archivdms/internal/audit/audit.go

// Package audit is a PostgreSQL-backed, append-only audit log, ported 1:1
// from archivmail's internal/audit pattern (including the DB-level
// immutability trigger and the tamper-evident JSON-Lines mirror file), with
// the mail-specific fields (mail_id, query) replaced by a generic document_id
// so the log fits archivdms's document-centric core model.
package audit

import (
	"context"
	"encoding/json"
	"fmt"
	"log/slog"
	"os"
	"path/filepath"
	"strings"
	"sync"
	"time"

	"github.com/jackc/pgx/v5/pgxpool"
)

// Event type constants.
const (
	EventLogin      = "login"
	EventLogout     = "logout"
	EventUserMgmt   = "user_mgmt"
	EventTenantMgmt = "tenant_mgmt"

	// EventDocumentCreate/Update/Delete cover document lifecycle changes.
	EventDocumentCreate = "document_create"
	EventDocumentUpdate = "document_update"
	EventDocumentDelete = "document_delete"

	// EventDocumentReprocessed records a re-processing run on an already-archived
	// document (POST /api/documents/{id}/reprocess): OCR is re-run on the stored
	// WORM file and ocr_text refreshed, followed by best-effort auto-assignment
	// and on_upload workflows. The file itself stays untouched (WORM).
	EventDocumentReprocessed = "document_reprocessed"

	// EventDocumentProcessed records the asynchronous post-upload processing
	// run of the tenant job queue (internal/jobqueue -> api.ProcessDocumentJob):
	// OCR extraction, title/belegdatum derivation, taxonomy auto-assignment and
	// on_upload workflows on an already-staged, already-archived document.
	// Logged on success AND failure (retry/backoff attempts included) so the
	// GoBD trail covers the whole ingest chain, not just the synchronous
	// staging step (EventDocumentCreate). The WORM file is never touched.
	EventDocumentProcessed = "document_processed"

	// Manual taxonomy assignment on a single document (doc_type/correspondent).
	// Auto-assignment during upload is covered by EventDocumentCreate.
	EventDocTypeSet       = "document_doctype_set"
	EventCorrespondentSet = "document_correspondent_set"

	// Trash / staged deletion workflow (Papierkorb + Vier-Augen-Prinzip).
	// Each phase produces its own append-only entry; confirm/execute emit
	// separate entries for requester (User A) and confirmer (User B).
	EventDocumentTrash         = "document_trash"                    // soft-delete into trash
	EventDocumentRestore       = "document_restore"                  // restored from trash
	EventDocumentDeleteRequest = "document_delete_request"           // final-deletion requested (User A)
	EventDocumentDeleteConfirm = "document_delete_confirm"           // final-deletion confirmed (User B)
	EventDocumentDeleteExecute = "document_delete_execute"           // WORM file removed, tombstone kept
	EventDocumentDeleteBlocked = "document_delete_blocked_retention" // blocked by retain_until

	// Reminder ("Wiedervorlage") events.
	EventReminderCreate       = "reminder_create"
	EventReminderStatusChange = "reminder_status_change"
	EventReminderDelete       = "reminder_delete"
	EventReminderNotify       = "reminder_notify" // cron: due-date notification sent

	// SFTP credential lifecycle + login events (internal/sftpserver,
	// internal/api/sftp_handlers.go).
	EventSFTPCredentialCreate = "sftp_credential_create"
	EventSFTPCredentialRevoke = "sftp_credential_revoke"
	EventSFTPLogin            = "sftp_login" // logged on every attempt, success and failure

	// Permission model (group-resolved document ACL, internal/storage/permissions.go,
	// internal/api/permission_handlers.go). Covers group/member changes and all
	// three grant layers (document-type / tag / per-document, incl. 'deny').
	EventPermissionGrantChanged = "permission_grant_changed"

	// External document share-links (internal/storage/shares.go,
	// internal/api/share_handlers.go + public_share_handlers.go). Create/Revoke
	// are authenticated tenant actions; Accessed is logged on the public
	// download endpoint (success and failure — see also the per-attempt
	// document_share_accesses table for the full public-access trail).
	EventShareCreated  = "share_created"
	EventShareRevoked  = "share_revoked"
	EventShareAccessed = "share_accessed"

	// Signed, time-limited storage download URLs (internal/objectstore,
	// internal/api/signed_url_handlers.go). Created is logged when a link is
	// issued, Accessed on every attempt to redeem one (success and failure).
	EventSignedURLCreated  = "signed_url_created"
	EventSignedURLAccessed = "signed_url_accessed"

