#ini-parser #minimalist #section #escaping #identifier #programming-language #value

mininip

A minimalist ini file parser (MinIniP stands for Minimalist Ini Parser). It is written in Rust but I will export its API to the C programming language in order to make various bindings

7 stable releases

1.3.1 Apr 9, 2020
1.3.0 Mar 21, 2020
1.2.3 Jan 19, 2020
1.2.2 Dec 28, 2019

#544 in Database interfaces

MPL-2.0 license

91KB
2K SLoC

MinIniP

What is MinIniP ?

Presentation

MinIniP stands for Minimalist INI Parser. It is a parser written in Rust to store datas in an easy and safe way. Currently, it is not intended to be extremly fast and there is not any benchmark of it. You can use it like you want in any project. Every change of one of the provided files must be published under the MPL-2.0.

Why MinIniP ?

Honnestly, there is not any particular reason to chose MinIniP. I just wrote it to play with Rust wich I learn a few months ago. I will use it in my personnal projects so it will be actively maintained for a long time.

You convinced me ! How to use it ?

Just add

mininip="1.2"

to your Cargo.toml in the dependencies section and you are right ! You can also download it at the official repository.

What is a valid INI file ?

A lack of standardisation

Since there is not any standard INI specification, each implementor writes its own. Here is mine. You can contribute to the project by extending this specification if you think something is missing. The only rule to follow is to not break backward compatibility, except in one case: adding a new INI type may break some use cases of the Raw type by moving a declaration of variable to the new type.

For instance, you may create a Set INI type which is like the sets in maths, and is enclosed by curly brackets {}. In this way, parsing this line

an INI key = { Hello, world, ! }

will no longer produce a Value::Raw value when parsing it but a Value::Set instead.

The specification followed by MinIniP

Identifiers

An identifier refers to either

  • A section name
  • A key name

An identifier must start with one of ., $, : or a-zA-Z. Since the second character, all of _, ~, -, ., :, $, a-zA-Z and 0-9 are allowed. In the API, an Identifier refers to a combination of a section name and a key name, so keep in mind it is not just a key name !

The specification above might be outdated, so refer to the generated documentation (Identifier::is_valid) to be aware of what is a valid INI identifier.

Values

Declaring a value

A value can be anything following the first = sign in an expression. It must be assigned to a key and not a section. In this code

key=value

key must be a valid identifier and value is defined as the value.

Types

A value can be either

  • Raw a valid value which does not match with any of the types below
  • Str a valid value inside two quotes ' or "
  • Int a 64-bits-sized integer
  • Float a 64-bits-sized floating-point number
  • Bool a boolean (either true (on, enabled, y or yes) or false (off, disabled, n or no))

The highest priority is for the type Str. Since quotes are forbidden in all the other use cases, a quoted value can only be a Str. Then, comes the Bool type which only allows a few values (see above). Then, comes Int and in case of failure while interpretting it as an integer, Float. If none of these types match with the given value, the value is Raw which is the value as written in the file (after unescaping, defined below).

Escape sequences

In an INI file, all the possible values are not accepted. For instance, you cannot store an emoji (like ☺ or ♥) or any other non-ASCII character as is in a file. It is even true for characters which may be part of the INI syntax like the semicolon ;, the colon :, the equal sign =... However, you have not to deal with these characters since MinIniP does it for you. The characters are escaped. Here is an exhaustive list of the recognized escapes sequences.

Non-escaped Escaped
\ \\
' \'
" \"
null character \0
bell / alert \a
backspace \b
tab character \t
carriage return \r
line feed \n
; \;
# \#
= \=
: \:
unicode character \xxxxxx with xxxxxx corresponding to its hexadecimal code (six digits)

Please note that escapes are not available for identifiers.

Sections

A section refers to what can be called in Rust a module, or a namespace in C++. In a few words, it is a named or anonymous set of keys. A section identifier must be a valid identifier or nothing at all. A section is declared by putting square brackets [] around its identifier on a line.

key_1 = value_1
key_2 = value_2

[section 1]
key_1 = value_3
key_2 = value_4

[section 2]
key_1 = value_5
key_2 = value_6

In this code, every occurences of key_1 and key_2 are different keys because they are not in the same section. The first ones are in the global / anonymous section, the two following ones in a section 1-named section, and the two last ones in a section 2-named section.

In this API, section names are the first value stored in an Identifier. It is a Option<String> since a section may be anonymous (the keys declared before the first section are in the anonymous section corresponding to None). All the named sections are represented as Some(name). The second value is the key, which is a String that must be a valid identifier.

No runtime deps