#!/usr/bin/env bash
# nc-find.sh — cross-site Cloverleaf thread/protocol search. Native v3.
#
# Replaces (in v3 terms) the v1 family `tbn`, `tbp`, `tbh`, `tbpr`, plus the
# v1 `<thread> where` command. Searches every NetConfig under $HCIROOT (or a
# passed list) without invoking v1/v2 wrappers, and emits site, thread, port,
# host, process, NetConfig path, and line number for each match.
#
# Usage:
#   nc-find.sh --name PATTERN            # like tbn: case-insensitive substring on thread name
#   nc-find.sh --port PORT               # like tbp: exact port match
#   nc-find.sh --host HOST               # like tbh: substring on host
#   nc-find.sh --process PROC            # like tbpr: substring on PROCESSNAME
#   nc-find.sh --where THREAD            # like `<thread> where`: file:line of the thread's declaration
#   nc-find.sh --xlate XLATENAME         # threads referencing a specific .xlt file
#   nc-find.sh --tclproc TCLPROC         # threads referencing a specific tclproc
#
# Common flags:
#   --hciroot DIR                        # default $HCIROOT
#   --netconfigs PATHS                   # colon-separated explicit list (overrides --hciroot)
#   --format tsv|table|jsonl             # default: table
#   --case-sensitive                     # default: case-insensitive for name/host/process
set -o pipefail

NC_SELF="$0"
LIB_DIR="$(cd "$(dirname "$NC_SELF")" && pwd)"
NCP="$LIB_DIR/nc-parse.sh"

die() { printf 'nc-find: %s\n' "$*" >&2; exit 1; }

MODE=""
QUERY=""
HCIROOT_OVERRIDE=""
NETCONFIGS_OVERRIDE=""
FORMAT="table"
CASE_SENSITIVE=0

while [ $# -gt 0 ]; do
  case "$1" in
    --name)            shift; MODE="name";    QUERY="$1" ;;
    --port)            shift; MODE="port";    QUERY="$1" ;;
    --host)            shift; MODE="host";    QUERY="$1" ;;
    --process)         shift; MODE="process"; QUERY="$1" ;;
    --where)           shift; MODE="where";   QUERY="$1" ;;
    --xlate)           shift; MODE="xlate";   QUERY="$1" ;;
    --tclproc)         shift; MODE="tclproc"; QUERY="$1" ;;
    --hciroot)         shift; HCIROOT_OVERRIDE="$1" ;;
    --netconfigs)      shift; NETCONFIGS_OVERRIDE="$1" ;;
    --format)          shift; FORMAT="$1" ;;
    --case-sensitive)  CASE_SENSITIVE=1 ;;
    -h|--help)         sed -n '2,22p' "$NC_SELF"; exit 0 ;;
    -*)                die "unknown flag: $1" ;;
    *)                 die "extra arg: $1" ;;
  esac
  shift
done

[ -n "$MODE" ] || die "specify a search mode: --name | --port | --host | --process | --where | --xlate | --tclproc"
case "$FORMAT" in tsv|table|jsonl) ;; *) die "bad --format: $FORMAT" ;; esac

# Resolve NetConfig list
NCONFIGS=()
if [ -n "$NETCONFIGS_OVERRIDE" ]; then
  IFS=':' read -ra NCONFIGS <<< "$NETCONFIGS_OVERRIDE"
else
  ROOT="${HCIROOT_OVERRIDE:-${HCIROOT:-}}"
  [ -n "$ROOT" ] || die "no \$HCIROOT and no --hciroot/--netconfigs"
  while IFS= read -r nc; do NCONFIGS+=("$nc"); done < <(find "$ROOT" -maxdepth 2 -name NetConfig -type f 2>/dev/null)
fi
[ ${#NCONFIGS[@]} -gt 0 ] || die "no NetConfig files found"

# Use a temp file to collect results, then format
RESULTS=$(mktemp)
trap 'rm -f "$RESULTS"' EXIT

# Helper: emit one result row to $RESULTS
emit() {
  # cols: site \t thread \t port \t host \t process \t direction \t file \t line
  printf "%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\n" "$@" >> "$RESULTS"
}

GREP_FLAGS=""
[ "$CASE_SENSITIVE" = "0" ] && GREP_FLAGS="-i"

