Convert XML to SQLite

Working from an .xml file

XML describes structure but not types, and nests arbitrarily deep, while a table does neither. The reader takes one specific slice of the document, and it is worth checking that the slice is the one you meant.

• →Only the direct children of the row element become columns. A grandchild such as <address><city> has no text of its own at that level, so it arrives as NULL - flatten one level before uploading.
• →Namespaces are reduced to local names, so <ns:order> and <order> are the same element here, and two elements differing only by prefix collide into one column.
• →Attributes are read first and child elements second, so where a row carries both an id attribute and an <id> child, the child's value is what lands in the column.
• →No XSD or DTD is consulted. Every value goes through the same inference a CSV gets, with the same refusal to guess at an ambiguous date.

What the .sqlite file looks like inside

The output here is a real database rather than a script, so the interesting question is not how to load it but what SQLite did with the types on the way in. Its storage classes are looser than the declarations the file ends up carrying.

• →Names are carried over exactly as they were, with none of the sanitising the script targets apply: a table called Sales Report keeps its space and a column called Order ID keeps its own, so your queries have to quote them.
• →Underneath the declared types sit only five storage classes, and NUMERIC affinity - what a decimal column gets - is the loosest of them: it normalises inserted text to whatever a plain INTEGER or REAL column would hold. The precision and scale in the schema are for a human reading it, not a guarantee SQLite enforces on the bytes; the Writing notes below cover what that means for money specifically.
• →Declared type and stored type are different things here: a BOOLEAN column holds the integers 1 and 0, and a DATE column holds ISO text, because those are the storage classes SQLite actually has. Every client reads them back the same way.

Reading XML

• →The row element is whichever child repeats — no configuration needed.
• →Attributes become columns alongside child elements, so nothing on the row itself is left behind. What a table cannot hold is depth: only the row element and its immediate children are read, so nested structures need flattening by one level first.
• →Empty and self-closing elements become NULL.

Writing SQLite

• →Columns get real declared types rather than everything being TEXT.
• →Primary keys and NOT NULL constraints are preserved.
• →A money column is declared NUMERIC(10,2), but that declaration does not stop SQLite’s own numeric affinity from converting the inserted value into a float once it lands, so 19.99 can come back with a rounding tail. Pick the DuckDB target instead when the exact decimal has to survive.

Convert XML to something else

• MySQL
• PostgreSQL
• Oracle
• SQL Server
• DuckDB
• CSV
• Excel

Convert something else to SQLite

• CSV
• Excel
• SQL dump
• DBF
• JSON
• Parquet
• DuckDB
• MS Access
• PostgreSQL backup