add script and README

README.md

@@ -0,0 +1,50 @@
+---
+language:
+- srn
+- nl
+multilinguality:
+- translation
+pretty_name: sranantongo
+task_categories:
+  - translation
+---
+## Dataset Description
+Dataset for Sranantongo, an English-based creole language spoken as lingua franca in Suriname. The dataset includes monolingual data as well as Sranantongo-Dutch parallel data.
+
+The following splits are available:
+* `srn`
+Monolingual data scraped from SIL. There is a `train` split (6570 sentences) available.
+* `srn-nl_jw`
+Parallel data srn-nl originating from Jehova Witnesses. There are `train` (299085 sentences), `validation` (256 sentences), and `test` (256 sentences) splits available.
+* `srn-nl_other`
+Parallel data srn-nl originating from Z-Library, Naks Sranan facebook and Dutch DOJ. There are `train` (3610 sentences), `validation` (256 sentences), and `test` (256 sentences) splits available.
+
+For more details see the accompanying paper https://arxiv.org/abs/2212.06383
+
+
+## Using dataset
+Example of loading monolingual data:
+```python
+dataset = load_dataset("davidstap/sranantongo", "srn", trust_remote_code=True)
+```
+
+Example of loading parallel JW data:
+```python
+dataset = load_dataset("davidstap/sranantongo", "srn-nl_jw", trust_remote_code=True)
+```
+
+Example of loading parallel other data:
+```python
+dataset = load_dataset("davidstap/sranantongo", "srn-nl_other", trust_remote_code=True)
+```
+
+### Citation Information
+
+```
+@article{zwennicker2022towards,
+  title={Towards a general purpose machine translation system for Sranantongo},
+  author={Zwennicker, Just and Stap, David},
+  journal={arXiv preprint arXiv:2212.06383},
+  year={2022}
+}
+```

sranantongo.py

@@ -0,0 +1,103 @@
+import datasets
+
+
+_DESCRIPTION = """\
+This datasets consists of monolingual (Sranantongo) and parallel (Sranantongo - Dutch) data.
+"""
+
+_CITATION = """\
+@article{zwennicker2022towards,
+  title={Towards a general purpose machine translation system for Sranantongo},
+  author={Zwennicker, Just and Stap, David},
+  journal={arXiv preprint arXiv:2212.06383},
+  year={2022}
+}
+"""
+
+_DATA_URL = "data/sranantongo.tar"
+
+_LANGUAGE2FILES = {
+    "srn":          {"train": "srn_mono_SIL.csv", "validation": None, "test": None}, 
+    "srn-nl_jw":    {split:f"srn-nl_JW_{split}.csv" for split in ["train", "validation", "test"]},
+    "srn-nl_other": {split:f"srn-nl_other_{split}.csv" for split in ["train", "validation", "test"]},
+}
+
+
+class SranantongoConfig(datasets.BuilderConfig):
+    """BuilderConfig for Sranantongo dataset."""
+
+    def __init__(self, name=str, **kwargs):
+        self.name = name
+        description = "Monolingual sentences in `Sranantongo`." if "mono" in self.name else f"Parallel sentences in `Sranantongo` and `Dutch`."
+        super(SranantongoConfig, self).__init__(name=self.name, description=description, **kwargs)
+
+
+class Sranantongo(datasets.GeneratorBasedBuilder):
+    """Sranantongo data from https://arxiv.org/abs/2212.06383"""    
+    BUILDER_CONFIGS = [
+        SranantongoConfig(name=name, version=datasets.Version("1.0.0", ""))
+        for name in _LANGUAGE2FILES.keys()
+    ]
+
+
+    def _info(self):
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            features=datasets.Features(
+                {
+                    "srn": datasets.Value("string"),
+                    **(
+                        {"nl": datasets.Value("string")}
+                        if "srn-nl" in self.config.name else {}
+                    )
+                }
+            ),
+            homepage="https://arxiv.org/abs/2212.06383",
+            citation=_CITATION,
+        )
+
+    def _split_generators(self, dl_manager):
+        files = dl_manager.iter_archive(dl_manager.download(_DATA_URL))
+
+        # Always generate the train split
+        generators = [
+            datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"files": files, "split": "train"})
+        ]
+        
+        # If the dataset configuration is for parallel data, add validation and test splits
+        if "srn-nl" in self.config.name:
+            for split in [datasets.Split.VALIDATION, datasets.Split.TEST]:
+                generators.append(
+                    datasets.SplitGenerator(name=split, gen_kwargs={"files": files, "split": split})
+                )
+
+        return generators
+
+
+    def _generate_examples(self, split, files):
+        """Returns examples as raw text."""
+
+        if "srn-nl" in self.config.name:
+            return self._generate_examples_parallel(split=split, files=files)
+        else:
+            return self._generate_examples_mono(split=split, files=files)
+
+
+    def _generate_examples_mono(self, split, files):
+        for path, file in files:
+            if path == _LANGUAGE2FILES[self.config.name][split]:
+                data = file.read().decode("utf-8").split("\n")
+
+        for idx, sentence in enumerate(data):
+            yield idx, {"srn": sentence}
+
+
+
+    def _generate_examples_parallel(self, split, files):
+        for path, file in files:
+            if path == _LANGUAGE2FILES[self.config.name][split]:
+                data = file.read().decode("utf-8").split("\n")
+
+        for idx, sentence in enumerate(data):
+            nl, srn = sentence.split("|")
+            yield idx, {"nl": nl, "srn": srn}