#!/usr/bin/env bash
# nc-document.sh — generate a v3 native markdown knowledge entry for a Cloverleaf
# subsystem identified by a name pattern. Walks every NetConfig under $HCIROOT
# (or a passed-in list), gathers config + flow + xlates + tclprocs, composes a
# markdown doc with placeholder context sections for humans to fill.
#
# Usage:
#   nc-document.sh --name <pattern> [options]
#
#   --name PATTERN       case-insensitive substring/regex to match protocol names
#   --hciroot DIR        defaults to $HCIROOT
#   --netconfigs PATHS   colon-separated explicit NetConfig list (overrides --hciroot scan)
#   --out PATH           output markdown path (default: stdout)
#   --title TITLE        doc title (default: derived from --name)
#   --poc-vendor TXT     Vendor POC content
#   --poc-internal TXT   Internal Owner content
#   --status TXT         e.g. production / test / decommissioning
#   --escalation TXT     Escalation path text
#   --open-items TXT     Open items text (bulleted by you if multi-line)
#   --notes TXT          freeform additional notes
#
# Any --poc/-status/--escalation/--open-items/--notes that you OMIT becomes an
# empty placeholder section in the doc, ready for someone to fill.
set -u
set -o pipefail

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

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

PATTERN=""
HCIROOT_OVERRIDE=""
NETCONFIGS_OVERRIDE=""
OUT=""
TITLE=""
POC_VENDOR=""
POC_INTERNAL=""
STATUS=""
ESCALATION=""
OPEN_ITEMS=""
NOTES=""

while [ $# -gt 0 ]; do
  case "$1" in
    --name)         shift; PATTERN="$1" ;;
    --hciroot)      shift; HCIROOT_OVERRIDE="$1" ;;
    --netconfigs)   shift; NETCONFIGS_OVERRIDE="$1" ;;
    --out)          shift; OUT="$1" ;;
    --title)        shift; TITLE="$1" ;;
    --poc-vendor)   shift; POC_VENDOR="$1" ;;
    --poc-internal) shift; POC_INTERNAL="$1" ;;
    --status)       shift; STATUS="$1" ;;
    --escalation)   shift; ESCALATION="$1" ;;
    --open-items)   shift; OPEN_ITEMS="$1" ;;
    --notes)        shift; NOTES="$1" ;;
    -h|--help)      sed -n '2,25p' "$NC_SELF"; exit 0 ;;
    -*)             die "unknown flag: $1" ;;
    *)              die "extra arg: $1" ;;
  esac
  shift
done

[ -n "$PATTERN" ] || die "missing --name PATTERN"
[ -z "$TITLE" ] && TITLE="$(printf '%s' "$PATTERN" | tr '[:upper:]' '[:lower:]')"

# Determine the NetConfig list
NCONFIGS=()
if [ -n "$NETCONFIGS_OVERRIDE" ]; then
  IFS=':' read -ra NCONFIGS <<< "$NETCONFIGS_OVERRIDE"
else
  ROOT="${HCIROOT_OVERRIDE:-${HCIROOT:-}}"
  [ -n "$ROOT" ] || die "no \$HCIROOT and no --hciroot; pass one or set the env var"
  [ -d "$ROOT" ] || die "hciroot not a directory: $ROOT"
  while IFS= read -r nc; do
    NCONFIGS+=("$nc")
  done < <(find "$ROOT" -maxdepth 2 -name NetConfig -type f 2>/dev/null)
fi
[ ${#NCONFIGS[@]} -gt 0 ] || die "no NetConfig files found"

# Emit to OUT or stdout
out_target() {
  if [ -n "$OUT" ]; then
    mkdir -p "$(dirname "$OUT")" 2>/dev/null
    cat > "$OUT"
  else
    cat
  fi
}

