Beginning with version 5.0, CSPro is Unicode compliant.
Unicode is a widely adopted system for representing characters for all languages currently in use. Early computer programs generally used only one byte to represent a character, which led to a limit in the number of characters that could be displayed on screen and used in computations. This limit of 256 characters was used effectively by people who only required English characters, or characters from most European languages, but it could not represent the languages used by more than a billion people, including most notably, many Asian languages.
The Unicode standard now defines more than 100,000 characters, but many of these characters represent extinct languages. For that reason, Windows only provides native support for a subset of Unicode characters. In a Unicode program, a character is represented by two bytes, which allows for up to 65,536 characters.
Whereas in English it is very clear what makes up one character—one keystroke—in other languages it is not as straightforward. For example, in Chinese, typing two characters, "天津," requires seven keystrokes ("Tianjin"). In Bangla, the character পি requires two keystrokes (প + ি), which combine to create one character. Both the Chinese and Bangla examples require four bytes in memory and return a string length of 2.
With each character stored as two bytes in memory, there are several ways to write characters to a disk. One way is to write two bytes for every character, but this is costly for most users, the bulk of whom only use characters that can be expressed properly with either the ASCII or ANSI encoding systems. The computer world has settled on using UTF-8, a variable-length encoding scheme that uses between one and four bytes to represent a character. ASCII characters are represented using one byte, other ANSI characters are represented using two bytes, and most Asian characters are represented using three bytes. To identify a file as encoded in UTF-8, a three-byte BOM (byte order mark) is placed at the beginning of a text file. For example, here is the UTF-8 representation of: "I am François from 法国."
With a UTF-8 encoding, a file's size is not equal to the number of characters in the file. An empty file is three bytes because of the BOM. (The fileempty
function can be used to determine whether a file is actually empty.) An ASCII file converted to UTF-8 will generally be the size of the original file plus three bytes. An ANSI file that uses non-English characters will be even larger. The name François takes eight bytes to represent in ANSI but uses nine bytes in UTF-8.
Versions of CSPro up to 4.1 used ANSI characters for data files and all specification files. Starting with version 5, CSPro uses the UTF-8 encoding scheme for representing Unicode characters. CSPro will always be able to read both ANSI and UTF-8 files, so any old files will work in the new version, but when files are modified and saved the files will be rewritten in UTF-8. All new data files and listing reports will be created in UTF-8. Because earlier versions of CSPro only supported ANSI files, any file created in CSPro 5.0 or later, including data files, is no longer automatically backwards compatible with older versions of CSPro.
If you only use ASCII characters in your specification and data files, your new files will be only three bytes larger than the CSPro 4.1 equivalents. These three bytes reflect the size of the UTF-8 BOM. If you use many accented characters that are valid in ANSI (like à and û), then your files will increase in size, though only slightly, because the vast majority of characters in any CSPro file are still digits and Latin characters. In general your files will increase in size by an inconsequential amount.
The CSPro installation includes a tool, the Unicode Text Converter, that allows you to change the encoding of text files. This utility allows you to convert ANSI files to UTF-8, though this is not necessary because CSPro 5.0 or later does this automatically. You will more likely use the tool to convert UTF-8 files to ANSI. If you use non-ANSI characters in your specification or data files, these characters will be converted to question marks during the conversion to ANSI. For instance, "I am François from 法国." will become "I am François from ??." Converting files from UTF-8 to ANSI is not necessary a lossless conversion, so it should be done only if absolutely necessary of if you are confident that you did not use any Unicode-only characters.
Some text editors, including Notepad, allow a user to change the encoding of a text file. These editors can be used to convert text files from UTF-8 to ANSI. If using Notepad, open the file and then select Save As. At the bottom of the dialog box is an option to change the encoding.
The important binary files created by CSPro are compiled data entry applications (.pen), file indices (.idx), and intermediary tabulation files (.tab and .tai). These formats have changed and CSPro can no longer read binary files written by CSPro 4.1 or previous versions. CSPro can determine if a compiled data entry application is of the correct format and will present an error message to an operator who tries to open an old version in CSEntry. CSPro can also determine if an old index is being used and will automatically generate a new index. If you use intermediary tabulation binary files (e.g., if you run tables in parts), you should delete and regenerate them when upgrading to a Unicode version of CSPro. The Unicode Text Converter will not work correctly on binary files.
Outside of the Export Data tool, CSPro no longer supports writing to ANSI format, though it can still read files encoded in ANSI. CSPro automatically converts files from ANSI to UTF-8 formats as needed. CSPro will convert only files that may be modified by a program. For example, if you use an external file in your application but only call fileread and never filewrite, CSPro will keep the input file in its original format. However, if there is a filewrite statement and the file is encoded in ANSI, CSPro will convert the file to UTF-8 before opening it. The old ANSI file will be sent to the Recycle Bin in case you want to recover it.
Most applications support Unicode so you may not have a problem using external applications with your data or specification files. However, if you must use an application that is not Unicode compliant, run the Unicode Text Converter on the necessary files.
If you are distributing files to a broad audience and if your data file does not contain any non-ANSI characters, you might consider converting it to ANSI for maximum compatibility. To illustrate an example of what might happen with a non-Unicode program reading a UTF-8 file, imagine a CSPro data file that has two record types, P (population) and H (housing). The ANSI file might look like this:
The UTF-8 file might look like this:
An older version of CSPro would regard the first row of data as an invalid record because the BOM was in the place of the record type. It would thus register the household as having two members instead of three. If any non-ASCII characters existed in the data file, it would further shift the data and then items would not be read correctly.
If you release a UTF-8-encoded data file widely, you may want to include a documentation file with text similar to the following:
Warning: This CSPro data file is encoded in UTF-8. UTF-8 is a text encoding format widely used to represent characters from any language. If you use this data file with a software package that does not support Unicode and UTF-8, you must first convert the data file to an ANSI encoding. Without this conversion there is a risk that the software package will misread parts of the data file.