#!/usr/bin/env bash
#
# sort-lines-group-by-start — cluster lines that share their first
# whitespace-separated field ($1). Keys appear in first-seen order, and lines
# within a key keep their original order. It groups; it does NOT sort.
#
# Pass the log file as an argument; result is written to stdout:
#
#     tools/sort-lines-group-by-start.sh filename.log
#
# Useful for a serial/boot log where several sources interleave and each line is
# prefixed with its source (the first field): this pulls every source's lines
# back together, in the order the sources first appeared, without reordering
# within a source.
#
#     input                       output
#     pci-bus: scan start         pci-bus: scan start
#     acpi: reported PNP0303      pci-bus: 5 functions
#     pci-bus: 5 functions        acpi: reported PNP0303
#     acpi: reported PNP0501      acpi: reported PNP0501

awk '{lines[$1] = lines[$1] ? lines[$1] ORS $0 : $0; if (!seen[$1]++) order[++count] = $1} END {for (i=1; i<=count; i++) print lines[order[i]]}' "$@"