Support streaming

multi_eurlex.py

@@ -16,10 +16,17 @@
 
 
 import json
-import os
 
 import datasets
 
+_HOMEPAGE = "https://github.io/iliaschalkidis"
+
+_DESCRIPTION = """\
+MultiEURLEX comprises 65k EU laws in 23 official EU languages (some low-ish resource).
+Each EU law has been annotated with EUROVOC concepts (labels) by the Publication Office of EU.
+As with the English EURLEX, the goal is to predict the relevant EUROVOC concepts (labels);
+this is multi-label classification task (given the text, predict multiple labels).
+"""
 
 _CITATION = """\
 @InProceedings{chalkidis-etal-2021-multieurlex,
@@ -35,13 +42,6 @@ _CITATION = """\
   location = {Punta Cana, Dominican Republic},
 }"""
 
-_DESCRIPTION = """\
-MultiEURLEX comprises 65k EU laws in 23 official EU languages (some low-ish resource).
-Each EU law has been annotated with EUROVOC concepts (labels) by the Publication Office of EU.
-As with the English EURLEX, the goal is to predict the relevant EUROVOC concepts (labels);
-this is multi-label classification task (given the text, predict multiple labels).
-"""
-
 DATA_URL = "https://zenodo.org/record/5363165/files/multi_eurlex.tar.gz"
 
 _LANGUAGES = [
@@ -8255,49 +8255,48 @@ class MultiEURLEX(datasets.GeneratorBasedBuilder):
             description=_DESCRIPTION,
             features=features,
             supervised_keys=None,
-            homepage="https://github.io/iliaschalkidis",
+            homepage=_HOMEPAGE,
             citation=_CITATION,
         )
 
     def _split_generators(self, dl_manager):
-        data_dir = dl_manager.download_and_extract(DATA_URL)
+        data_dir = dl_manager.download(DATA_URL)
         return [
             datasets.SplitGenerator(
                 name=datasets.Split.TRAIN,
                 # These kwargs will be passed to _generate_examples
-                gen_kwargs={"filepath": os.path.join(data_dir, "train.jsonl"), "split": "train"},
+                gen_kwargs={"archive": dl_manager.iter_archive(data_dir), "filepath": "train.jsonl"},
             ),
             datasets.SplitGenerator(
                 name=datasets.Split.TEST,
                 # These kwargs will be passed to _generate_examples
-                gen_kwargs={"filepath": os.path.join(data_dir, "test.jsonl"), "split": "test"},
+                gen_kwargs={"archive": dl_manager.iter_archive(data_dir), "filepath": "test.jsonl"},
             ),
             datasets.SplitGenerator(
                 name=datasets.Split.VALIDATION,
                 # These kwargs will be passed to _generate_examples
-                gen_kwargs={"filepath": os.path.join(data_dir, "dev.jsonl"), "split": "dev"},
+                gen_kwargs={"archive": dl_manager.iter_archive(data_dir), "filepath": "dev.jsonl"},
             ),
         ]
 
-    def _generate_examples(self, filepath, split):
+    def _generate_examples(self, archive, filepath):
         """This function returns the examples in the raw (text) form."""
-
-        if self.config.language == "all_languages":
-            with open(filepath, encoding="utf-8") as f:
-                for id_, row in enumerate(f):
-                    data = json.loads(row)
-                    yield id_, {
-                        "celex_id": data["celex_id"],
-                        "text": {lang: data["text"][lang] for lang in self.config.languages},
-                        "labels": data["eurovoc_concepts"][self.config.label_level],
-                    }
-        else:
-            with open(filepath, encoding="utf-8") as f:
-                for id_, row in enumerate(f):
-                    data = json.loads(row)
-                    if data["text"][self.config.language] is not None:
+        for path, f in archive:
+            if path == filepath:
+                if self.config.language == "all_languages":
+                    for id_, row in enumerate(f):
+                        data = json.loads(row)
                         yield id_, {
                             "celex_id": data["celex_id"],
-                            "text": data["text"][self.config.language],
+                            "text": {lang: data["text"][lang] for lang in self.config.languages},
                             "labels": data["eurovoc_concepts"][self.config.label_level],
                         }
+                else:
+                    for id_, row in enumerate(f):
+                        data = json.loads(row)
+                        if data["text"][self.config.language] is not None:
+                            yield id_, {
+                                "celex_id": data["celex_id"],
+                                "text": data["text"][self.config.language],
+                                "labels": data["eurovoc_concepts"][self.config.label_level],
+                            }