Project

Profile

Help

Should PyDocumentBuilder have exception_occurred/error_message properties to report parse errors?

Added by Martin Honnen about 1 year ago

I am wondering how to handle errors occurring during XML parsing with PyDocumentBuilder in my code, it seems the document builder itself has no properties like exception_occurred or error_message, those on the PySaxonProcessor don't seem to be set if a parse error occurs, only the standard err output gets an error message, at least as far as I can tell from the following code:

from saxonche import *

with PySaxonProcessor(license=True) as proc:
    print(proc.version)

    proc.exception_clear()
    doc_builder = proc.new_document_builder()

    xdm_node = doc_builder.parse_xml(xml_text='<root><foo></root>')

    if proc.exception_occurred:
        print('Error parsing XML: ' + proc.error_message)
        proc.exception_clear()
        exit(1)
    else:
        print(xdm_node)

Output:

SaxonC-HE 12.0 from Saxonica
Error on line 1 column 14 
  SXXP0003   Error reported by XML parser: The element type "foo" must be terminated by the
  matching end-tag "</foo>".: The element type "foo" must be terminated by the matching
  end-tag "</foo>".

Process finished with exit code -1073741819 (0xC0000005)

Shouldn't my Python code be able to handle/read out the parse error other than it being written to standard err by the C++ code?


Replies (3)

Please register to reply

RE: Should PyDocumentBuilder have exception_occurred/error_message properties to report parse errors? - Added by O'Neil Delpratt about 1 year ago

The exception handling methods are available in the C++ code, missing from the Python PyDocumentBuilder class. I will file a bug issue for this.

RE: Should PyDocumentBuilder have exception_occurred/error_message properties to report parse errors? - Added by O'Neil Delpratt about 1 year ago

The issue with the the example code is the PySaxonProcessor needs proc.set_cwd('file://example'). This workaround still does not look correct, but for a well-formed XML text it will work. I have patched the problems for this bug.

    (1-3/3)

    Please register to reply