fix: temp-fix

[Dhairya3391]

Fixes the HiAnime provider after recent server changes, cleans up subtitle handling, and resolves several edge-case bugs.

Summary of Changes

• Fix HiAnime Server Rename: HiAnime renamed its video player server from HD-1 to ZokoAnime. The scraper now dynamically finds the ZokoAnime stream, with a fallback if server names change again.
• Clean Search Results: Strips out the site's 'Top 10' sidebar so searches only return matching titles instead of appending ~30 unrelated shows. Invalid queries now correctly show 'No results found!'.
• Subtitles Without Temp Files: Passes remote subtitle links directly to players (mpv, IINA, VLC, Syncplay) instead of downloading temp files. Colon-escapes URLs for IINA and uses :input-slave for VLC.
• Subtitle Language Preference: Selects English by default (configurable via ANI_CLI_SUB_LANG) instead of always grabbing the last track in the list.
• Intro Skipping: Ensures the anime ID is loaded before ani-skip runs so --skip works as expected.
• History Fixes: Validates the full title slug on HiAnime so older v5.0 history won't play the wrong show. Passes explicit arguments to background history workers to eliminate variable races.
• Decimal Episode Numbers: Uses exact text matching instead of regular expressions for episode numbers like 12.5 so the dot isn't treated as a wildcard.
• Speed & Cleanup: Replaces subprocess calls (printf | sed) with built-in shell parameter expansions, removes the unused description endpoint request, and increases curl max-time to 15s for large episode lists like One Piece.
• Documentation: Updates man pages and documentation files from anidb to hianime.

[Copilot]

🟡 Changes recommended

The history listing and episode-matching logic has confirmed correctness issues (subshell/background wait misuse, racey globals, and regex-based episode matching) that can produce wrong or incomplete results.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Pull request overview

This PR updates ani-cli’s scraping layer to work with HiAnime again, preserves season switching, and adds a Miruro-backed variant script with improved stream selection/referrer handling.

Changes:

• Replaces the previous provider scraping logic in ani-cli with a HiAnime-based flow (search/desc/episodes/servers → m3u8 selection).
• Adds a new ani-cli-miruro script that uses Miruro’s API backend and implements provider prioritization + HLS/MP4 probing.
• Adds a .gitignore entry to avoid committing scraper/node_modules/.

File summaries

File	Description
ani-cli-miruro	New Miruro-backed CLI script including episode/source fetching and playback logic.
ani-cli	Updates main scraper to HiAnime, adds season switching, refactors menus, and adjusts history/playback flows.
.gitignore	Ignores scraper/node_modules/.

Review details

Suppressed comments (4)

ani-cli:340

• backup_old_hist appends to ${histfile}.new without truncating/creating it first, and then unconditionally mvs it. If ${histfile}.new already exists (e.g., from a previous interrupted run) or the history is empty, this can duplicate entries or fail the migration.

backup_old_hist() {
    backupfile="${histfile}.v4"
    while IFS="	" read -r ep_no anime_id anime_title; do
        case "$anime_id" in
            *-*) printf "%s\t%s\t%s\n" "$ep_no" "$anime_id" "$anime_title" >>"${histfile}.new" ;;

ani-cli:435

• canon_ep_no is derived via a regex-based sed match on $ep_no. For episode numbers like 12.5, . is a regex wildcard and can produce an incorrect line number, which then breaks ani-skip episode mapping.

This issue also appears on line 441 of the same file.

            canon_ep_no="$(printf "%s" "$ep_list" | sed -n "/^${ep_no}$/=")"

ani-cli:442

• Episode validation and canon_ep_no calculation use regex-based grep/sed matching on $ep_no. With non-integer episode numbers (e.g. 12.5), this can accept the wrong episode and compute an incorrect canon_ep_no. Prefer fixed-string matching.

        printf "%s" "$ep_list" | grep -q "^$ep_no$" || die "Invalid episode!"
        canon_ep_no="$(printf "%s" "$ep_list" | sed -n "/^${ep_no}$/=")"

ani-cli:599

• In the history path, process_hist_entry is started as a background job inside a command substitution, but wait is executed outside that subshell. The parent shell cannot wait for jobs started in the command-substitution subshell, so the list can be incomplete/truncated. Move wait inside the same command substitution (and pass args to avoid data races).

