chore: refactor languages parameter for text_type functions (#1399)

### Summary
In order to support language functionality other than Tesseract OCR, we
want to represent languages provided for either partitioning accuracy or
OCR as a standard list of langcodes as strings. To identify element
types such as NarrativeText and Title, continue the refactor into
functions that use language checks to determine those potential
classifications.

### Details
Replaces `language` with `languages` (a list of strings) as a parameter
to `is_possible_narrative_text` and `is_possible_title`.


### Test
Call `is_possible_narrative_text` and `is_possible_title` with text in a
variety of languages and different inputs for `languages`. The resulting
element classifications should be no different from the current outputs.

ex: see `test_text_type_handles_multi_language_examples` in
`test_unstructured/partition/test_text_type.py`.

CHANGELOG.md

@@ -1,4 +1,4 @@
-## 0.10.15-dev8
+## 0.10.15-dev9
 
 ### Enhancements
 
@@ -9,6 +9,8 @@
 * Adds `xlsx` and `xls` to `skip_infer_table_types` default list in `partition`
 * Adds `languages` as an input parameter and marks `ocr_languages` kwarg for deprecation in image partitioning functions
 * Adds `languages` as an input parameter and marks `ocr_languages` kwarg for deprecation in auto partition
+* Replaces `language` with `languages` as an input parameter to unstructured-partition-text_type functions
+* Removes `UNSTRUCTURED_LANGUAGE` env var. To skip English specific checks, set the `languages` parameter to non-English language(s).
 
 ### Features
 

test_unstructured/partition/test_text_type.py

@@ -67,25 +67,32 @@ def test_text_type_handles_non_english_examples(monkeypatch):
     narrative_text = "Я говорю по-русски. Вы тоже?"
     title = "Риски"
 
-    assert text_type.is_possible_narrative_text(narrative_text, language="en") is False
-    assert text_type.is_possible_narrative_text(narrative_text, language="") is True
+    assert text_type.is_possible_narrative_text(narrative_text, languages=["eng"]) is False
+    assert text_type.is_possible_narrative_text(narrative_text, languages=[]) is True
 
-    assert text_type.is_possible_narrative_text(title, language="en") is False
-    assert text_type.is_possible_narrative_text(title, language="") is False
+    assert text_type.is_possible_narrative_text(title, languages=["eng"]) is False
+    assert text_type.is_possible_narrative_text(title, languages=[]) is False
 
-    assert text_type.is_possible_title(title, language="en") is False
-    assert text_type.is_possible_title(title, language="") is True
+    assert text_type.is_possible_title(title, languages=["eng"]) is False
+    assert text_type.is_possible_title(title, languages=[]) is True
 
 
-def test_text_type_handles_non_english_examples_with_env_var(monkeypatch):
-    monkeypatch.setenv("UNSTRUCTURED_LANGUAGE", "")
+def test_text_type_handles_multi_language_examples(monkeypatch):
+    monkeypatch.setenv("UNSTRUCTURED_LANGUAGE_CHECKS", "true")
+    narrative_text = "Я говорю по-русски. Вы тоже? 不，我不会说俄语。"
+    title = "Риски (Riesgos)"
 
-    narrative_text = "Я говорю по-русски. Вы тоже?"
-    title = "Риски"
+    assert text_type.is_possible_narrative_text(narrative_text, languages=["eng"]) is False
+    assert text_type.is_possible_narrative_text(narrative_text, languages=["chi", "rus"]) is True
+    assert text_type.is_possible_narrative_text(narrative_text, languages=[]) is True
 
-    assert text_type.is_possible_narrative_text(narrative_text) is True
-    assert text_type.is_possible_narrative_text(title) is False
-    assert text_type.is_possible_title(title) is True
+    assert text_type.is_possible_narrative_text(title, languages=["eng"]) is False
+    assert text_type.is_possible_narrative_text(title, languages=["spa", "rus"]) is False
+    assert text_type.is_possible_narrative_text(title, languages=[]) is False
+
+    assert text_type.is_possible_title(title, languages=["eng"]) is False
+    assert text_type.is_possible_title(title, languages=["spa", "rus"]) is True
+    assert text_type.is_possible_title(title, languages=[]) is True
 
 
 @pytest.mark.parametrize(

unstructured/__version__.py

@@ -1 +1 @@
-__version__ = "0.10.15-dev8"  # pragma: no cover
+__version__ = "0.10.15-dev9"  # pragma: no cover

unstructured/partition/text_type.py

@@ -30,7 +30,7 @@ def is_possible_narrative_text(
     text: str,
     cap_threshold: float = 0.5,
     non_alpha_threshold: float = 0.5,
-    language: str = "en",
+    languages: List[str] = ["eng"],
     language_checks: bool = False,
 ) -> bool:
     """Checks to see if the text passes all of the checks for a narrative text section.
@@ -47,8 +47,8 @@ def is_possible_narrative_text(
     non_alpha_threshold
         The minimum proportion of alpha characters the text needs to be considered
         narrative text
-    language
-        The two letter language code for the text. defaults to "en" for English
+    languages
+        The list of languages present in the document. Defaults to ["eng"] for English
     language_checks
         If True, conducts checks that are specific to the chosen language. Turn on for more
         accurate partitioning and off for faster processing.
@@ -65,8 +65,7 @@ def is_possible_narrative_text(
         trace_logger.detail(f"Not narrative. Text is all numeric:\n\n{text}")  # type: ignore
         return False
 
-    language = os.environ.get("UNSTRUCTURED_LANGUAGE", language)
-    if language == "en" and language_checks and not contains_english_word(text):
+    if "eng" in languages and language_checks and not contains_english_word(text):
         return False
 
     # NOTE(robinson): it gets read in from the environment as a string so we need to
@@ -84,7 +83,7 @@ def is_possible_narrative_text(
     if under_non_alpha_ratio(text, threshold=non_alpha_threshold):
         return False
 
-    if (sentence_count(text, 3) < 2) and (not contains_verb(text)) and language == "en":
+    if "eng" in languages and (sentence_count(text, 3) < 2) and (not contains_verb(text)):
         trace_logger.detail(f"Not narrative. Text does not contain a verb:\n\n{text}")  # type: ignore # noqa: E501
         return False
 
@@ -96,7 +95,7 @@ def is_possible_title(
     sentence_min_length: int = 5,
     title_max_word_length: int = 12,
     non_alpha_threshold: float = 0.5,
-    language: str = "en",
+    languages: List[str] = ["eng"],
     language_checks: bool = False,
 ) -> bool:
     """Checks to see if the text passes all of the checks for a valid title.
@@ -111,8 +110,8 @@ def is_possible_title(
         The maximum number of words a title can contain
     non_alpha_threshold
         The minimum number of alpha characters the text needs to be considered a title
-    language
-        The two letter language code for the text. defaults to "en" for English
+    languages
+        The list of languages present in the document. Defaults to ["eng"] for English
     language_checks
         If True, conducts checks that are specific to the chosen language. Turn on for more
         accurate partitioning and off for faster processing.
@@ -146,8 +145,7 @@ def is_possible_title(
     if text.endswith(","):
         return False
 
-    language = os.environ.get("UNSTRUCTURED_LANGUAGE", language)
-    if language == "en" and not contains_english_word(text) and language_checks:
+    if "eng" in languages and not contains_english_word(text) and language_checks:
         return False
 
     if text.isnumeric():