Revision of rename-subtitles

Grégoire Détrez revised this gist 3 months ago. Go to revision

1 file changed, 3 insertions

README.md

			@@ -9,6 +9,9 @@ files, letting players (GNOME Videos, VLC) auto-load them.
9	9		`Season 1 Episode 2`, or a lone `E02`).
10	10		- Falls back to 1-to-1 matching when a folder has a single video and a
11	11		single subtitle with no episode markers.
	12	+	- Handles still-downloading videos: a `Show.S01E02.mkv.part` is matched as
	13	+	if already complete, so the subtitle is named for the finished file
	14	+	(`Show.S01E02.srt`, not `Show.S01E02.mkv.srt`).
12	15		- Preserves language codes: `sub.en.srt` matched to `Movie.mkv` →
13	16		`Movie.en.srt`.
14	17		- Never overwrites an existing file — such cases are reported as skipped.

Grégoire Détrez revised this gist 3 months ago. Go to revision

1 file changed, 30 insertions, 6 deletions

rename-subtitles

			@@ -4,7 +4,9 @@
4	4		Designed as a GNOME/Nautilus right-click script, but also runnable from a
5	5		terminal. Subtitles are paired to videos by episode number (S01E02, 1x02, ...)
6	6		with a fallback to 1-to-1 matching when a folder holds a single video + single
7		-	subtitle. Trailing language codes (e.g. `.en`, `.eng`) are preserved and
	7	+	subtitle. Still-downloading videos with a trailing `.part` suffix (e.g.
	8	+	`myfile.mkv.part`) are matched as if already complete. Trailing language codes
	9	+	(e.g. `.en`, `.eng`) are preserved and
8	10		existing files are never overwritten. A Zenity dialog lists the planned renames
9	11		for confirmation before anything happens.
10	12
			@@ -29,6 +31,10 @@ SUBTITLE_EXTS = {
29	31		".srt", ".sub", ".ssa", ".ass", ".vtt", ".idx", ".sup",
30	32		}
31	33
	34	+	# Suffix appended to still-downloading files (Firefox, wget, ...). A video
	35	+	# named `Show.S01E02.mkv.part` is treated as `Show.S01E02.mkv`.
	36	+	PART_EXT = ".part"
	37	+
32	38		# Language codes we recognise as a trailing token and preserve.
33	39		LANG_CODES = {
34	40		# ISO 639-1 (2-letter) — common subset
			@@ -101,14 +107,32 @@ def gather_files(paths):
101	107		return files
102	108
103	109
	110	+	def video_stem(name):
	111	+	"""Filename stem for a (possibly still-downloading) video.
	112	+
	113	+	Drops a trailing `.part` download suffix and then the real extension, so
	114	+	`Show.S01E02.mkv.part` and `Show.S01E02.mkv` both yield `Show.S01E02`.
	115	+	"""
	116	+	base = os.path.basename(name)
	117	+	root, ext = os.path.splitext(base)
	118	+	if ext.lower() == PART_EXT:
	119	+	base = root # drop '.part', exposing the real extension
	120	+	return os.path.splitext(base)[0]
	121	+
	122	+
104	123		def classify(files):
105	124		videos, subs = [], []
106	125		for f in files:
107		-	ext = os.path.splitext(f)[1].lower()
	126	+	root, ext = os.path.splitext(f)
	127	+	ext = ext.lower()
	128	+	if ext in SUBTITLE_EXTS:
	129	+	subs.append(f)
	130	+	continue
	131	+	# A still-downloading video keeps its real extension before `.part`.
	132	+	if ext == PART_EXT:
	133	+	ext = os.path.splitext(root)[1].lower()
108	134		if ext in VIDEO_EXTS:
109	135		videos.append(f)
110		-	elif ext in SUBTITLE_EXTS:
111		-	subs.append(f)
112	136		return videos, subs
113	137
114	138
			@@ -158,8 +182,8 @@ def plan_renames(files):
158	182		continue
159	183
160	184		_, lang, sub_ext = split_lang(subname)
161		-	video_stem = os.path.splitext(os.path.basename(video))[0]
162		-	new_name = video_stem + (("." + lang) if lang else "") + sub_ext
	185	+	stem = video_stem(video)
	186	+	new_name = stem + (("." + lang) if lang else "") + sub_ext
163	187		dst = os.path.join(os.path.dirname(sub), new_name)
164	188
165	189		if os.path.abspath(dst) == os.path.abspath(sub):

