Added OLE file checking to identify xlsx and encrypted files.

lib/Excel/Reader/XLSX.pm

@@ -16,6 +16,7 @@ use strict;
 use warnings;
 use Exporter;
 use Archive::Zip;
+use OLE::Storage_Lite;
 use File::Temp  qw(tempdir);
 use Excel::Reader::XLSX::Workbook;
 use Excel::Reader::XLSX::Package::ContentTypes;
@@ -31,18 +32,24 @@ our $VERSION = '0.00';
 # Error codes for some common errors.
 our $ERROR_none                        = 0;
 our $ERROR_file_not_found              = 1;
-our $ERROR_file_zip_error              = 2;
-our $ERROR_file_missing_subfile        = 3;
-our $ERROR_file_has_no_content_types   = 4;
-our $ERROR_file_missing_workbook       = 5;
+our $ERROR_file_is_xls                 = 2;
+our $ERROR_file_is_encrypted           = 3;
+our $ERROR_file_is_unknown_ole         = 4;
+our $ERROR_file_zip_error              = 5;
+our $ERROR_file_missing_subfile        = 6;
+our $ERROR_file_has_no_content_types   = 7;
+our $ERROR_file_missing_workbook       = 8;
 
 our @error_strings = (
     '',                                                 # 0
     'File not found',                                   # 1
-    'File has zip error',                               # 2
-    'File missing subfile',                             # 3
-    'File has no [Content_Types].xml',                  # 4
-    'File is missing workbook.xml',                     # 5
+    'File is xls not xlsx',                             # 2
+    'File is encrypted xlsx',                           # 3
+    'File is unknown OLE doc type',                     # 4
+    'File has zip error',                               # 5
+    'File missing subfile',                             # 6
+    'File has no [Content_Types].xml',                  # 7
+    'File is missing workbook.xml',                     # 8
 );
 
 
@@ -90,6 +97,14 @@ sub read_file {
         return;
     }
 
+    # Check for xls or encrypted OLE files.
+    my $ole_file = $self->_check_if_ole_file( $filename );
+    if ( $ole_file ) {
+        $self->{_error_status}     = $ole_file;
+        $self->{_error_extra_text} = $filename;
+        return;
+    }
+
     # Create a, locally scoped, temp dir to unzip the XLSX file into.
     my $tempdir  = File::Temp->newdir( DIR => $self->{_tempdir} );
 
@@ -130,19 +145,22 @@ sub read_file {
     # Read the filenames from the [Content_Types].
     my %files = $content_types->_get_files();
 
-    # Create a reader object to read the sharedStrings.xml file.
 
+    # Check that the files actually exist.
+    my $files_exist = $self->_check_files_exist( $tempdir, %files );
+
+    if ( !$files_exist ) {
+        $self->{_error_status} = $ERROR_file_missing_subfile;
+        return;
+    }
+
+
+    # Create a reader object to read the sharedStrings.xml file.
     my $shared_strings = Excel::Reader::XLSX::Package::SharedStrings->new();
 
     # Read the sharedStrings if present. Only files with strings have one.
     if ( $files{_shared_strings} ) {
 
-        # Check that the file exists even if it is listed in [Content_Types].
-        if ( !-e $tempdir . $files{_shared_strings} ) {
-            $self->{_error_status} = $ERROR_file_missing_subfile;
-            return;
-        }
-
         $shared_strings->_read_file( $tempdir . $files{_shared_strings} );
         $shared_strings->_read_all_nodes();
     }
@@ -155,11 +173,6 @@ sub read_file {
 
     );
 
-    # Check that the file exists even if it is listed in [Content_Types].
-    if ( !-e $tempdir . $files{_workbook} ) {
-        $self->{_error_status} = $ERROR_file_missing_subfile;
-        return;
-    }
 
     # Read data from the workbook.xml file.
     $workbook->_read_file( $tempdir . $files{_workbook} );
@@ -175,6 +188,88 @@ sub read_file {
 }
 