# Gather all matching protocols across all NetConfigs
declare -a MATCHES
for nc in "${NCONFIGS[@]}"; do
  site=$(basename "$(dirname "$nc")")
  while IFS= read -r prot; do
    [ -z "$prot" ] && continue
    MATCHES+=("$site|$nc|$prot")
  done < <("$NCP" list-protocols "$nc" 2>/dev/null | grep -i -- "$PATTERN" || true)
done

if [ ${#MATCHES[@]} -eq 0 ]; then
  printf 'No protocols matching "%s" found in %d NetConfig(s).\n' "$PATTERN" "${#NCONFIGS[@]}" >&2
  exit 2
fi

# ─────────────────────────────────────────────────────────────────────────────
# Compose markdown
# ─────────────────────────────────────────────────────────────────────────────
{
  printf '# %s — Cloverleaf System Knowledge Entry\n\n' "$TITLE"
  printf '_Auto-generated by Larry-Anywhere v3 nc-document.sh on %s. Auto-derived facts are below; context fields are for humans to fill or refine._\n\n' "$(date -Iseconds 2>/dev/null || date)"

  printf '## Context\n\n'
  printf -- '- **Vendor POC:** %s\n'   "${POC_VENDOR:-_(unfilled — add vendor contact name + email/phone)_}"
  printf -- '- **Internal Owner:** %s\n' "${POC_INTERNAL:-_(unfilled — add the internal owner / engineer)_}"
  printf -- '- **Status:** %s\n'        "${STATUS:-_(unfilled — production / test / decommissioning / on hold)_}"
  printf -- '- **Escalation:** %s\n'    "${ESCALATION:-_(unfilled — on-call path, ticket queue, etc.)_}"
  printf '\n### Open items\n'
  if [ -n "$OPEN_ITEMS" ]; then
    printf '%s\n\n' "$OPEN_ITEMS"
  else
    printf '_(unfilled — add open items / known issues / pending work)_\n\n'
  fi
  printf '### Notes\n'
  if [ -n "$NOTES" ]; then
    printf '%s\n\n' "$NOTES"
  else
    printf '_(unfilled — add any free-form context)_\n\n'
  fi

  # ─── Threads inventory ───
  # v0.7.5: tr -cd '0-9' instead of tr -d ' ' — Cygwin wc.exe CR-taint would
  # otherwise crash `printf '%d'` with "invalid number".
  printf '## Threads (%d matched in %d site(s))\n\n' "${#MATCHES[@]}" "$(printf '%s\n' "${MATCHES[@]}" | awk -F'|' '{print $1}' | sort -u | wc -l | tr -cd '0-9')"
  printf '| Site | Thread | Process | Direction | Port | Host | Type |\n'
  printf '|---|---|---|---|---|---|---|\n'
  for line in "${MATCHES[@]}"; do
    IFS='|' read -r site nc prot <<< "$line"
    pname=$("$NCP" protocol-field "$nc" "$prot" PROCESSNAME 2>/dev/null | head -1)
    obib=$("$NCP" protocol-field "$nc" "$prot" OBWORKASIB 2>/dev/null | head -1)
    outonly=$("$NCP" protocol-field "$nc" "$prot" OUTBOUNDONLY 2>/dev/null | head -1)
    ptype=$("$NCP" protocol-nested "$nc" "$prot" PROTOCOL.TYPE 2>/dev/null | head -1)
    phost=$("$NCP" protocol-nested "$nc" "$prot" PROTOCOL.HOST 2>/dev/null | head -1)
    pport=$("$NCP" protocol-nested "$nc" "$prot" PROTOCOL.PORT 2>/dev/null | head -1)
    isserver=$("$NCP" protocol-nested "$nc" "$prot" PROTOCOL.ISSERVER 2>/dev/null | head -1)

    direction="?"
    [ "$isserver" = "1" ]  && direction="inbound (TCP listener)"
    [ "$obib" = "1" ] && [ "$direction" = "?" ] && direction="inbound (ICL/file)"
    [ "$outonly" = "1" ] && [ "$direction" = "?" ] && direction="outbound"

    phost_clean=$(printf '%s' "$phost" | sed 's/^{}$//')
    pport_clean=$(printf '%s' "$pport" | sed 's/^{}$//')

    printf '| `%s` | `%s` | `%s` | %s | %s | %s | %s |\n' \
      "$site" "$prot" "${pname:-?}" "$direction" "${pport_clean:-—}" "${phost_clean:-—}" "${ptype:-?}"
  done
  printf '\n'

