trim-icalendar-ics

#!/bin/bash
set -euo pipefail

usage() {
    cat <<EOF
Usage: $(basename "$0") <calendar.ics> <year> <month> [--rewrite-domain OLD NEW]

Removes all events before the specified year/month and saves the result
to a new file with .trimmed.ics suffix.

Arguments:
    calendar.ics    Path to the iCalendar file (absolute or relative)
    year            4-digit year (e.g., 2024)
    month           2-digit month (01-12)

Options:
    --rewrite-domain OLD NEW
        Replace email domain OLD with NEW in all events.
        Handles iCalendar line folding automatically.

Examples:
    $(basename "$0") ~/calendars/work.ics 2024 06
    # Creates ~/calendars/work.trimmed.ics with events from June 2024 onwards

    $(basename "$0") ~/calendars/work.ics 2024 06 --rewrite-domain utilimarc.com smith-system.com
    # Also rewrites all @utilimarc.com emails to @smith-system.com

Requirements (macOS):
    brew install coreutils   # provides gcsplit

EOF
    exit 1
}

ics_unfold() {
    awk '{
        if (substr($0, 1, 1) == " " || substr($0, 1, 1) == "\t") {
            printf "%s", substr($0, 2)
        } else {
            if (NR > 1) print ""
            printf "%s", $0
        }
    }
    END { print "" }'
}

ics_fold() {
    awk '{
        line = $0
        if (length(line) <= 75) {
            print line
        } else {
            print substr(line, 1, 75)
            line = substr(line, 76)
            while (length(line) > 74) {
                print " " substr(line, 1, 74)
                line = substr(line, 75)
            }
            if (length(line) > 0) {
                print " " line
            }
        }
    }'
}

cleanup() {
    if [[ -n "${temp_dir:-}" && -d "$temp_dir" ]]; then
        rm -rf "$temp_dir"
    fi
}

trap cleanup EXIT

if [[ $# -lt 3 ]]; then
    usage
fi

calendar_input="$1"
year="$2"
month="$3"
shift 3

old_domain=""
new_domain=""

while [[ $# -gt 0 ]]; do
    case "$1" in
        --rewrite-domain)
            if [[ $# -lt 3 ]]; then
                echo "Error: --rewrite-domain requires OLD and NEW arguments" >&2
                exit 1
            fi
            old_domain="$2"
            new_domain="$3"
            shift 3
            ;;
        *)
            echo "Error: Unknown option: $1" >&2
            usage
            ;;
    esac
done

if [[ ! "$year" =~ ^[0-9]{4}$ ]]; then
    echo "Error: Year must be 4 digits (e.g., 2024)" >&2
    exit 1
fi

if [[ ! "$month" =~ ^(0[1-9]|1[0-2])$ ]]; then
    echo "Error: Month must be 01-12" >&2
    exit 1
fi

if [[ "$calendar_input" = /* ]]; then
    calendar_file="$calendar_input"
else
    calendar_file="$(pwd)/$calendar_input"
fi

if [[ ! -f "$calendar_file" ]]; then
    echo "Error: Calendar file not found: $calendar_file" >&2
    exit 1
fi

calendar_dir="$(dirname "$calendar_file")"
calendar_basename="$(basename "$calendar_file" .ics)"
output_file="${calendar_dir}/${calendar_basename}.trimmed.ics"

if command -v gcsplit &>/dev/null; then
    CSPLIT=gcsplit
elif command -v csplit &>/dev/null; then
    if csplit --version 2>&1 | grep -q GNU; then
        CSPLIT="csplit"
    else
        echo "Error: GNU csplit required. On macOS, install with: brew install coreutils" >&2
        exit 1
    fi
else
    echo "Error: csplit not found. On macOS, install with: brew install coreutils" >&2
    exit 1
fi

temp_dir=$(mktemp -d)
temp_files_digits=5

cd "$temp_dir"

total_files=$($CSPLIT -n "$temp_files_digits" "$calendar_file" '/BEGIN:VEVENT/' '{*}' 2>/dev/null | wc -l)
total_files=$((total_files))

if [[ $total_files -lt 2 ]]; then
    echo "Error: No events found in calendar file" >&2
    exit 1
fi

echo "Found $((total_files - 1)) events in calendar"

grep -l DTSTART xx* 2>/dev/null | while read -r file; do
    dtstart=$(grep -m1 'DTSTART' "$file" | head -1)
    if [[ "$dtstart" =~ ([0-9]{4})([0-9]{2}) ]]; then
        event_year="${BASH_REMATCH[1]}"
        event_month="${BASH_REMATCH[2]}"
        if (( 10#$event_year > 10#$year || (10#$event_year == 10#$year && 10#$event_month >= 10#$month) )); then
            cat "$file"
        fi
    fi
done | grep -v '^END:VCALENDAR$' > events

header_file="xx$(printf '%0*d' "$temp_files_digits" 0)"

{
    cat "$header_file"
    cat events
    echo "END:VCALENDAR"
} > "$output_file"

if [[ -n "$old_domain" ]]; then
    echo "Rewriting @${old_domain} to @${new_domain}..."
    escaped_old=$(printf '%s' "$old_domain" | sed 's/[.[\*^$/]/\\&/g')
    escaped_new=$(printf '%s' "$new_domain" | sed 's/[&/\]/\\&/g')
    tr -d '\r' < "$output_file" \
        | ics_unfold \
        | sed "s/@${escaped_old}/@${escaped_new}/g" \
        | ics_fold > "${output_file}.tmp"
    mv "${output_file}.tmp" "$output_file"
fi

kept_events=$(grep -c 'BEGIN:VEVENT' "$output_file" 2>/dev/null || echo 0)
echo "Kept $kept_events events from ${year}-${month} onwards"
echo "Output saved to: $output_file"