malformed XML generated from malformed marc

[jakub-id]

We've received records that contain subfield delimiters (1F) in the subfield contents. While this is wrong according to MARC spec, yaz-macdump will process the record without error. While the generated line and JSON representations are valid, XML will be malformed:

yaz-marcdump -f marc8 -o marcxml malformed.mrc | xmllint --noout -
-:148: parser error : Unescaped '<' not allowed in attributes values
    <subfield code="树上又有什么让人惊讶的景象呢</subfield>
                                                              ^
-:148: parser error : attributes construct error
    <subfield code="树上又有什么让人惊讶的景象呢</subfield>
                                                              ^
-:148: parser error : Couldn't find end of Start Tag subfield line 148
    <subfield code="树上又有什么让人惊讶的景象呢</subfield>
                                                              ^
-:148: parser error : Opening and ending tag mismatch: datafield line 145 and subfield
    <subfield code="树上又有什么让人惊讶的景象呢</subfield>
                                                                         ^
-:150: parser error : Opening and ending tag mismatch: record line 2 and datafield
  </datafield>
              ^
-:200: parser error : Opening and ending tag mismatch: collection line 1 and record
</record>
         ^
-:201: parser error : Extra content at the end of the document
</collection>
^

Expected result: yaz-marcdump handles the records but warns and generates well-formed XML.

malformed.mrc.gz