#!/bin/bash
# Script: 01_estrai_e_hash_debug.sh
# Estrae i MDB dalla directory configurata in agent_config.json (ricorsivo!), salva CSV e hash, salta i file già estratti se non modificati.
# Debug avanzato con echo/log, progress bar e contatori precisi, errori sempre tracciati mai bloccanti.
# Autore: Pikappa2 – 2025-05-28

set -u

# --- Funzione log avanzata ---
logmsg() {
  # $1 = livello (INFO, WARN, ERR, REPORT, DEBUG)
  # $2 = messaggio
  local LEVEL="$1"
  local MSG="$2"
  local TS
  TS=$(date "+%Y-%m-%d %H:%M:%S")
  echo "[$TS][$LEVEL] $MSG" | tee -a "$LOG"
}

# --- Funzione debug condizionale ---
debugmsg() {
  # $1 = messaggio
  if [[ "$DEBUG" == "true" ]]; then
    logmsg "DEBUG" "$1"
  fi
}

# --- Lettura configurazione ---
CONFIG="$HOME/netgescon/agent_config.json"

if ! command -v jq &>/dev/null; then
  echo "[ERRORE] jq non trovato! Installa con: sudo apt-get install jq"
  exit 1
fi

if [[ ! -f "$CONFIG" ]]; then
  echo "[ERRORE] File di configurazione $CONFIG non trovato!"
  exit 1
fi

IN=$(jq -r '.InputDirectory // empty' "$CONFIG")
OUT=$(jq -r '.OutputDirectory // empty' "$CONFIG")
DEBUG=$(jq -r '.Debug // "false"' "$CONFIG" | tr '[:upper:]' '[:lower:]')

if [[ -z "$IN" || -z "$OUT" ]]; then
  echo "[ERRORE] InputDirectory o OutputDirectory non valorizzati nel file di configurazione!"
  exit 1
fi

HASH="$OUT/hash"
LOG="$HOME/netgescon/log/estrazione.log"

mkdir -p "$OUT" "$HASH" "$(dirname "$LOG")" 2>/dev/null || {
  echo "[ERRORE] Impossibile creare cartelle di output."
  exit 1
}

logmsg "INFO" "==== AVVIO ESTRAZIONE DATI ===="
logmsg "INFO" "Input:   $IN"
logmsg "INFO" "Output:  $OUT"
logmsg "INFO" "Hash:    $HASH"
logmsg "INFO" "Log:     $LOG"
logmsg "INFO" "Debug:   $DEBUG"

MDB_FOUND=0
MDB_PROCESSED=0
TBL_EXTRACTED=0
TBL_SKIPPED=0
PDF_FOUND=0
PDF_COPIED=0
PDF_SKIPPED=0
ERROR_MDB=0

# --- Funzione estrazione tabelle MDB ---
extract_tables() {
  local MDB="$1"
  local MDBNAME="$2"
  local DESTDIR="$3"
  local TBL_FOUND=0
  local TBL_EXTR=0
  local TBL_SKIP=0
  local CHANGED=0

  debugmsg "Entra in extract_tables con MDB=$MDB, MDBNAME=$MDBNAME, DESTDIR=$DESTDIR"
  mkdir -p "$DESTDIR" 2>/dev/null || {
    logmsg "ERR" "Impossibile creare cartella $DESTDIR"
    return 1
  }

  local TBL_LIST
  TBL_LIST=$(mdb-tables -1 "$MDB" 2>&1)
  if [[ $? -ne 0 || -z "$TBL_LIST" ]]; then
    logmsg "ERR" "Impossibile leggere tabelle da $MDB: $TBL_LIST"
    return 1
  fi

  while read -r TBL; do
    [[ -z "$TBL" ]] && continue
    ((TBL_FOUND++))
    local CSV="$DESTDIR/$TBL.csv"
    local HASHFILE="$HASH/$MDBNAME.$TBL.csv.md5"
    TMP_CSV="$(mktemp)"
    debugmsg "Estrazione tabella $TBL da $MDB"
    local EXPORT_OUT
    EXPORT_OUT=$(mdb-export "$MDB" "$TBL" 2>&1 > "$TMP_CSV")
    if [[ $? -ne 0 ]]; then
      logmsg "WARN" "Errore nell'estrazione della tabella $TBL da $MDBNAME: $EXPORT_OUT"
      rm -f "$TMP_CSV"
      continue
    fi
    HASHVAL=$(md5sum "$TMP_CSV" | cut -d' ' -f1)
    if [[ -f "$CSV" && -f "$HASHFILE" ]]; then
      OLDHASH=$(cat "$HASHFILE")
      if [[ "$OLDHASH" == "$HASHVAL" ]]; then
        ((TBL_SKIPPED++))
        ((TBL_SKIP++))
        rm -f "$TMP_CSV"
        debugmsg "Tabella $TBL ($MDBNAME) invariata, saltata."
        continue
      fi
    fi
    cp "$TMP_CSV" "$CSV" 2>/dev/null || logmsg "WARN" "Impossibile copiare $TMP_CSV -> $CSV"
    echo "$HASHVAL" > "$HASHFILE"
    ((TBL_EXTRACTED++))
    ((TBL_EXTR++))
    CHANGED=1
    logmsg "INFO" "Estratta tabella $TBL da $MDBNAME (hash $HASHVAL)"
    rm -f "$TMP_CSV"
  done <<< "$TBL_LIST"

  if [[ $TBL_EXTR -gt 0 ]]; then
    logmsg "REPORT" "MDB $MDBNAME: $TBL_EXTR tabelle estratte/aggiornate, $TBL_SKIP skippate."
  fi
  return $CHANGED
}

