cjk enhancement, #505

.github/workflows/batch-check.yaml

@@ -168,8 +168,12 @@ jobs:
         run: |
           ./latexindent.exe -l test-cases/replacements/issue-503b.yaml -r test-cases/replacements/issue-503a.tex 
           cat test-cases/replacements/indent.log
-      ### - name: latexindent.exe cjk test, issue 505
-      ###   # https://github.com/cmhughes/latexindent.pl/issues/505
-      ###   run: |
-      ###     ./latexindent.exe -w test-cases/back-up-tests/新建.tex -y 'onlyOneBackUp:1'
-      ###     cat test-cases/back-up-tests/indent.log
+      - name: latexindent.exe cjk test, issue 505
+        # https://github.com/cmhughes/latexindent.pl/issues/505
+        run: |
+          ./latexindent.exe -w --encoding "GB2312" test-cases/back-up-tests/新建.tex -y 'onlyOneBackUp:1' 
+          cat test-cases/back-up-tests/indent.log
+          ./latexindent.exe --encoding "GB2312" test-cases/back-up-tests/新建.tex -o=+-mod1 -y 'onlyOneBackUp:1'
+          cat test-cases/back-up-tests/indent.log
+          ./latexindent.exe --encoding "GB2312" test-cases/back-up-tests/新建.tex -g=新建.log
+          cat 新建.log

LatexIndent/BackUpFileProcedure.pm

@@ -23,7 +23,7 @@ use LatexIndent::LogFile         qw/$logger/;
 use File::Basename;    # to get the filename and directory path
 use File::Copy;        # to copy the original file to backup (if overwrite option set)
 use Exporter qw/import/;
-use Encode   qw/decode/;
+use Encode   qw/decode encode/;
 our @EXPORT_OK = qw/create_back_up_file check_if_different/;
 
 # copy main file to a backup in the case of the overwrite switch being active
@@ -36,7 +36,13 @@ sub create_back_up_file {
     # if we want to overwrite the current file create a backup first
     $logger->info("*Backup procedure (-w flag active):");
 
-    my $fileName = decode( "utf-8", ${$self}{fileName} );
+    my $fileName;
+    if ( ${$self}{encoding} and ref( find_encoding( ${$self}{encoding} ) ) ) {
+        $fileName = ${$self}{fileName};
+    }
+    else {
+        $fileName = decode( "utf-8", ${$self}{fileName} );
+    }
 
     # grab the file extension preferences
     my %fileExtensionPreference = %{ $mainSettings{fileExtensionPreference} };
@@ -49,7 +55,13 @@ sub create_back_up_file {
     my $backupFileNoExt = basename( ${$self}{fileName}, @fileExtensions );
 
     # add the user's backup directory to the backup path
-    $backupFileNoExt = decode( "utf-8", "${$self}{cruftDirectory}/$backupFileNoExt" );
+    if ( ${$self}{encoding} and ref( find_encoding( ${$self}{encoding} ) ) ) {
+        my $encodingObject = find_encoding( ${$self}{encoding} );
+        $backupFileNoExt = $encodingObject->encode("${$self}{cruftDirectory}/$backupFileNoExt");
+    }
+    else {
+        $backupFileNoExt = decode( "utf-8", "${$self}{cruftDirectory}/$backupFileNoExt" );
+    }
 
     # local variables, determined from the YAML settings
     my $onlyOneBackUp       = $mainSettings{onlyOneBackUp};

LatexIndent/Document.pm

@@ -20,7 +20,7 @@ use warnings;
 use Data::Dumper;
 use File::Basename;    # to get the filename and directory path
 use open ':std', ':encoding(UTF-8)';
-use Encode qw/decode/;
+use Encode;
 
 # gain access to subroutines in the following modules
 use LatexIndent::Switches
@@ -98,7 +98,12 @@ sub latexindent {
 
     # one-time operations
     $self->store_switches;
+    if ( ${$self}{encoding} and ref( find_encoding( ${$self}{encoding} ) ) ) {
+        my $encodingObject = find_encoding( ${$self}{encoding} );
+        ${$self}{fileName} = $encodingObject->encode( ${$self}{fileName} );
+    } else {
     ${$self}{fileName} = decode( "utf-8", $fileNames[0] );
+}
     $self->process_switches( \@fileNames );
     $self->yaml_read_settings;
 

