initial implementation of -recorder (for latexmlpost)

brucemiller · Jan 7, 2024 · e2a1eb1 · e2a1eb1
1 parent 340840b
commit e2a1eb1
Show file tree

Hide file tree

Showing 6 changed files with 43 additions and 7 deletions.
diff --git a/bin/latexmlpost b/bin/latexmlpost
@@ -30,7 +30,7 @@ use LaTeXML::Util::ObjectDB;
 #======================================================================
 # undef => unspecified; 0 = NO, 1 = YES
 my ($help, $showversion, $verbosity, $validate, $omit_doctype) = (0, 0, 0, 1, 0);
-my ($sourcedir, $destination, $logfile) = (undef, undef, undef);
+my ($sourcedir, $destination, $logfile, $recorder, $recordfile) = (undef, undef, undef, undef, undef);
 my @paths = ();
 my ($format, $extension, $is_html, $urlstyle) = (undef, undef, undef, 'server');
 my ($numbersections) = (1);
@@ -58,11 +58,13 @@ my $splitback = [qw(bibliography appendix index)];
 
 # Get the command line arguments.
 GetOptions("quiet" => sub { $verbosity--; },
-  "verbose" => sub { $verbosity++; },
-  "VERSION" => \$showversion,
-  "help|?"  => \$help,
-  "log=s"   => \$logfile,
-  "debug=s" => sub { no strict 'refs'; $LaTeXML::DEBUG{ lc($_[1]) } = 1; },
+  "verbose"      => sub { $verbosity++; },
+  "VERSION"      => \$showversion,
+  "help|?"       => \$help,
+  "log=s"        => \$logfile,
+  "recorder"     => \$recorder,
+  "recordfile=s" => \$recordfile,
+  "debug=s"      => sub { no strict 'refs'; $LaTeXML::DEBUG{ lc($_[1]) } = 1; },
   # Source Options
   "sourcedirectory=s" => \$sourcedir,
   "validate!"         => \$validate,
@@ -165,6 +167,7 @@ if (my @fails = validate_args()) {
 my $starttime = StartTime();
 my $latexmlpost;
 eval {    # Catch errors
+  UseRecordFile($recordfile);
   UseLog($logfile);
   Note($LaTeXML::IDENTITY . ($prescan ? " scanning " : " paginating ") . $xmlfile);
 
@@ -214,6 +217,7 @@ Note("Postprocessing " . ($code == 3 ? 'failed' : 'complete') . " " . $status .
 UseLog(undef);
 CheckDebuggable();
 UseSTDERR(undef);
+UseRecordFile(undef);
 exit($code == 3 ? 1 : 0);
 
 #======================================================================
@@ -247,6 +251,13 @@ sub validate_args {
       $logfile = pathname_make(dir => pathname_cwd(), name => $name, type => 'latexmlpost.log'); } }
   $logfile = 'latexmlpost.log' unless $logfile;
 
+  # Optionally create a .fls file for tracking the opening of files (option --recorder)
+  # Use $jobname.latexmlpost.fls, latexmlpost.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 => 'latexmlpost.fls') : 'latexmlpost.fls'; }
+
   #======================================================================
   # Sanity check and Completion of options.
   #======================================================================

diff --git a/lib/LaTeXML/Common/XML.pm b/lib/LaTeXML/Common/XML.pm
@@ -228,6 +228,7 @@ sub initialize_catalogs {
   return if $catalogs_initialized;
   $catalogs_initialized = 1;
   foreach my $catalog (pathname_findall('LaTeXML.catalog', installation_subdir => '.')) {
+    RecordInput($catalog);
     XML::LibXML->load_catalog($catalog); }
   return; }
 

diff --git a/lib/LaTeXML/Common/XML/XSLT.pm b/lib/LaTeXML/Common/XML/XSLT.pm
@@ -12,6 +12,9 @@
 package LaTeXML::Common::XML::XSLT;
 use strict;
 use warnings;
+use LaTeXML::Common::Error;
+use URI;
+use URI::Escape;
 use XML::LibXSLT;
 
