Archives: ZIP, gzip and tar¶
Since 1.4. Suppliers often send a products.zip or a stock.csv.gz; OpenImport unpacks it and reads the file
inside, and can pack an export before it is delivered.
Reading an archive¶
Every source (upload, server folder, web address, SFTP, FTP, S3, cloud drives) may deliver an archive. Step 3 of the profile form has two settings:
| Setting | Meaning |
|---|---|
| Archives (ZIP, gzip, tar) | Unpack archives (default): a .zip, .gz, .tar, .tar.gz or .tgz file (by name, else by its first bytes) is unpacked and the file inside is read. Read the file as it is: nothing is unpacked. |
| File inside the archive | Which file to take out of a ZIP or tar: a name or a pattern (products-*.csv, data/*.xml, matched against the file name and its path inside the archive, case-insensitive). Empty: the first CSV, TSV, TXT, Excel, ODS, JSON or XML file by name. Hidden files and macOS __MACOSX folders are ignored. |
A .gz file holds one file, named after the archive (stock.csv.gz → stock.csv); a .tar.gz holds a tar.
The Format setting and the sample in the form apply to the file inside; the run page names it
("Read products.csv from feed.zip.") and the run's copy is that file, not the archive. After success (move or
delete) acts on the archive, as it does on any other source file. Excel (.xlsx) and OpenDocument files are ZIP
files too; they are never taken for archives.
Writing an archive¶
Compress in step 3 of an export profile packs the written file as .zip, .gz or .tar.gz before it is
delivered (download, folder, SFTP, FTP, S3, cloud drive, REST, e-mail). The file name gets the archive's extension
(products-2026-10-21.csv.zip); the archive holds the file under its own name.
Limits and safety¶
- Entry names never become paths: the chosen file is written under a fixed name in the run folder, so an archive
with
../../or absolute names (zip-slip) cannot write anywhere else. Links and devices inside tar files and ZIPs are skipped. - The unpacked size is counted while writing, not taken from the archive's headers: more than 1 GB stops the run with a message (zip bombs); the partial file is removed. An archive with more than 10,000 entries is refused.
- Archives inside archives are not unpacked; password-protected entries are refused; a damaged archive fails the run
with a plain message. The limits are arguments of
Iranimij\OpenImport\Model\Archive\Unpackerindi.xml.