On 21 Jan 2024, at 12:42, Stefan Behnel wrote: Hi Stefan,
If you want to avoid creating element objects all together, maybe even don't need a full (sub-)tree structure to get all relevant information, I suggest you try the low-level SAX interface.
https://lxml.de/parsing.html#the-target-parser-interface
It's quite efficient and usable for locally constrained XML transformations, e.g. filtering elements or attributes.
And you can still parse input chunk by chunk, if you need that:
Yes, I've read about both of those but always shied away from them. The ETree interface really is a join to work with and in most cases all you need but I guess I've come across the two edge-cases where the performance overhead can be considered an issue. On a slightly related note, is there anyway getting the parser to treat some attributes as numbers to avoid casting in Python? Charlie -- Charlie Clark Managing Director Clark Consulting & Research German Office Sengelsweg 34 Düsseldorf D- 40489 Tel: +49-203-3925-0390 Mobile: +49-178-782-6226