LatexIndent/FileExtension.pm

@@ -25,6 +25,7 @@ use Encode                       qw/decode/;
 use LatexIndent::GetYamlSettings qw/%mainSettings/;
 use LatexIndent::Switches        qw/%switches $is_check_switch_active/;
 use LatexIndent::LogFile         qw/$logger/;
+use Encode;
 our @EXPORT_OK = qw/file_extension_check/;
 
 sub file_extension_check {

LatexIndent/LogFile.pm

@@ -23,7 +23,7 @@ use File::Path            qw(make_path);
 use Exporter              qw/import/;
 use LatexIndent::Switches qw/%switches/;
 use LatexIndent::Version  qw/$versionNumber $versionDate/;
-use Encode                qw/decode/;
+use Encode;
 our @EXPORT_OK = qw/process_switches $logger/;
 our $logger;
 
@@ -116,6 +116,9 @@ usage: latexindent.pl [options] [file]
       --GCString
           loads the Unicode::GCString module for the align-at-ampersand routine
           Note: this requires the Unicode::GCString module to be installed on your system
+      -e, --encoding=<value>
+          specify encoding, sample usage
+                latexindent.pl --encoding gb2312 myfile.tex
 ENDQUOTE
             ;
         exit(0);
@@ -147,7 +150,13 @@ ENDQUOTE
     my $logfileName = ( $switches{cruftDirectory} ? ${$self}{cruftDirectory} . "/" : '' )
         . ( $switches{logFileName} || "indent.log" );
 
-    $logfileName = decode( "utf-8", $logfileName );
+    if ( ${$self}{encoding} and ref( find_encoding( ${$self}{encoding} ) ) ) {
+        my $encodingObject = find_encoding( ${$self}{encoding} );
+        $logfileName = $encodingObject->encode($logfileName);
+    }
+    else {
+        $logfileName = decode( "utf-8", $logfileName );
+    }
 
     # details of the script to log file
     $logger->info("*$FindBin::Script version $versionNumber, $versionDate, a script to indent .tex files");

LatexIndent/Switches.pm

@@ -51,6 +51,8 @@ sub store_switches {
     $is_check_verbose_switch_active = defined $switches{checkverbose} ? $switches{checkverbose} : 0;
     $is_check_switch_active
         = $is_check_verbose_switch_active ? $is_check_verbose_switch_active : $is_check_switch_active;
+    ${$self}{encoding} = ( $switches{encoding} ? $switches{encoding} : 0 );
+
     delete ${$self}{switches};
 }
 1;

latexindent.pl

@@ -54,6 +54,7 @@
     "checkv|kv"                 => \$switches{checkverbose},
     "lines|n=s"                 => \$switches{lines},
     "GCString"                  => \$switches{GCString},
+    "encoding|e=s"              => \$switches{encoding},
 );
 
 # conditionally load the GCString module

test-cases/back-up-tests/back-up-tests.sh

@@ -22,6 +22,11 @@ latexindent.pl -s -w 新建.tex -g issue-505.log -y 'onlyOneBackUp:1'
 cp issue-505.log issue-505.txt
 perl -p0i -e 's/.*?(INFO:\s*Backup)/$1/s' issue-505.txt
 
+latexindent.pl -s -w 新建.tex -g=新建.log -y 'onlyOneBackUp:1'
+cp 新建.log issue-505-1.txt
+perl -p0i -e 's/.*?(INFO:\s*Backup)/$1/s' issue-505-1.txt
+
+latexindent.pl -s -w --encoding "GB2312" 新建.tex -y 'onlyOneBackUp:1' 
 set +x
 [[ $gitStatus == 1 ]] && git status
 [[ $noisyMode == 1 ]] && makenoise

test-cases/back-up-tests/issue-505-1.txt

@@ -0,0 +1,14 @@
+INFO:  Backup procedure (-w flag active):
+       ./新建.bak will be overwritten (see onlyOneBackUp)
+       Backing up 新建.tex to ./新建.bak...
+       新建.tex will be overwritten after indentation
+INFO:  Phase 1: searching for objects
+       No objects found.
+INFO:  Phase 2: finding surrounding indentation
+INFO:  Phase 3: indenting objects
+INFO:  Phase 4: final indentation check
+INFO:  Output routine:
+       Overwriting file 新建.tex
+       --------------
+INFO:  Please direct all communication/issues to:
+        https://github.com/cmhughes/latexindent.pl

test-cases/back-up-tests/issue-505.yaml

@@ -0,0 +1 @@
+encoding: GB2312