ada_caser is a tool to adjust the casing in an Ada source code to a "normal" form.
The normal casing form in Ada source is
- keywords are in lower case
- identifiers are in title case (
Title_Case).
There are some exceptions to the normal rule: for example, IO in Ada.Text_IO, and GNAT.
ada_caser [switches] files
| Switch | |
|---|---|
-v, --verbose |
Report progress |
-c, --charset=ARG |
Specify the character set (default "utf-8") |
-l, --language=ARG |
Specify the language version (default "Ada_2022") |
-p, --pipe |
Send the output to standard output |
-D, --dictionary=ARG |
Add a casing dictionary |
-P, --project=ARG |
Use a GNAT Project (default none) |
-V, --version |
Report the version |
If a project file is supplied, ada_caser reads the project file, and the tree of referenced projects. It then uses Libadalang to find the declaration (if any) of each identifier in the input source code and replaces the source text with that of the declaration.
If no project file is supplied, ada_caser uses a default project which allows it to access the standard library (Ada*, GNAT*, Interfaces*).
So, for example, ada.text_io.fixed_io will be replaced by Ada.Text_IO.Fixed_IO.
If you're using ada_caser with a project in an Alire crate, you should run it in an alr exec context so that it can find all the referenced projects.
For previously-declared identifiers, ada_caser will convert the identifier to the casing with which it was declared.
The default processing of undeclared identifiers is to convert the first character of each word to upper-case (reserved words are converted to lower-case, other tokens are left unchanged).
Considering proj.regular_io, proj is a single-word identifier and will be rendered as Proj, while regular_io is a two-word identifier which will be rendered as Regular_Io.
For identifiers, the default behaviour can be altered using exception dictionaries.
Exception dictionaries are line-oriented. Blank lines and comments (lines whose first non-blank character is #) are ignored. Other lines are exceptions; only the first word on a line is counted as an exception.
A casing exception can either be a full or a part exception.
- A full exception matches a whole identifier.
- A part exception matches any word in an identifier.
A part exception is marked by a leading *.
An exception dictonary containing
GNAT
*IO
would render gnat.command_line as GNAT.Command_Line, gnat_support a s Gnat_Support, and ada.text_io as Ada.Text_IO. Note, this casing is in fact provided by ada_caser for standard library projects (which contain the Ada, GNAT and Interfaces package hierarchies), but probably not for embedded projects.
It is an error to provide differing exceptions, for example Id and *ID.
Build with alr build, install with alr install.
Testing is done using Black Box Testing (bbt), available via Alire. See the test scenarios.
These scenarios also demonstrate ada_caser's features.
Run using
bbt --exact_match --yes --cleanup test.md