x revised this gist 3 months ago. Go to revision

1 file changed, 25 insertions, 9 deletions

rename-subtitles

			@@ -19,6 +19,7 @@ import re
19	19		import shutil
20	20		import subprocess
21	21		import sys
	22	+	from xml.sax.saxutils import escape as _xml_escape
22	23
23	24		VIDEO_EXTS = {
24	25		".mkv", ".mp4", ".avi", ".m4v", ".mov", ".wmv",
			@@ -54,8 +55,6 @@ _EP_PATTERNS = [
54	55		re.compile(r"[Ss]eason[\._ -]?(\d{1,2})[\._ -]?"
55	56		r"[Ee]pisode[\._ -]?(\d{1,3})"), # Season 1 Episode 2
56	57		]
57		-	# Lighter fallback: a lone episode marker, treated as season 1.
58		-	_EP_ONLY = re.compile(r"(?:^\|[\._ -])[Ee](\d{1,3})(?:[\._ -]\|$)")
59	58
60	59
61	60		def episode_key(name):
			@@ -65,9 +64,6 @@ def episode_key(name):
65	64		m = pat.search(stem)
66	65		if m:
67	66		return (int(m.group(1)), int(m.group(2)))
68		-	m = _EP_ONLY.search(stem)
69		-	if m:
70		-	return (1, int(m.group(1)))
71	67		return None
72	68
73	69
			@@ -139,9 +135,11 @@ def plan_renames(files):
139	135		for k in dup_keys:
140	136		video_by_key.pop(k, None)
141	137
	138	+	# One video + one subtitle: pair them directly when the subtitle carries no
	139	+	# episode marker to match on. The video may still have one (e.g. a lone
	140	+	# `Show.S01E01.mkv` next to a bare `english.srt`).
142	141		single_fallback = (len(videos) == 1 and len(subs) == 1
143		-	and episode_key(os.path.basename(subs[0])) is None
144		-	and episode_key(os.path.basename(videos[0])) is None)
	142	+	and episode_key(os.path.basename(subs[0])) is None)
145	143
146	144		planned_dsts = set()
147	145		for sub in subs:
			@@ -190,6 +188,12 @@ def build_message(renames, skips):
190	188		return "\n".join(lines)
191	189
192	190
	191	+	def _desktop_notify(title, body):
	192	+	"""Best-effort desktop notification when no dialog/console is available."""
	193	+	if shutil.which("notify-send"):
	194	+	subprocess.run(["notify-send", title, body])
	195	+
	196	+
193	197		def confirm(message, use_zenity):
194	198		"""Return True if the user approves the plan."""
195	199		if use_zenity:
			@@ -197,7 +201,7 @@ def confirm(message, use_zenity):
197	201		["zenity", "--question", "--no-wrap",
198	202		"--title", "Rename subtitles to match videos",
199	203		"--ok-label", "Rename", "--cancel-label", "Cancel",
200		-	"--text", message],
	204	+	"--text", _xml_escape(message)],
201	205		).returncode
202	206		return rc == 0
203	207		print(message)
			@@ -212,7 +216,7 @@ def notify(message, use_zenity, kind="info"):
212	216		subprocess.run(
213	217		["zenity", "--" + kind, "--no-wrap",
214	218		"--title", "Rename subtitles to match videos",
215		-	"--text", message],
	219	+	"--text", _xml_escape(message)],
216	220		)
217	221		else:
218	222		print(message)
			@@ -229,6 +233,18 @@ def main():
229	233
230	234		use_zenity = shutil.which("zenity") is not None and bool(os.environ.get("DISPLAY") or os.environ.get("WAYLAND_DISPLAY"))
231	235
	236	+	# Without zenity we prompt on the console, but a GUI launch (e.g. the
	237	+	# Nautilus Scripts menu) has no console: confirm() would read EOF and abort
	238	+	# with no feedback at all. Surface a desktop notification instead of
	239	+	# silently doing nothing.
	240	+	if not use_zenity and not sys.stdin.isatty():
	241	+	_desktop_notify(
	242	+	"Rename subtitles to match videos",
	243	+	"This script needs 'zenity' to show its confirmation dialog.\n"
	244	+	"Install it (e.g. 'sudo dnf install zenity') and try again.",
	245	+	)
	246	+	return 1
	247	+
232	248		files = gather_files(paths)
233	249		if not files:
234	250		notify("No files found to process.", use_zenity, "warning")

x revised this gist 3 months ago. Go to revision

2 files changed, 318 insertions

README.md(file created)

