add tests for languages arg and raising TypeError

test_unstructured/partition/csv/test_csv.py

@@ -215,3 +215,9 @@ def test_partition_csv_element_metadata_has_languages():
     filename = "example-docs/stanley-cups.csv"
     elements = partition_csv(filename=filename, strategy="fast")
     assert elements[0].metadata.languages == ["eng"]
+
+
+def test_partition_csv_element_metadata_accepts_languages_arg():
+    filename = "example-docs/stanley-cups.csv"
+    elements = partition_csv(filename=filename, strategy="fast", languages=["deu"])
+    assert elements[0].metadata.languages == ["deu"]

test_unstructured/partition/docx/test_docx.py

@@ -534,6 +534,12 @@ def test_partition_docx_detects_multiple_elements_in_other_language():
     assert langs == [["eng"], ["spa", "eng"], ["eng"], ["eng"], ["spa"]]
 
 
+def test_partition_docx_element_metadata_accepts_languages_arg():
+    filename = "example-docs/handbook-1p.docx"
+    elements = partition_docx(filename=filename, languages=["deu"])
+    assert elements[0].metadata.languages == ["deu"]
+
+
 def test_invalid_languages_arg_raises_TypeError():
     with pytest.raises(TypeError):
         filename = "example-docs/handbook-1p.docx"

test_unstructured/partition/odt/test_odt.py

@@ -1,8 +1,6 @@
 import os
 import pathlib
 
-import pytest
-
 from unstructured.chunking.title import chunk_by_title
 from unstructured.documents.elements import Table, TableChunk, Title
 from unstructured.partition.json import partition_json
@@ -210,9 +208,3 @@ def test_partition_odt_detects_multiple_elements_in_other_language():
     elements = partition_odt(filename=filename, detect_language_per_element=True)
     langs = [element.metadata.languages for element in elements]
     assert langs == [["eng"], ["spa", "eng"], ["eng"], ["eng"], ["spa"]]
-
-
-def test_invalid_languages_arg_raises_TypeError():
-    with pytest.raises(TypeError):
-        filename = "example-docs/fake.odt"
-        partition_odt(filename=filename, languages="eng")

test_unstructured/partition/test_text.py

@@ -525,27 +525,25 @@ def test_partition_text_element_metadata_has_languages():
     assert elements[0].metadata.languages == ["eng"]
 
 
-# def test_partition_text_detects_quote_in_other_language():
-#     filename = "example-docs/language-docs/eng_spa.txt"
-#     elements = partition_text(filename=filename, detect_language_per_element=True)
-#     langs = list({element.metadata.languages[0] for element in elements})
-#     assert set(langs) == set(["eng", "spa"])
-
-
-# def test_partition_text_detects_quotes_in_multiple_languages():
-#     filename = "example-docs/language-docs/eng_afr_spa.txt"
-#     elements = partition_text(filename=filename, detect_language_per_element=True)
-#     langs = list({element.metadata.languages[0] for element in elements})
-#     assert set(langs) == set(["eng", "spa"])
-
-
 def test_partition_text_detects_multiple_elements_in_other_language():
     filename = "example-docs/language-docs/eng_spa_mult.txt"
     elements = partition_text(filename=filename, detect_language_per_element=True)
     langs = [element.metadata.languages for element in elements]
     assert langs == [["eng"], ["spa", "eng"], ["eng"], ["eng"], ["spa"]]
 
 
+def test_partition_text_element_metadata_accepts_languages_arg():
+    filename = "example-docs/norwich-city.txt"
+    elements = partition_text(filename=filename, languages=["deu"])
+    assert elements[0].metadata.languages == ["deu"]
+
+
+def test_partition_text_element_metadata_raises_TypeError():
+    with pytest.raises(TypeError):
+        filename = "example-docs/norwich-city.txt"
+        partition_text(filename=filename, languages="eng")
+
+
 def test_partition_text_detects_more_than_3_languages():
     filename = "example-docs/language-docs/UDHR_first_article_all.txt"
     elements = partition_text(filename=filename, detect_language_per_element=True)