How do I use Oracle utl_file to write an image clob - jpeg

Question

I have an Oracle Apex application which generates automated eMails. In Apex, the user inserts a JPG image into a rich text field. That image is saved into a CLOB field. When the stored procedure is called, it reads the JPG image and stores it into a local variable called l_image_clob. The program sends the embedded image (note: this an embedded image and it is not an eMail attachment) along with the rest of the eMail body to a list of users. That's all working fine.
Now I'm attempting to save the contents of the JPG image stored in l_image_clob to a JPG file on the windows server. The following code produces a file, named properly and the size is correct, but it isn't readable by the system. I get the error "this is not a valid bitmap file" when I try to open it with Microsoft Paint. How to I use utl_file to do this?
Here's the code which creates the file that is "not a valid bitmap file"
-- Create a file based on the content of l_image_clob
l_image_filename := 'image_' || p_event_pkey || '_' || i ||
'.' || l_image_ext;
l_file_handle := utl_file.fopen(l_dirname , l_image_filename, 'wb');
-- wb is write byte. This returns file handle
<<inner_loop>>
for i in 1 .. ceil( length( l_image_clob ) / chnksz )
loop
utl_file.put_raw( l_file_handle,
utl_raw.cast_to_raw( substr( l_image_clob, (i-1) * chnksz + 1, chnksz )));
utl_file.fflush(l_file_handle);
end loop inner_loop;
utl_file.fclose(l_file_handle);
Thanks for looking at this.

Answer 1

I found an answer. The Apex-initiated image was base64 encoded. Therefore I had to Decode it. Someone helped me with a procedure to do this. My modified code now looks like this:
-- Create a file based on the content of l_image_clob
l_image_filename := 'image_' || p_event_pkey || '_' || i ||
'.' || l_image_ext;
clob_base64_to_file(l_image_clob, l_dirname, l_image_filename);
The procedure that's called is as follows:
create or replace procedure clob_base64_to_file(
p_clob in clob,
p_dir in varchar2,
p_filename in varchar2
)
is
t_buffer varchar2(32767);
t_pos number := 1;
t_len number;
t_fh utl_file.file_type;
t_size number := nls_charset_decl_len( 32764,
nls_charset_id( 'char_cs' ) );
begin
t_fh := utl_file.fopen( p_dir, p_filename, 'wb', 32767 );
t_len := length( p_clob );
loop
exit when t_pos > t_len;
t_buffer := replace( replace( substr( p_clob, t_pos, t_size ),
chr(10) ), chr(13) );
t_pos := t_pos + t_size;
while t_pos <= t_len and mod( length( t_buffer ), 4 ) > 0
loop
t_buffer := t_buffer || replace( replace( substr( p_clob, t_pos, 1 ),
chr(10) ), chr(13) );
t_pos := t_pos + 1;
end loop;
utl_file.put_raw( t_fh,
utl_encode.base64_decode( utl_raw.cast_to_raw( t_buffer ) ) );
end loop;
utl_file.fclose( t_fh );
end;
When I call the clob_base64_to_file procedure it decodes the image and creates a file based on the directory and filename I supply in the call.