Search translation memory

Overview

• Runs an asynchronous TMX export based on wildcard/phrase queries and optional metadata filters. The job streams filtered translation units into TMX via the downstream /export_tmx_by_query endpoint. Use downloadExport/{asyncRequestId} to fetch the finished file.

Flow

1. Caller posts ExportByQueryDto.
2. Service queues backend export (job metadata marks it asynchronous).
3. Response contains asyncRequest + asyncExport descriptors so you can poll or receive callbacks.

Queries & locales

• queries align one-to-one with queryLangs. Each entry represents a search expression (supports * and ? wildcards, max 5/10 respectively) applied to a locale.
• Query semantics mirror the wildcard search endpoint: unquoted text = contains mode; quoted text = phrase mode; Boolean operators are ignored; * or empty string means match-all (the backend may drop redundant source-only pairs).
• exportTargetLangs sets which target TUVs are written to TMX. If omitted, all TM target locales are exported.

Filtering

• Timestamp bounds (createdAtMin/Max, modifiedAtMin/Max) and author filters (createdBy, modifiedBy) narrow the TU set before export.
• Optional project reference provides additional downstream scoping or naming context.

Output & limits

• Output is always TMX, streamed in chunks; filenames follow dto.filename when provided, otherwise backend defaults.
• Requests with excessive locales or wildcard limit violations reject with 400.
• If no matches remain after filtering, the job still produces an empty TMX.

Callbacks & polling

• Provide callbackUrl to receive completion notification from the async mediator; otherwise poll asyncRequest via standard async APIs and download through /downloadExport/{asyncRequestId} when ready.