Weird error related to \section and the number of the surrounding paragraphs

[e-kwsm]

This problem is hard to explain...

---

pdfLaTeX, LuaLaTeX, and XeLaTeX fail to compile a code like \section–wrapfigure environment–multiple paragraphs.

MWE is

\documentclass{article}
%\usepackage{wrapfig}
\usepackage{exwrapfig}

\begin{document}
\section{Section}
\begin{wrapfigure}{r}{0.3\textwidth}
\end{wrapfigure}
lorem

ipsum
\end{document}

whose error message is

! Extra \else.
\l_exwf_everypar_tl ...ox \z@ \lastbox }\fi \else 
                                                  \clubpenalty \@clubpenalty...

l.11 i
    psum
?

---

The engines also fails to compile a code like \section–A SINGLE paragraph–wrapfigure environment–one or more paragraphs like

\section{Section}
lorem
\begin{wrapfigure}{r}{0.3\textwidth}
\end{wrapfigure}
ipsum

with another error:

! TeX capacity exceeded, sorry [input stack size=50000].
\everypar ->\clubpenalty \@clubpenalty 
                                       \everypar {}\fi 
l.9 \end{wrapfigure}
                  
 725 words of node memory still in use:
   7 hlist, 1 vlist, 2 rule, 1 disc, 3 local_par, 1 dir, 18 glue, 2 kern, 4 pen
alty, 14 glyph, 13 attribute, 49 glue_spec, 13 attribute_list, 1 temp, 1 if_sta
ck, 2 write nodes
   avail lists: 1:1,2:5,3:1,4:1,5:6,7:1,8:1
!  ==> Fatal error occurred, no output PDF file produced!

---

Compilation failure of wrapfigure inside section depends on the numbers of the paragraphs around the environment.
Here is the summary:

num of above paragraphs	num of below paragraphs	result
0	≤ 1	OK
0	≥ 2	error (! Extra \else)
1	0	OK
1	≥ 1	error (! TeX capacity exceeded)
≥ 2	≥ 0	OK

Compilation succeeds if \section is commented out (\subsection and \subsubsection does not affect).

wrapfig has no error in all of the cases.

[wtsnjp]

Seems to be a bug…

Thanks for reporting. I'll look into this later.

[wtsnjp]

It seems the commit b70266e sucessfully fixed this bug. I hope this works fine with you too.

[e-kwsm]

@wtsnjp Thank you, it works.

[e-kwsm]

Oops, it doesn't.

A problem still occurs if numbers of the above/below paragraphs are 0 and greater than 4, resp.

\documentclass{article}
\usepackage{graphicx}
\usepackage{exwrapfig}

\begin{document}
\section{Section}
\begin{wrapfigure}{r}{0.3\textwidth}
  \includegraphics[height=1ex]{example-image}
\end{wrapfigure}
lorem

ipsum

dolor

sit

amet
\end{document}

gives

.................................................
. exwrapfig info: "putting-here"
. 
. Putting wrapfigure here: on line 10.
.................................................
.................................................
. exwrapfig info: "finish"
. 
. Finish wrapping text on line 15.
.................................................

! Extra \else.
<everypar> ...se {\setbox \z@ \lastbox }\fi \else 
                                                  \clubpenalty \@clubpenalty...
l.18 a
      met
? 
! Extra \fi.
<everypar> ...nalty \@clubpenalty \everypar {}\fi 
                                                  
l.18 a
      met
? 

If I ignore them, a PDF is generated but its indentation is odd:

left: wrapfig, right: exwrapfig

[wtsnjp]

I think the commit 45479b1 prevent the error. However, the output is not fully compatible with the wrapfig package (maybe I have to insert \par token to the right place). I will contiune to work on this.

[e-kwsm]

I think the commit 45479b1 prevent the error.

Yes, it works.

---

FYI, another error occurs if \section, \subsection, or \paragraph follows; e.g.

\documentclass{article}
\usepackage{graphicx}
\usepackage{exwrapfig}

\begin{document}
\section{Section}
lorem

\begin{wrapfigure}{r}{0.3\textwidth}
  \includegraphics[height=1ex]{example-image}
\end{wrapfigure}

ipsum

\section{Section2}

\end{document}

gives

! TeX capacity exceeded, sorry [input stack size=5000].
\everypar ...\fi \else \clubpenalty \@clubpenalty 
                                                  \everypar {}\fi 
l.15 \section{Section2}
                       
!  ==> Fatal error occurred, no output PDF file produced!

