Record export to CSV #
Record export to CSV prepares a file from selected records. It is a different function from exporting the whole Database as a ZIP package.
CSV is useful when you want to review data in a spreadsheet, pass a selected record list for further work, or prepare data for a report or publication.
What can be exported #
In User data reports, select one Dataset or explicitly choose All available Datasets. The report includes your own records within that scope, not other members' entire collections. Search further limits the exported rows. A full SQLite backup still contains the entire private database; the new offline Viewer also requires a Dataset selection before record export.
Record export can include a single record, records selected on the list, or records matched by the current search filters.
Record set and CSV export #
This workflow can also be connected with the idea of a record set. First you select or filter the records that form a working set, then you export that set to a CSV file.
When to use CSV #
Use record export to CSV when you need a table for analysis, checking, correction or sharing a selected part of the data.
If you want to keep a full package of a Database, use Database export to ZIP. ZIP is intended for broader archiving, while CSV is intended for working with a concrete record list.
Import exported records #
The preview lists source Datasets with record counts and one target Dataset. You must explicitly confirm adding all records as new records to that target. A file from several Datasets can therefore be combined into one target; original records are not moved or deleted. To preserve separate destinations, export and import each Dataset separately. Unknown source Datasets are marked explicitly.
Open Records → Import CSV records, choose the target Dataset and upload the file. The preview lists the target and records. Import … records as new into: … is the single confirmation, including repeat imports. Check the Dataset name on this button. There are no two additional consent boxes. Import always creates new records; it never updates existing ones. Relationship links are optional and do not block import. Missing life stage or time zone is only requested after reading the file; answers can be supplied without uploading again.
The entomone.records/1 profile is shared by selection export, all filtered results and the Records tab of the current online/offline report. Export includes all matching pages. Files use UTF-8, semicolons and technical column names. Keep the version, metadata and empty-value marker columns. Existing offline Viewer files need a refreshed backup to receive the new format.
The importing account becomes the observer. The original observer and source information remain in private import history, re-exports and the private backup/Viewer. Site, visit and plant matches are suggestions requiring individual confirmation; these entities are not created. Custom fields require matching UUIDs, types, versions and options in the target Dataset.
Photos and active iNaturalist publication links are not transferred. New records are visible by default and unpublished; after import, the owner can make them Hidden. CSV does not replace Dataset ZIP or a full database backup.
Input limits are 10 MiB and 10,000 records. Preview errors block the entire write. Older files may require an explicit life stage or source timezone; ambiguous times are not guessed. Reimporting identical contents requires confirmation. After an interrupted write or an uncertain-outcome message, do not retry under a new filename: an administrator must reconcile the operation. Apply is disabled when distributed writes are unavailable.
Duplicate check #
If matching records have no media or publication, the recommendation keeps the older record with the lowest ID and suggests moving newer records with higher IDs to Trash. Dates and import history do not change this order. Retaining a record with media still takes priority.
After importing, use Duplicate check next to the success message or in the Records menu. It compares your active records in the selected Dataset, not other members' collections. Differences in substantive record data exclude a shared group. Technical identifiers, creation dates, import history and media do not determine equality.
Media counts appear separately. If only one record has media, it is recommended for retention. Import provenance is supporting information, not proof of identical content.
Select copies To Trash, use Apply recommendation in this group, or Apply safe recommendations - all groups. Bulk recommendations cover the entire selected Dataset and skip ambiguous groups. Manual selection applies to the current page. Unselected records remain; at least one must remain in each group. Records with media, publication, or awaiting publication are protected in this tool.
Review the copy IDs and retained records in the summary, then confirm the move. Each operation is limited to 100 records. The server rechecks permissions and data; changed records stop the operation. A batch is not one transaction: the result reports how many records were already moved if subsequent moves failed. Trash supports restoring records. This tool does not permanently delete or merge data, or remove media.
The scan is limited to 20,000 records in the private database. Exceeding the limit or encountering an error displays an incomplete-check message, never a false “no duplicates” result.