Skip to content

Latest commit

 

History

History
122 lines (81 loc) · 5.87 KB

Unicode_and_Rust.md

File metadata and controls

122 lines (81 loc) · 5.87 KB

Unicode and Rust

The Unicode Standard, and related specifications, are a complex system with interdependent terms and properties. Here’s a summary for working with Unicode when programming in Rust.

Basic Unicode Concepts

  • Unicode Abstract Characters are abstract units of information used for the organization, control, or representation of textual data.

  • Unicode Code Points are integer values in the Unicode codespace: numbers between 0 (zero) and 0x10_FFFF, inclusive.

  • Unicode Scalar Values are integer values in a subset of Unicode Code Points: the Unicode codespace excluding high-surrogate and low-surrogate code points: U+D800..U+DFFF, inclusive.

  • Unicode Encoded Characters are Unicode Scalar Values assigned to a Unicode Abstract Characters by the Unicode Standard. Some Unicode Abstract Characters are represented with a sequence of Unicode Encoded Characters.

Unicode Scalar Values marked as noncharacters or reserved (a.k.a unassigned) are not considered Unicode Encoded Characters. Therefore, Unicode Scalar Values can have one of the following assignment statuses:

  • assigned character, code points that are marked to be an Encoded Character,

    Character*, or

  • unassigned or reserved, code points that can become an Encoded Character in the future.

In contrast to Unicode Code Points and Unicode Scalar Values (which are sets of numbers written in stone), the set of Unicode Encoded Characters (a subset of Scalar Values) expands with every version of Unicode. Figure below shows the number of Unicode Assigned Characters over time, from 1991 to 2017.

Number of Unicode Assigned Characters over time

In the Unicode Standard, it is recommended that processes do not interpret unassigned or noncharacter values as Encoded Characters. This is a general recommendation and many application decide to pass through such code points.

Rust Compiler and Core Libraries

The Rust Programming language has support for Unicode characters in its core with the char primitive type, which represents a Unicode Scalar Values, that is a Unicode-version-agnostic type.

Since the list of assigned characters changes with Unicode versions, Unicode Encoded Characters are not hard-coded in the Rust language, and only made available in internal and core libraries for compilation tasks, like when parsing a string in a macro in compile-time, and basic input and output handling.

Need for Third-Party Libraries

What is needed for a programming language compiler and its extensions can never be enough for all other application of text processing. The needs of applications also varies in size and shape, and most importantly, the version of Unicode and related data being used for processing.

This is why extended Unicode data and algorithms, along with other internationalization-related functionalities, are provided as third-party libraries in any major programming language.

UNIC, is such a third-party library for the Rust programming language, which provides functionalities for Unicode and related standards atop the basic char and str primitive types in Rust. UNIC provides access to all kinds of data and algorithms that are Unicode-version-dependent.

Unicode Version: code vs. data

An important point to remember here is that the version of Unicode data provided in a third-party library, like UNIC, may be different with the one provided in Rust.

This is critical for some applications, specially those depending on processed data like search engines. Such applications should not be forced to update all stored data when their code compiler receives a newer version of Unicode.

Another reason for the separation of Unicode version of the code (the compiler) from the data is that a data processing codebase should not be held back from upgrading its Unicode version because of any limitations with upgrading its code to a newer version of the compiler.

And a peculiar case for keeping the Unicode version of the data separate from the code is when it gets to Private-Use Code Points. These code points (like any other code point) receive default character properties, which are specified by the Unicode Standard. However, they can be assigned non-default property values, as deemed by the application. This is something that should never affect how source code is being compiled, or a compiler being deployed, but only the data being processed an application with clear boundaries for the Private-Use code points.

Technical Challenges

A big technical challenge with having more than one library (a language core library and some third-party libraries) for basic character/string functionalities and text processing for handling text is the possibility of them getting mixed up. This is where every programmer need to pay attention to which library to use in every situation.

That is why having one library that provides majority of functionalities and enforces consistency between its components is important in this area. UNIC has a goal to enforce this consistency during its own development time (via integration tests), as well as in application runtime (assertion in debug mode).

Still, there is one situation that is not avoidable by third-parties: mixing core library and third-party library in one process. Here, the burden is on the application developer to avoid any mixing, or ensure and verify that it is safe to do so.

See Also