 
+###############################################################################
+#
+# _check_files_exist()
+#
+# Verify that the subfiles read from the Content_Types actually exist;
+#
+sub _check_files_exist {
+
+    my $self    = shift;
+    my $tempdir = shift;
+    my %files   = @_;
+    my @filenames;
+
+    # Get the filenames for the files hash.
+    for my $key ( keys %files ) {
+        my $filename = $files{$key};
+
+        # Worksheets are stored in an aref.
+        if ( ref $filename ) {
+            push @filenames, @$filename;
+        }
+        else {
+            push @filenames, $filename;
+        }
+    }
+
+    # Verify that the files exist.
+    for my $filename ( @filenames ) {
+        if ( !-e $tempdir . $filename ) {
+            $self->{_error_extra_text} = $filename;
+            return;
+        }
+    }
+
+    return 1;
+}
+
+
+###############################################################################
+#
+# _check_if_ole_file()
+#
+# Check if the file in an OLE compound doc. This can happen in a few cases.
+# This first is when the file is xls and not xlsx. The second is when the
+# file is an encrypted xlsx file. We also handle the case of unknown OLE
+# file types.
+#
+# Porting note. As a lightweight test you can check for OLE files by looking
+# for the magic number 0xD0CF11E0 (docfile0) at the start of the file.
+#
+sub _check_if_ole_file {
+
+    my $self     = shift;
+    my $filename = shift;
+    my $ole      = OLE::Storage_Lite->new( $filename );
+    my $pps      = $ole->getPpsTree();
+
+    # If getPpsTree() failed then this isn't an OLE file.
+    return if !$pps;
+
+    # Loop throught the PPS children below the root.
+    for my $child_pps ( @{ $pps->{Child} } ) {
+
+        my $pps_name = OLE::Storage_Lite::Ucs2Asc( $child_pps->{Name} );
+
+        # Match an Excel xls file.
+        if ($pps_name eq 'Workbook' || $pps_name eq 'Book' ) {
+            return $ERROR_file_is_xls;
+        }
+
+        # Match an encrypted Excel xlsx file.
+        if ($pps_name eq 'EncryptedPackage') {
+            return $ERROR_file_is_encrypted;
+        }
+    }
+
+    return $ERROR_file_is_unknown_ole;
+}
+
+
+
+
 ###############################################################################
 #
 # error().
@@ -188,7 +283,7 @@ sub error {
     my $error       = $error_strings[$error_index];
 
     if ($self->{_error_extra_text}) {
-        $error .= '. ' . $self->{_error_extra_text};
+        $error .= ': ' . $self->{_error_extra_text};
     }
 
     return $error;

lib/Excel/Reader/XLSX/Package/ContentTypes.pm

@@ -140,6 +140,9 @@ sub _read_node {
     my $content_type = $node->getAttribute('ContentType');
 
 
+    # Strip leading directory separator from filename.
+    $part_name =~ s{^/}{};
+
     if ( $part_name =~ /app\.xml$/ ) {
         $self->{_files}->{_app} = $part_name;
         return;
@@ -161,7 +164,14 @@ sub _read_node {
     }
 
     if ( $part_name =~ /workbook\.xml$/ ) {
-        $self->{_files}->{_workbook} = $part_name;
+
+        # The workbook.xml.rels file isn't included in the ContentTypes but
+        # it is usually in the _rels dir at the same level at the workbook.xml.
+        my $workbook_rels = $part_name;
+        $workbook_rels =~ s{(workbook.xml)}{_rels/$1.rels};
+
+        $self->{_files}->{_workbook}      = $part_name;
+        $self->{_files}->{_workbook_rels} = $workbook_rels;
         return;
     }
 

t/package/content_types/content_types01.t

@@ -34,12 +34,13 @@ $reader->_read_all_nodes();
 $caption = " \tContentTypes: _strings";
 
 $expected = {
-    '_workbook'       => '/xl/workbook.xml',
-    '_app'            => '/docProps/app.xml',
-    '_styles'         => '/xl/styles.xml',
-    '_worksheets'     => [ '/xl/worksheets/sheet1.xml' ],
-    '_core'           => '/docProps/core.xml',
-    '_shared_strings' => '/xl/sharedStrings.xml'
+    '_workbook'       => 'xl/workbook.xml',
+    '_workbook_rels'  => 'xl/_rels/workbook.xml.rels',
+    '_app'            => 'docProps/app.xml',
+    '_styles'         => 'xl/styles.xml',
+    '_worksheets'     => [ 'xl/worksheets/sheet1.xml' ],
+    '_core'           => 'docProps/core.xml',
+    '_shared_strings' => 'xl/sharedStrings.xml'
 };
 
 

t/package/content_types/content_types02.t

@@ -35,12 +35,13 @@ $reader->_read_all_nodes();
 $caption = " \tContentTypes: _strings";
 
 $expected = {
-    '_workbook'       => '/xl/workbook.xml',
-    '_app'            => '/docProps/app.xml',
-    '_styles'         => '/xl/styles.xml',
-    '_worksheets'     => [ '/xl/worksheets/sheet1.xml' ],
-    '_core'           => '/docProps/core.xml',
-    '_shared_strings' => '/xl/sharedStrings.xml'
+    '_workbook'       => 'xl/workbook.xml',
+    '_workbook_rels'  => 'xl/_rels/workbook.xml.rels',
+    '_app'            => 'docProps/app.xml',
+    '_styles'         => 'xl/styles.xml',
+    '_worksheets'     => [ 'xl/worksheets/sheet1.xml' ],
+    '_core'           => 'docProps/core.xml',
+    '_shared_strings' => 'xl/sharedStrings.xml'
 };
 
 

t/package/content_types/content_types03.t

@@ -34,16 +34,18 @@ $reader->_read_all_nodes();
 $caption = " \tContentTypes: _strings";
 