	// LDAP directory integration (internal/ldapstore, internal/ldapauth,
	// internal/api/ldap_handlers.go). ConfigChanged covers create/update/delete
	// and the test action; LoginSuccess/Failed are logged on every LDAP bind
	// attempt (incl. JIT provisioning); RoleSync records a role change applied by
	// group membership re-synchronisation (including downgrades).
	EventLdapConfigChanged = "ldap_config_changed"
	EventLdapLoginSuccess  = "ldap_login_success"
	EventLdapLoginFailed   = "ldap_login_failed"
	EventLdapRoleSync      = "ldap_role_sync"

	// EventOllamaConfigUpdate records a change to a tenant's external-Ollama
	// connection config (GET/PUT /api/ollama-config). Logged on every attempt,
	// success and failure (GoBD-Nachvollziehbarkeit).
	EventOllamaConfigUpdate = "ollama_config_update"

	// Classification templates (Klassifizierungsvorlagen,
	// internal/storage/classification_templates.go,
	// internal/api/classification_template_handlers.go). Create/Update/Delete
	// cover template administration (incl. tag / field-default bulk replace);
	// Applied records applying a template to a document — logged on success and
	// failure, and also when a retain_until shortening attempt was rejected
	// (RetainUntilBlocked) so GoBD traceability shows the rejection explicitly.
	EventTemplateCreate  = "classification_template_create"
	EventTemplateUpdate  = "classification_template_update"
	EventTemplateDelete  = "classification_template_delete"
	EventTemplateApplied = "classification_template_applied"

	// Workflows / Consumption-Regeln (internal/storage/workflows.go,
	// internal/api/workflow_handlers.go). Create/Update/Delete cover workflow
	// administration (incl. action bulk replace and dry-run test). EventWorkflowRun
	// records a manual (test-endpoint) or automatic (on_upload) evaluation. Note
	// automatic runs are additionally recorded document-scoped in the
	// workflow_runs table for GoBD reproducibility.
	EventWorkflowCreate = "workflow_create"
	EventWorkflowUpdate = "workflow_update"
	EventWorkflowDelete = "workflow_delete"
	EventWorkflowRun    = "workflow_run"

	// Heuristische Metadaten-Vorschläge (internal/storage/metadata_suggestions.go,
	// internal/api/metadata_suggestion_handlers.go). Records a (non-binding)
	// suggestion run for a document; accepting a suggested field goes through the
	// normal edit endpoints, not through this event.
	EventSuggestionGenerated = "metadata_suggestion_generated"

	// Freitext-Notizen pro Dokument (internal/storage/document_notes.go,
	// internal/api/document_note_handlers.go). Create/Delete cover the note
	// lifecycle; pure reads (listing notes) are not audited, consistent with the
	// rest of this project. Notes are hard-deleted (not GoBD documents), but the
	// deletion itself is still recorded for Nachvollziehbarkeit.
	EventNoteCreate = "document_note_create"
	EventNoteDelete = "document_note_delete"

	// EventMLRetrain records a Naive-Bayes classifier retraining run
	// (internal/classifier, cmd/archivdms/cmd_classify_retrain.go, cron-driven).
	// Logged once per tenant per retrain, success and failure — a per-tenant
	// failure is isolated and does not block the other tenants. Detail carries
	// the per-kind document counts / skip reasons for GoBD-Nachvollziehbarkeit.
	EventMLRetrain = "ml_classifier_retrain"

	// Gespeicherte Suchansichten (SavedViews, Paperless-ngx inspiriert —
	// internal/storage/saved_views.go, internal/api/saved_view_handlers.go).
	// Create/Update/Delete cover a user's named, reusable search/filter view.
	// Views can be private (own) or shared tenant-wide (is_shared); only the
	// creator may update or delete a view. Pure reads (listing views) are not
	// audited, consistent with the rest of this project.
	EventSavedViewCreate = "saved_view_create"
	EventSavedViewUpdate = "saved_view_update"
	EventSavedViewDelete = "saved_view_delete"

	// EventRetentionApplied records a batch run of the GoBD retention-rules
	// engine (internal/storage/retention_rules.go ApplyRetentionRules,
	// cmd/archivdms/cmd_retention_apply.go, cron-driven). One summary entry per
	// run per tenant (tenant + count of documents whose retain_until was
	// computed and set), NOT per document — batch-summary style to avoid audit
	// log spam. Logged on success and failure.
	EventRetentionApplied = "retention_applied"

