Merge pull request #19 from brc-dd/fix/18

Fix character decoding issues with text-like files

pyproject.toml

@@ -38,6 +38,7 @@ dependencies = [
   "youtube-transcript-api",
   "SpeechRecognition",
   "pathvalidate",
+  "charset-normalizer",
 ]
 
 [project.urls]

src/markitdown/_markitdown.py

@@ -26,6 +26,7 @@ import pptx
 import puremagic
 import requests
 from bs4 import BeautifulSoup
+from charset_normalizer import from_path
 
 # Optional Transcription support
 try:
@@ -161,9 +162,7 @@ class PlainTextConverter(DocumentConverter):
         elif "text/" not in content_type.lower():
             return None
 
-        text_content = ""
+        text_content = str(from_path(local_path).best())
-        with open(local_path, "rt", encoding="utf-8") as fh:
-            text_content = fh.read()
         return DocumentConverterResult(
             title=None,
             text_content=text_content,

tests/test_files/test_mskanji.csv

@@ -0,0 +1,4 @@
+–¼‘O,”N—î,<2C>Z<EFBFBD>Š
+<EFBFBD>²“¡‘¾˜Y,30,“Œ‹ž
+ŽO–؉pŽq,25,‘å<E28098>ã
+îà‹´<EFBFBD>~,35,–¼ŒÃ‰®

tests/test_markitdown.py

@@ -87,6 +87,13 @@ SERP_TEST_EXCLUDES = [
     "data:image/svg+xml,%3Csvg%20width%3D",
 ]
 
+CSV_CP932_TEST_STRINGS = [
+    "名前,年齢,住所",
+    "佐藤太郎,30,東京",
+    "三木英子,25,大阪",
+    "髙橋淳,35,名古屋",
+]
+
 
 @pytest.mark.skipif(
     skip_remote,
@@ -164,6 +171,12 @@ def test_markitdown_local() -> None:
     for test_string in SERP_TEST_STRINGS:
         assert test_string in text_content
 
+    ## Test non-UTF-8 encoding
+    result = markitdown.convert(os.path.join(TEST_FILES_DIR, "test_mskanji.csv"))
+    text_content = result.text_content.replace("\\", "")
+    for test_string in CSV_CP932_TEST_STRINGS:
+        assert test_string in text_content
+
 
 @pytest.mark.skipif(
     skip_exiftool,