 $expected = {
-    '_workbook'   => '/xl/workbook.xml',
-    '_app'        => '/docProps/app.xml',
-    '_styles'     => '/xl/styles.xml',
-    '_worksheets' => [
-        '/xl/worksheets/sheet1.xml',
-        '/xl/worksheets/sheet2.xml',
-        '/xl/worksheets/sheet3.xml'
+    '_workbook'      => 'xl/workbook.xml',
+    '_workbook_rels' => 'xl/_rels/workbook.xml.rels',
+    '_app'           => 'docProps/app.xml',
+    '_styles'        => 'xl/styles.xml',
+    '_worksheets'    => [
+        'xl/worksheets/sheet1.xml',
+        'xl/worksheets/sheet2.xml',
+        'xl/worksheets/sheet3.xml'
+
     ],
-    '_core'           => '/docProps/core.xml',
-    '_shared_strings' => '/xl/sharedStrings.xml'
+    '_core'           => 'docProps/core.xml',
+    '_shared_strings' => 'xl/sharedStrings.xml'
 };
 
 

t/regression/json_files/read_error01.json

@@ -4,6 +4,6 @@
    "xlsx_file"   : "error01.xlsx",
    "expected" : {
       "error_code" : 1,
-      "error_text" : "File not found. t/regression/xlsx_files/error01.xlsx"
+      "error_text" : "File not found: t/regression/xlsx_files/error01.xlsx"
    }
 }

t/regression/json_files/read_error02.json

@@ -1,9 +1,9 @@
 {
    "caption"     : "Test error handling.",
-   "description" : "Returns a zip error when not a valid zip file.",
+   "description" : "Returns a error if the file is an xls file.",
    "xlsx_file"   : "error02.xlsx",
    "expected" : {
       "error_code" : 2,
-      "error_text" : "File has zip error. format error: can\'t find EOCD signature "
+      "error_text" : "File is xls not xlsx: t/regression/xlsx_files/error02.xlsx"
    }
 }

t/regression/json_files/read_error03.json

@@ -0,0 +1,9 @@
+{
+   "caption"     : "Test error handling.",
+   "description" : "Returns an error when the file is encrypted xlsx.",
+   "xlsx_file"   : "error03.xlsx",
+   "expected" : {
+      "error_code" : 3,
+      "error_text" : "File is encrypted xlsx: t/regression/xlsx_files/error03.xlsx"
+   }
+}

t/regression/json_files/read_error04.json

@@ -0,0 +1,9 @@
+{
+   "caption"     : "Test error handling.",
+   "description" : "Returns a error if the file is an unknown OLE doc.",
+   "xlsx_file"   : "error04.xlsx",
+   "expected" : {
+      "error_code" : 4,
+      "error_text" : "File is unknown OLE doc type: t/regression/xlsx_files/error04.xlsx"
+   }
+}

t/regression/json_files/read_error05.json

@@ -0,0 +1,9 @@
+{
+   "caption"     : "Test error handling.",
+   "description" : "Returns a zip error when not a valid zip file.",
+   "xlsx_file"   : "error05.xlsx",
+   "expected" : {
+      "error_code" : 5,
+      "error_text" : "File has zip error: format error: can\'t find EOCD signature "
+   }
+}

t/regression/json_files/read_error06.json

@@ -0,0 +1,9 @@
+{
+   "caption"     : "Test error handling.",
+   "description" : "Returns an error when a subfile is missing.",
+   "xlsx_file"   : "error06.xlsx",
+   "expected" : {
+      "error_code" : 6,
+      "error_text" : "File missing subfile: docProps/core.xml"
+   }
+}

t/regression/read_error03.t

@@ -0,0 +1,43 @@
+###############################################################################
+#
+# Tests for Excel::Writer::XLSX.
+#
+# reverse('©'), February 2012, John McNamara, jmcnamara@cpan.org
+#
+
+use lib 't/lib';
+use TestFunctions qw(_is_deep_diff _read_json);
+use strict;
+use warnings;
+use Excel::Reader::XLSX;
+
+use Test::More tests => 1;
+
+###############################################################################
+#
+# Test setup.
+#
+my $json_filename = 't/regression/json_files/read_error03.json';
+my $json          = _read_json( $json_filename );
+my $caption       = $json->{caption};
+my $expected      = $json->{expected};
+my $xlsx_file     = 't/regression/xlsx_files/' . $json->{xlsx_file};
+my $got;
+
+
+###############################################################################
+#
+# Test error handling when reading data from an Excel file.
+#
+use Excel::Reader::XLSX;
+
+my $reader   = Excel::Reader::XLSX->new();
+my $workbook = $reader->read_file( $xlsx_file );
+
+$got = {
+    error_code => $reader->error_code(),
+    error_text => $reader->error(),
+};
+
+# Test the results.
+_is_deep_diff( $got, $expected, $caption );