# Per-NetConfig scanning
for nc in "${NCONFIGS[@]}"; do
  site=$(basename "$(dirname "$nc")")

  case "$MODE" in
    name|where)
      # Find protocol declarations matching the pattern
      if [ "$MODE" = "where" ]; then
        # Exact match (one specific thread name)
        line=$(grep -nE "^protocol[[:space:]]+${QUERY}[[:space:]]+\{" "$nc" 2>/dev/null | head -1 | cut -d: -f1)
        if [ -n "$line" ]; then
          pname=$("$NCP" protocol-field "$nc" "$QUERY" PROCESSNAME 2>/dev/null | head -1)
          pport=$("$NCP" protocol-nested "$nc" "$QUERY" PROTOCOL.PORT 2>/dev/null | head -1 | sed 's/^{}$//')
          phost=$("$NCP" protocol-nested "$nc" "$QUERY" PROTOCOL.HOST 2>/dev/null | head -1 | sed 's/^{}$//')
          isserver=$("$NCP" protocol-nested "$nc" "$QUERY" PROTOCOL.ISSERVER 2>/dev/null | head -1)
          obib=$("$NCP" protocol-field "$nc" "$QUERY" OBWORKASIB 2>/dev/null | head -1)
          outonly=$("$NCP" protocol-field "$nc" "$QUERY" OUTBOUNDONLY 2>/dev/null | head -1)
          direction="?"
          [ "$isserver" = "1" ] && direction="inbound-tcp"
          [ "$obib" = "1" ] && [ "$direction" = "?" ] && direction="inbound-icl"
          [ "$outonly" = "1" ] && [ "$direction" = "?" ] && direction="outbound"
          emit "$site" "$QUERY" "${pport:-—}" "${phost:-—}" "${pname:-?}" "$direction" "$nc" "$line"
        fi
      else
        # Partial match (substring)
        while IFS= read -r raw; do
          line=$(printf '%s' "$raw" | cut -d: -f1)
          thread_name=$(printf '%s' "$raw" | sed -n 's/^[0-9]*:protocol[[:space:]]\+\([A-Za-z0-9_]\+\)[[:space:]]*{.*$/\1/p')
          [ -z "$thread_name" ] && continue
          pname=$("$NCP" protocol-field "$nc" "$thread_name" PROCESSNAME 2>/dev/null | head -1)
          pport=$("$NCP" protocol-nested "$nc" "$thread_name" PROTOCOL.PORT 2>/dev/null | head -1 | sed 's/^{}$//')
          phost=$("$NCP" protocol-nested "$nc" "$thread_name" PROTOCOL.HOST 2>/dev/null | head -1 | sed 's/^{}$//')
          isserver=$("$NCP" protocol-nested "$nc" "$thread_name" PROTOCOL.ISSERVER 2>/dev/null | head -1)
          obib=$("$NCP" protocol-field "$nc" "$thread_name" OBWORKASIB 2>/dev/null | head -1)
          outonly=$("$NCP" protocol-field "$nc" "$thread_name" OUTBOUNDONLY 2>/dev/null | head -1)
          direction="?"
          [ "$isserver" = "1" ] && direction="inbound-tcp"
          [ "$obib" = "1" ] && [ "$direction" = "?" ] && direction="inbound-icl"
          [ "$outonly" = "1" ] && [ "$direction" = "?" ] && direction="outbound"
          emit "$site" "$thread_name" "${pport:-—}" "${phost:-—}" "${pname:-?}" "$direction" "$nc" "$line"
        done < <(grep -nE $GREP_FLAGS "^protocol[[:space:]]+[A-Za-z0-9_]*${QUERY}[A-Za-z0-9_]*[[:space:]]*\{" "$nc" 2>/dev/null)
      fi
      ;;

    port)
      # Find protocols whose inner PROTOCOL.PORT equals QUERY.
      "$NCP" list-protocols "$nc" 2>/dev/null | while IFS= read -r tname; do
        p=$("$NCP" protocol-nested "$nc" "$tname" PROTOCOL.PORT 2>/dev/null | head -1)
        if [ "$p" = "$QUERY" ]; then
          pname=$("$NCP" protocol-field "$nc" "$tname" PROCESSNAME 2>/dev/null | head -1)
          phost=$("$NCP" protocol-nested "$nc" "$tname" PROTOCOL.HOST 2>/dev/null | head -1 | sed 's/^{}$//')
          line=$("$NCP" protocol-line "$nc" "$tname" 2>/dev/null)
          isserver=$("$NCP" protocol-nested "$nc" "$tname" PROTOCOL.ISSERVER 2>/dev/null | head -1)
          obib=$("$NCP" protocol-field "$nc" "$tname" OBWORKASIB 2>/dev/null | head -1)
          outonly=$("$NCP" protocol-field "$nc" "$tname" OUTBOUNDONLY 2>/dev/null | head -1)
          direction="?"
          [ "$isserver" = "1" ] && direction="inbound-tcp"
          [ "$obib" = "1" ] && [ "$direction" = "?" ] && direction="inbound-icl"
          [ "$outonly" = "1" ] && [ "$direction" = "?" ] && direction="outbound"
          emit "$site" "$tname" "$p" "${phost:-—}" "${pname:-?}" "$direction" "$nc" "${line:-?}"
        fi
      done
      ;;

