Add hour durations and allow hyphens in English

tests/test_en_durations.py

@@ -0,0 +1,57 @@
+import pytest
+import spacy
+from spacy.language import Language
+
+label = "timexy_label"
+lang = "en"
+
+
+@pytest.fixture()
+def nlp() -> Language:
+    nlp = spacy.blank(lang)
+    nlp.add_pipe("timexy", config={"label": label})
+    return nlp
+
+
+def test_digit_years(nlp: Language) -> None:
+    doc = nlp("I will try that in 1 year and 2 years")
+
+    for e in doc.ents:
+        assert e.label_ == label
+
+
+def test_word_years(nlp: Language) -> None:
+    doc = nlp("I will try that in one year and two years")
+
+    for e in doc.ents:
+        assert e.label_ == label
+
+
+def test_digit_hyphen_years(nlp: Language) -> None:
+    doc = nlp("1-year repeat is ideal.")
+
+    assert len(doc.ents) == 1
+    assert doc.ents[0].label_ == label
+
+
+def test_word_hyphen_years(nlp: Language) -> None:
+    doc = nlp("one-year repeat is ideal.")
+
+    assert len(doc.ents) == 1
+    assert doc.ents[0].label_ == label
+
+
+def test_digit_hours(nlp: Language) -> None:
+    doc = nlp("repeat in 48-72 hours")
+
+    assert len(doc.ents) == 1
+    assert doc.ents[0].label_ == label
+
+
+def test_word_hours(nlp: Language) -> None:
+    doc = nlp("I will try that in one hour and two hours")
+    print(doc.ents)
+
+    assert len(doc.ents) == 2
+    for e in doc.ents:
+        assert e.label_ == label

timexy/languages/en.py

@@ -8,6 +8,7 @@
         "M": ["month", "months"],
         "W": ["week", "weeks"],
         "D": ["day", "days"],
+        "H": ["hour", "hours"],
     },
     num_words=[
         "zero",

timexy/timexy.py

@@ -75,14 +75,15 @@ def __init__(
             for val in vals:
                 self.matcher.add(
                     key,
-                    [[{"IS_DIGIT": True}, {"TEXT": val}]],
+                    [[{"IS_DIGIT": True}, {"TEXT": "-", "OP": "?"}, {"TEXT": val}]],
                 )
 
                 self.matcher.add(
                     key,
                     [
                         [
                             {"LOWER": {"IN": self.timexy_lang.num_words}},
+                            {"TEXT": "-", "OP": "?"},
                             {"LOWER": val.lower()},
                         ]
                     ],