netgescon-master/scripts/01_estrai_e_hash.sh

#!/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 e log.
# Autore: Pikappa2 – 2025-05-27

set -e

CONFIG="$HOME/netgescon/agent_config.json"

# Controllo jq
if ! command -v jq &> /dev/null; then
    echo "[ERRORE] Il programma 'jq' non è installato. Installa con: sudo apt-get install jq"
    exit 1
fi

# Lettura parametri da config
IN=$(jq -r '.InputDirectory' "$CONFIG")
OUT=$(jq -r '.OutputDirectory' "$CONFIG")
HASH="$OUT/hash"
LOG="$HOME/netgescon/log/estrazione.log"

mkdir -p "$IN" "$OUT" "$HASH" "$(dirname "$LOG")"

# Funzione di log avanzato
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"
}

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

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

    echo "DEBUG: [extract_tables] MDB=$MDB, MDBNAME=$MDBNAME, DESTDIR=$DESTDIR"
    logmsg "DEBUG" "Entra in extract_tables con MDB=$MDB, MDBNAME=$MDBNAME, DESTDIR=$DESTDIR"

    mkdir -p "$DESTDIR"
    if ! mdb-tables -1 "$MDB" > /dev/null 2>&1; then
        logmsg "ERR" "Impossibile leggere tabelle da $MDB"
        echo "DEBUG: [extract_tables] ERRORE su mdb-tables $MDB"
        return 1
    fi

    mdb-tables -1 "$MDB" | while read TBL; do
        [ -z "$TBL" ] && continue
        ((TBL_FOUND++))
        local CSV="$DESTDIR/$TBL.csv"
        local HASHFILE="$HASH/$MDBNAME.$TBL.csv.md5"
        TMP_CSV="$(mktemp)"
        echo "DEBUG: [extract_tables] Estrai tabella $TBL da $MDB" | tee -a "$LOG"
        if ! mdb-export "$MDB" "$TBL" > "$TMP_CSV" 2>/dev/null; then
            logmsg "WARN" "Errore nell'estrazione della tabella $TBL da $MDBNAME"
            rm -f "$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"
                logmsg "DEBUG" "Tabella $TBL ($MDBNAME) invariata, saltata."
                continue
            fi
        fi
        cp "$TMP_CSV" "$CSV"
        echo "$HASHVAL" > "$HASHFILE"
        ((TBL_EXTRACTED++))
        ((TBL_EXTR++))
        CHANGED=1
        logmsg "INFO" "Estratta tabella $TBL da $MDBNAME (hash $HASHVAL)"
        rm -f "$TMP_CSV"
    done
    if [ $TBL_EXTR -gt 0 ]; then
        logmsg "REPORT" "MDB $MDBNAME: $TBL_EXTR tabelle estratte/aggiornate, $TBL_SKIP skippate."
    fi
    return $CHANGED
}

# --- FUNZIONE PER COPIA E HASH 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")"
    HASHVAL=$(md5sum "$SRC" | cut -d' ' -f1)
    if [ -f "$DST" ] && [ -f "$HASHFILE" ]; then
        OLDHASH=$(cat "$HASHFILE")
        if [ "$OLDHASH" = "$HASHVAL" ]; then
            ((PDF_SKIPPED++))
            logmsg "DEBUG" "PDF $REL invariato, saltato."
            return
        fi
    fi
    cp "$SRC" "$DST"
    echo "$HASHVAL" > "$HASHFILE"
    ((PDF_COPIED++))
    logmsg "INFO" "PDF copiato/aggiornato: $REL (hash $HASHVAL)"
}

# --- MDB: ESTRAZIONE RICORSIVA ---
find "$IN" -type f -iname '*.mdb' | while IFS= read -r MDB; do
    ((MDB_FOUND++))
    RELPATH="${MDB#$IN/}"
    MDBNAME=$(basename "$MDB" .mdb)
    DESTDIR="$OUT/$(dirname "$RELPATH")/$MDBNAME"
    echo "DEBUG: [main loop] Trovato MDB: $MDB RELPATH=$RELPATH MDBNAME=$MDBNAME DESTDIR=$DESTDIR"
    logmsg "DEBUG" "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"
        echo "DEBUG: [main loop] ERRORE su extract_tables $MDB"
    fi
done

# --- PDF: RICERCA E COPIA RICORSIVA ---
find "$IN" -type f -iname '*.pdf' | while IFS= read -r PDF; do
    ((PDF_FOUND++))
    RELPATH="${PDF#$IN/}"
    copy_and_hash_pdf "$PDF" "$RELPATH"
done

# --- 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 ===="