Question about speed of findAllElements

[bubnenkoff]

I need to find only first matched element. Now I am doing it:

final title =  document.findAllElements('title', namespace: '*').first.text;

But my XML docs are very big, and I suppose that findAllElements are doing full docs scanning. I have heard that some parsers can read only parts of docs, Can dart-xml do that?

[renggli]

Yes, check out the event-driven API: https://github.com/renggli/dart-xml#event-driven

Note: findAllElements is not slow. What is not efficient is to read gigabytes of text and to build a DOM with millions of objects that are not used in the end. With the event-driven API you can avoid that and incrementally scan through the input only building and keeping the parts of the tree in memory that you actually need.

[bubnenkoff]

@renggli thanks!
Could you show an example how to use stream parsing to extract title value? And how to use namespaces *?

parseEvents(bookshelfXml)
    .whereType<XmlTextEvent>()
    .map((event) => event.text.trim()) // what should be here? stmth like: event.tag == 'title'
    .where((text) => text.isNotEmpty)
    .forEach(print);

[bubnenkoff]

  final file = File(file_name);
  await file.openRead()
  .transform(utf8.decoder)
  .toXmlEvents()
  .selectSubtreeEvents((event) => event == 'ns6:title')
  .toXmlNodes()
  .forEach((node) => print(node));

This code is from example.
The first problem that node do not have method innerText
The second problem -- it's do not clear how to ignore nasmaspeces. Could you help me please?

I tried to do:

  final file = File(file_name);
  await file.openRead()
  .transform(utf8.decoder)
  .toXmlEvents()
  .selectSubtreeEvents((event) => event.name == 'title' && event.namespacePrefix == '*' )
  .toXmlNodes()
  .forEach((node) => print(node));

But it does not find tag value

[renggli]

Regarding #113 (comment): This is not using a Stream, but a lazy Iterable. You need to keep track of the events and the nesting yourself. Typical strategy would be to use Iterable. skipWhile until the XmlStartElementEvent with the name title and then Iterable.takeWhile until the corresponding XmlEndElementEvent. From that sub-iterable you then extract the XmlTextEvent as in the example above.

Regarding #113 (comment): You seem to miss the import of the XML library, the innerText method is an extension method defined in the base package. For the matching remove the condition on the namespace prefix, you are comparing string fields. See the code and class documentation for other accessors.

[bubnenkoff]

And analyzer show that node is list, not single node:

So only:
.forEach((node) => print(node[0].innerText));
works

[bubnenkoff]

It's seems that this code are work:

    final file = File(file_name);
    await file.openRead()
    .transform(utf8.decoder)
    .toXmlEvents()
    .selectSubtreeEvents((event) => event.localName == 'docPublishDate')
    .toXmlNodes()
    .forEach((node) => print(node[0].innerText));

Only question is node[0]. It's error in docs? But by idea node should not have any index.

[bubnenkoff]

In result I wrote next code.
The task sound like: to take any datetime from document:

  Future<String?> extractValue(List<String> variantList) async  {
    String? result;
      for(var el in variantList) {  

          final file = File(file_name);
          await file.openRead()
          .transform(utf8.decoder)
          .toXmlEvents()
          .selectSubtreeEvents((event) => variantList.contains(event.localName))
          .toXmlNodes()
          .forEach((node) {
             result = node[0].innerText;
          } );

          return result;

      }


    return result; 
  }

  print(await extractValue(["docPublishDate", "createDate", "plannedPublishDate", "docDT", "protocolDate", "publishDTInEIS", "publishDate"]));

It this code is ok? It's working. But maybe I should add any interruption/break? I need to find only first match. I do not need full scan of document.

[renggli]

You are right, before the .forEach there is a .expand((nodes) => nodes) missing in the tutorial.

[renggli]

For the rest have a look at Dart tutorials on Streams, finding the first element of a stream is a standard operation unrelated to this library.