chore: update tests

tests/test_api.py

@@ -0,0 +1,23 @@
+import pytest
+import pathlib
+import cophi
+
+
+DOCUMENT = "A B C D E F"
+
+def make_file(tmpdir, fname, content):
+    p = tmpdir.mkdir("sub").join(fname)
+    p.write(content)
+    return p
+
+def test_document(tmpdir):
+    filepath = make_file(tmpdir, "document.txt", DOCUMENT)
+    document = cophi.document(str(filepath), token_pattern=r"\w")
+    assert document.text == DOCUMENT
+
+def test_corpus(tmpdir):
+    p = make_file(tmpdir, "document.txt", DOCUMENT)
+    directory = pathlib.Path(str(p)).parent
+    corpus, metadata = cophi.corpus(directory)
+    assert metadata["parent"].iloc[0] == str(directory)
+    assert corpus.documents[0].text == DOCUMENT

tests/test_complexity.py

@@ -1,10 +1,9 @@
 import collections
-
 import pandas as pd
 import pytest
-
 import cophi
 
+
 P_STAR = 1
 Z = 2
 NUM_TYPES = 5
@@ -13,7 +12,6 @@
             "tuldava_ln", "brunet_w", "cttr", "summer_s", "sichel_s", "michea_m",
             "honore_h", "entropy", "yule_k", "simpson_d", "herdan_vm", "orlov_z"}
 
-
 @pytest.fixture
 def frequency_spectrum():
     tokens = ["A", "A", "A", "B", "B", "C", "D", "E"]

tests/test_model.py

@@ -2,9 +2,9 @@
 import lxml
 import numpy as np
 import pandas as pd
-
 import cophi
 
+
 DOCUMENT = "AAABBCCCDEF"
 TOKENS = list(DOCUMENT)
 LOWERCASE_TOKENS =  [token.lower() for token in TOKENS]

tests/test_utils.py

@@ -1,9 +1,8 @@
 import collections
-
 import pytest
-
 import cophi
 
+
 PARAGRAPHS = [["A B C D E F", "G H I J K L"]]
 DOCUMENT = PARAGRAPHS[0][0]
 TOKENS = DOCUMENT.split(" ")
@@ -27,14 +26,12 @@ def test_segment_fuzzy():
     segments = cophi.utils.segment_fuzzy(PARAGRAPHS, 1)
     assert list(segments) == [[["A B C D E F"]], [["G H I J K L"]]]
 
-"""
 def test_parameter():
     parameter = cophi.utils._parameter(TOKENS, "sichel_s")
-    assert parameter == {"num_types": 6, "freq_spectrum": collections.Counter({1: 6})}
+    assert len(parameter) == 2
     parameter = cophi.utils._parameter(TOKENS, "honore_h")
-    assert parameter == {'num_types': 6, 'num_tokens': 6, 'freq_spectrum': Counter({1: 6})}
+    assert len(parameter) == 3
     parameter = cophi.utils._parameter(TOKENS, "entropy")
-    assert parameter == 
+    assert len(parameter) == 2
     parameter = cophi.utils._parameter(TOKENS, "ttr")
-    assert parameter == 
-"""
+    assert len(parameter) == 2