From d06cace2d6246745720e50675e03bb5c6ae9f76a Mon Sep 17 00:00:00 2001
From: josdion <josdion@users.noreply.github.com>
Date: Tue, 9 Jun 2020 22:20:43 +0300
Subject: [PATCH] Fix #886

Use the whole subtitle file to detect encoding. Seems like if trying to detect encoding only by a part of the file, is not always working. This modification will lead to worse performance, but should be more reliable.
---
 bazarr/list_subtitles.py      | 13 ++++---------
 libs/subliminal_patch/core.py |  2 +-
 2 files changed, 5 insertions(+), 10 deletions(-)

diff --git a/bazarr/list_subtitles.py b/bazarr/list_subtitles.py
index 20bddc995..1cf1f0494 100644
--- a/bazarr/list_subtitles.py
+++ b/bazarr/list_subtitles.py
@@ -364,8 +364,8 @@ def guess_external_subtitles(dest_folder, subtitles):
                 logging.debug("BAZARR falling back to file content analysis to detect language.")
                 detected_language = None
 
-                # to improve performance, skip detection of files larger that 5M
-                if os.path.getsize(subtitle_path) > 5*1024*1024:
+                # to improve performance, skip detection of files larger that 1M
+                if os.path.getsize(subtitle_path) > 1*1024*1024:
                     logging.debug("BAZARR subtitles file is too large to be text based. Skipping this file: " +
                                   subtitle_path)
                     continue
@@ -374,16 +374,11 @@ def guess_external_subtitles(dest_folder, subtitles):
                     text = f.read()
 
                 try:
-                    # to improve performance, use only the first 32K to detect encoding
-                    guess = chardet.detect(text[:32768])
+                    guess = chardet.detect(text)
                     logging.debug('BAZARR detected encoding %r', guess)
-                    if guess["confidence"] < 0.6:
-                        raise UnicodeError
-                    if guess["encoding"] == "ascii":
-                        guess["encoding"] = "utf-8"
                     text = text.decode(guess["encoding"])
                     detected_language = guess_language(text)
-                except UnicodeError:
+                except (UnicodeDecodeError, TypeError):
                     logging.exception("BAZARR subtitles file doesn't seems to be text based. Skipping this file: " +
                                       subtitle_path)
                 except:
diff --git a/libs/subliminal_patch/core.py b/libs/subliminal_patch/core.py
index 397effcfe..f854390cf 100644
--- a/libs/subliminal_patch/core.py
+++ b/libs/subliminal_patch/core.py
@@ -606,7 +606,7 @@ def _search_external_subtitles(path, languages=None, only_one=False, scandir_gen
             continue
 
         if p_root.lower() == fn_no_ext_lower:
-            # skip check for language code is the subtitle file name is the same as the video name
+            # skip check for language code if the subtitle file name is the same as the video name
             subtitles[p] = None
             continue