Aksharamukha Script Converter

The backend is a Python 3 application using Flask. To run it, install the required dependencies and execute the main entry point.

python3 main.py

The easiest way to run the full Aksharamukha stack (frontend, backend, and local font server) is via Docker Compose. This orchestrates all necessary containers and binds the web application to http://localhost:12345.

# Inside the project's root folder
docker compose up

The project provides several JSON-based data resources located in aksharamukha-back/resources for script mappings, matrices, and syllabaries:

• Script Mapping: Found in script_mapping/. Characters with approximate mappings to the generic Indic scheme are marked with ʽ (U+02BD). You may need to strip this character during post-processing.
• Script Matrix: Found in script_matrix/. Filenames indicate the guiding romanization and chunk size (e.g., script_matrix_IAST5.json uses IAST and chunks of 5).
• Syllabaries: Found in syllabary/. Contains lists of vowels, consonants, and consonant-vowel compounds for each script.
• Sanskrit/Pali Conjuncts: Found in conjuncts1/ and conjuncts2/. Files are split by the number of consonants and include a suffix for the appended vowel (e.g., conjuncts_Gujarati_o.json contains Gujarati conjuncts with the vowel /o/).

The /api/convert_docx endpoint handles DOCX files. It expects the input text to be a comma-separated string where the first part is a Base64 encoded version of the file content.

Request Body (JSON):

• source: Source script.
• target: Target script.
• text: A string in the format [base64_encoded_content],[file_part_identifier].
• nativize: Boolean.
• preOptions: List of strings.
• postOptions: List of strings.

Response:

• A JSON string containing the converted file in the format [base64_encoded_content],[file_part_identifier].

POST /api/convert_docx
{
  "source": "Devanagari",
  "target": "Latin",
  "text": "base64_data,file_id",
  "nativize": true,
  "preOptions": [],
  "postOptions": []
}

The /api/convert_html endpoint uses transliterate_file.convert_html to process HTML strings, preserving the structure while converting the text content.

Request Body (JSON):

• source: Source script.
• target: Target script.
• text: The HTML string.
• nativize: Boolean.
• preOptions: List of strings.
• postOptions: List of strings.

Response:

• A JSON string containing the converted HTML.

POST /api/convert_html
{
  "source": "Devanagari",
  "target": "Latin",
  "text": "<p>नमस्ते</p>",
  "nativize": true,
  "preOptions": [],
  "postOptions": []
}

The /api/conjuncts endpoint returns lists of conjunct characters for a specific script and vowel.

Request Body (JSON):

• script1: The source script.
• script2: The target script.
• vowel: The vowel to associate with the conjuncts.
• postoptions: List of post-processing options.

Response:

• A JSON object containing multiple levels of conjuncts (e.g., conjuncts1S1, conjuncts1S2, etc.) representing different complexity levels or groupings.

POST /api/conjuncts
{
  "script1": "Devanagari",
  "script2": "Latin",
  "vowel": "a",
  "postoptions": []
}

Use the /api/detectpre endpoint to find appropriate pre-processing options for a specific script and text. This helps in preparing text for accurate transliteration.

Request Body (JSON):

• text: The text to analyze.
• source: The detected or known source script name.

Response:

• A JSON object containing recommended pre-processing options.

POST /api/detectpre
{
  "text": "your text here",
  "source": "Tamil"
}

The core functionality of the API is provided via the /api/convert endpoint. It allows transliterating text from a source script to a target script with optional pre/post-processing and nativization.

Request Body (JSON):

• source: The source script name.
• target: The target script name.
• text: The text to be converted.
• nativize: Boolean (true/false) to determine if the output should be nativized.
• preOptions: List of strings representing pre-processing steps.
• postOptions: List of strings representing post-processing steps.

Response:

• The converted text (newlines are replaced with <br/>).

POST /api/convert
{
  "source": "Devanagari",
  "target": "Latin",
  "text": "नमस्ते",
  "nativize": true,
  "preOptions": [],
  "postOptions": []
}