Changeset 1224

Timestamp:

2000-06-21T10:26:01+12:00 (24 years ago)

Author:

sjboddie

Message:

added handling of arabic encoding and ability to read in an entire
file to a string

File:

• trunk/gsdl/perllib/multiread.pm (modified) (5 diffs)

trunk/gsdl/perllib/multiread.pm

@@ -29 +29 @@
 # unicode  - just unicode (doesn't currently do endian detection)
 # gb       - GB
-# extended - extended ascii
-
+# latin1   - extended ascii (iso-8859-1)
+# arabic   - 8 bit arabic (iso-8859-6)
 
 package multiread;
@@ -169 +169 @@
     }
 
-    if ($self->{'encoding'} eq "extended") {
-    # extended ascii
+    if ($self->{'encoding'} eq "latin1") {
+    # Latin 1 extended ascii (ISO-8859-1)
     return undef if (eof ($handle));
     return &unicode::ascii2utf8 (getc ($handle));
+    }
+
+    if ($self->{'encoding'} eq "arabic") {
+    # 8 bit Arabic (IOS-8859-6)
+    return undef if (eof ($handle));
+    return &unicode::unicode2utf8(&unicode::arabic2unicode (getc ($handle)));
     }
 
@@ -236 +242 @@
     }
     
-    if ($self->{'encoding'} eq "extended") {
-    # extended ascii
+    if ($self->{'encoding'} eq "latin1") {
+    # extended ascii (ISO-8859-1)
     my $line = "";
     if (defined ($line = <$handle>)) {
@@ -244 +250 @@
     return undef;
     }
+    
+    if ($self->{'encoding'} eq "arabic") {
+    # 8 bit arabic (ISO-8859-6)
+    my $line = "";
+    if (defined ($line = <$handle>)) {
+        return &unicode::unicode2utf8(&unicode::arabic2unicode ($line));
+    }
+    return undef;
+    }
 
     # unknown encoding
@@ -250 +265 @@
 
 
+# will convert entire contents of file to utf8 and append result to $outputref
+sub read_file {
+    my $self = shift (@_);
+    my ($outputref) = @_;
+
+    # make sure we have a file handle
+    return if ($self->{'handle'} eq "");
+
+    my $handle = $self->{'handle'};
+
+    if ($self->{'first'} && $self->{'encoding'} eq "utf8") {
+    # special case for the first line of utf8 text to detect whether
+    # the file is in utf8 or unicode
+    $$text .= $self->read_line ();
+    }
+
+    if ($self->{'encoding'} eq "utf8") {
+    undef $/;
+    $$outputref .=  <$handle>;
+    $/ = "\n";
+    return;
+    }
+
+    if ($self->{'encoding'} eq "unicode") {
+    my $line = "";
+    while (defined ($line = $self->read_line())) {
+        $$outputref .= $line;
+    }
+    return;
+    }
+
+    if ($self->{'encoding'} eq "gb") {
+    undef $/;
+    $$outputref .= &unicode::unicode2utf8 (&gb::gb2unicode (<$handle>));
+    $/ = "\n";
+    return;
+    }
+    
+    if ($self->{'encoding'} eq "latin1") {
+    undef $/;
+    $$outputref .= &unicode::ascii2utf8 (<$handle>);
+    $/ = "\n";
+    return;
+    }
+    
+    if ($self->{'encoding'} eq "arabic") {
+    undef $/;
+    $$outputref .= &unicode::unicode2utf8(&unicode::arabic2unicode (<$handle>));
+    $/ = "\n";
+    return;
+    }
+}
+
+
 1;