        anime_list=$(while read -r ep_no anime_id anime_title; do process_hist_entry & done <"$histfile")
        wait

• Files reviewed: 1/1 changed files
• Comments generated: 3
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[port19x]

Why is copilot AI here??? Wtf man

[port19x]

@Dhairya3391 if this fixes the outage, bump it to v5.1 pls

[port19x]

Oh, you brought your own copilot, that's fair

[Dhairya3391]

@port19x anything else needed? it works as of now as fix but can't rely on this because site is not the best looking.
edit : misclicked on tag

[port19x]

@Dhairya3391 why ping pystardust lol.
We can keep this around for now, but we'll wait a little for if anidb comes back up after maintenance

[port19x]

Please add your used AI tool as a coauthor, I'll give this a review pass in codex.
Too fat and too early for me to manually read all the code

[71zenith]

I don't think anidb is coming back. This site also doesn't look very reliable. This is the other candidate that I could successfully scrape without being hls-blocked https://anizone.to/
But this is also another medium sized self hosted library and could go down as easily as anidb.

[port19x]

Hey @71zenith, can you look at my PR over at #1895

[port19x]

Want me to come into the discord again for the next provider?

[Dhairya3391]

yeah we'll need to look at reliable providers again

[71zenith]

I dont know how much productive it would be but we could communicate better i guess.

[port19x]

I'm billionairelover

[Tsagar3]

Fix: subtitle tracks are dropped in the HiAnime flow

I tested this PR and found that episodes play without subtitles. Here's the root cause and a minimal fix.

Root cause

The HiAnime embed payload (the blurb decoded from window.__P) carries the subtitles outside the HLS playlist:

{
  "download_url": "/download/mal/21/1/sub",
  "src": "https://hls2.aniwatchtv.uk/v/.../master.m3u8",
  "subtitles": [
    { "lang": "en", "label": "English", "default": true, "src": "https://hls2.aniwatchtv.uk/v/.../subs/6e4a0p3k8ehe9eee.vtt" }
  ],
  "sprite_vtt": ""
}

The master.m3u8 (and the resulting index.m3u8) only contains #EXT-X-STREAM-INF variants — there are no #EXT-X-MEDIA:TYPE=SUBTITLES tracks. ffprobe on a segment confirms only h264 video + aac audio.

deobfuscate_blob only extracts the stream src (the .m3u8), so the separate .vtt file referenced by "subtitles" is discarded. mpv then never receives a subtitle track, hence no subs.

Fix

• deobfuscate_blob also extracts the .vtt src from the subtitles array.
• hianime_m3u8 persists it to a temp file (reusing the existing histfile-based tempfile pattern), cleared when the episode has no subs.
• get_video_link reads it into sub_link.
• play_episode passes --sub-file="$sub_link" to mpv (flatpak too) when present.

Verified with mpv against the extracted VTT:
● Subs --sid=1 '6e4a0p3k8ehe9eee.vtt' (webvtt) [external] and it fetches the .vtt with the correct Referer.

Shell quirk worth knowing

The decoded JSON has no trailing newline, and a lone printf "%b" | sed -n 's/.../p' may then emit its output without a trailing newline. Two consecutive extracts in a subshell can therefore concatenate into a single line (master.m3u8https://...vtt). The patch wraps each extract in printf "%s\n" "$( ... )" and filters blank lines downstream so the two URLs always land on their own lines.

