecordOutput($destination);
   open($OUTPUT, ">", $destination) or die "Couldn't open output file $destination: $!";
   print $OUTPUT $serialized;
   close($OUTPUT); }
 else {
   print $serialized; }
 
+UseRecordFile(undef);
+
 # ========================================
 # Now, unbind stuff, so we can clear memory
 $latexml    = undef;
@@ -186,6 +192,13 @@ sub validate_args {
       $logfile = pathname_make(dir => pathname_cwd(), name => $name, type => 'latexml.log'); } }
   $logfile = 'latexml.log' unless $logfile;
 
+  # Optionally create a .fls file for tracking the opening of files (option --recorder)
+  # Use $jobname.latexml.fls, latexml.fls, or argument of --recordfile
+  $recordfile = $recorder ? $recordfile : undef;
+  if ($pathname && $recorder && !$recordfile) {
+    my ($dir, $name, $ext) = pathname_split($pathname);
+    $recordfile = $name ? pathname_make(dir => pathname_cwd(), name => $name, type => 'latexml.fls') : 'latexml.fls'; }
+
   # Check that destination is valid before wasting any time...
   if ($destination) {
     $destination = pathname_canonical($destination);

lib/LaTeXML/Common/Error.pm

@@ -46,6 +46,8 @@ our @EXPORT = (
   qw(&MergeStatus),
   # Run time reporting
   qw(&StartTime &RunTime),
+  # Recording API
+  qw(&UseRecordFile &RecordInput &RecordOutput),
 );
 
 our $VERBOSITY   = 0;
@@ -134,11 +136,49 @@ sub UseLog {
     $log_count++;
     return if $LOG or not($path);                 # already opened?
     pathname_mkdir(pathname_directory($path));    # and hopefully no errors! :>
+    RecordOutput($path) if !$append || $log_count == 1;
     open($LOG, ($append ? '>>' : '>'), $path) or die "Cannot open log file $path for writing: $!";
     $LOG_PATH = $path;
     binmode($LOG, ":encoding(UTF-8)"); }
   return; }
 
+#%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+# Record file
+our $RECORDFILE;
+our $RECORDFILE_PATH;
+# NOTE: same strategy as for $LOG
+our $recordfile_count = 0;
+
+sub UseRecordFile {
+  my ($path, $append) = @_;
+  if (!$path) {    # Single false argument? Turn OFF and Close
+    $recordfile_count--;
+    return if !$RECORDFILE || $recordfile_count;
+    close($RECORDFILE) or die "Cannot close record file: $!";
+    $RECORDFILE = undef; }
+  else {
+    $recordfile_count++;
+    return if $RECORDFILE or not($path);          # already opened?
+    pathname_mkdir(pathname_directory($path));    # and hopefully no errors! :>
+    my $is_empty = -z $path || !$append;
+    open($RECORDFILE, ($append ? '>>' : '>'), $path) or die "Cannot open record file $path for writing: $!";
+    $RECORDFILE_PATH = $path;
+    binmode($RECORDFILE, ":encoding(UTF-8)");
+    # first line must be current working directory
+    print $RECORDFILE "PWD " . pathname_cwd() . "\n" if $is_empty; }
+  return; }
+
+# Recording API
+sub RecordInput {
+  my ($filename) = @_;
+  print $RECORDFILE "INPUT $filename\n" unless !$RECORDFILE;
+  return; }
+
+sub RecordOutput {
+  my ($filename) = @_;
+  print $RECORDFILE "OUTPUT $filename\n" unless !$RECORDFILE;
+  return; }
+
 #%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 # Low-level I/O
 

lib/LaTeXML/Common/Font/Metric.pm

@@ -47,6 +47,7 @@ sub read_tfm {
   my $pathname = $$self{file};
   # Read the TFM raw data.
   my $TFM;
+  RecordInput($pathname);
   if (!open($TFM, '<', $pathname)) {
     Error('expected', $pathname, undef,
       "Couldn't open TFM files '$pathname': $!");

