manga109

This library is a type-based API client for Manga109.

Manga109 is a dataset of manga (Japanese comics) for academic research. An API client already exists (here) for the dataset, but the dictionary-based data structures require users to have a detailed understanding of the dataset (key names, value types, etc.). This library makes active use of dataclass, introduced in Python 3.7 (there is also a backport for earlier versions), to allow users to access any data while benefiting from completions and type hints.

Show data structures as YAML.

books:
  - book:
      title: str
      characters:
        - character:
            id: str
            name: str
      pages:
        - page:
            index: int
            width: int
            height: int
            bodies:
              - body:
                  id: str
                  xmin: int
                  ymin: int
                  xmax: int
                  ymax: int
                  character: str
            faces:
              - face:
                  id: str
                  xmin: int
                  ymin: int
                  xmax: int
                  ymax: int
                  character: str
            frames:
              - frame:
                  id: str
                  xmin: int
                  ymin: int
                  xmax: int
                  ymax: int
            texts:
              - text:
                  id: str
                  xmin: int
                  ymin: int
                  xmax: int
                  ymax: int
                  text: str
            image_path: str

Usage

Basic Usage

Make instance

from manga109 import Manga109

client = Manga109("path/to/manga109")

# You can use specific titles.
client = Manga109("path/to/manga109", titles=["ARMS", "GakuenNoise", "PrayerHaNemurenai"])

Access any information

print(client.books[0].title)  # ARMS
print(client.books[0].characters[14].name)  # タイロン
print(client.books[0].pages[80].texts[5].text)  # 返事して

Get the path to the image

print(client.books[0].pages[17].image_path)  # path/to/manga109/images/ARMS/017.jpg

Advanced Usage

Get the number of characters per title

print([{b.title: len(b.characters)} for b in client.books])
# [{'ARMS': 23}, {'AisazuNihaIrarenai': 17}, {'AkkeraKanjinchou': 16}, ...]

Get a list of texts per page with one or more texts

print([[t.text for t in p.texts] for p in client.books[0].pages if p.texts])
# [['あ'], ['キャーッ', 'はやく逃げないとまきぞえくっちゃう', ...], ...]

Installation

• Python 3.7+

pip install git+https://github.com/skmatz/manga109.git