scan-it

#!/bin/bash

# Scanner Configuration
SCANNER_DEVICE="airscan:e0:Canon TS7400 series"
RESOLUTION=300
MODE="Color"

# Output Configuration
OUTPUT_FORMAT="jpeg"
OUTPUT_PREFIX="out%d" # scanimage adds extension automatically usually, but let's be safe. 
# actually scanimage --batch=format%d.jpg works best. 
# The original script used OUTPUT_PREFIX="out%d.$OUTPUT_FORMAT" which is correct for --batch argument format specifier

BATCH_PREFIX="out"
FILE_PATTERN="${BATCH_PREFIX}%d.${OUTPUT_FORMAT}"

# Helper to print steps
function log() {
    echo -e "\033[1;32m[SCAN]\033[0m $1"
}

function error() {
    echo -e "\033[1;31m[ERROR]\033[0m $1"
}

scan_batch() {
    local start=$1
    local inc=$2
    
    # Capture stderr to count pages, while also showing it to user
    # We use tee to show output, but we need to capture it.
    # A temp file is cleaner than a pipe for capturing reliably while streaming.
    local logfile=$(mktemp)
    
    scanimage -d "$SCANNER_DEVICE" \
              --source "ADF" \
              --format "$OUTPUT_FORMAT" \
              --resolution "$RESOLUTION" \
              --mode "$MODE" \
              --batch="${BATCH_PREFIX}%d.${OUTPUT_FORMAT}" \
              --batch-start="$start" \
              --batch-increment="$inc" \
              -x 215 -y 297 2> >(tee "$logfile" >&2)

    # Count how many "Scanned page" or scanning events occurred.
    # scanimage output format: "Scanned page 1. (scanner specific)"
    # Reliability depends on scanimage version/driver.
    # Alternative: Count created files.
    
    rm -f "$logfile"
}

scan_single_sided() {
    log "Starting Single Sided Scan..."
    scan_batch 1 1
}

scan_double_sided() {
    log "Phase 1: Scanning Front Sides (Odds)..."
    scan_batch 1 2

    # Robust page counting by counting odd-numbered files generated
    # (Assuming we started at 1 and incremented by 2)
    # We look for the highest numbered odd file to determine count, 
    # or just count how many files look like out*.jpeg
    
    # Let's simple-count the files that match the pattern for this session.
    # Warning: this counts all jpegs.
    local count=$(ls ${BATCH_PREFIX}*.${OUTPUT_FORMAT} 2>/dev/null | wc -l)

    if [ "$count" -eq 0 ]; then
        error "No pages detected. Exiting."
        exit 1
    fi

    log "Batch 1 complete. Detected $count pages."
    log "Please flip the document stack."
    log "Ensure the page that was scanned LAST (top of output) is fed FIRST (top of input)."
    read -p "Press Enter to continue..."

    # Calculate start page for reverse side
    # Fronts: 1, 3, 5 (3 pages)
    # Backs should be: 6, 4, 2
    # Start = count * 2
    local start_back=$((count * 2))

    log "Phase 2: Scanning Back Sides (Evens)..."
    scan_batch "$start_back" "-2"
}

check_blank_pages() {
    log "Checking for potentially blank pages..."
    # Check if tesseract is installed
    if ! command -v tesseract &> /dev/null; then
        echo "Tesseract not found, skipping blank page check."
        return
    fi

    for file in ${BATCH_PREFIX}*.${OUTPUT_FORMAT}; do
        if [ -f "$file" ]; then
            text=$(tesseract "$file" stdout 2>/dev/null)
            # If text is empty or very short? 
            # Empty check:
            if [ -z "${text//[[:space:]]/}" ]; then
                echo "  ? Potential blank page: $file"
            fi
        fi
    done
}

# CLI Argument Parsing
MODE_FLAG="double" # default

while getopts ":hsd" opt; do
  case $opt in
    h)
      echo "Usage: $0 [-h] [-s (single)] [-d (double, default)]"
      exit 0
      ;;
    s)
      MODE_FLAG="single"
      ;;
    d)
      MODE_FLAG="double"
      ;;
    \?)
      echo "Invalid option: -$OPTARG"
      exit 1
      ;;
  esac
done

if [ "$MODE_FLAG" == "single" ]; then
    scan_single_sided
else
    scan_double_sided
fi

check_blank_pages
log "Done."