I tested several conditions by

#!/usr/bin/env python3
import pathlib
import jinja2
import subprocess
from typing import Iterable, Dict


def main(texs: Iterable[str] = None, max_num_above_pars: int = 3, max_num_below_pars: int = 3,
        next_structures: Dict[str, str] = None, clean_success: bool = True,
        clean_ignorable: bool = False, clean_fail: bool = False):
    if texs is None:
        texs = [
            "pdflatex",
            "lualatex",
            "xelatex",
        ]

    if next_structures is None:
        next_structures = {
            "none": "",
            "section": r"\section{Section}",
            "subsection": r"\subsection{Subsection}",
            "paragraph": r"\paragraph{Paragraph}",
        }

    cmd = jinja2.Template(r"""{{tex}} -jobname={{jobname}} << EOF
\documentclass{article}
\usepackage{graphicx}
\usepackage{exwrapfig}

\begin{document}
\section{Section 1}

{{above_pars}}

\begin{wrapfigure}{r}{0.1\textwidth}
  \includegraphics[height=1ex]{example-image}
\end{wrapfigure}

{{below_pars}}

{{next_structure}}

\end{document}
EOF""")

    lorem_ipsum = """Lorem ipsum dolor sit amet, consectetur adipiscing elit,
sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad
minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea
commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit
esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat
non proident, sunt in culpa qui officia deserunt mollit anim id est
laborum."""

    possible_errors = {
        "capacity": "TeX capacity exceeded",
        "lastbox": "lastbox",
        "svsechd": "@svsechd",
    }

    lines = {
        "short": lorem_ipsum[:5],
        "long":  lorem_ipsum,
    }

    print("engine", "length", "above", "below", "next struct", "error", "PDF generated?", sep="\t")
    for tex in texs:
        for length, line in sorted(lines.items(), reverse=True):
            for num_above_pars in range(max_num_above_pars):
                for num_below_pars in range(max_num_below_pars):
                    if num_above_pars == 0 and num_below_pars == 0 and length == "long":
                        continue
                    for next_structure in sorted(next_structures.items()):
                        cond = [tex, length, num_above_pars, num_below_pars, next_structure[0]]
                        jobname = "-".join(map(str, cond))
                        subprocess.run("rm -f {}.*".format(jobname), shell=True)

                        p = subprocess.run(cmd.render(tex=tex, jobname=jobname,
                            above_pars="\n\n".join([line] * num_above_pars),
                            below_pars="\n\n".join([line] * num_below_pars),
                            next_structure=next_structure[1]),
                            stdout=subprocess.DEVNULL, shell=True)
                        r = p.returncode
                        if r == 0:
                            if clean_success:
                                subprocess.run("rm -f {}.*".format(jobname), shell=True)
                        else:
                            subprocess.run("rm -f {}.aux".format(jobname), shell=True)
                            errors = []
                            for k, v in possible_errors.items():
                                pg = subprocess.run("grep -Fq '{}' {}.log".format(v, jobname), shell=True)
                                if pg.returncode == 0:
                                    errors += [k]
                            if not errors:
                                errors = ["UNKNOWN"]
                            is_pdf_gen = pathlib.Path("{}.pdf".format(jobname)).exists()
                            if is_pdf_gen:
                                if clean_ignorable:
                                    subprocess.run("rm -f {}.*".format(jobname), shell=True)
                            else:
                                if clean_fail:
                                    subprocess.run("rm -f {}.*".format(jobname), shell=True)
                            print(*cond, " ".join(sorted(errors)), is_pdf_gen, sep="\t")


if __name__ == "__main__":
    main(texs=["pdflatex"])

which gives (PDFLaTeX)

length	above	below	next struct	error	PDF generated?
short	0	0	paragraph	capacity	False
short	0	0	section	capacity lastbox	False
short	0	0	subsection	capacity lastbox	False
short	0	1	paragraph	capacity	False
short	0	1	section	capacity lastbox	False
short	0	1	subsection	capacity lastbox	False
short	0	2	paragraph	capacity	False
short	0	2	section	lastbox	True
short	0	2	subsection	lastbox	True
short	1	0	paragraph	capacity	False
short	1	0	section	capacity	False
short	1	0	subsection	capacity	False
short	1	1	paragraph	capacity	False
short	1	1	section	capacity	False
short	1	1	subsection	capacity	False
short	1	2	paragraph	capacity	False
long	1	0	paragraph	capacity	False
long	1	0	section	capacity	False
long	1	0	subsection	capacity	False

[wtsnjp]

I gave up to stay in the expl3 world and replace a tl variable (i.e., a normal macro) \l_exwf_everypar_tl to a toks register. At least it seems to pass all the tests above.