Support train dict format as JSONL (#4471)

* Support train dict format as JSONL

* Add (overly simple) check for dict vs. tuple to read JSONL lines as
either train dicts or train tuples

* Extend JSON/JSONL roundtrip conversion tests using `docs_to_json()`
and `GoldCorpus.train_tuples`

* Revert docs to default JSON output with convert

spacy/gold.pyx

@@ -176,6 +176,11 @@ class GoldCorpus(object):
                 gold_tuples = read_json_file(loc)
             elif loc.parts[-1].endswith("jsonl"):
                 gold_tuples = srsly.read_jsonl(loc)
+                first_gold_tuple = next(gold_tuples)
+                gold_tuples = itertools.chain([first_gold_tuple], gold_tuples)
+                # TODO: proper format checks with schemas
+                if isinstance(first_gold_tuple, dict):
+                    gold_tuples = read_json_object(gold_tuples)
             elif loc.parts[-1].endswith("msg"):
                 gold_tuples = srsly.read_msgpack(loc)
             else:

spacy/tests/test_gold.py

@@ -99,14 +99,23 @@ def test_iob_to_biluo():
 
 def test_roundtrip_docs_to_json():
     text = "I flew to Silicon Valley via London."
+    tags = ['PRP', 'VBD', 'IN', 'NNP', 'NNP', 'IN', 'NNP', '.']
+    heads = [1, 1, 1, 4, 2, 1, 5, 1]
+    deps = ['nsubj', 'ROOT', 'prep', 'compound', 'pobj', 'prep', 'pobj', 'punct']
+    biluo_tags = ["O", "O", "O", "B-LOC", "L-LOC", "O", "U-GPE", "O"]
     cats = {"TRAVEL": 1.0, "BAKING": 0.0}
     nlp = English()
     doc = nlp(text)
+    for i in range(len(tags)):
+        doc[i].tag_ = tags[i]
+        doc[i].dep_ = deps[i]
+        doc[i].head = doc[heads[i]]
+    doc.ents = spans_from_biluo_tags(doc, biluo_tags)
     doc.cats = cats
-    doc[0].is_sent_start = True
+    doc.is_tagged = True
-    for i in range(1, len(doc)):
+    doc.is_parsed = True
-        doc[i].is_sent_start = False
 
+    # roundtrip to JSON
     with make_tempdir() as tmpdir:
         json_file = tmpdir / "roundtrip.json"
         srsly.write_json(json_file, [docs_to_json(doc)])
@@ -116,6 +125,52 @@ def test_roundtrip_docs_to_json():
 
     assert len(doc) == goldcorpus.count_train()
     assert text == reloaded_doc.text
+    assert tags == goldparse.tags
+    assert deps == goldparse.labels
+    assert heads == goldparse.heads
+    assert biluo_tags == goldparse.ner
+    assert "TRAVEL" in goldparse.cats
+    assert "BAKING" in goldparse.cats
+    assert cats["TRAVEL"] == goldparse.cats["TRAVEL"]
+    assert cats["BAKING"] == goldparse.cats["BAKING"]
+
+    # roundtrip to JSONL train dicts
+    with make_tempdir() as tmpdir:
+        jsonl_file = tmpdir / "roundtrip.jsonl"
+        srsly.write_jsonl(jsonl_file, [docs_to_json(doc)])
+        goldcorpus = GoldCorpus(str(jsonl_file), str(jsonl_file))
+
+    reloaded_doc, goldparse = next(goldcorpus.train_docs(nlp))
+
+    assert len(doc) == goldcorpus.count_train()
+    assert text == reloaded_doc.text
+    assert tags == goldparse.tags
+    assert deps == goldparse.labels
+    assert heads == goldparse.heads
+    assert biluo_tags == goldparse.ner
+    assert "TRAVEL" in goldparse.cats
+    assert "BAKING" in goldparse.cats
+    assert cats["TRAVEL"] == goldparse.cats["TRAVEL"]
+    assert cats["BAKING"] == goldparse.cats["BAKING"]
+
+    # roundtrip to JSONL tuples
+    with make_tempdir() as tmpdir:
+        jsonl_file = tmpdir / "roundtrip.jsonl"
+        # write to JSONL train dicts
+        srsly.write_jsonl(jsonl_file, [docs_to_json(doc)])
+        goldcorpus = GoldCorpus(str(jsonl_file), str(jsonl_file))
+        # load and rewrite as JSONL tuples
+        srsly.write_jsonl(jsonl_file, goldcorpus.train_tuples)
+        goldcorpus = GoldCorpus(str(jsonl_file), str(jsonl_file))
+
+    reloaded_doc, goldparse = next(goldcorpus.train_docs(nlp))
+
+    assert len(doc) == goldcorpus.count_train()
+    assert text == reloaded_doc.text
+    assert tags == goldparse.tags
+    assert deps == goldparse.labels
+    assert heads == goldparse.heads
+    assert biluo_tags == goldparse.ner
     assert "TRAVEL" in goldparse.cats
     assert "BAKING" in goldparse.cats
     assert cats["TRAVEL"] == goldparse.cats["TRAVEL"]

website/docs/api/cli.md

@@ -155,21 +155,14 @@ $ python -m spacy convert [input_file] [output_dir] [--file-type] [--converter]
 
 ### Output file types {new="2.1"}
 
-> #### Which format should I choose?
->
-> If you're not sure, go with the default `jsonl`. Newline-delimited JSON means
-> that there's one JSON object per line. Unlike a regular JSON file, it can also
-> be read in line-by-line and you won't have to parse the _entire file_ first.
-> This makes it a very convenient format for larger corpora.
-
 All output files generated by this command are compatible with
 [`spacy train`](/api/cli#train).
 
-| ID      | Description                       |
+| ID      | Description                |
-| ------- | --------------------------------- |
+| ------- | -------------------------- |
-| `jsonl` | Newline-delimited JSON (default). |
+| `json`  | Regular JSON (default).    |
-| `json`  | Regular JSON.                     |
+| `jsonl` | Newline-delimited JSON.    |
-| `msg`   | Binary MessagePack format.        |
+| `msg`   | Binary MessagePack format. |
 
 ### Converter options