Figure number reset for figures in different headings

[pwolfram]

Figures numbers are reset under different headings, e.g., for a file tester.md,

# head 1                             

![Fig1](){#fig:fig1}

blah blah Fig @fig:fig1.

![Fig2](){#fig:fig2}

blah blah Fig @fig:fig2.

# head 2

![Fig3](){#fig:fig3}

blah blah Fig @fig:fig3.

built by pandoc -o tester.docx -f markdown -t docx tester.md --filter=pandoc-xnos; open tester.docx
produces output like

Is this expected behavior?

Version info:

pandoc --version
pandoc 2.2.3.2
Compiled with pandoc-types 1.17.5.1, texmath 0.11.0.1, skylighting 0.7.2
Default user data directory: /Users/pwolfram/.pandoc
Copyright (C) 2006-2018 John MacFarlane
Web:  http://pandoc.org
This is free software; see the source for copying conditions.
There is no warranty, not even for merchantability or fitness
for a particular purpose.

pip install  pandoc-xnos
Requirement already satisfied: pandoc-xnos in /Users/pwolfram/miniconda3/lib/python3.7/site-packages (2.3.0)
Requirement already satisfied: psutil<6,>=4.1.0 in /Users/pwolfram/miniconda3/lib/python3.7/site-packages (from pandoc-xnos) (5.6.3)
Requirement already satisfied: pandocfilters<2,>=1.4.2 in /Users/pwolfram/miniconda3/lib/python3.7/site-packages (from pandoc-xnos) (1.4.2)

[p3palazzo]

Happens only in the docx/odt writers for me. PDF is correctly numbered.

[pwolfram]

Thanks @p3palazzo, I also didn't check latex or html compatibility either because my use case is docx for this application.

Minimally, I would expect design behavior to be consistent between docx/odt/pdf/latex/html/etc unless I've missed something in terms of design criteria.

Advice on how to dig in is appreciated. My work-around will be to just hard code figure numbers for now.

[jschlatow]

@pwolfram @p3palazzo The reason why PDF is correctly numbered is that LaTeX/pdf does the numbering itself whereas for other output formats, pandoc-fignos must do hard-code the figure numbers.

I had a quick look at the code and believe the reason for this is that the figure number is reset in every section irrespective of whether fignos-number-by-section is set or not. A patch for this might be:

diff --git a/pandoc_fignos.py b/pandoc_fignos.py
index da1039a..9d21696 100644
--- a/pandoc_fignos.py
+++ b/pandoc_fignos.py
@@ -83,7 +83,7 @@ warninglevel = 2        # 0 - no warnings; 1 - some warnings; 2 - all warnings
 
 # Processing state variables
 cursec = None  # Current section
-Ntargets = 0   # Number of targets in current section (or document)
+Ntargets = 1   # Number of targets in current section (or document)
 targets = {}   # Global targets tracker
 
 # Processing flags
@@ -172,7 +172,8 @@ def _process_figure(key, value, fmt):
     # Update the current section number
     if attrs['secno'] != cursec:  # The section number changed
         cursec = attrs['secno']   # Update the global section tracker
-        Ntargets = 1              # Resets the global target counter
+        if numbersections:
+            Ntargets = 1              # Resets the global target counter
 
     # Pandoc's --number-sections supports section numbering latex/pdf, html,
     # epub, and docx

By the way, this will also affect the other pandoc-*nos filters.

[p3palazzo]

Thank you @ValiValpas. The patch works in all common use cases (PDF, DOCX, ODT, HTML). When setting fignos-number-by-section: True in DOCX output, the sections themselves are not numbered (a problem with Pandoc's section numbering? using pandoc's --number-sections, of course), but the figures are correctly numbered as 1.1, 1.2, 2.1, etc.

[pwolfram]

What is the culture here for PRs/patches? Don't know what my responsibility as a contributor is in this venue but can help on this later next week.

Thanks everyone for digging in and for the solution!

[uthpalaherath]

This happens when converting markdown to HTML. Works fine for PDF.

[p3palazzo]

What is the culture here for PRs/patches? Don't know what my responsibility as a contributor is in this venue but can help on this later next week.

@ValiValpas should be the one making the pull request because he came up with the solution, otherwise we'd be taking credit for his work.

[Dabsen]

Thanks! But what is the status of this fix? @ValiValpas, did you actually make the pull request? I can't see it.

[jschlatow]

@Dabsen, I referenced this issue in my commit so that my commit is shown in this issue. A pull request is not necessary, it would just create another issue that needs to be referenced and synced with this one.

[debug-ito]

I'm having this issue. A workaround (by the user) is to add "secno=1" attribute to every figures. E.g.

![Figure One](fig_one.png){#fig:one secno=1}

[tomduck]

Thank you for this report and discussion, which were helpful. Sorry for the long wait.

The problem is now fixed in the nextrelease branch. If problems persist beyond the next release, please re-open this Issue.

Tom

[tomduck]

pandoc-fignos 2.3.0 is now released. --Tom