  # ─── Per-thread detail ───
  for line in "${MATCHES[@]}"; do
    IFS='|' read -r site nc prot <<< "$line"
    printf '## `%s` (site: `%s`)\n\n' "$prot" "$site"

    printf '### Sources (what feeds this thread)\n\n'
    sources=$("$NCP" sources "$nc" "$prot" 2>/dev/null)
    if [ -n "$sources" ]; then
      printf '%s\n' "$sources" | awk '{print "- `" $0 "`"}'
    else
      printf '_(none found in `%s`; may be fed via TCP from outside, or from another site via ICL)_\n' "$site"
    fi
    printf '\n'

    printf '### Destinations (where this thread routes to)\n\n'
    dests=$("$NCP" destinations "$nc" "$prot" 2>/dev/null)
    if [ -n "$dests" ]; then
      printf '%s\n' "$dests" | awk '{print "- `" $0 "`"}'
    else
      printf '_(no DEST entries in DATAXLATE block)_\n'
    fi
    printf '\n'

    printf '### Xlates referenced\n\n'
    xlates=$("$NCP" xlate-refs "$nc" "$prot" 2>/dev/null)
    if [ -n "$xlates" ]; then
      printf '%s\n' "$xlates" | awk -v site="$site" '{print "- `" site "/Xlate/" $0 "`"}'
    else
      printf '_(no xlates — pass-through or raw routing only)_\n'
    fi
    printf '\n'

    printf '### TCL procs referenced\n\n'
    tcls=$("$NCP" tclproc-refs "$nc" "$prot" 2>/dev/null)
    if [ -n "$tcls" ]; then
      printf '%s\n' "$tcls" | awk -v site="$site" '{print "- `" site "/tclprocs/" $0 ".tcl`"}'
    else
      printf '_(no TCL procs referenced)_\n'
    fi
    printf '\n'
  done

  # ─── Sources outside the matched set (the "fed by" landscape) ───
  printf '## Adjacent threads (the network this subsystem talks to)\n\n'
  printf '_All threads that **either feed** matched threads **or are fed by** matched threads. These are the immediate operational neighbors._\n\n'
  printf '| Site | Thread | Relationship to matched set |\n'
  printf '|---|---|---|\n'
  declare -A SEEN
  for line in "${MATCHES[@]}"; do
    IFS='|' read -r site nc prot <<< "$line"
    while IFS= read -r src; do
      key="$site|$src"
      [ -n "${SEEN[$key]:-}" ] && continue
      SEEN[$key]=1
      printf '| `%s` | `%s` | feeds `%s` |\n' "$site" "$src" "$prot"
    done < <("$NCP" sources "$nc" "$prot" 2>/dev/null)
    while IFS= read -r dst; do
      key="$site|$dst"
      [ -n "${SEEN[$key]:-}" ] && continue
      SEEN[$key]=1
      printf '| `%s` | `%s` | receives from `%s` |\n' "$site" "$dst" "$prot"
    done < <("$NCP" destinations "$nc" "$prot" 2>/dev/null)
  done
  printf '\n'

  # ─── Footer ───
  printf '---\n\n'
  printf '_Generated: %s · NetConfigs scanned: %d · Pattern: `%s`_\n' \
    "$(date -Iseconds 2>/dev/null || date)" "${#NCONFIGS[@]}" "$PATTERN"
} | out_target

[ -n "$OUT" ] && printf 'nc-document: wrote %s (%d matched threads across %d site(s))\n' \
  "$OUT" "${#MATCHES[@]}" "$(printf '%s\n' "${MATCHES[@]}" | awk -F'|' '{print $1}' | sort -u | wc -l | tr -cd '0-9')" >&2