#!/bin/bash

# Script to auto-crop and scale WebM files based on content analysis of all frames
# Usage: ./webm_auto_crop.sh [directory_path]

# Function to display usage
usage() {
    echo "Usage: $0 [directory_path]"
    echo "  directory_path: Path to directory containing WebM files (default: current directory)"
    echo ""
    echo "This script will:"
    echo "  - Analyze frames in each WebM file to detect content boundaries"
    echo "  - Calculate unified crop region that covers all frames"
    echo "  - Crop and scale the entire video to fill original dimensions"
    echo "  - Preserve audio when available"
    echo "  - Save results in a 'crop' subdirectory"
    exit 1
}

# Check if ffmpeg is installed
if ! command -v ffmpeg &> /dev/null; then
    echo "Error: ffmpeg is not installed or not in PATH"
    echo "Please install ffmpeg to use this script"
    exit 1
fi

# Check if ffprobe is installed
if ! command -v ffprobe &> /dev/null; then
    echo "Error: ffprobe is not installed or not in PATH"
    echo "Please install ffprobe (part of ffmpeg) to use this script"
    exit 1
fi

# Check if bc is installed (needed for calculations)
if ! command -v bc &> /dev/null; then
    echo "Error: bc (basic calculator) is not installed or not in PATH"
    echo "Please install bc to use this script"
    echo "  macOS: brew install bc"
    echo "  Ubuntu/Debian: sudo apt install bc"
    exit 1
fi

echo "Content detection method:"
if command -v identify &> /dev/null && command -v convert &> /dev/null; then
    echo "  ✓ Using ImageMagick (optimal for transparency)"
else
    echo "  ⚠ Using ffmpeg cropdetect (fallback method)"
    echo "    For better results with transparent content, install ImageMagick:"
    echo "    macOS: brew install imagemagick"
    echo "    Ubuntu/Debian: sudo apt install imagemagick"
fi
echo ""

# Set directory path (use current directory if not specified)
if [ $# -eq 0 ]; then
    DIRECTORY="."
elif [ $# -eq 1 ]; then
    DIRECTORY="$1"
else
    echo "Error: Too many arguments"
    usage
fi

# Check if directory exists
if [ ! -d "$DIRECTORY" ]; then
    echo "Error: Directory '$DIRECTORY' does not exist"
    exit 1
fi

# Convert to absolute path
DIRECTORY=$(realpath "$DIRECTORY")
echo "Processing WebM files in: $DIRECTORY"

# Create crop output directory
CROP_DIR="$DIRECTORY/crop"
if [ ! -d "$CROP_DIR" ]; then
    mkdir -p "$CROP_DIR"
    echo "Created output directory: $CROP_DIR"
else
    echo "Using existing output directory: $CROP_DIR"
fi

# Find all WebM files in the directory
WEBM_FILES=($(find "$DIRECTORY" -maxdepth 1 -name "*.webm" -type f))

# Check if any WebM files were found
if [ ${#WEBM_FILES[@]} -eq 0 ]; then
    echo "No WebM files found in '$DIRECTORY'"
    exit 0
fi

echo "Found ${#WEBM_FILES[@]} WebM file(s)"
echo ""

# Counter for processed files
PROCESSED=0
FAILED=0

# Function to detect content bounding box for a frame with transparency
detect_content_bounds() {
    local frame_file="$1"
    local bounds_file="$2"

    # Use ImageMagick to find the bounding box of non-transparent content
    if command -v identify &> /dev/null && command -v convert &> /dev/null; then
        # Get the trim bounding box (removes transparent borders) - works with WebP
        TRIM_INFO=$(convert "$frame_file" -format "%@" info: 2>/dev/null)
        if [ -n "$TRIM_INFO" ] && [[ "$TRIM_INFO" =~ ^([0-9]+)x([0-9]+)\+([0-9]+)\+([0-9]+)$ ]]; then
            # Parse geometry: WxH+X+Y
            echo "${BASH_REMATCH[1]}:${BASH_REMATCH[2]}:${BASH_REMATCH[3]}:${BASH_REMATCH[4]}" > "$bounds_file"
        else
            echo "" > "$bounds_file"
        fi
    else
        # Fallback: use ffmpeg cropdetect with adjusted threshold for transparency
        ffmpeg -i "$frame_file" \
               -vf "cropdetect=limit=0.1:round=2" \
               -f null \
               - 2>&1 | grep "crop=" | tail -1 | sed 's/.*crop=\([0-9:]*\).*/\1/' > "$bounds_file" 2>/dev/null
    fi
}