	// EventRetentionRuleCreate/Update/Delete record CRUD changes to GoBD
	// retention rules (internal/storage/retention_rules.go,
	// internal/api/retention_rule_handlers.go). Compliance-critical: changing a
	// retention period alters how long documents must be kept, so every mutation
	// — success and failure — is audit-logged.
	EventRetentionRuleCreate = "retention_rule_create"
	EventRetentionRuleUpdate = "retention_rule_update"
	EventRetentionRuleDelete = "retention_rule_delete"

	// Digitale Akten (digitaler Aktenordner — internal/storage/akten.go,
	// internal/api/akte_handlers.go). Create/Update/Close/Delete cover the akte
	// lifecycle; DocumentAdd/DocumentRemove record assigning/removing a document
	// to/from an akte (PUT /api/documents/{id}/akte). An akte has no own ACL —
	// its visibility derives from the documents it contains (see
	// project_akte_konzept_plan.md).
	EventAkteCreate         = "akte_create"
	EventAkteUpdate         = "akte_update"
	EventAkteClose          = "akte_close"
	EventAkteDelete         = "akte_delete"
	EventAkteDocumentAdd    = "akte_document_add"
	EventAkteDocumentRemove = "akte_document_remove"

	// EventDocumentSplit records a barcode-separator-page split at ingest
	// (internal/pagesplit, internal/api/document_handlers.go storeUploadedFile).
	// GoBD-Nachvollziehbarkeit: the uploaded multi-page original is NOT archived
	// as such — it is replaced by N part documents — so the split itself is the
	// only record tying the parts back to the original upload. Detail therefore
	// carries the original filename, its SHA-256, the page count, the separator
	// page numbers and the resulting document IDs. Logged on success and on
	// failure (Success:false when the split was attempted but aborted, in which
	// case the original is archived unsplit).
	EventDocumentSplit = "document_split"

	// EventDocumentExport records a single-document export
	// (GET /api/documents/{id}/export): a ZIP containing the original WORM file,
	// metadata.json and, if present, ocr_text.txt. Read-only, but archived
	// content plus its full metadata leaves the system as a package, so — like
	// EventComplianceExport / EventAccountingPull — it is logged like a mutation,
	// including failures (Success:false).
	EventDocumentExport = "document_export"

	// EventDocumentBulkExport records a multi-document export
	// (POST /api/documents/export): one ZIP with a doc-<id>/ folder per
	// document plus index.csv. Exactly ONE entry per request (not per
	// document) — Detail carries "exported=<n> skipped=<m>", where skipped
	// counts documents the caller could not see or that failed to read (also
	// listed in the archive's errors.txt). Logged on failure too
	// (Success:false), including rejected requests (too many IDs, invalid
	// filter) and aborted streams.
	EventDocumentBulkExport = "document_bulk_export"
)

// Compliance-Export (internal/api/compliance_handlers.go).
const (
	// EventComplianceExport records a generated GoBD-Verfahrensdokumentation
	// draft. Read-only, but exported tenant configuration leaves the system, so
	// it is logged like a mutation (including failures) — and for a superadmin
	// cross-tenant export the Detail carries the target tenant_id.
	EventComplianceExport = "compliance_procedure_doc_export"
)

// Buchhaltungs-Pull-API (internal/api/accounting_handlers.go,
// internal/storage/accounting_api_keys.go).
const (
	// EventAccountingKeyCreated/Revoked record the lifecycle of a per-tenant
	// accounting API key. The plaintext key is NEVER part of Detail — only the
	// key id and its label.
	EventAccountingKeyCreated = "accounting_key_created"
	EventAccountingKeyRevoked = "accounting_key_revoked"

	// EventAccountingPull records machine-to-machine reads through an
	// accounting API key. Read-only, but archived content leaves the system via
	// a non-browser path, so it is logged like a mutation (including failures):
	// every file download individually, and every list query with its result
	// range (first/last document id).
	EventAccountingPull = "accounting_pull"
)

// Entry is a single audit log record.
type Entry struct {
	ID         int64     `json:"id"`
	Timestamp  time.Time `json:"timestamp"`
	EventType  string    `json:"event_type"`
	Username   string    `json:"username"`
	IPAddress  string    `json:"ip_address"`
	DocumentID string    `json:"document_id"`
	Success    bool      `json:"success"`
	Detail     string    `json:"detail"`
	// TenantID, when set, records which tenant this event belongs to. nil means
	// a tenant-less / system-wide event (e.g. superadmin actions, cron jobs).
	TenantID *int64 `json:"tenant_id,omitempty"`
}