		@@ -0,0 +1,47 @@
1	+	# rename-subtitles
2	+
3	+	A GNOME/Nautilus script that renames subtitle files so they match their video
4	+	files, letting players (GNOME Videos, VLC) auto-load them.
5	+
6	+	## What it does
7	+
8	+	- Pairs subtitles to videos by episode number (`S01E02`, `1x02`,
9	+	`Season 1 Episode 2`, or a lone `E02`).
10	+	- Falls back to 1-to-1 matching when a folder has a single video and a
11	+	single subtitle with no episode markers.
12	+	- Preserves language codes: `sub.en.srt` matched to `Movie.mkv` →
13	+	`Movie.en.srt`.
14	+	- Never overwrites an existing file — such cases are reported as skipped.
15	+	- Shows a Zenity confirmation dialog listing every planned rename before
16	+	touching anything.
17	+
18	+	Video extensions: mkv mp4 avi m4v mov wmv flv webm mpg mpeg ts m2ts
19	+	Subtitle extensions: srt sub ssa ass vtt idx sup
20	+
21	+	## Install (Nautilus right-click)
22	+
23	+	```sh
24	+	cp rename-subtitles ~/.local/share/nautilus/scripts/"Rename subtitles to match videos"
25	+	chmod +x ~/.local/share/nautilus/scripts/"Rename subtitles to match videos"
26	+	nautilus -q # reload the Scripts menu (quits open Files windows)
27	+	```
28	+
29	+	Then in Files: select the folder (or the video + subtitle files),
30	+	right-click → Scripts → Rename subtitles to match videos.
31	+
32	+	### Dependency
33	+
34	+	The confirmation dialog needs zenity:
35	+
36	+	```sh
37	+	sudo dnf install zenity # Fedora
38	+	```
39	+
40	+	Without zenity the script still works from a terminal (it prompts on the
41	+	console), but the right-click dialog will not appear.
42	+
43	+	## Command-line usage
44	+
45	+	```sh
46	+	./rename-subtitles [PATH ...] # files and/or folders; defaults to cwd
47	+	```

rename-subtitles(file created)