# Function to parse crop parameters
parse_crop() {
    local crop_str="$1"
    if [[ "$crop_str" =~ ^([0-9]+):([0-9]+):([0-9]+):([0-9]+)$ ]]; then
        echo "${BASH_REMATCH[1]} ${BASH_REMATCH[2]} ${BASH_REMATCH[3]} ${BASH_REMATCH[4]}"
    else
        echo "0 0 0 0"
    fi
}

# Process each WebM file
for webm_file in "${WEBM_FILES[@]}"; do
    # Get the base filename without path and extension
    base_name=$(basename "$webm_file" .webm)

    # Output WebM file path
    webm_output="$CROP_DIR/${base_name}.webm"

    echo "Processing: $(basename "$webm_file")"

    # Detect codec type (VP8 or VP9) to use correct decoder
    CODEC_INFO=$(ffprobe -v quiet -select_streams v:0 -show_entries stream=codec_name -of csv=p=0 "$webm_file" 2>/dev/null)

    # Set appropriate decoder based on codec
    if [ "$CODEC_INFO" = "vp9" ]; then
        DECODER="libvpx-vp9"
        ENCODER="libvpx-vp9"
    else
        DECODER="libvpx"
        ENCODER="libvpx"
    fi

    # Get video properties - use EXACT same method as webm_to_webp.sh
    ORIG_WIDTH=$(ffprobe -v quiet -select_streams v:0 -show_entries stream=width -of csv=p=0 "$webm_file" 2>/dev/null)
    ORIG_HEIGHT=$(ffprobe -v quiet -select_streams v:0 -show_entries stream=height -of csv=p=0 "$webm_file" 2>/dev/null)

    if [ -z "$ORIG_WIDTH" ] || [ -z "$ORIG_HEIGHT" ]; then
        echo "  ✗ Could not get video dimensions"
        ((FAILED++))
        continue
    fi

    echo "  → Original dimensions: ${ORIG_WIDTH}x${ORIG_HEIGHT}"

    # Get total frame count - EXACT same method as webm_to_webp.sh
    FRAME_COUNT=$(ffprobe -v quiet -select_streams v:0 -count_frames -show_entries stream=nb_read_frames -of csv=p=0 "$webm_file" 2>/dev/null)

    if [ -z "$FRAME_COUNT" ] || [ "$FRAME_COUNT" -eq 0 ]; then
        echo "  ✗ Could not determine frame count"
        ((FAILED++))
        continue
    fi

    echo "  → Total frames: $FRAME_COUNT"

    # Check if video has audio
    AUDIO_STREAMS=$(ffprobe -v quiet -select_streams a -show_entries stream=index -of csv=p=0 "$webm_file" 2>/dev/null | wc -l)

    # Create temporary directory for frame analysis
    TEMP_DIR=$(mktemp -d)

    echo "  → Analyzing frames to find optimal crop region..."

    # Calculate sample count - sample every 3rd frame to ensure we catch all extreme positions
    SAMPLE_COUNT=$((FRAME_COUNT / 3))
    if [ "$SAMPLE_COUNT" -gt 300 ]; then
        SAMPLE_COUNT=300  # Cap at 300 to avoid excessive processing
    elif [ "$SAMPLE_COUNT" -lt 50 ]; then
        SAMPLE_COUNT=$FRAME_COUNT  # For very short videos, sample all frames
    fi

    # Initialize crop bounds (will find union of all bounds)
    MIN_X=$ORIG_WIDTH
    MIN_Y=$ORIG_HEIGHT
    MAX_X=0
    MAX_Y=0
    VALID_FRAMES=0

    echo "  → Analyzing $SAMPLE_COUNT sample frames from $FRAME_COUNT total frames..."

