A Language Module is a Data Object that represents a language—either internally within Jitran or externally—in a way that is semantically and syntactically compatible with other languages. A language module exists externally as a file or directory and internally within Jitran as a data object or a Java object.

IR is the bridge between the languages. This means that the semantic information of the Input Language is contained within the IR nodes. The IRs are therefore the Data Object that contain all the semantic (1.3.1/semantics) information of the input language. There are different types of IRs, each of which may contain different IR nodes; for this reason, the entire Data Object is also called an IR Tree.

JTX stands for Jitran Syntax and is the language that defines the Syntax of a language. This information, which comes from JTX, is used by the Frontend and Backend to translate the language.

The Frontend and Backend are programs, which may be programmed to handle more complex logic and integrated into syntax.jtx. The frontend is used during the Transformation Stage, and the backend is used during the Generation Stage.

Jitran's pipeline consists of various steps in which the source code is analyzed, converted to IR, and then the output is generated from the IR. The process of analyzing the source code written in the Input Language and converting it into IR nodes is described as the Transformation Stage. The process of generating the output code in the Output Language from the IR is referred to as the Generation Stage.

Definition

The lexer converts a sequence of characters into a sequence of tokens based on the syntax of the input language.

Input

A sequence of characters representing the source code.

Output

A sequence of tokens.

Semantics

• The lexer must process characters in order.
• The lexer must produce tokens according to the defined token types.
• The lexer must not modify the input sequence.

Definition

The parser consumes tokens and generates IRs based on syntax elements. The syntax elements are based on the language module of the input language.

Input

A sequence of tokens.

Output

A IR Tree.

Semantics

• The parser must process tokens in order (priority order).
• The parser must be depending on syntax elements defined in the input language.

Definition

The renderer gets the ir tree and generates a output string based on syntax elements. The syntax elements are based on the language module of the output language.

Input

A ir tree.

Output

Code as string

Semantics

• The renderer must process irs in order (hierarchical direction).
• The renderer must be depending on syntax elements defined in the output language.

Definition

The Transformation Stage is the first part of the Jitran pipeline after the language modules and source code have been loaded; it converts the source code into IR. It consists of the lexer, the parser, and the frontend.

Input

A sequence of characters representing the source code.

Output

A IR Tree.

Semantics

• The transformation stage must go through individual steps in a strict sequence.
• The transformation stage must be based on the input language.
• The transformation stage must be deterministic.
• The transformation stage may use the Language Module Metadata of the output language.
• The transformation stage must not depend on syntax elements of the output language.

Definition

The Generation Stage is the second part of the Jitran pipeline after the transformation stage has generated IR Nodes; it generates the output based on IR. It consists of the renderer, and the backend.

Input

A IR Tree

Output

An output code.

Semantics

• The generation stage must execute its components in a defined order.
• The generation stage must be based on the output language.
• The generation stage must be deterministic.
• The generation stage may use the Language Module Metadata of the input language.
• The generation stage must not depend on syntax elements of the input language.

A language module is designed to represent a language for Jitran. It must be in either directory format or internal format. It contains various data components.

Components List

The following components must not be more than once.

Semantics

• Manifest, Syntax and Native Functions must defined.

• Frontend and Backend may defined.

Definition

The Directory Format is used for developing a language module and includes the components listed above (2.2.1 Components). Each component listed above is represented by a file in a main directory. To use the Language Module it must compress the directories into a language module file (language-id.jtlm).

Semantics

• All files must be located in the root directory.

• Manifest, Syntax and Natives files must exists.

• All files should have a valid filepath (no characters like '\s', '\n', '"'...).

Definition

The Internal Format means that the language module resides in the JVM as a java object, and all data is initialized as fields of that java object.

Internal Fields

All fields are final so they can not modify by other Language Modules.

All of the following fields refer to the language modules and represent java objects:

The IR (Intermediate Representation) serves as the sole medium of exchange between two language modules. In this process, the source program is passed semantically from the transformation stage to the generation stage.

The goal of IR is to capture the semantics of as many programs as possible written in as many languages as possible.

Frontend

The frontend is a subprogram of Jitran that serves as userdefined logic for translating any logic in the transformation stage.

Backend

The backend is a subprogram of Jitran that serves as userdefined logic for translating any logic in the generation stage.

Language

Both are write in JigsawScript but are compiled in java bytecode.

The execution model of Jitran defines how the pipeline is executed to transform source code into output code.

Jitran executes its pipeline in a strictly ordered sequence of stages. Each stage consumes the output of the previous stage and produces a new intermediate result.

The execution is deterministic, meaning that the same input always produces the same output.

Overview

The Manifestdata is a component of a language module and contains the metadata of the. There are only one per language module.

In jitran the manifest data has an directory format and an internal format.

General

The Directory format consists of a file named manifest.jtm. The file is located in the root directory of the language module (in Directory Format).

Semantics

• The manifest filename must be end with .jtm.
• The manifest filename should be manifest.jtm.
• There must be only 1 manifest per language module.

Internal Format

In the Internal Format, the manifest data is stored as a data object in Jitran Core.

Components

All following components representing java fields of the java class.

All components are final so they can not modify by other Language Modules.

All of the following components refer to the manifest and represent java objects:

Semantics

• The Manifest must defined before the transformation stage.

Overview

The Manifest file contains the manifest data of the language module.

• The manifest filename must be end with .jtm.
• The manifest filename should be manifest.jtm because the name has the highest priority in case of more than one file.

Structure

A metadata consists on an metadata-tag followed by an colon and a value (<tag>:<value>).

Values

Every tag is assigned to a value.

Values (Semantics)

• A boolean must writen in lower case.
• A boolean has two states writen true (comparable with yes) and false (comparable with false).
• A string must start and end with an " (dopple quode) or ' (quode).
• A string may use escape sequences.
• An integer must writen as an single decimal number.
• A float must starts with a decimal number the integer part.
• A float must ends with a decimal number the float part after a . (dot).
• A float with a float part with value 0 may omit the float part and the . (dot).

Metadata Tag