Download youtube video and caption using yt-dlp utility.

yt-int.sh

#!/bin/bash

# Usage: ./yt-int.sh <youtube_url>

if [ -z "$1" ]; then
    echo "Usage: $0 <youtube_url>"
    exit 1
fi

URL="$1"

# Temporary file to hold format list
formats_file=$(mktemp)

# Get all formats for the video
yt-dlp -F "$URL" > "$formats_file"

declare -A video_choices
declare -a res_order=(1080 720 360)
idx=1

echo "Searching for common video formats (1080p, 720p, 360p) ..."

for res in "${res_order[@]}"; do
    # Find video-only formats in mp4 or webm with matching resolution in "p" or WxH format
    line=$(grep -E "^[0-9]+.*(mp4|webm).*video only" "$formats_file" | grep -E "[[:space:]]${res}p?[[:space:]]" | head -n 1)
    if [ -z "$line" ]; then
        line=$(grep -E "^[0-9]+.*(mp4|webm).*video only" "$formats_file" | grep -E "x$res" | head -n 1)
    fi
    if [ -n "$line" ]; then
        code=$(echo "$line" | awk '{print $1}')
        desc=$(echo "$line" | cut -c 5- | awk '{$1=$1};1')  # Trim leading spaces
        echo "[$idx] $desc"
        video_choices[$idx]="$code"
        ((idx++))
    fi
done

# If no standard formats found, fallback to top 3 video-only mp4/webm formats
if [ ${#video_choices[@]} -eq 0 ]; then
    echo "No standard 1080p, 720p, or 360p video-only formats detected"
    echo "Showing top 3 video-only mp4/webm formats instead:"
    mapfile -t lines < <(grep -E "^[0-9]+.*(mp4|webm).*video only" "$formats_file" | head -n 3)
    idx=1
    for line in "${lines[@]}"; do
        code=$(echo "$line" | awk '{print $1}')
        desc=$(echo "$line" | cut -c 5- | awk '{$1=$1};1')
        echo "[$idx] $desc"
        video_choices[$idx]="$code"
        ((idx++))
    done
fi

# Prompt user for video format choice
read -p "Enter the number of the video format you want to download: " vnum

if [[ -z "${video_choices[$vnum]}" ]]; then
    echo "Invalid choice or no formats available. Exiting."
    rm -f "$formats_file"
    exit 1
fi

vcode="${video_choices[$vnum]}"

# Automatically pick best m4a audio format
acode=$(grep -E "^[0-9]+.*m4a.*audio only" "$formats_file" | head -n 1 | awk '{print $1}')
if [ -z "$acode" ]; then
    echo "No m4a audio format found, selecting best audio instead."
    acode=$(grep -E "^[0-9]+.*audio only" "$formats_file" | head -n 1 | awk '{print $1}')
fi

echo "Selected video format code: $vcode"
echo "Selected audio format code: $acode"

# Output template with title and video ID
OUTTEMPLATE='%(title)s [%(id)s].%(ext)s'

echo "Downloading auto-generated English captions (subtitles)..."
# Try SRT first, fallback to VTT if SRT unavailable
if yt-dlp --write-auto-subs --sub-lang en --sub-format srt --skip-download -o "$OUTTEMPLATE" "$URL"; then
    echo "Auto-generated English subtitles downloaded in SRT format."
else
    echo "Failed to download subtitles in SRT format, trying VTT format..."
    yt-dlp --write-auto-subs --sub-lang en --sub-format vtt --skip-download -o "$OUTTEMPLATE" "$URL"
fi

echo "Downloading video and audio streams and merging into MP4..."
if yt-dlp -f "${vcode}+${acode}" --remux-video mp4 -o "$OUTTEMPLATE" "$URL"; then
    # Download succeeded - rename files removing trailing [ID] and fix double .en in subtitles

    video_file0=$(yt-dlp --get-filename -o "$OUTTEMPLATE" "$URL")
    video_ext="${video_file0##*.}"
    video_base="${video_file0%.*}"

    # Remove trailing spaces + [ID]
    echo "video_base: $video_base"
    base_no_id=$(echo "$video_base" | sed -E 's/ *\[[^]]+\]$//')
    final_video="${base_no_id}.mp4"
    video_file="${video_base}.mp4"

    if [[ "$video_file" != "$final_video" ]]; then
        echo "video_file: $video_file"
        if [ -f "$video_file" ]; then
            mv -- "$video_file" "$final_video"
            echo "Renamed video file to: $final_video"
        else
            echo "Video file '$video_file' not found after download."
        fi
    else
        echo "Video filename did not require renaming."
    fi

    # For subtitles:
    # Fix the subtitle file variable to avoid duplicated ".en"
    # Original filenames have pattern: 'title [ID].en.srt' (or .vtt)
    # So subtitle basename = video_base (which includes [ID]), append '.en.srt' duplicates '.en'
    # Instead, detect subtitles by replacing extension in video_file from 'ext' to 'en.srt' or 'en.vtt' after removing [ID]

    # Prepare possible subtitle filenames by replacing video extension to srt/vtt with .en
    # First, build subtitle base without [ID]
    base_no_id_sub="${base_no_id}"  # Already stripped [ID]

    # Possible subtitle filenames
	orig_sub_srt="${video_base}.en.srt"  # original file with [ID]
	sub_file_srt="${base_no_id}.en.srt"  # cleaned final filename without [ID]
	
	if [[ "$orig_sub_srt" != "$sub_file_srt" && -f "$orig_sub_srt" ]]; then
    	mv -- "$orig_sub_srt" "$sub_file_srt"
    	echo "Renamed subtitle file to: $sub_file_srt"
	elif [ -f "$sub_file_srt" ]; then
    	echo "Subtitle filename did not require renaming."
	else
    	echo "Subtitle file not found."
	fi

else
    echo "ERROR: Video+audio download failed. No video file created."
    rm -f "$formats_file"
    exit 1
fi

# Cleanup
rm -f "$formats_file"