Quick dirty hack to find the location of a timestamp in a huge log file (assuming logs are roughly in order)

ak_locate_in_log

#!/bin/bash
#
# Quick dirty hack to let me find the location of a timestamp in a huge log file

file="$1"
date="$2"

if [ -z "$file" ] || [ -z "$date" ]; then
  echo 'Invalid arguments! Example use:'
  echo '  ak_locate_in_log ./path_to_file "2023-08-13 01:00:02.085"'
  exit 1
fi


size=$(stat --format=%s "$file")

# work in 1M blocks
blocks=$(( $size / (1024 * 1024) ))

# skip to the middle and grab a block
skip=$(( $blocks / 2 ))
range_size=$skip


while : ; do
  first=$(dd status=none if="$file" bs=1M skip=$skip count=1 | egrep -o '^[0-9-]{10} [0-9:]+' | sort | head -n1)
  last=$(dd status=none  if="$file" bs=1M skip=$skip count=1 | egrep -o '^[0-9-]{10} [0-9:]+' | sort | tail -n1)

  echo first=$first
  echo last=$last
  echo date=$date

  range_size=$(( $range_size / 2 ))

  if [[ $range_size = 0 ]]; then
    break
  fi

  if [[ "$date" > "$last" ]]; then
    echo "Going up"
    skip=$(( "$skip" + "$range_size" ))
    continue
  fi

  if [[ $date < $first ]]; then
    echo "Going down"
    skip=$(( $skip - $range_size ))
    continue
  fi

  # done
  break

  # TODO what about no first / last?
done

echo "Stopped at block: $skip"
echo
echo dd status=none if="$file" bs=1M skip="$skip" count=1