// QueryFilter specifies filtering options for audit log queries.
type QueryFilter struct {
	Username   string
	EventType  string
	DocumentID string
	From       *time.Time
	To         *time.Time
	TenantID   *int64
	PageSize   int
	Page       int
}

// Logger is a PostgreSQL-backed, append-only audit log mirrored to a
// tamper-evident JSON-Lines file opened in append-only mode.
type Logger struct {
	pool   *pgxpool.Pool
	logger *slog.Logger

	fileMu  sync.Mutex
	file    *os.File
	logPath string
}

type fileEntry struct {
	Timestamp  string `json:"timestamp"`
	EventType  string `json:"event_type"`
	Username   string `json:"username"`
	IPAddress  string `json:"ip_address"`
	DocumentID string `json:"document_id,omitempty"`
	Success    bool   `json:"success"`
	Detail     string `json:"detail,omitempty"`
	TenantID   *int64 `json:"tenant_id,omitempty"`
}

// New connects to PostgreSQL using the given DSN and initialises the schema.
func New(dsn, logPath string, logger *slog.Logger) (*Logger, error) {
	ctx := context.Background()
	pool, err := pgxpool.New(ctx, dsn)
	if err != nil {
		return nil, fmt.Errorf("audit: connect: %w", err)
	}
	if err := initSchema(ctx, pool); err != nil {
		pool.Close()
		return nil, fmt.Errorf("audit: create schema: %w", err)
	}
	l := &Logger{pool: pool, logger: logger, logPath: logPath}
	l.openLogFile()
	return l, nil
}

// initSchema creates the audit_log table and installs the immutability
// trigger. Both operations are idempotent and safe on existing databases.
func initSchema(ctx context.Context, pool *pgxpool.Pool) error {
	if _, err := pool.Exec(ctx, `
		CREATE TABLE IF NOT EXISTS audit_log (
			id          BIGSERIAL    PRIMARY KEY,
			timestamp   TIMESTAMPTZ  NOT NULL DEFAULT NOW(),
			event_type  VARCHAR(50)  NOT NULL,
			username    VARCHAR(255) NOT NULL DEFAULT '',
			ip_address  VARCHAR(45)  NOT NULL DEFAULT '',
			document_id VARCHAR(64)  NOT NULL DEFAULT '',
			success     BOOLEAN      NOT NULL DEFAULT true,
			detail      TEXT         NOT NULL DEFAULT '',
			tenant_id   BIGINT
		);
	`); err != nil {
		return err
	}

	// Append-only enforcement at the database level: any UPDATE/DELETE raises.
	if _, err := pool.Exec(ctx, `
		CREATE OR REPLACE FUNCTION audit_log_no_mutation()
		RETURNS trigger AS $$
		BEGIN
			RAISE EXCEPTION 'audit_log is append-only: % is not permitted', TG_OP
				USING ERRCODE = 'integrity_constraint_violation';
		END;
		$$ LANGUAGE plpgsql;
	`); err != nil {
		return fmt.Errorf("create trigger function: %w", err)
	}
	if _, err := pool.Exec(ctx, `DROP TRIGGER IF EXISTS audit_log_immutable ON audit_log;`); err != nil {
		return fmt.Errorf("drop trigger: %w", err)
	}
	if _, err := pool.Exec(ctx, `
		CREATE TRIGGER audit_log_immutable
			BEFORE UPDATE OR DELETE ON audit_log
			FOR EACH ROW EXECUTE FUNCTION audit_log_no_mutation();
	`); err != nil {
		return fmt.Errorf("create trigger: %w", err)
	}
	return nil
}

func (l *Logger) openLogFile() {
	if l.logPath == "" {
		l.logger.Warn("audit: log_path not configured, file logging disabled")
		return
	}
	if dir := filepath.Dir(l.logPath); dir != "" && dir != "." {
		_ = os.MkdirAll(dir, 0o750)
	}
	f, err := os.OpenFile(l.logPath, os.O_CREATE|os.O_WRONLY|os.O_APPEND, 0o640)
	if err != nil {
		l.logger.Warn("audit: audit log file not writable, continuing with DB-only logging",
			"path", l.logPath, "err", err)
		return
	}
	l.file = f
}

