sane-tag-music -> sane-tag-media, and formally support pdf

TODO.md

@@ -149,7 +149,7 @@
 - have xdg-open parse `<repo:...> URIs (or adjust them so that it _can_ parse)
 - sane-bt-search: show details like 5.1 vs stereo, h264 vs h265
   - maybe just color these "keywords" in all search results?
-- transmission: apply `sane-tag-music` path fix in `torrent-done` script
+- transmission: apply `sane-tag-media` path fix in `torrent-done` script
   - many .mkv files do appear to be tagged: i'd just need to add support in my own tooling
 - uninsane.org: make URLs relative to allow local use (and as offline homepage)
 - email: fix so that local mail doesn't go to junk

hosts/common/programs/sane-scripts.nix

@@ -38,7 +38,7 @@ in
       "sane-scripts.secrets-update-keys"
       "sane-scripts.shutdown"
       "sane-scripts.sudo-redirect"
-      "sane-scripts.tag-music"
+      "sane-scripts.tag-media"
       "sane-scripts.vpn"
       "sane-scripts.which"
       "sane-scripts.wipe"
@@ -218,10 +218,10 @@ in
     "sane-scripts.sync-music" = {};
     "sane-scripts.sync-from-iphone" = {};
 
-    "sane-scripts.tag-music".suggestedPrograms = [
+    "sane-scripts.tag-media".suggestedPrograms = [
       "exiftool"  #< for (slightly) better sandboxing than default exiftool
     ];
-    "sane-scripts.tag-music".sandbox = {
+    "sane-scripts.tag-media".sandbox = {
       method = "bwrap";
       autodetectCliPaths = "existing";
       whitelistPwd = true;  # for music renaming

pkgs/additional/sane-scripts/default.nix

@@ -197,8 +197,8 @@ let
       srcRoot = ./src;
       pkgs = [ "ffmpeg" "python3.pkgs.unidecode" "sox" ];
     };
-    tag-music = static-nix-shell.mkPython3 {
-      pname = "sane-tag-music";
+    tag-media = static-nix-shell.mkPython3 {
+      pname = "sane-tag-media";
       srcRoot = ./src;
       pkgs = [ "python3.pkgs.mutagen" "python3.pkgs.pyexiftool" "python3.pkgs.pykakasi" "python3.pkgs.unidecode" ];
     };

pkgs/additional/sane-scripts/src/sane-tag-media

@@ -61,7 +61,7 @@
 # mutagen docs:
 # - <https://mutagen.readthedocs.io/en/latest/>
 """
-tool which runs over a complete music library or a subset of it and:
+tool which runs over a complete music library (or audiobooks, videos, ebooks (limited)) or a subset of it and:
 - detect tags which are missing or likely incorrect
 - write new tags to existing media
   - new tags are specified manually (--artist, --album, ...)
@@ -69,9 +69,9 @@ tool which runs over a complete music library or a subset of it and:
 
 this tool does NOT move or rename files. it only edits tags.
 
-USAGE: cd MUSIC_LIBRARY_TOP && sane-tag-music [options] fix-paths|fix-tags|print-missing|show [cmd-specific options] DIRECTORY [DIRECTORY ...]
+USAGE: cd MEDIA_LIBRARY_TOP && sane-tag-media [options] fix-paths|fix-tags|print-missing|show [cmd-specific options] DIRECTORY [DIRECTORY ...]
 
-scans DIRECTORY and guesses artist/album/title for each track, based on path relative to pwd.
+scans DIRECTORY and guesses artist/album/title for each file, based on path relative to pwd.
 if the guessed tags look more correct than the existing tags (i.e. if the existing file is missing a tag),
 then this updates the tags on-disk to reflect their path.
 
@@ -87,7 +87,7 @@ options:
   --title TITLE
   --trackno TRACK_NUMBER
   --derive                apply existing tags (e.g. album) found in the file set to any files in the set missing such tags
-  --type audio|image      skip files which aren't of some specific media type
+  --type audio|image|text skip files which aren't of some specific media type
 
 fix-tags options:
   --force: apply path-based tag to each file, even those which already have tags
@@ -130,6 +130,9 @@ IMAGE_EXTENSIONS = [
   "jpeg",
   "png",
 ]
+TEXT_EXTENSIONS = [
+  "pdf",
+]
 
 # some software encodes its name in the "producer" metadata field,
 # whereas i use that field to mean "producer" as "which studio produced this".
@@ -145,6 +148,7 @@ KNOWN_SOFTWARE_PRODUCERS = [
 class MediaType(Enum):
     Audio = "audio"
     Image = "image"
+    Text = "text"
     Other = "other"
 
 def maybe_romanize(a: str) -> str|None:
@@ -746,10 +750,12 @@ class MediaFile:
     def is_type(self, ty: MediaType) -> bool:
         is_audio = self.ext in AUDIO_EXTENSIONS
         is_image = self.ext in IMAGE_EXTENSIONS
+        is_text = self.ext in TEXT_EXTENSIONS
         want_audio = ty == MediaType.Audio
         want_image = ty == MediaType.Image
+        want_text = ty == MediaType.Text
 
-        return is_audio == want_audio and is_image == want_image
+        return is_audio == want_audio and is_image == want_image and is_text == want_text
 
     def tags_on_disk(self) -> Tags:
         return Tags(