DataMosaicDocs · Connecting to data
Files
CSV, JSON, XML and Excel — uploaded, on a path, over HTTP, or fetched from FTP or SFTP.
A File source reads a file and turns it into rows. Four formats: CSV, JSON, XML, XLSX.
#Where the file comes from
The From field offers seven transports:
| Transport | What it means |
|---|---|
| Uploaded file | Stored with Mosaic and read on every run |
| Local path | On the machine running Mosaic, read fresh every run |
| Pasted in | The content typed straight into Mosaic |
| HTTP(S) | Fetched from a URL |
| FTP / FTPS / SFTP | Fetched from a server |
Uploaded file is the default. Dropping a file anywhere on the panel uploads it and switches the transport, and the extension sets the format if it says anything useful. Accepted: .csv, .tsv, .txt, .json, .xml, .xlsx. The upload ceiling is 64 MB.
Local path is the one to use for a file a nightly job drops on the server. It is re-read on every run, so the Check follows the file.
FTP, FTPS and SFTP take a host, port, username and password, plus a remote path. SFTP can use a private key with an optional passphrase instead of a password.
The remote path has one feature worth knowing:
A trailing
*takes the newest match —sales_*finds today's export.
That is what makes a dated export usable on a schedule without editing the source each morning.
#Parsing
Only the options the chosen format uses are shown.
| Format | Options |
|---|---|
| CSV | Delimiter (blank detects between ; , tab and |), First row is the header |
| XLSX | First row is the header |
| JSON | Rows path — "Dotted path to the array of rows. Blank means the document is the array." |
| XML | Rows path, typically rows.row |
| All | Max rows, 50000 |
Quoted cells and the doubled-quote escape are handled in CSV.
An XML document with no rows path usually yields nothing, and Mosaic says so: "No rows found. An XML document usually needs a rows path pointing at the repeating element, such as "rows.row"."
#Three limits worth knowing before you plan around them
Excel reads the first sheet. There is no sheet selector. A workbook whose data is on the second tab has to be saved differently, or converted to CSV.
Files are read as UTF-8. There is no encoding option. A Latin-1 export with accented characters will come through mangled — convert it before Mosaic sees it.
There is no timeout on the file panel. A slow FTP fetch is not bounded here.
#Size
The 64 MB ceiling applies to uploads only. A local path, a URL or an FTP fetch has no size limit — but Max rows still truncates, with a warning saying how many rows were left.
For a genuinely large export, prefer a local path over an upload: nothing is copied, and the file can be replaced without touching Mosaic.