Merge pull request #5264 from BoLiuV5/boliuv5-add-copyright-test-cases-2299

Fix copyright detection for text-classified files and restore fixtures(#2299)

Author: AyanSinhaMahapatra

src/textcode/analysis.py

@@ -93,8 +93,11 @@ def numbered_text_lines(
         logger_debug('numbered_text_lines: T.is_text_with_long_lines:', T.is_text_with_long_lines)
         logger_debug('numbered_text_lines: T.is_binary:', T.is_binary)
 
-    # TODO: we should have a command line to force digging inside binaries
-    if not T.contains_text:
+    # Some text-like files (notably source files misidentified by libmagic as
+    # image/data with a "text" payload) still contain extractable content even
+    # when `contains_text` is false. Allow text-classified files through to the
+    # normal extraction path instead of dropping them prematurely.
+    if not (T.contains_text or T.is_text):
         return iter([])
 
     # Should we read this as some markup, pdf office doc, text or binary?

tests/cluecode/data/copyrights/misco4/to_improve/junk-copyright-98.txt

@@ -0,0 +1 @@
+(c) 2007 James Turnbull <james@lovedthanlost.net>

tests/cluecode/data/copyrights/ort/james_turnbull.txt

@@ -0,0 +1,8 @@
+what:
+  - copyrights
+  - holders
+copyrights:
+  - "(c) 2007 James Turnbull <james@lovedthanlost.net>"
+holders:
+  - "James Turnbull"
+notes: "ORT-derived fixture; see ORT test assets for provenance"