From a7d4b9f1efdba4bff6fb5710ef846931260d4de4 Mon Sep 17 00:00:00 2001
From: joan <68073738+joanitolopo@users.noreply.github.com>
Date: Sat, 24 Feb 2024 21:24:43 +0700
Subject: [PATCH] Closes #10 | Create beaye_lexicon dataset loader (#320)

* Create beaye_lexicon dataset loader

* add implementation of eng-day word pairs
---
 .../sea_datasets/beaye_lexicon/__init__.py    |   0
 .../beaye_lexicon/beaye_lexicon.py            | 116 ++++++++++++++++++
 2 files changed, 116 insertions(+)
 create mode 100644 seacrowd/sea_datasets/beaye_lexicon/__init__.py
 create mode 100644 seacrowd/sea_datasets/beaye_lexicon/beaye_lexicon.py

diff --git a/seacrowd/sea_datasets/beaye_lexicon/__init__.py b/seacrowd/sea_datasets/beaye_lexicon/__init__.py
new file mode 100644
index 000000000..e69de29bb
diff --git a/seacrowd/sea_datasets/beaye_lexicon/beaye_lexicon.py b/seacrowd/sea_datasets/beaye_lexicon/beaye_lexicon.py
new file mode 100644
index 000000000..01695249f
--- /dev/null
+++ b/seacrowd/sea_datasets/beaye_lexicon/beaye_lexicon.py
@@ -0,0 +1,116 @@
+from pathlib import Path
+from typing import Dict, List, Tuple
+
+import datasets
+import pandas as pd
+
+from seacrowd.utils.configs import SEACrowdConfig
+from seacrowd.utils.constants import Licenses
+
+_CITATION = """\
+@misc{beayelexicon2024,
+  author    = {Lopo, Joanito Agili and Moeljadi, David and Cahyawijaya, Samuel and Aji, Alham Fikri and Sommerlot, 
+  Carly J. and Jacob, June},
+  title     = {Penyusunan Korpus Paralel Bahasa Indonesia–Bahasa Melayu Ambon, Melayu Kupang, Beaye, dan Uab Meto},
+  year      = {2024},
+  howpublished = {Online},
+  url       = {https://github.com/joanitolopo/makalah-kongresxii},
+  note      = {Manuscript in preparation},
+}
+"""
+
+_DATASETNAME = "beaye_lexicon"
+_DESCRIPTION = """The Beaye Lexicon is a lexicon resource encompassing translations between Indonesian, English, and 
+Beaye words. Developed through a collaborative effort involving two native Beaye speakers and evaluated by linguistic 
+experts, this lexicon comprises 984 Beaye vocabularies. The creation of the Beaye Lexicon marks the inaugural effort in 
+documenting the previously unrecorded Beaye language."""
+
+_HOMEPAGE = "https://github.com/joanitolopo/bhinneka-korpus/tree/main/lexicon"
+_LICENSE = Licenses.APACHE_2_0.value
+_URLS = "https://raw.githubusercontent.com/joanitolopo/bhinneka-korpus/main/lexicon"
+_SUPPORTED_TASKS = []
+_SOURCE_VERSION = "1.0.0"
+_SEACROWD_VERSION = "1.0.0"
+_LOCAL = False
+
+_LANGUAGES = ["ind", "day", "eng"]
+
+class BeayeLexicon(datasets.GeneratorBasedBuilder):
+    """Beaye Lexicon is a lexicon resource encompassing translations between Indonesian, English, and Beaye words"""
+
+    SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)
+    SEACROWD_VERSION = datasets.Version(_SEACROWD_VERSION)
+
+    BUILDER_CONFIGS = (
+        [
+            SEACrowdConfig(
+                name=f"{_DATASETNAME}_{lang}_source",
+                version=datasets.Version(_SOURCE_VERSION),
+                description=f"beaye lexicon with source schema for {lang} language",
+                schema="source",
+                subset_id="beaye_lexicon",
+            )
+            for lang in _LANGUAGES if lang != "eng"
+        ]
+        + [
+            SEACrowdConfig(
+                name=f"{_DATASETNAME}_ext_{lang}_source",
+                version=datasets.Version(_SOURCE_VERSION),
+                description=f"beaye lexicon with source schema for extensive definiton of beaye language",
+                schema="source",
+                subset_id="beaye_lexicon",
+            )
+            for lang in _LANGUAGES if lang != "ind"
+        ]
+    )
+
+    DEFAULT_CONFIG_NAME = f"{_DATASETNAME}_ind_source"
+
+    def _info(self) -> datasets.DatasetInfo:
+        schema = self.config.schema
+        if schema == "source":
+            features = datasets.Features({"id": datasets.Value("string"), "word": datasets.Value("string")})
+        else:
+            raise NotImplementedError()
+
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            features=features,
+            homepage=_HOMEPAGE,
+            license=_LICENSE,
+            citation=_CITATION,
+        )
+
+    def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
+        """Returns SplitGenerators."""
+        if "ext" in self.config.name.split("_"):
+            data_dir = Path(dl_manager.download(_URLS + "/english.xlsx"))
+        else:
+            data_dir = Path(dl_manager.download(_URLS + "/lexicon.xlsx"))
+
+        return [
+            datasets.SplitGenerator(
+                name=datasets.Split.TRAIN,
+                gen_kwargs={
+                    "filepath": data_dir,
+                    "split": "train",
+                }
+            )
+        ]
+
+    def _generate_examples(self, filepath: Path, split: str) -> Tuple[int, Dict]:
+        """Yields examples as (key, example) tuples."""
+        dfs = pd.read_excel(filepath, engine="openpyxl")
+        if "ext" in self.config.name.split("_"):
+            lang = self.config.name.split("_")[3]
+        else:
+            lang = self.config.name.split("_")[2]
+
+        text = dfs[lang]
+
+        if self.config.schema == "source":
+            for idx, word in enumerate(text.values):
+                row = {"id": str(idx), "word": word}
+                yield idx, row
+        else:
+            raise ValueError(f"Invalid config: {self.config.name}")