Fortran Wiki
ucs4

The optional Fortran 2003 ISO_10646 standard

The Fortran 2003 standard first defines support for processing of Unicode UTF-8-encoded files. There are three simple main points:

1. The option ENCODING=“UTF8” on OPEN() statements indicates to automatically encode and decode formatted data from UTF-8 files to binary UCS-4 internal values.

2. This new UCS-4 type is expressed as a CHARACTER variable declared to have KIND=“ISO_10646”,

3. The functionality provided includes overloading the ASCII character-related intrinsics, comparitive operators, and assigment with support for processing of UCS-4 encoded data.

These three simple additions combined conveniently make it so the processing of UCS-4 encoded data can be coded with the same methods as used for the ASCII-7 encoded data historically supported by the default intrinsic CHARACTER type.

That is, these features make interacting wth UTF-8 files virtually effortless.

The default CHARACTER kind is still required

One cannot quite completely quit using ASCII yet. Even though the Fortran standard allows processor-dependent characters to appear in comments and character constants (e.g. quoted strings) it is not mandated what encoding of characters are allowed there. UTF-8 is largely a de-facto standard for file encoding so it is extremely likely you can compile UTF-8 Fortran source files as long as the multi-byte characters are restricted to comments and constant strings, but it is not guaranteed by the Fortran Standard.

Indeed, the Fortran standard defines most interactions with operating systems such as filenames and command line character encoding as implementation-dependent and requiring the use of a default CHARACTER kind.

So in practice almost all processors require data passed to and from the system to be encoded as byte streams of UTF-8 characters, not as UCS-4 data. This includes arguments passed in from command lines and environment variables, filenames on INQUIRE and OPEN statements and string constants.

However, since that is all processor-dependent as far as the standard is concerned Fortran does not support intrinsics that convert to and from the internal UCS-4 representation and UTF-8 byte streams other than the afore-mentioned automatic conversion on READ and WRITE statements where the ENCODING=“UTF-8” option has been used on an OPEN() statement.

Further details follow:

Introduction to Fortran ISO_10646 (UCS-4-encoded Unicode) support

• Lesson I: reading and writing UTF-8 Unicode files
• Lesson II: creating Unicode strings in ASCII Fortran source files
• Lesson III: mixing ASCII and UCS4 kinds as regards
  • assignments
  • concatenation
  • passing arguments to external ASCII libraries
  • mixing kinds on I/O argument lists
• Lesson IV: what is and is not supported with internal READ and WRITE statements
• Lesson V: processing Unicode file names on OPEN() statements
• Lesson VI: reading UTF-8 strings from command lines and environment variables.
• Lesson VII: passing Unicode strings to and from C
• Summary putting it all together

• NEXT