Issue Number | 4445 |
---|---|
Summary | [Summaries] Reference numbers for placeholder CitationLink elements |
Created | 2018-03-28 16:01:47 |
Issue Type | Bug |
Submitted By | Juthe, Robin (NIH/NCI) [E] |
Assigned To | Englisch, Volker (NIH/NCI) [C] |
Status | Closed |
Resolved | 2018-05-24 13:27:50 |
Resolution | Fixed |
Path | /home/bkline/backups/jira/ocecdr/issue.223245 |
We often add empty CitationLink elements as placeholders when we're entering proposed summary changes. The reference list in the summary QC reports used to preserve those references in sequence and have a blank line beside the placeholder reference. For example, if the placeholder CitationLink element was located where the 20th ref would occur, then there would be a blank line beside ref 20 in the reference list, and all other references would be numbered accordingly.
We noticed that the "blank" reference is now showing up as ref 1 in the reference list, which is misleading and typically inaccurate (unless the placeholder reference is actually the first reference in the section). We think this is a new problem possibly introduced in Gauss.
I'll add the CDR ID and version number of a doc on PROD where you can observe this issue.
To see an example of this issue, check out V250 of CDR 62911 (Gastric Cancer) on PROD. In the Stage IV/Recurrent section, Immunotherapy section, Pembrolizumab subsection, there are empty brackets in #1.
Do I understand this correctly? The problem is not with the citation within the text. The problem is only the sort order of where that blank "citation" appears. It should be listed as citation #22 instead of #1 and all citations up to #22 are now shifted by one, right?
By the way, this looks like a typo: Fuchs absract
It seems very likely that this change was introduced with
Gauss. I found a work-around one could use until the filters
have been adjusted - though there was no filter change causing this
problem:
When entering an empty CitationLink go to the @cdr:ref attribute and
enter a blank space. This will sort the empty citation properly but only
if you have no more than one empty citation within a section because our
filters are removing duplicates.
Yes, your description is correct.
We often have more than one empty CitationLink, but I'll let Victoria know that this is a potential work-around in the instances where she has just one missing citation. Thanks. Should this go in the release independent queue or does it require a release?
The following filter has been updated to fix the citation sorting issue:
CDR0000335418.xml: Denormalization Filter: Summary InLine Numbering
Please take a look on DEV.
For my information:
Changes are in the local branch cdr4445-citsort.
~juther, were you or ~vshields able to take a look at these changes to the citation sorting?
I have not looked at this yet. ~vshields, maybe we can review this together tomorrow?
~vshields and I both reviewed this today and it looks pretty good on DEV. Victoria noticed something about the placement of the empty reference in a string of references and we weren't sure if it had always been that way of if this is a new "feature". 🙂 We can certainly live with it as is, and it's definitely an improvement, but I thought I'd mention it in case it's a simple fix. Victoria summed this all up much better than I could so I'm going to paste her comments below:
"So it’s better, and certainly we can live with it, but I don’t think it’s working quite like it used to. I used everyone’s favorite test summary, Bladder, for this. On DEV. I added citation links in the Treatment Option Overview and the Stage 0 Treatment section. I added a sentence near each ref so it could be found easier.
Where I added the ref by itself, the report looks good.
Where the ref was added as the last ref in a group of refs, the text is marked in a way that looks like the first ref is the missing ref. But the references list is correct. In the TOO, the first ref I added will be #10 and there’s a blank by #10 in the refs list, but the space in the text is at the beginning of the group, not the end, and looks like #5 is missing.
I’m perfectly happy with this, and it might even have worked this way before."
So basically, the sort order in the reference section is now correct but the order in which the empty reference is displayed within a reference range is still broken. I would have to turn in my German passport if I were to let this go through.
It is definitely thorough testing! I'll take a look on Monday.
The way the citation numbering is done in the filters is something like this:
Collect all the CitationLink elements in the Reference section
Sort all of the citations
De-dup all of the citations - Now we have the correct sort-order in the Ref-Section.
Go back to the text and assign the citation number assigned in the Ref-Section to the citation within the text. Matching the citations is done by matching the CDR-IDs of the citation.
Since the empty CitationLink elements don't include a CDR-ID yet
we're unable to link multiple empty citations to the appropriate in-line
CitationLinks. I have modified the filters to link any empty in-line
CitationLink element within a section to the sequence number of the
first of (possibly) multiple empty CitationLink
elements.
Based on a discussion with ~juther and ~vshields this approach will work because a
situation when two empty citations are added to the same section is not
common.
I think I finished all that's needed in order to display the citations with approved (red) and proposed (green) markup.
The following filters have been updated:
CDR0000339576: Module: InLine Markup Formatter
CDR0000335417: Denormalization Filter: Summary Reference Numbering
CDR0000335418: Denormalization Filter: Summary InLine Numbering
CDR0000335169: Module: STYLE Default
CDR0000321373: Revision Markup Filter for QC Report
CDR0000315892: Clean up Insertion and Deletion
CDR0000000079: Health Professional Summary Report
This is ready for review on DEV.
The modified filters have been copied to QA. I'm currently running before/after diff reports.
There were no surprises running the diff reports for all document
types on QA. All of the changes between the before and
after reports were expected.
~juther, please take a look
and let me know when you'd like me to copy these changes to PROD.
This looks good on QA. You can promote it to PROD. Thanks!
The filter changes have been copied to STAGE and PROD.
https://github.com/NCIOCPL/cdr-server/commit/41d02f3
Verified on PROD - thanks!
Elapsed: 0:00:00.001557