Poll a Bulk API job until it finishes, with a hard deadline. Each status check is already HTTP-timeout-bounded; this bounds the overall wait so a job

Build S3 key for ContentVersion file using checksum (content-addressable storage) All files with same checksum share same S3 location for true dedupli

Count archived records for an object type (uses pre-computed stats)

Count deleted records for an object type

Create a new archive run in the database with negative period and archive expression.

Delete all active backups (period > 0) older than cutoffPeriod from the DB. Archives (period < 0) are untouched. The huge objects table is deleted in

Download query results from Bulk API without processing to storage. This method only downloads and concatenates CSV chunks, extracting checkpoint info

Download one batch of records using pipeline optimization. If a preparedQuery is provided, uses that instead of creating a new one. After downloading

Fetch a record from the sandbox by ID. Returns field map, or null if not found.

Get the current archive period as negative YYMM (e.g. -2602 for February 2026).

Determine MIME type based on file extension

Get list of objects that have archived records (uses pre-computed stats)

Get all unique records for a specific object type (for browsing)

Get reference field names for an object (cached). Uses describe metadata to determine which fields are lookups.

Load all unique ContentVersion checksums from database for cache preloading. Returns set of "checksum.extension" keys for fast deduplication. @return

Mark a file as completed (successful download or skip) Updates the sliding window and triggers checkpoint if needed @param recordId ContentVersion Id

Marks the job as complete and sets the end time.

Updates backup statistics based on the result of a backup run. This enables adaptive optimization over time. @param backup The Heimdall_Backup_Config

Upload a CSV file to S3 with partitioned path pattern: OrgId={x}/Period={yyyymm}/Year={y}/Month={m}/Date={d}/{ObjectName}/{file}.csv @param csvPath P

Adjust batch size based on how long the batch took. - If timeout/error: halve the batch size - If fast (<30s) and full batch: increase by 50% - Otherw

Concatenate CSV file by appending rows from source to destination (skipping header)

Create a Bulk Query using a raw SOQL string. Uses "query" operation (not queryAll).

Download all ContentVersion files from CSV and upload to S3 in parallel. Uses 10 concurrent threads for high-throughput processing. Only downloads fil

Escape special characters for JSON

Formats bytes into human-readable format (B, KB, MB, GB, TB).

Get archived records for a specific object type

Compute what a retention cleanup would remove for periods older than cutoffPeriod. Counts CSV files (one per S3 object) from the small csvfiles table

Build the field list for SOQL queries, with SystemModstamp and Id last.

Return the S3 keys (one per CSV file) to delete for periods older than cutoffPeriod. Same filter as the preview; used by execute to drive the S3 delet

Insert batch of objects into the objects table