# --- Funzione PDF ---
copy_and_hash_pdf() {
  local SRC="$1"
  local REL="$2"
  local DST="$OUT/$REL"
  local HASHFILE="$HASH/$(echo "$REL" | sed 's/\//_/g').pdf.md5"
  mkdir -p "$(dirname "$DST")" 2>/dev/null
  HASHVAL=$(md5sum "$SRC" | cut -d' ' -f1)
  if [[ -f "$DST" && -f "$HASHFILE" ]]; then
    OLDHASH=$(cat "$HASHFILE")
    if [[ "$OLDHASH" == "$HASHVAL" ]]; then
      ((PDF_SKIPPED++))
      debugmsg "PDF $REL invariato, saltato."
      return
    fi
  fi
  cp "$SRC" "$DST" 2>/dev/null || logmsg "WARN" "Impossibile copiare $SRC -> $DST"
  echo "$HASHVAL" > "$HASHFILE"
  ((PDF_COPIED++))
  logmsg "INFO" "PDF copiato/aggiornato: $REL (hash $HASHVAL)"
}

# --- MDB: ESTRAZIONE RICORSIVA CON PROGRESS BAR ---
debugmsg "Inizio ricerca file MDB..."
MDB_LIST=()
while IFS= read -r line; do MDB_LIST+=("$line"); done < <(find "$IN" -type f -iname '*.mdb')
TOTAL_MDB=${#MDB_LIST[@]}
CURRENT_MDB=0

for MDB in "${MDB_LIST[@]}"; do
  ((CURRENT_MDB++))
  # Barra di avanzamento testuale
  BAR_WIDTH=30
  PERCENT=$((CURRENT_MDB * 100 / TOTAL_MDB))
  BAR_DONE=$((CURRENT_MDB * BAR_WIDTH / TOTAL_MDB))
  BAR_LEFT=$((BAR_WIDTH - BAR_DONE))
  BAR="["
  for ((i=0;i<BAR_DONE;i++)); do BAR+="#"; done
  for ((i=0;i<BAR_LEFT;i++)); do BAR+=" "; done
  BAR+="]"
  echo -ne "$BAR $PERCENT% ($CURRENT_MDB/$TOTAL_MDB) MDB\r"

  ((MDB_FOUND++))
  RELPATH="${MDB#$IN/}"
  MDBNAME=$(basename "$MDB" .mdb)
  DESTDIR="$OUT/$(dirname "$RELPATH")/$MDBNAME"
  debugmsg "Trovato MDB: $MDB RELPATH=$RELPATH MDBNAME=$MDBNAME DESTDIR=$DESTDIR"
  if extract_tables "$MDB" "$MDBNAME" "$DESTDIR"; then
    ((MDB_PROCESSED++))
  else
    ((ERROR_MDB++))
    logmsg "ERR" "File MDB problematico: $MDB"
    debugmsg "Errore su extract_tables $MDB"
  fi
done
echo # newline dopo barra

# --- PDF: RICERCA E COPIA RICORSIVA CON PROGRESS BAR ---
debugmsg "Inizio ricerca file PDF..."
PDF_LIST=()
while IFS= read -r line; do PDF_LIST+=("$line"); done < <(find "$IN" -type f -iname '*.pdf')
TOTAL_PDF=${#PDF_LIST[@]}
CURRENT_PDF=0

for PDF in "${PDF_LIST[@]}"; do
  ((CURRENT_PDF++))
  BAR_WIDTH=30
  PERCENT=$((CURRENT_PDF * 100 / TOTAL_PDF))
  BAR_DONE=$((CURRENT_PDF * BAR_WIDTH / TOTAL_PDF))
  BAR_LEFT=$((BAR_WIDTH - BAR_DONE))
  BAR="["
  for ((i=0;i<BAR_DONE;i++)); do BAR+="#"; done
  for ((i=0;i<BAR_LEFT;i++)); do BAR+=" "; done
  BAR+="]"
  echo -ne "$BAR $PERCENT% ($CURRENT_PDF/$TOTAL_PDF) PDF\r"

  ((PDF_FOUND++))
  RELPATH="${PDF#$IN/}"
  copy_and_hash_pdf "$PDF" "$RELPATH"
done
echo # newline dopo barra

# --- REPORT FINALE ---
logmsg "REPORT" "==== REPORT ESTRAZIONE ===="
logmsg "REPORT" "MDB trovati: $MDB_FOUND"
logmsg "REPORT" "MDB processati: $MDB_PROCESSED"
logmsg "REPORT" "MDB con errori: $ERROR_MDB"
logmsg "REPORT" "Tabelle estratte/aggiornate: $TBL_EXTRACTED"
logmsg "REPORT" "Tabelle skippate: $TBL_SKIPPED"
logmsg "REPORT" "PDF trovati: $PDF_FOUND"
logmsg "REPORT" "PDF copiati/aggiornati: $PDF_COPIED"
logmsg "REPORT" "PDF skippati: $PDF_SKIPPED"
if [[ $TBL_EXTRACTED -eq 0 && $PDF_COPIED -eq 0 ]]; then
  logmsg "REPORT" "Nessuna tabella estratta/aggiornata e nessun PDF copiato (tutto invariato)."
fi
logmsg "INFO" "==== FINE ESTRAZIONE ===="