1 files changed, 85 insertions, 0 deletions
diff --git a/.local/bin/srt-lines-matching b/.local/bin/srt-lines-matching
new file mode 100755
index 0000000..accabb1
--- /dev/null
+++ b/.local/bin/srt-lines-matching
@@ -0,0 +1,85 @@
+#!/usr/local/bin/python3
+
+"""Filter subtitles that match or don't match a particular pattern."""
+
+import importlib
+import srt_tools.utils
+import logging
+
+log = logging.getLogger(__name__)
+
+
+def strip_to_matching_lines_only(subtitles, imports, func_str, invert, per_sub):
+    for import_name in imports:
+        real_import = importlib.import_module(import_name)
+        globals()[import_name] = real_import
+
+    raw_func = eval(func_str)  # pylint: disable-msg=eval-used
+
+    if invert:
+        func = lambda line: not raw_func(line)
+    else:
+        func = raw_func
+
+    for subtitle in subtitles:
+        if per_sub:
+            if not func(subtitle.content):
+                subtitle.content = ""
+        else:
+            subtitle.content = "\n".join(
+                line for line in subtitle.content.splitlines() if func(line)
+            )
+
+        yield subtitle
+
+
+def parse_args():
+    examples = {
+        "Only include Chinese lines": "srt lines-matching -m hanzidentifier -f hanzidentifier.has_chinese",
+        "Exclude all lines which only contain numbers": "srt lines-matching -v -f 'lambda x: x.isdigit()'",
+    }
+    parser = srt_tools.utils.basic_parser(description=__doc__, examples=examples)
+    parser.add_argument(
+        "-f", "--func", help="a function to use to match lines", required=True
+    )
+    parser.add_argument(
+        "-m",
+        "--module",
+        help="modules to import in the function context",
+        action="append",
+        default=[],
+    )
+    parser.add_argument(
+        "-s",
+        "--per-subtitle",
+        help="match the content of each subtitle, not each line",
+        action="store_true",
+    )
+    parser.add_argument(
+        "-v",
+        "--invert",
+        help="invert matching -- only match lines returning False",
+        action="store_true",
+    )
+    return parser.parse_args()
+
+
+def main():
+    args = parse_args()
+    logging.basicConfig(level=args.log_level)
+    srt_tools.utils.set_basic_args(args)
+    matching_subtitles_only = strip_to_matching_lines_only(
+        args.input, args.module, args.func, args.invert, args.per_subtitle
+    )
+    output = srt_tools.utils.compose_suggest_on_fail(
+        matching_subtitles_only, strict=args.strict
+    )
+
+    try:
+        args.output.write(output)
+    except (UnicodeEncodeError, TypeError):  # Python 2 fallback
+        args.output.write(output.encode(args.encoding))
+
+
+if __name__ == "__main__":  # pragma: no cover
+    main()