DefaultHome returns the default msgvault home directory. Respects MSGVAULT_HOME environment variable and expands ~ in its value.

DeleteAllDeduped permanently deletes every dedup-hidden row regardless of batch. Returns the number of rows deleted and the number of distinct batches

DeleteCollection drops the collection. Sources and messages untouched. Rejects deletion of the auto-managed default collection.

DeleteDedupedBatch permanently deletes all hidden rows associated with a dedup batch. Only deletes rows where deleted_at IS NOT NULL AND delete_batch_

DeleteToken revokes the refresh token at Microsoft (best-effort) and removes the local token file. Revocation failures are logged but do not prevent l

DriverName returns the driver name to pass to sql.Open.

FormatSize formats bytes as a human-readable string.

GetAccountSchedule returns the schedule for a specific account email. Returns nil if the account is not configured for scheduling. The returned value

GetMessage retrieves a full message from SQLite. Uses direct SQLite connection when available for better BLOB handling.

GetMessageBody returns the body_text and body_html for a message.

GetMessageSummariesByIDs returns summary-level rows (no body, no raw MIME) for the supplied IDs in the same order as ids. Missing IDs are silently dro

GetProfile returns the authenticated user's profile.

GetTextStats returns aggregate stats for text messages.

IngestRawMessage parses and stores a raw MIME message into the database. This is the shared ingestion path for all importers (mbox, emlx, etc.). Para

InspectParticipantDisplayName returns the display_name from the participants table for an email.

IsBusyError returns true if the error indicates the database is held by another connection, either busy (SQLITE_BUSY) or locked (SQLITE_LOCKED). Used

IsEmpty reports whether the filter has no restrictions. A zero-value Filter is empty and backends should skip filter resolution entirely.

ListInProgress returns all in-progress deletion manifests.

MarkMessageDeleted marks a message as deleted from the source.

MarkMessagesDeletedBatch marks multiple messages as deleted from the source in a single transaction.

MarkMigrationApplied records that a migration has run. Idempotent.

NewClient creates a new Gmail API client.

NewRateLimiter creates a new rate limiter. rps is requests per second, burst is the maximum burst size.

NewReaderWithMaxMessageBytes creates a new MBOX reader that rejects messages larger than maxMessageBytes. If maxMessageBytes <= 0, no limit is enforce

ReclaimStale clears the claim on any pending row whose claimed_at is older than olderThan. Returns the number of rows reclaimed.

SearchByDomains returns messages where any participant (from, to, cc, or bcc) belongs to one of the given domains.

SetupMessages adds multiple pre-built RawMessage values to the mock store in a thread-safe manner. Nil entries in the input slice are silently skipped

Shutdown gracefully shuts down the server.

SimulateTrashError injects a trash error for a specific message ID.

StripDomain returns the local part of a synthetic "<slug>@facebook.messenger" email. If the input lacks the expected domain suffix, it is returned as-

TextSearch performs plain full-text search over text messages. Uses FTS5 if available; otherwise returns empty results.

TimerCount returns the number of pending timers.

ToJSON serializes the config to JSON.

TruncateRunes truncates a string to maxRunes runes (not bytes), adding "..." if truncated. This is UTF-8 safe and won't split multi-byte characters.

UpdateSourceSyncConfig updates the JSON sync configuration for an IMAP source. The sync_config column is JSONB on PG; the dialect supplies the appropr

ValidateCronExpr validates a cron expression without scheduling anything.

WithLogger sets the logger for the client.

WithProgress sets the progress reporter.

WithReason adds an entry to the errors[].reason array.

WithSummary sets the manifest summary.

WriteAndVerifyFile writes content to a file, asserts it exists, and verifies its content matches. Returns the full path to the written file.

aggRowMap builds a map from key to count, failing on duplicate keys.

appleTimestampToTime converts an Apple epoch timestamp to time.Time. macOS High Sierra+ stores dates as nanoseconds since Apple epoch; older versions

assertMessageNotHasLabel checks that a message does NOT have a specific label.

assertRow finds a single key in the aggregate rows and asserts its count is 1.

assertRowsContain verifies that a subset of expected key/count pairs exist in the aggregate rows (order-independent). Also checks for duplicate keys.

assertTempDirSecured checks that a temp dir has permissions no more permissive than 0700. This is umask-tolerant (stricter is fine).

batchPopulate batch-loads recipients and labels for a slice of messages.

buildRecipientSet deduplicates addresses and returns a RecipientSet ready for store.PersistMessage.

keyColumns are passed through to buildAggregateSearchConditions to control which columns text search terms filter on.

confirmEmbed reads a y/N answer from the command input. Default is no.

connect establishes and authenticates the IMAP connection. Caller must hold mu.

--------------------------------------------------------------------------- Shared helpers -----------------------------------------------------------

drain sets tokens to zero.

ensureAddressUTF8 validates and converts address names to valid UTF-8 in place.

executeQuery opens an in-memory DuckDB, registers views over the Parquet files in analyticsDir, runs the SQL, and writes the results in the requested

exitInlineSearchMode resets inline search UI state without changing filter state.

extractBaseSemver extracts the base semver from a version string.

extractValue extracts the value part from a vCard line, handling both "KEY:value" and "KEY;params:value" formats.

formatAddresses formats a slice of addresses as a comma-separated string.

formatCount formats a number with thousand separators.

getIDArg extracts a required positive integer ID from the arguments map.

hasAttr checks whether attr is in the attrs list.

idsToJSON encodes an int64 slice as a JSON array for the json_each pattern used in the fused CTE. Returns a NULL NullString when the slice is empty, s

imapScopeForTenant returns the correct IMAP scope based on the tenant ID. The consumer tenant has a fixed, well-known ID; all others are org tenants.

inClause returns "col IN (?,?,?)" for len(ids) placeholders. Caller must append the ids to the args slice in the same order.

indexNames returns the names of every index on the embeddings table in the connection's own (per-test) schema. Used by the V3 and V5 tests to assert w

inferTimeGranularity adjusts the granularity based on the time period string length.

initSyncState initializes sync state, resuming from checkpoint if possible.

isExchangeDN reports whether s looks like an Exchange Distinguished Name.

isOwnerOnly returns true if the permission mode grants nothing to group or other.

keyLeft returns a KeyMsg for the Left arrow key.

keyRight returns a KeyMsg for the Right arrow key.

loadTextConversations fetches text conversations matching the current filter.

makeRealStreamtypedBlob builds a blob matching the actual macOS Sequoia format where the multi-byte length has extra framing bytes (0x81, 0x92, 0x00)

makeRows creates 10 AggregateRows with sequential keys ("row-0", "row-1", ...).

mapConversation maps a WhatsApp chat to a msgvault conversation. Returns the source_conversation_id, conversation_type, and title.