// Log appends an entry to the audit log. Errors are logged but not returned.
func (l *Logger) Log(entry Entry) {
	ts := entry.Timestamp
	if ts.IsZero() {
		ts = time.Now().UTC()
	}
	ctx := context.Background()
	_, err := l.pool.Exec(ctx,
		`INSERT INTO audit_log (timestamp, event_type, username, ip_address, document_id, success, detail, tenant_id)
		 VALUES ($1, $2, $3, $4, $5, $6, $7, $8)`,
		ts.UTC(), entry.EventType, entry.Username, entry.IPAddress, entry.DocumentID,
		entry.Success, entry.Detail, entry.TenantID,
	)
	if err != nil {
		l.logger.Error("audit: insert failed", "err", err)
	}
	l.writeFile(entry, ts.UTC())
}

func (l *Logger) writeFile(entry Entry, ts time.Time) {
	l.fileMu.Lock()
	defer l.fileMu.Unlock()
	if l.file == nil {
		return
	}
	line, err := json.Marshal(fileEntry{
		Timestamp:  ts.Format(time.RFC3339),
		EventType:  entry.EventType,
		Username:   entry.Username,
		IPAddress:  entry.IPAddress,
		DocumentID: entry.DocumentID,
		Success:    entry.Success,
		Detail:     entry.Detail,
		TenantID:   entry.TenantID,
	})
	if err != nil {
		l.logger.Error("audit: marshal log line failed", "err", err)
		return
	}
	if _, err := l.file.Write(append(line, '\n')); err != nil {
		l.logger.Error("audit: write to log file failed", "path", l.logPath, "err", err)
	}
}

// Query retrieves audit entries matching the given filter.
func (l *Logger) Query(filter QueryFilter) ([]Entry, int, error) {
	pageSize := filter.PageSize
	if pageSize <= 0 {
		pageSize = 50
	}
	where, args := buildWhere(filter)
	ctx := context.Background()

	countSQL := "SELECT COUNT(*) FROM audit_log" + where
	var total int
	if err := l.pool.QueryRow(ctx, countSQL, args...).Scan(&total); err != nil {
		return nil, 0, fmt.Errorf("audit: count: %w", err)
	}

	offset := filter.Page * pageSize
	limitArg := len(args) + 1
	offsetArg := len(args) + 2
	querySQL := fmt.Sprintf(
		"SELECT id, timestamp, event_type, username, ip_address, document_id, success, detail FROM audit_log%s ORDER BY timestamp DESC LIMIT $%d OFFSET $%d",
		where, limitArg, offsetArg,
	)
	allArgs := append(args, pageSize, offset)

	rows, err := l.pool.Query(ctx, querySQL, allArgs...)
	if err != nil {
		return nil, 0, fmt.Errorf("audit: query: %w", err)
	}
	defer rows.Close()

	var entries []Entry
	for rows.Next() {
		var e Entry
		if err := rows.Scan(&e.ID, &e.Timestamp, &e.EventType, &e.Username, &e.IPAddress, &e.DocumentID, &e.Success, &e.Detail); err != nil {
			return nil, 0, fmt.Errorf("audit: scan: %w", err)
		}
		entries = append(entries, e)
	}
	return entries, total, rows.Err()
}

// Close closes the audit log file and the connection pool.
func (l *Logger) Close() error {
	l.fileMu.Lock()
	if l.file != nil {
		_ = l.file.Sync()
		_ = l.file.Close()
		l.file = nil
	}
	l.fileMu.Unlock()
	l.pool.Close()
	return nil
}

func buildWhere(f QueryFilter) (string, []interface{}) {
	var clauses []string
	var args []interface{}
	n := 1
	if f.Username != "" {
		clauses = append(clauses, fmt.Sprintf("username = $%d", n))
		args = append(args, f.Username)
		n++
	}
	if f.EventType != "" {
		clauses = append(clauses, fmt.Sprintf("event_type = $%d", n))
		args = append(args, f.EventType)
		n++
	}
	if f.DocumentID != "" {
		clauses = append(clauses, fmt.Sprintf("document_id = $%d", n))
		args = append(args, f.DocumentID)
		n++
	}
	if f.From != nil {
		clauses = append(clauses, fmt.Sprintf("timestamp >= $%d", n))
		args = append(args, f.From.UTC())
		n++
	}
	if f.To != nil {
		clauses = append(clauses, fmt.Sprintf("timestamp <= $%d", n))
		args = append(args, f.To.UTC())
		n++
	}
	if f.TenantID != nil {
		clauses = append(clauses, fmt.Sprintf("tenant_id = $%d", n))
		args = append(args, *f.TenantID)
		n++
	}
	if len(clauses) == 0 {
		return "", args
	}
	return " WHERE " + strings.Join(clauses, " AND "), args
}