Discrepancy with genhtml

[letmaik]

When I generate an HTML report with genhtml I get 5623/7013 lines covered in total, which is 80%.

Looking at the xml file produced by this library the root entry has lines-covered="10233" lines-valid="23111", which is 44%.

How can such a situation occur?

[Natsuki-gh]

I have found a discrepancy where the xml and genhtml have a difference about paths taken.
With genhtml all lines in my function were executed however in the lcov-to-cobertura-xml not all lines were so marked.
looking at the info file there are multiple SF records for the same file:
SF:/code/example.c
...
DA:100,1
DA:101,1
end_of_record
SF:/code/example.c
...
DA:100,0
DA:101,0
end_of_record
The python script line clears file_lines every SF record. This then replaces file_dict['lines'] at line 120 on end_of_record.
i.e. the information is from the last SF record and the information from the previous SF records is lost.
I suspect the processing of end_of_record needs to integrate the new data with the existing.