indent_tokenizer 0.4.0

Generate tokens based on indentation
Documentation

= Identation tokenizer

A small an simple indentation tokenizer

A similar project is https://github.com/jleahred/indentation_flattener[indentation_flattener]. In this case we flatten the input adding PUSH_INDENT, and POP_INDENT. This looks better for PEG grammars.

== Usage

Add to cargo.toml [source, toml]

[dependencies] indent_tokenizer = "0.2.0"

See example below

== Modifs

0.2.0:: Removed general types (as String, u32 or usize) + Using concrete types (new types)

== Indentation format

Tabs are no valid on indentation grouping.

Let's see by example.

.Simple valid input

..... ... .... .... .... .... .... .... .... ....

Indentation groups can have any number of spaces

.Valid indentation different spaces

..... level0 .... level1 <-- .... level2 .... level1 <-- .... level1 <-- .... level0 .... level0 .... level1 <--

It's not a good idea to have same level with different spaces, but it's allowed when you are creating a new level.

In this example, last level1 is idented with more spaces than previus ones

.Invalid indentation

..... ... .... .... .... <-- incorrect indentation .... <-- correct previous ident level .... .... .... ....

In order to go back a level, the indentation has to match with the previous on this level.

As we saw in previous example, increasing level is free indentation.

.Start line indicator

|..... |..... |...... |......

You can start lines with |, but it's optional.

.Indentation indicator is optional

..... |..... ...... ...... ......

Look that | is one position previous to indentation level.

It is usefull when you need to start with spaces.

.I want to start a line with spaces

..... | ..... <- This line starts with an space | ...... <- Starting with 2 spaces |..... <- starts with no spaces ..... <- starting with no spaces ... <- starting with no spaces

.My line starts with a |

..... ||..... This line starts with a | |...... This one starts with .

A line is empty when there are no content or it only has spaces.

.Empty lines

..... ..... ..... ..... ..... next line is empty

.....     next line is empty

..... ..... next line is empty


What if I want represent empty lines?

.Representing empty lines

..... ..... ..... ..... ..... I want new line after this line |

.....     and three new lines, please

| | |


What if I want to represent spaces at end of line?

Spaces at end of line will not be erased, therefore, you don't need to do anything about it.

But could be intesting to represent it because some editors can run trailing or just because you can visualize it.

.Representing spaces at end line

..... ..... ..... ..... This line keeps 2 spaces and end | and you know it

Next line is properly indented and only has spaces

| |


In fact, you can write | at end of all lines. It will be removed.

Next strings, are equivalent.

.| it's optional at end of line

.....| .....| .....| .....|

..... ..... ..... .....


But I could need a pipe | at end of line

.pipe at end of line

..... ..... ..... ..... This line ends with a pipe||


.Pitfall

|..... ..... <- Invalid, remember, indentation mark | is previus to real indentation

|..... ..... <- This is OK, but not elegant

| .... <- I want to start with an space |..... <- This is redundant, but more clear


== Tokens

  • Each change of leven represent an end of token.
  • An empty line, is used to separate tokens on same level
  • A token contain lines and a list of tokens

.Tokens

This is the first token This is another token, because it's on a different level And another token This is also a different token

A token can contain multiple lines This is another token with three lines

Empty lines can be used to separate tokens This is a token, that continues here. Next empty line define a token division

And this is a different one
with a couple of lines

== Identation tokenizer API

Version 0.2 removed general types as String, usize, u32...

Instead, it's created an specific type on each context.

Concrete types:: [source, rust]

#[derive(Debug, PartialEq, Copy, Clone)] pub struct LineNum(u32);

#[derive(Debug, PartialEq, Clone, Eq)] pub struct SLine(String);

  • LineNum to represent the line number
  • SLine to respresent the line string

Internally, the system uses more new types as NSpaces to represent number of spaces

Function to call:: [source, rust]

pub fn tokenize(input: &str) -> Result<Vec, Error>

Token type:: [source, rust]

#[derive(Debug, PartialEq)] pub struct Token { pub lines: Vec, pub tokens: Vec, }

Error type:: [source, rust]

#[derive(Debug, PartialEq)] pub struct Error { pub line: LineNum, pub desc: String, }

Thats all

Look into lib.rs to see the api and tests.rs to se examples

== Examples

You can look into tests.rs, there are several tests.

.Complex example [source, rust]

let tokens = tokenize("

0 || 01a 01b 01c

 02a
 02b

    |020a
    ||020b

    |  021a
    |021b

1a 1b 11a ||11b 11c

12a  ||

|12b || 2a 21a 21b | |

")

The result will be

[source, rust]

vec![Token { lines: vec![SLine::from("0")], tokens: vec![Token { lines: vec![SLine::from("| 01a"), SLine::from("01b"), SLine::from("01c")], tokens: vec![], }, Token { lines: vec![SLine::from("02a"), SLine::from("02b")], tokens: vec![Token { lines: vec![SLine::from("020a"), SLine::from("|020b")], tokens: vec![], }, Token { lines: vec![SLine::from(" 021a"), SLine::from("021b")], tokens: vec![], }], }], }, Token { lines: vec![SLine::from("1a"), SLine::from("1b")], tokens: vec![Token { lines: vec![SLine::from("11a"), SLine::from("|11b"), SLine::from("11c")], tokens: vec![], }, Token { lines: vec![SLine::from("12a |"), SLine::from("12b |")], tokens: vec![], }], }, Token { lines: vec![SLine::from("2a")], tokens: vec![Token { lines: vec![SLine::from("21a"), SLine::from("21b"), SLine::from(""), SLine::from("")], tokens: vec![], }], }];