    # Sample frames evenly throughout the video with much higher density
    for ((i=0; i<SAMPLE_COUNT; i++)); do
        # Calculate frame position (evenly distributed)
        FRAME_POS=$((i * FRAME_COUNT / SAMPLE_COUNT))

        # Extract frame at specific position
        TEMP_FRAME="$TEMP_DIR/frame_${FRAME_POS}.webp"
        BOUNDS_FILE="$TEMP_DIR/bounds_${FRAME_POS}.txt"

        echo "    Extracting frame #${FRAME_POS}..."

        # Extract frame - EXACT same command as webm_to_webp.sh
        TEMP_FRAME="$TEMP_DIR/frame_${FRAME_POS}.webp"
        ffmpeg -c:v "$DECODER" \
               -i "$webm_file" \
               -vf "select=eq(n\\,$FRAME_POS)" \
               -vframes 1 \
               -c:v libwebp \
               -lossless 1 \
               -y \
               "$TEMP_FRAME" \
               -v quiet 2>/dev/null

        if [ -f "$TEMP_FRAME" ]; then
            # Check if frame was extracted successfully
            FRAME_SIZE=$(stat -f%z "$TEMP_FRAME" 2>/dev/null || stat -c%s "$TEMP_FRAME" 2>/dev/null)
            echo "    Frame file size: $FRAME_SIZE bytes"

            # Detect content bounds
            detect_content_bounds "$TEMP_FRAME" "$BOUNDS_FILE"

            if [ -f "$BOUNDS_FILE" ]; then
                CROP_PARAMS=$(cat "$BOUNDS_FILE")
                echo "    Crop params: '$CROP_PARAMS'"

                if [ -n "$CROP_PARAMS" ] && [[ "$CROP_PARAMS" =~ ^([0-9]+):([0-9]+):([0-9]+):([0-9]+)$ ]]; then
                    read -r CROP_W CROP_H CROP_X CROP_Y <<< $(parse_crop "$CROP_PARAMS")

                    if [ "$CROP_W" -gt 0 ] && [ "$CROP_H" -gt 0 ]; then
                        # Calculate bounds of this frame's content
                        FRAME_MIN_X=$CROP_X
                        FRAME_MIN_Y=$CROP_Y
                        FRAME_MAX_X=$((CROP_X + CROP_W))
                        FRAME_MAX_Y=$((CROP_Y + CROP_H))

                        # Update union bounds
                        if [ "$FRAME_MIN_X" -lt "$MIN_X" ]; then MIN_X=$FRAME_MIN_X; fi
                        if [ "$FRAME_MIN_Y" -lt "$MIN_Y" ]; then MIN_Y=$FRAME_MIN_Y; fi
                        if [ "$FRAME_MAX_X" -gt "$MAX_X" ]; then MAX_X=$FRAME_MAX_X; fi
                        if [ "$FRAME_MAX_Y" -gt "$MAX_Y" ]; then MAX_Y=$FRAME_MAX_Y; fi

                        ((VALID_FRAMES++))
                        echo "    ✓ Frame #${FRAME_POS}: ${CROP_W}x${CROP_H} at (${CROP_X},${CROP_Y})"
                    else
                        echo "    ✗ Frame #${FRAME_POS}: Invalid crop dimensions"
                    fi
                else
                    echo "    ✗ Frame #${FRAME_POS}: No valid crop params detected"
                fi
            else
                echo "    ✗ Frame #${FRAME_POS}: Bounds file not created"
            fi
        else
            echo "    ✗ Frame #${FRAME_POS}: Extraction failed"
        fi
    done

    # Clean up temp directory
    rm -rf "$TEMP_DIR"

    echo "  → Valid frames found: $VALID_FRAMES out of $SAMPLE_COUNT"