 sub new {
@@ -20,6 +23,10 @@ sub new {
   LaTeXML::Common::XML::initialize_catalogs();
   #  LaTeXML::Common::XML::initialize_input_callbacks($xslt,installation_subdir => 'resources/XSLT');
   # Do we still need this logic, if callbacks work?
+  my $input_callbacks = XML::LibXML::InputCallback->new();
+  $input_callbacks->register_callbacks([sub { RecordInput(uri_unescape(URI->new($_[0])->path)); return 0; }, undef, undef, undef]);
+  $xslt->input_callbacks($input_callbacks);
+
   if (!ref $stylesheet) {
     $stylesheet = LaTeXML::Common::XML::Parser->new()->parseFile($stylesheet); }
   #    $stylesheet = $xslt->parse_stylesheet_file($stylesheet); }
@@ -31,5 +38,9 @@ sub transform {
   my ($self, $document, %params) = @_;
   return $$self{stylesheet}->transform($document, %params); }
 
+sub security_callbacks {
+  my ($self, $security) = @_;
+  return $$self{stylesheet}->security_callbacks($security); }
+
 #======================================================================
 1;
diff --git a/lib/LaTeXML/Post/MakeBibliography.pm b/lib/LaTeXML/Post/MakeBibliography.pm
@@ -13,6 +13,7 @@ package LaTeXML::Post::MakeBibliography;
 use strict;
 use warnings;
 use LaTeXML::Util::Pathname;
+use LaTeXML::Common::Error;
 use LaTeXML::Common::XML;
 use LaTeXML::Util::Radix;
 use charnames qw(:full);
@@ -155,6 +156,7 @@ sub getBibliographies {
           $raw .= $rawbib; }
         else {
           # TODO: Is this a memory concern for large bib files?
+          RecordInput($rawbib);
           if (open(my $bibfh, '<', $rawbib)) {
             $raw .= join("", <$bibfh>);
             close $bibfh; }

diff --git a/lib/LaTeXML/Post/Writer.pm b/lib/LaTeXML/Post/Writer.pm
@@ -14,6 +14,7 @@ package LaTeXML::Post::Writer;
 use strict;
 use warnings;
 use LaTeXML::Util::Pathname;
+use LaTeXML::Common::Error;
 use LaTeXML::Common::XML;
 use LaTeXML::Post;
 use base qw(LaTeXML::Post::Processor);
@@ -54,6 +55,7 @@ sub process {
       or return Fatal('I/O', $destdir, undef, "Couldn't create directory '$destdir'",
       "Response was: $!");
     my $OUT;
+    RecordOutput($destination);
     open($OUT, '>', $destination)
       or return Fatal('I/O', $destdir, undef, "Couldn't write '$destination'", "Response was: $!");
     print $OUT $serialized;

diff --git a/lib/LaTeXML/Post/XSLT.pm b/lib/LaTeXML/Post/XSLT.pm
@@ -76,6 +76,12 @@ sub process {
   if (my $icon = $params{ICON}) {
     # Hmm.... what type? could be various image types
     $params{ICON} = '"' . $self->copyResource($doc, $icon, undef) . '"'; }
+
+  my $security = XML::LibXSLT::Security->new;
+  $security->register_callback(write_file => sub { RecordOutput($_[1]); return 1; });
+  $security->register_callback(create_dir => sub { RecordOutput($_[1]); return 1; });
+  $$self{stylesheet}->security_callbacks($security);
+
   my $newdoc = $doc->new($$self{stylesheet}->transform($doc->getDocument, %params));
   return $newdoc; }
 
@@ -115,7 +121,10 @@ sub copyResource {
       if (!pathname_is_contained($dest, $doc->getSiteDirectory)) {
         $dest = pathname_make(dir => $doc->getSiteDirectory, name => $name, type => $ex); } }
     # Now, copy (unless in same place! happens a lot during testing!!!!)
-    pathname_copy($path, $dest) unless $path eq $dest;
+    if (!($path eq $dest)) {
+      RecordInput($path);
+      pathname_copy($path, $dest);
+      RecordOutput($dest); }
     # and return the relative path from the dest doc to the resource
     return pathname_relative($dest, $doc->getDestinationDirectory); }
   else {