release: v1.1.0

Testing

src/cophi/__init__.py

@@ -46,4 +46,4 @@
 :module:`complexity` module.
 """
 
-from cophi.api import document, corpus
+from cophi.api import document, corpus, export

src/cophi/__version__.py

@@ -1,3 +1,3 @@
-VERSION = (1, 0, 10)
+VERSION = (1, 1, 0)
 
 __version__ = ".".join(map(str, VERSION))

src/cophi/api.py

@@ -81,3 +81,20 @@ def lazy_reading(filepaths):
                                     ignore_index=True)
     logger.info("Constructing Corpus object ...")
     return cophi.model.Corpus(documents), metadata
+
+
+def export(dtm, filepath, format="text"):
+    """Export a document-term matrix.
+
+    Parameters:
+        dtm: A document-term matrix.
+        filepath: Path to output file. Possibel values are `plaintext`/`text` or
+            `svmlight`.
+        format: File format.
+    """
+    if format.lower() in {"plaintext", "text"}:
+        cophi.model.Corpus.plaintext(dtm, filepath)
+    elif format.lower() in {"svmlight"}:
+        cophi.model.Corpus.svmlight(dtm, filepath)
+    else:
+        raise ValueError("'{}' is no supported file format.".format(format))

tests/test_api.py

@@ -10,6 +10,14 @@ def make_file(tmpdir, fname, content):
     p.write(content)
     return p
 
+@pytest.fixture
+def document():
+    return cophi.model.Document(DOCUMENT, "document", r"\w")
+
+@pytest.fixture
+def corpus(document):
+    return cophi.model.Corpus([document])
+
 def test_document(tmpdir):
     filepath = make_file(tmpdir, "document.txt", DOCUMENT)
     document = cophi.document(str(filepath), token_pattern=r"\w")
@@ -21,3 +29,19 @@ def test_corpus(tmpdir):
     corpus, metadata = cophi.corpus(directory)
     assert metadata["parent"].iloc[0] == str(directory)
     assert corpus.documents[0].text == DOCUMENT
+
+def test_export(corpus):
+    output = pathlib.Path("corpus.svmlight")
+    cophi.export(corpus.dtm, output, "svmlight")
+    assert output.exists()
+    with output.open("r", encoding="utf-8") as file:
+        assert file.read() == "document document a:1 b:1 c:1 d:1 e:1 f:1\n"
+
+    output = pathlib.Path("corpus.txt")
+    cophi.export(corpus.dtm, output, "text")
+    assert output.exists()
+    with output.open("r", encoding="utf-8") as file:
+        assert file.read() == "document document a b c d e f\n"
+
+    with pytest.raises(ValueError):
+        cophi.export(corpus.dtm, output, "unknown")

tests/test_utils.py

@@ -34,4 +34,4 @@ def test_parameter():
     parameter = cophi.utils._parameter(TOKENS, "entropy")
     assert len(parameter) == 2
     parameter = cophi.utils._parameter(TOKENS, "ttr")
-    assert len(parameter) == 2
+    assert len(parameter) == 2