Suggested diff

@@ cleanup() @@
 cleanup() {
-    tput sgr0               # clears colors
-    rm -f "${histfile}.new" # remove temp logfile
+    tput sgr0                    # clears colors
+    rm -f "${histfile}.new"      # remove temp logfile
+    rm -f "${histfile}.subs"     # remove temp subtitle url
 }

@@ deobfuscate_blob() @@
-    printf "%b" "$_output" | sed -n 's|.*"src":"\([^"]*\.m3u8[^"]*\)".*|\1|p'
+    printf "%s\n" "$(printf "%b" "$_output" | sed -n 's|.*"src":"\([^"]*\.m3u8[^"]*\)".*|\1|p')"
+    printf "%s\n" "$(printf "%b" "$_output" | sed -n 's|.*"subtitles":\[[^]]*"src":"\([^"]*\.vtt[^"]*\)".*|\1|p')"
 )

@@ hianime_m3u8() - "5. Deobfuscate" @@
-    _m3u8_master="$(deobfuscate_blob "$_blob")"
+    _deobfuscated="$(printf "%s\n" "$(deobfuscate_blob "$_blob")" | sed '/^$/d')"
+    _m3u8_master="$(printf "%s\n" "$_deobfuscated" | sed -n '1p')"
+    _subs="$(printf "%s\n" "$_deobfuscated" | sed -n '2p')"
     [ -z "$_m3u8_master" ] && return 1
+    if [ -n "$_subs" ]; then
+        printf "%s\n" "$_subs" >"${histfile}.subs"
+    else
+        rm -f "${histfile}.subs"
+    fi

@@ get_video_link() @@
     links=$(hianime_m3u8 "$ep_no" "$mode" | sort -g -r -s)
     [ -z "$links" ] && die "No sources found for $mode!"
     info "hianime.at links fetched"
+    sub_link="$(cat "${histfile}.subs" 2>/dev/null)"
     select_quality "$quality"

@@ play_episode() @@
     [ -z "$video_link" ] && get_video_link
+    sub_file_flag=""
+    [ -n "$sub_link" ] && sub_file_flag="--sub-file=$sub_link"
     # wait for the previous player to close (in range selection)
     wait
     # shellcheck disable=SC2086
     case "$player_function" in
-        debug) printf "All links:\n%s\nSelected link:\n%s\n" "$links" "$video_link" ;;
+        debug) printf "All links:\n%s\nSelected link:\n%s\nSubtitles:\n%s\n" "$links" "$video_link" "$sub_link" ;;
         ...
-        *flatpak*mpv*) flatpak run io.mpv.Mpv --http-header-fields="Referer: https://zokoanime.video/" $skip_flag $player_extra_flags --force-media-title="${anime_title} Episode ${ep_no}" "$video_link" >/dev/null 2>&1 & ;;
+        *flatpak*mpv*) flatpak run io.mpv.Mpv --http-header-fields="Referer: https://zokoanime.video/" $sub_file_flag $skip_flag $player_extra_flags --force-media-title="${anime_title} Episode ${ep_no}" "$video_link" >/dev/null 2>&1 & ;;
         *mpv*)
             if [ "$no_detach" = 0 ]; then
-                nohup $player_function --http-header-fields="Referer: https://zokoanime.video/" $skip_flag $player_extra_flags --force-media-title="${anime_title} Episode ${ep_no}" "$video_link" >/dev/null 2>&1 &
+                nohup $player_function --http-header-fields="Referer: https://zokoanime.video/" $sub_file_flag $skip_flag $player_extra_flags --force-media-title="${anime_title} Episode ${ep_no}" "$video_link" >/dev/null 2>&1 &
             else
-                $player_function --http-header-fields="Referer: https://zokoanime.video/" $skip_flag $player_extra_flags --force-media-title="${anime_title} Episode ${ep_no}" "$video_link"
+                $player_function --http-header-fields="Referer: https://zokoanime.video/" $sub_file_flag $skip_flag $player_extra_flags --force-media-title="${anime_title} Episode ${ep_no}" "$video_link"

Note: subs are currently only wired into the mpv paths; VLC and Syncplay would need --sub-file equivalents if so desired.

Happy to turn this into a commit on the branch if it helps.

