SVG Translation Tool

This tool extracts multilingual text pairs from SVG files and applies translations to other SVG files by inserting missing <text systemLanguage="XX"> blocks.

Installation

This tool requires Python 3.10+. Install the lightweight core dependencies with:

pip install CopySVGTranslation

Usage

Extracting and injecting in a single step

from pathlib import Path
from CopySVGTranslation import svg_extract_and_inject

tree = svg_extract_and_inject(
    extract_file=Path("examples/source_multilingual.svg"),
    inject_file=Path("examples/target_missing_translations.svg"),
    data_output_file = Path("examples/data.json"),
    save_result=True,
)

if tree is not None:
    print("Injection completed!")

The helper stores the extracted phrases under Path("examples/data.json") and, when save_result=True, writes the translated SVG to output_dir=Path("./translated"). If you also need statistics about how many translations were inserted, call the lower level injector with return_stats=True:

from CopySVGTranslation.injection import inject

tree, stats = inject(
    inject_file="examples/target_missing_translations.svg",
    mapping_files=["CopySVGTranslation/data/source_multilingual.svg.json"],
    output_dir=Path("./translated"),
    save_result=True,
    return_stats=True,
)

print(stats)

Injecting with pre-translated data

When you already have the translation JSON, load it and use inject directly. Important parameters include overwrite to update existing translations and output_dir to control where translated files are written.

from pathlib import Path
from CopySVGTranslation import inject

translations = {
    "new": {
        "Hello": {"ar": "مرحبًا", "fr": "Bonjour"},
    }
}

tree, stats = inject(
    inject_file=Path("examples/target_missing_translations.svg"),
    all_mappings=translations,
    output_dir=Path("./translated"),
    overwrite=True,
    save_result=True,
    return_stats=True,
)

print("Saved to", Path("./translated/target_missing_translations.svg"))
print(stats)

Data Model

The extractor writes a JSON document rooted under the "new" key. Each entry maps normalized English text to a dictionary of language codes and translations. An example of the modern format:

{
    "new": {
        "but are connected in anti-phase": {
            "ar": "لكنها موصولة بمرحلتين متعاكستين."
        }
    }
}

Older exports may omit the wrapper and look like {"english": {"ar": "…"}}. The injector transparently accepts both structures, but the recommended format is the nested "new" layout shown above.

Extract Example

Input SVG (arabic.svg)

<?xml version="1.0"?>
<svg xmlns="http://www.w3.org/2000/svg">
  <switch>
      <text id="t0-ar" systemLanguage="ar">
          <tspan id="t0-ar">الموسيقى في عام 2020</tspan>
      </text>
      <text id="t0-fr" systemLanguage="fr">
          <tspan id="t0-fr">La musique en 2020</tspan>
      </text>
      <text id="t0">
          <tspan id="t0">Music in 2020</tspan>
      </text>
  </switch>
  <switch>
      <text id="t0-ar" systemLanguage="ar">
          <tspan id="t0-ar">مرحبا</tspan>
      </text>
      <text id="t0-fr" systemLanguage="fr">
          <tspan id="t0-fr">Bonjour</tspan>
      </text>
      <text id="t0">
          <tspan id="t0">Hello</tspan>
      </text>
  </switch>
</svg>

Python code

from pathlib import Path
from CopySVGTranslation import extract

translations = extract(
    svg_file_path=Path("arabic.svg"),
    case_insensitive=True,
)

Extracted JSON

{
    "new": {
        "music in 2020": {
            "ar": "الموسيقى في عام 2020",
            "fr": "La musique en 2020"
        },
        "hello": {
            "ar": "مرحبا",
            "fr": "Bonjour"
        }
    },
    "title": {
        "music in": {
            "ar": "الموسيقى في عام",
            "fr": "La musique en"
        }
    },
    "tspans_by_id": {
        "t0": "Hello"
    }
}

Injection Example

• TODO

Testing

Run the unit tests:

python -m pytest tests -v

Implementation Details

Text Normalization

The tool normalizes text by:

• Trimming leading and trailing whitespace
• Replacing multiple internal whitespace characters with a single space
• Optionally converting to lowercase for case-insensitive matching

ID Generation

When adding new translation nodes, the tool generates unique IDs by:

• Taking the existing ID and appending the language code (e.g., text2213 becomes text2213-ar)
• If the generated ID already exists, appending a numeric suffix until unique (e.g., text2213-ar-1)

Error Handling

The tool includes comprehensive error handling for:

• Missing input files
• Invalid XML structure
• Missing required attributes
• File permission issues