New translation modules architecture

[Ph0enixKM]

#582

Changes

Each translate method of every SyntaxModule now generates a new pseudo AST that represents Bash output. The root and the spirit of the new AST is TranslationFragmentable (located in src/translate/fragments/fragment.rs) that is a trait that implements each Translation Fragment.

New components

There are nine (or so) new implementations of said TranslationFragment:

• raw - The most atomic translation fragment. It has been introduces so that we don't have to implement all of the Bash syntax and keep the pseudo AST minimal as possible. It represents some textual Bash code and renders to verbatim the same value that it stores.
• block - Represents a sequence of TranslationFragment separated with new line character. Handles indentation and proper bash scoping.
• comment - Represents a Bash comment (needed for correctly detecting if there is any logic is in a block)
• compound - Represents a sequence of Translation Fragments. It can represent an expression, statement, or even a part of statement.
• eval - Represents an evaluation statement which handles escaping of special symbols inside of the nested expression eval "$variable=\${variable}".
• interpolable - Represents a region that can be interpolated with other values. In case of Amber this type of fragment is generated by command and text literal.
• list - Represents a sequence of translation fragments separated by a custom separator (in Bash we separate values and arguments with spaces)
• subprocess - Represents a correctly handled and escaped bash subprocess $(...)
• var - Represents variable's value access, handles length getting, index getting, and array slicing. I works in two modes:
  • Name - renders raw name of the variable (used by builtin nameof)
  • BashRef - renders variable when passing it by reference (used by function invocation when passing ref variables)
  • BashValue - renders variable when passing by copy accordingly to its type.

New compiler workflow

The compiler workflow now looks as the following:

1. Parse code to Amber's AST
2. Translate Amber's AST to Bash pseudo AST
3. Render Bash pseudo AST to Bash code

---

If you have any questions, please let me know in the comments and I'll update this description accordingly.

[Ph0enixKM]

In the meantime I'd like to show you that I made the bash script much more readable with this PR in terms of fixed indentation errors

[hdwalters]

Some minor additional comments; otherwise looks good.

[hdwalters]

May I suggest either changing this instance to Nameof, or all other instances to NameOf (personally I prefer the second option) to assist case sensitive searches?

[hdwalters]

This looks better, but the reason I suggested the uncollapsed version is that it has both if let TranslationFragment::Empty = failed predicates at the same level, highlighting that you have different code branches doing similar things.

[hdwalters]

This file does not appear to serve a purpose. Was it added by accident?

[b1ek]

lets merge