		@@ -0,0 +1,271 @@
1	+	#!/usr/bin/env python3
2	+	"""Rename subtitle files to match their corresponding video files.
3	+
4	+	Designed as a GNOME/Nautilus right-click script, but also runnable from a
5	+	terminal. Subtitles are paired to videos by episode number (S01E02, 1x02, ...)
6	+	with a fallback to 1-to-1 matching when a folder holds a single video + single
7	+	subtitle. Trailing language codes (e.g. `.en`, `.eng`) are preserved and
8	+	existing files are never overwritten. A Zenity dialog lists the planned renames
9	+	for confirmation before anything happens.
10	+
11	+	Invocation:
12	+	* From Nautilus: reads NAUTILUS_SCRIPT_SELECTED_FILE_PATHS.
13	+	* From a terminal: `./rename-subtitles [PATH ...]` (files and/or folders).
14	+	With no arguments it operates on the current directory.
15	+	"""
16	+
17	+	import os
18	+	import re
19	+	import shutil
20	+	import subprocess
21	+	import sys
22	+
23	+	VIDEO_EXTS = {
24	+	".mkv", ".mp4", ".avi", ".m4v", ".mov", ".wmv",
25	+	".flv", ".webm", ".mpg", ".mpeg", ".ts", ".m2ts",
26	+	}
27	+	SUBTITLE_EXTS = {
28	+	".srt", ".sub", ".ssa", ".ass", ".vtt", ".idx", ".sup",
29	+	}
30	+
31	+	# Language codes we recognise as a trailing token and preserve.
32	+	LANG_CODES = {
33	+	# ISO 639-1 (2-letter) — common subset
34	+	"en", "fr", "es", "de", "it", "nl", "pt", "sv", "da", "no", "fi",
35	+	"pl", "ru", "cs", "hu", "ro", "tr", "el", "he", "ar", "ja", "ko",
36	+	"zh", "hi", "th", "vi", "id", "uk", "bg", "hr", "sr", "sk", "sl",
37	+	"et", "lv", "lt", "fa",
38	+	# ISO 639-2 (3-letter) and common English names
39	+	"eng", "english", "fre", "fra", "french", "spa", "esp", "spanish",
40	+	"ger", "deu", "german", "ita", "italian", "dut", "nld", "dutch",
41	+	"por", "portuguese", "swe", "swedish", "dan", "danish", "nor",
42	+	"norwegian", "fin", "finnish", "pol", "polish", "rus", "russian",
43	+	"cze", "ces", "czech", "hun", "hungarian", "rum", "ron", "romanian",
44	+	"tur", "turkish", "gre", "ell", "greek", "heb", "hebrew", "ara",
45	+	"arabic", "jpn", "japanese", "kor", "korean", "chi", "zho", "chinese",
46	+	"hin", "hindi", "tha", "thai", "vie", "vietnamese", "ind", "indonesian",
47	+	"ukr", "ukrainian",
48	+	}
49	+
50	+	# Episode-key regexes, tried in order. Each must yield (season, episode).
51	+	_EP_PATTERNS = [
52	+	re.compile(r"[Ss](\d{1,2})[\._ -]?[Ee](\d{1,3})"), # S01E02, s1e2, S01.E02
53	+	re.compile(r"(?<!\d)(\d{1,2})[xX](\d{2,3})(?!\d)"), # 1x02, 12x05
54	+	re.compile(r"[Ss]eason[\._ -]?(\d{1,2})[\._ -]?"
55	+	r"[Ee]pisode[\._ -]?(\d{1,3})"), # Season 1 Episode 2
56	+	]
57	+	# Lighter fallback: a lone episode marker, treated as season 1.
58	+	_EP_ONLY = re.compile(r"(?:^\|[\._ -])[Ee](\d{1,3})(?:[\._ -]\|$)")
59	+
60	+
61	+	def episode_key(name):
62	+	"""Return (season, episode) parsed from name, or None."""
63	+	stem = os.path.splitext(name)[0]
64	+	for pat in _EP_PATTERNS:
65	+	m = pat.search(stem)
66	+	if m:
67	+	return (int(m.group(1)), int(m.group(2)))
68	+	m = _EP_ONLY.search(stem)
69	+	if m:
70	+	return (1, int(m.group(1)))
71	+	return None
72	+
73	+
74	+	def split_lang(subname):
75	+	"""Split a subtitle filename into (stem_without_lang, lang_or_None, ext).
76	+
77	+	The token immediately before the extension is treated as a language code
78	+	only if it appears in LANG_CODES.
79	+	"""
80	+	root, ext = os.path.splitext(subname)
81	+	ext = ext.lower()
82	+	stem, sep, last = root.rpartition(".")
83	+	if sep and last.lower() in LANG_CODES:
84	+	return stem, last, ext
85	+	return root, None, ext
86	+
87	+
88	+	def gather_files(paths):
89	+	"""Resolve paths (files/dirs) into a flat, de-duplicated file list."""
90	+	files = []
91	+	seen = set()
92	+
93	+	def add(p):
94	+	ap = os.path.abspath(p)
95	+	if ap not in seen and os.path.isfile(ap):
96	+	seen.add(ap)
97	+	files.append(ap)
98	+
99	+	for p in paths:
100	+	if os.path.isdir(p):
101	+	for entry in sorted(os.listdir(p)):
102	+	add(os.path.join(p, entry))
103	+	elif os.path.isfile(p):
104	+	add(p)
105	+	return files
106	+
107	+
108	+	def classify(files):
109	+	videos, subs = [], []
110	+	for f in files:
111	+	ext = os.path.splitext(f)[1].lower()
112	+	if ext in VIDEO_EXTS:
113	+	videos.append(f)
114	+	elif ext in SUBTITLE_EXTS:
115	+	subs.append(f)
116	+	return videos, subs
117	+
118	+
119	+	def plan_renames(files):
120	+	"""Return (renames, skips).
121	+
122	+	renames: list of (src_abs, dst_abs, old_name, new_name).
123	+	skips: list of (name, reason).
124	+	"""
125	+	videos, subs = classify(files)
126	+	renames, skips = [], []
127	+