lib/LaTeXML/Common/Model.pm

@@ -138,6 +138,7 @@ sub loadCompiledSchema {
   my ($self, $file) = @_;
   ProgressSpinup("Loading compiled schema $file");
   my $MODEL;
+  RecordInput($file);
   open($MODEL, '<', $file) or Fatal('I/O', $file, undef, "Cannot open Compiled Model $file for reading", $!);
   my $line;
   while ($line = <$MODEL>) {

lib/LaTeXML/Common/XML.pm

@@ -36,6 +36,7 @@ use strict;
 use warnings;
 use XML::LibXML qw(:all);
 use XML::LibXML::XPathContext;
+use LaTeXML::Common::Error;
 use LaTeXML::Util::Pathname;
 use Encode;
 use Carp;

lib/LaTeXML/Common/XML/Parser.pm

@@ -12,12 +12,18 @@
 package LaTeXML::Common::XML::Parser;
 use strict;
 use warnings;
+use LaTeXML::Common::Error;
+use URI;
+use URI::Escape;
 use XML::LibXML;
 
 sub new {
   my ($class) = @_;
   my $parser = XML::LibXML->new();
   $parser->validation(0);
+  my $input_callbacks = XML::LibXML::InputCallback->new();
+  $input_callbacks->register_callbacks([sub { RecordInput(uri_unescape(URI->new($_[0])->path)); return 0; }, undef, undef, undef]);
+  $parser->input_callbacks($input_callbacks);
   return bless { parser => $parser }, $class; }
 
 sub parseFile {

lib/LaTeXML/Core/Mouth/Binding.pm

@@ -27,6 +27,7 @@ sub new {
   my ($class, $pathname) = @_;
   my ($dir, $name, $ext) = pathname_split($pathname);
   my $self = bless { source => $pathname, shortsource => "$name.$ext" }, $class;
+  RecordInput($pathname);
   ProgressSpinup("Loading $$self{source}");
   return $self; }
 
@@ -65,7 +66,7 @@ sub stringify {
 
 __END__
 
-=pod 
+=pod
 
 =head1 NAME
 

lib/LaTeXML/Core/Mouth/file.pm

@@ -32,6 +32,7 @@ sub new {
 sub openFile {
   my ($self, $pathname) = @_;
   my $IN;
+  RecordInput($pathname);
   if (!-r $pathname) {
     Error('I/O', 'unreadable', $self, "File $pathname is not readable. Ignoring."); }
   elsif ((!-z $pathname) && (-B $pathname)) {

lib/LaTeXML/Package.pm

@@ -2660,6 +2660,7 @@ sub maybeRequireDependencies {
   if (my $path = FindFile($file, type => $type, noltxml => 1)) {
     local $/ = undef;
     my $IN;
+    RecordInput($path);
     if (open($IN, '<', $path)) {
       my $code = <$IN>;
       close($IN);

lib/LaTeXML/Package/listings.sty.ltxml

@@ -14,6 +14,7 @@ package LaTeXML::Package::Pool;
 use strict;
 use warnings;
 use LaTeXML::Package;
+use LaTeXML::Common::Error;
 use MIME::Base64;
 use Encode qw(is_utf8 encode);
 
@@ -321,6 +322,7 @@ sub listingsReadRawFile {
   my $path     = FindFile($filename);
   my $text;
   my $LST_FH;
+  RecordInput($filename) if $path;
   if ($path && open($LST_FH, '<', $path)) {
     { local $/ = undef;
       $text = <$LST_FH>;

lib/LaTeXML/Pre/BibTeX.pm

@@ -67,6 +67,7 @@ sub newFromFile {
     "SEACHPATHS is " . join(', ', @$paths)) unless $file;
   my $BIB;
   $$self{file} = $bibname;
+  RecordInput($file);
   open($BIB, '<', $file) or Fatal('I/O', $file, undef, "Can't open BibTeX $file for reading", $!);
   $$self{lines} = [<$BIB>];
   close($BIB);