    # Calculate final crop dimensions
    FINAL_CROP_W=$((MAX_X - MIN_X))
    FINAL_CROP_H=$((MAX_Y - MIN_Y))

    # Validate crop dimensions
    if [ "$VALID_FRAMES" -eq 0 ]; then
        echo "  ✗ No content detected in any frame - video may be fully transparent"
        ((FAILED++))
        continue
    elif [ "$FINAL_CROP_W" -le 0 ] || [ "$FINAL_CROP_H" -le 0 ] || [ "$MIN_X" -lt 0 ] || [ "$MIN_Y" -lt 0 ]; then
        echo "  ✗ Could not determine valid crop region (bounds: ${MIN_X},${MIN_Y} to ${MAX_X},${MAX_Y})"
        ((FAILED++))
        continue
    elif [ "$FINAL_CROP_W" -ge "$ORIG_WIDTH" ] && [ "$FINAL_CROP_H" -ge "$ORIG_HEIGHT" ]; then
        echo "  ⚠ Content already fills entire frame - skipping crop"
        ((FAILED++))
        continue
    fi

    echo "  → Crop region: ${FINAL_CROP_W}x${FINAL_CROP_H} at offset (${MIN_X},${MIN_Y})"

    # Calculate scale to fit inside original dimensions while maintaining aspect ratio
    SCALE_W=$(echo "scale=6; $ORIG_WIDTH / $FINAL_CROP_W" | bc)
    SCALE_H=$(echo "scale=6; $ORIG_HEIGHT / $FINAL_CROP_H" | bc)

    # Use the smaller scale factor to ensure content fits completely inside
    if (( $(echo "$SCALE_W < $SCALE_H" | bc -l) )); then
        SCALE_FACTOR=$SCALE_W
    else
        SCALE_FACTOR=$SCALE_H
    fi

    # Calculate scaled dimensions
    SCALED_W=$(echo "$FINAL_CROP_W * $SCALE_FACTOR" | bc | cut -d. -f1)
    SCALED_H=$(echo "$FINAL_CROP_H * $SCALE_FACTOR" | bc | cut -d. -f1)

    echo "  → Scale factor: ${SCALE_FACTOR}x (content will be ${SCALED_W}x${SCALED_H})"

    # Build ffmpeg filter chain - crop, scale to fit, then pad to original size
    FILTER_CHAIN="crop=${FINAL_CROP_W}:${FINAL_CROP_H}:${MIN_X}:${MIN_Y},scale=${SCALED_W}:${SCALED_H}:flags=lanczos,pad=${ORIG_WIDTH}:${ORIG_HEIGHT}:(ow-iw)/2:(oh-ih)/2:color=0x00000000"

    # Build ffmpeg command
    if [ "$AUDIO_STREAMS" -gt 0 ]; then
        echo "  → Processing with audio..."
        FFMPEG_CMD="ffmpeg -c:v $DECODER -i \"$webm_file\" -vf \"$FILTER_CHAIN\" -c:v $ENCODER -b:v 1M -c:a libopus -b:a 128k -y \"$webm_output\""
    else
        echo "  → Processing without audio..."
        FFMPEG_CMD="ffmpeg -c:v $DECODER -i \"$webm_file\" -vf \"$FILTER_CHAIN\" -c:v $ENCODER -b:v 1M -an -y \"$webm_output\""
    fi

    # Execute ffmpeg command
    echo "  → Running command: $FFMPEG_CMD"
    if eval "$FFMPEG_CMD" -v error -stats; then
        echo "  ✓ Successfully created: ${base_name}.webm"
        ((PROCESSED++))
    else
        echo "  ✗ Failed to process: $(basename "$webm_file")"
        ((FAILED++))
    fi
    echo ""
done

# Summary
echo "=== Processing Complete ==="
echo "Successfully processed: $PROCESSED files"
if [ $FAILED -gt 0 ]; then
    echo "Failed to process: $FAILED files"
fi
echo "Output directory: $CROP_DIR"
echo ""

# List generated files
if [ $PROCESSED -gt 0 ]; then
    echo "Generated cropped WebM files:"
    ls -la "$CROP_DIR"/*.webm 2>/dev/null | while read -r line; do
        echo "  $line"
    done
fi