128	+	# Map episode key -> video (skip ambiguous duplicate keys).
129	+	video_by_key = {}
130	+	dup_keys = set()
131	+	for v in videos:
132	+	k = episode_key(os.path.basename(v))
133	+	if k is None:
134	+	continue
135	+	if k in video_by_key:
136	+	dup_keys.add(k)
137	+	else:
138	+	video_by_key[k] = v
139	+	for k in dup_keys:
140	+	video_by_key.pop(k, None)
141	+
142	+	single_fallback = (len(videos) == 1 and len(subs) == 1
143	+	and episode_key(os.path.basename(subs[0])) is None
144	+	and episode_key(os.path.basename(videos[0])) is None)
145	+
146	+	planned_dsts = set()
147	+	for sub in subs:
148	+	subname = os.path.basename(sub)
149	+	subkey = episode_key(subname)
150	+
151	+	if subkey is not None and subkey in video_by_key:
152	+	video = video_by_key[subkey]
153	+	elif subkey is not None and subkey in dup_keys:
154	+	skips.append((subname, "multiple videos share this episode number"))
155	+	continue
156	+	elif single_fallback:
157	+	video = videos[0]
158	+	else:
159	+	skips.append((subname, "no matching video found"))
160	+	continue
161	+
162	+	_, lang, sub_ext = split_lang(subname)
163	+	video_stem = os.path.splitext(os.path.basename(video))[0]
164	+	new_name = video_stem + (("." + lang) if lang else "") + sub_ext
165	+	dst = os.path.join(os.path.dirname(sub), new_name)
166	+
167	+	if os.path.abspath(dst) == os.path.abspath(sub):
168	+	skips.append((subname, "already named correctly"))
169	+	continue
170	+	if os.path.exists(dst) or dst in planned_dsts:
171	+	skips.append((subname, "target '%s' already exists" % new_name))
172	+	continue
173	+
174	+	planned_dsts.add(dst)
175	+	renames.append((sub, dst, subname, new_name))
176	+
177	+	return renames, skips
178	+
179	+
180	+	def build_message(renames, skips):
181	+	lines = []
182	+	if renames:
183	+	lines.append("The following subtitles will be renamed:\n")
184	+	for _, _, old, new in renames:
185	+	lines.append(" %s\n → %s\n" % (old, new))
186	+	if skips:
187	+	lines.append("\nSkipped:")
188	+	for name, reason in skips:
189	+	lines.append(" %s (%s)" % (name, reason))
190	+	return "\n".join(lines)
191	+
192	+
193	+	def confirm(message, use_zenity):
194	+	"""Return True if the user approves the plan."""
195	+	if use_zenity:
196	+	rc = subprocess.run(
197	+	["zenity", "--question", "--no-wrap",
198	+	"--title", "Rename subtitles to match videos",
199	+	"--ok-label", "Rename", "--cancel-label", "Cancel",
200	+	"--text", message],
201	+	).returncode
202	+	return rc == 0
203	+	print(message)
204	+	try:
205	+	return input("\nProceed with rename? [y/N] ").strip().lower() in ("y", "yes")
206	+	except EOFError:
207	+	return False
208	+
209	+
210	+	def notify(message, use_zenity, kind="info"):
211	+	if use_zenity:
212	+	subprocess.run(
213	+	["zenity", "--" + kind, "--no-wrap",
214	+	"--title", "Rename subtitles to match videos",
215	+	"--text", message],
216	+	)
217	+	else:
218	+	print(message)
219	+
220	+
221	+	def main():
222	+	env = os.environ.get("NAUTILUS_SCRIPT_SELECTED_FILE_PATHS", "")
223	+	if env.strip():
224	+	paths = [p for p in env.split("\n") if p]
225	+	elif len(sys.argv) > 1:
226	+	paths = sys.argv[1:]
227	+	else:
228	+	paths = [os.getcwd()]
229	+
230	+	use_zenity = shutil.which("zenity") is not None and bool(os.environ.get("DISPLAY") or os.environ.get("WAYLAND_DISPLAY"))
231	+
232	+	files = gather_files(paths)
233	+	if not files:
234	+	notify("No files found to process.", use_zenity, "warning")
235	+	return 1
236	+
237	+	renames, skips = plan_renames(files)
238	+
239	+	if not renames:
240	+	msg = "No subtitles needed renaming."
241	+	if skips:
242	+	msg += "\n\n" + build_message([], skips)
243	+	notify(msg, use_zenity, "info")
244	+	return 0
245	+
246	+	if not confirm(build_message(renames, skips), use_zenity):
247	+	return 0
248	+
249	+	done, failed = 0, []
250	+	for src, dst, old, new in renames:
251	+	try:
252	+	# Final guard against overwriting a file created in the meantime.
253	+	if os.path.exists(dst):
254	+	failed.append("%s (target appeared)" % old)
255	+	continue
256	+	os.rename(src, dst)
257	+	done += 1
258	+	except OSError as exc:
259	+	failed.append("%s (%s)" % (old, exc))
260	+
261	+	summary = "Renamed %d subtitle%s." % (done, "" if done == 1 else "s")
262	+	if skips:
263	+	summary += "\nSkipped %d." % len(skips)
264	+	if failed:
265	+	summary += "\nFailed %d:\n %s" % (len(failed), "\n ".join(failed))
266	+	notify(summary, use_zenity, "error" if failed else "info")
267	+	return 1 if failed else 0
268	+
269	+
270	+	if __name__ == "__main__":
271	+	sys.exit(main())

Newer Older