[Dhairya3391]

added sub fix, it's bit janky but works for now

[jakobi]

Still a bug in deobfuscate, last line:

The sed regex stretches is greedy, ignores languages of interest and thus extends to the end of array of subtitles: it always and only reports the vtt of the last-listed language:

printf "%s\n" "$(printf "%b" "$_output" | \
   sed -n 's|.*"subtitles" \[[^]]*"src":"\([^"]*\.vtt[^"]*\)".*|\1|p')"

To instead force english, this scrap works (plus some debug on stderr; line-splitting in sed really isn't my forte):

# match against label, do avoid capturing parens in the regex
export _langre="english|german" 
printf "%b" "$_output" | perl -e '
   undef $/; $_=<>; 
   s/.*"subtitles":\[([^]]*).*/$1/s and
     /"label":"[^"]*(?:$ENV{_langre}).*?"src":"([^"]+)/si and 
    printf "%s\n", "$1" and
    printf MAIN::STDERR "%s\n", "$1";
'

Likely the language selection grep should be moved to the caller, rather than hidden in deobfuscate, i.e. deobfuscate in that case would return lines 2..n being the languages to egrep | head -1 against.

Maybe an ordered list of regexes, (say ':' separated like in $PATH) would be better, as that would permit the user to specify the languages with an order of precedence.

OT: how does language selection work anyway in ani-cli, say for a dub? I didn't see anything when skimming for any use of $lang/$_lang or http headers (and $_lang is the dub/sub selection instead)?

[Tsagar3]

Good catch @jakobi — I can confirm the greedy match. I reproduced it with a JSON mirroring the real embed (3 languages, English marked "default":true):

# current sed
.*"subtitles":\[[^]]*"src":"..."
        -> always returns the LAST-listed language (es.vtt in my test)

The problem is [^]]* swallows the whole array and the trailing .* backtracks to the last "src":" before the closing ], so language order (and "default") is ignored. It just happened to work in my earlier test because One Piece ep 1 lists a single language.

The clean fix is to emit one line per subtitle entry and let the caller pick:

--- deobfuscate_blob
-    printf "%s\n" "$(printf "%b" "$_output" | sed -n 's|.*"subtitles":\[[^]]*"src":"\([^"]*\.vtt[^"]*\)".*|\1|p')"
+    printf "%s\n" "$(printf "%b" "$_output" | tr '}' '\n' | sed -n 's|.*"lang":"\([^"]*\)".*"label":"\([^"]*\)".*"src":"\([^"]*\.vtt[^"]*\)".*|\1\t\2\t\3|p')"
+
+# now emits: lang<TAB>label<TAB>src , one per language

--- hianime_m3u8 (deobfuscate step)
     _deobfuscated="$(printf "%s\n" "$(deobfuscate_blob "$_blob")" | sed '/^$/d')"
     _m3u8_master="$(printf "%s\n" "$_deobfuscated" | sed -n '1p')"
-    _subs="$(printf "%s\n" "$_deobfuscated" | sed -n '2p')"
+    _sub_entries="$(printf "%s\n" "$_deobfuscated" | sed -n '2,$p')"
+    _sub_langs="${ANI_CLI_SUB_LANG:-en}"
+    _subs="$(printf "%s\n" "$_sub_entries" | grep -iE "^$(printf '%s' "$_sub_langs" | tr ',:' '|')[[:space:]]" | head -n1 | cut -f 3)"
+    [ -z "$_subs" ] && _subs="$(printf "%s\n" "$_sub_entries" | head -n1 | cut -f 3)"

Selection logic (verified):

• picks the first entry whose lang matches ANI_CLI_SUB_LANG (default en, comma/colon-separated list gives ordered precedence, i.e. @jakobi's PATH-style idea for free)
• falls back to the first-listed entry when nothing matches (keeps the pre-fix default=true behaviour when English is listed first, like the real embed)
• everything downstream (sub_link, local VTT download, --sub-file/VLC/Syncplay flags) is untouched since it just consumes _subs

Re grep -E/case-insensitivity: the script already uses grep -E for quality selection, so this is consistent. If you want label-based matching too, extending the alternation to label is trivial.

And to answer your OT: language selection in ani-cli today is only sub vs dub — --dub sets mode=dub, which picks the dub server/stream via data-type on the servers list. The subtitle language itself wasn't user-selectable before this; ANI_CLI_SUB_LANG adds exactly that knob. For dub streams the .vtt is the dialogue subtitle track (usually en), which the default covers.