Translate TRA Files for Infinity Engine Mods

This skill helps you work with TRA localization files and dialogue scripts for Infinity Engine games (Baldur's Gate, Icewind Dale, etc.). It provides translation, text processing, and dialogue management capabilities for game modding projects.

What This Skill Does

Translates TRA localization files using multiple engines (Google Translate, DeepL, Google Cloud)

Processes Russian text with proper ё/е handling using dictionaries

Parses D dialogue files using ANTLR4 grammar

Extracts and translates game dialogues with gender-specific variants

Updates TRA files with translations while preserving formatting

Manages file encoding detection and conversion

Handles dictionary-based context-aware translations

Instructions

Step 1: Understand the TRA File Format

TRA files contain localization strings in these formats:

```

@1 = ~English text here~

@2 = "Another format with quotes"

```

Scripts automatically detect delimiters (`~~` or `""`) and preserve formatting. Encoding is auto-detected using chardet (UTF-8, CP1251, CP1252, etc.).

Step 2: Set Up Python Environment

Before working with TRA files, ensure all dependencies are installed:

```bash

pip install googletrans==4.0.0rc1 textblob pycld2 google-cloud-translate deepl colorama click chardet levenshtein nltk antlr4-python3-runtime

```

If a virtual environment exists in `venv/`, activate it first:

```bash

source venv/bin/activate

```

Step 3: Translate TRA Files

Use `translate_tra.py` for basic TRA file translation:

```bash

python3 translate_tra.py infile.tra [--engine=googletrans|deepl|googlecloud] [--lang=ru] [--out=output.tra]

```

**Options:**

`--engine`: Translation engine (googletrans [free], deepl [API key required], googlecloud [auth required])

`--lang`: Target language (default: ru for Russian)

`--out`: Output file path (default: overwrites input)

**Default engine is DeepL**, which provides high-quality translations but requires an API key.

Step 4: Process Russian Text (ё/е Conversion)

For Russian translations, use `yo_making.py` to convert е to ё based on dictionaries:

```bash

python3 yo_making.py infile.tra [--out=output.tra] [--always-no]

```

This uses two dictionaries:

`russian_yo_words.txt` - Words that always use ё

`russian_yo_words_both.txt` - Words that can use either (interactive mode)

Use `--always-no` to skip interactive prompts for ambiguous cases.

Step 5: Parse Dialogue Files (D Files)

For dialogue-heavy mods, use the grammar-based parser to extract dialogues:

```bash

python3 parse_d_grammar.py d_folder tra_folder output_folder

```

**What this does:**

Parses D dialogue files using ANTLR4 grammar (`DParser.g4`, `DLexer.g4`)

Handles all dialogue constructs: `IF-THEN`, `CHAIN`, `APPEND`, `EXTERN`

Extracts SAY lines (NPC dialogue) and REPLY lines (player responses)

Resolves `@number` references from TRA files

Creates one CSV per connected dialogue with proper flow tracking

Generates `_unused_tra_lines.csv` for lines not used in dialogues

**D File Constructs Supported:**

`APPEND ~NPC~ ... END` - Append to existing NPC dialogue

`IF ~condition~ THEN BEGIN state_id ... END` - Conditional dialogue states

`SAY @number` - NPC dialogue line (references TRA)

`++ @number GOTO target` - Player reply options

`CHAIN ... EXIT` - Sequential dialogue chains

`EXTERN ~NPC~ state` - Jump to another NPC's dialogue

Step 6: Translate Dialogue CSV Files

Translate extracted dialogues using local LLM (Ollama):

```bash

python3 translate_dialogs.py [source_folder] [--target target_folder] [--model model_name]

```

**Features:**

Translates CSV dialog files from English to Russian

Supports gender-specific translations for Player1

Handles `_unused_tra_lines.csv` with batch translation (10 lines per request)

Uses `actors.csv` for character information

Uses `dict.csv` for context-aware translation dictionary

**Dictionary Features:**

Word boundary matching (avoids partial matches)

Prioritizes longer phrases over shorter ones

Supports multiple translations with context

Step 7: Update TRA Files with Translations

Apply translated dialogues back to TRA files:

```bash

python3 update_tra_from_csv.py tra_folder csv_folder [--output output_folder]

```

**Behavior:**

Reads translated dialogue CSVs (male/female variants) and unused lines CSV

Updates original TRA files with Russian translations

Preserves original formatting, comments, and sound references

Gender handling: same translation → single line; different → `~male~ ~female~` format

Only updates lines with TRA references (ignores D file direct text)

Output encoding: UTF-8

Step 8: Additional Utilities

**Find Similar Lines:**

```bash

python3 find_closest_lines_in_tra.py

```

Finds similar lines between TRA files using Levenshtein distance.

**Extract Specific Lines:**

```bash

python3 extract_lines_tra.py

```

Extracts specific lines from TRA files based on criteria.

**Extract Actor Names:**

```bash

python3 extract_actors_using_grammar.py

```

Extracts actor names from D files using the grammar parser.

**Generate Renaming Tables:**

```bash

python3 generate_global_renaming_table.py

```

Creates renaming tables for game files with prefixes.

**Encoding Conversion:**

```bash

./utf8_2_cp1251.sh # Convert UTF-8 to CP1251

./cp1251_2_utf8.sh # Convert CP1251 to UTF-8

```

Important Notes

**Always use `parse_d_grammar.py`** for dialogue parsing (robust ANTLR4-based parser)

All scripts auto-detect file encoding using chardet

TRA file formatting and comments are preserved during updates

Translation engines require proper authentication (DeepL API key, Google Cloud auth)

For Russian mods, run `yo_making.py` after translation for proper ё/е handling

Dictionary-based translations (`dict.csv`) support context-aware word/phrase replacement

Gender-specific dialogue variants are handled automatically in the translation pipeline

Typical Workflow

1. Extract dialogues: `parse_d_grammar.py d_folder tra_folder csv_output`

2. Translate dialogues: `translate_dialogs.py csv_output --target translated_csv`

3. Update TRA files: `update_tra_from_csv.py tra_folder translated_csv --output updated_tra`

4. Fix Russian ё: `yo_making.py updated_tra/*.tra`

This workflow ensures proper dialogue flow tracking, context-aware translations, and correct formatting preservation.

Translate TRA Files for Infinity Engine Mods

Translate TRA Files for Infinity Engine Mods

What This Skill Does

Instructions

Step 1: Understand the TRA File Format

Step 2: Set Up Python Environment

Step 3: Translate TRA Files

Step 4: Process Russian Text (ё/е Conversion)

Step 5: Parse Dialogue Files (D Files)

Step 6: Translate Dialogue CSV Files

Step 7: Update TRA Files with Translations

Step 8: Additional Utilities

Important Notes

Typical Workflow

Reviews (0)