    host)
      "$NCP" list-protocols "$nc" 2>/dev/null | while IFS= read -r tname; do
        h=$("$NCP" protocol-nested "$nc" "$tname" PROTOCOL.HOST 2>/dev/null | head -1 | sed 's/^{}$//')
        if [ "$CASE_SENSITIVE" = "1" ]; then
          [[ "$h" == *"$QUERY"* ]] || continue
        else
          shopt -s nocasematch 2>/dev/null
          [[ "$h" == *"$QUERY"* ]] || { shopt -u nocasematch 2>/dev/null; continue; }
          shopt -u nocasematch 2>/dev/null
        fi
        [ -z "$h" ] && continue
        pname=$("$NCP" protocol-field "$nc" "$tname" PROCESSNAME 2>/dev/null | head -1)
        p=$("$NCP" protocol-nested "$nc" "$tname" PROTOCOL.PORT 2>/dev/null | head -1 | sed 's/^{}$//')
        line=$("$NCP" protocol-line "$nc" "$tname" 2>/dev/null)
        emit "$site" "$tname" "${p:-—}" "$h" "${pname:-?}" "?" "$nc" "${line:-?}"
      done
      ;;

    process)
      "$NCP" list-protocols "$nc" 2>/dev/null | while IFS= read -r tname; do
        pname=$("$NCP" protocol-field "$nc" "$tname" PROCESSNAME 2>/dev/null | head -1)
        if [ "$CASE_SENSITIVE" = "1" ]; then
          [[ "$pname" == *"$QUERY"* ]] || continue
        else
          shopt -s nocasematch 2>/dev/null
          [[ "$pname" == *"$QUERY"* ]] || { shopt -u nocasematch 2>/dev/null; continue; }
          shopt -u nocasematch 2>/dev/null
        fi
        p=$("$NCP" protocol-nested "$nc" "$tname" PROTOCOL.PORT 2>/dev/null | head -1 | sed 's/^{}$//')
        h=$("$NCP" protocol-nested "$nc" "$tname" PROTOCOL.HOST 2>/dev/null | head -1 | sed 's/^{}$//')
        line=$("$NCP" protocol-line "$nc" "$tname" 2>/dev/null)
        emit "$site" "$tname" "${p:-—}" "${h:-—}" "$pname" "?" "$nc" "${line:-?}"
      done
      ;;

    xlate|tclproc)
      # threads that reference a given xlate or tclproc
      need_pattern="$QUERY"
      "$NCP" list-protocols "$nc" 2>/dev/null | while IFS= read -r tname; do
        if [ "$MODE" = "xlate" ]; then
          hits=$("$NCP" xlate-refs "$nc" "$tname" 2>/dev/null | grep -F -- "$need_pattern" || true)
        else
          hits=$("$NCP" tclproc-refs "$nc" "$tname" 2>/dev/null | grep -F -- "$need_pattern" || true)
        fi
        [ -z "$hits" ] && continue
        pname=$("$NCP" protocol-field "$nc" "$tname" PROCESSNAME 2>/dev/null | head -1)
        line=$("$NCP" protocol-line "$nc" "$tname" 2>/dev/null)
        emit "$site" "$tname" "—" "—" "${pname:-?}" "?" "$nc" "${line:-?}"
      done
      ;;
  esac
done

# Format output
case "$FORMAT" in
  tsv)
    printf "site\tthread\tport\thost\tprocess\tdirection\tfile\tline\n"
    cat "$RESULTS"
    ;;
  table)
    {
      printf "site\tthread\tport\thost\tprocess\tdirection\tline\n"
      awk -F'\t' '{printf "%s\t%s\t%s\t%s\t%s\t%s\t%s\n", $1, $2, $3, $4, $5, $6, $8}' "$RESULTS"
    } | awk -F'\t' '
      { for (i=1;i<=NF;i++){ if (length($i)>w[i]) w[i]=length($i); cell[NR,i]=$i }; rows=NR; cols=NF }
      END {
        for (r=1; r<=rows; r++) {
          for (c=1; c<=cols; c++) printf "%-*s  ", w[c], cell[r,c]
          printf "\n"
          if (r==1) { for (c=1; c<=cols; c++) for (k=0;k<w[c];k++) printf "-"; for (c=1; c<=cols; c++) printf "  "; printf "\n" }
        }
      }'
    ;;
  jsonl)
    awk -F'\t' '
      function esc(s) { gsub(/\\/, "\\\\", s); gsub(/"/, "\\\"", s); return s }
      { printf "{\"site\":\"%s\",\"thread\":\"%s\",\"port\":\"%s\",\"host\":\"%s\",\"process\":\"%s\",\"direction\":\"%s\",\"file\":\"%s\",\"line\":\"%s\"}\n",
          esc($1),esc($2),esc($3),esc($4),esc($5),esc($6),esc($7),esc($8) }
    ' "$RESULTS"
    ;;
esac

# Emit count to stderr
# v0.7.5: tr -cd '0-9' instead of tr -d ' ' — Cygwin wc.exe CR-taint defense
# (printf '%d' "5\r" fails with "invalid number").
n=$(wc -l < "$RESULTS" | tr -cd '0-9')
[ "$FORMAT" = "table" ] && printf '\n%d match(es)\n' "${n:-0}" >&2