segmentation fault with tlfp-parse-many

[olivierlacan]

With pypy3.5-6.0.0 the following command segfaults after a few minutes:

senapy-cli doslegs_urls --min-year=2019 | tlfp-parse-many data/

Last few lines before the segfault:

https://www.senat.fr/dossier-legislatif/pjl16-537.html
 -= DOSLEG URL: https://www.senat.fr/dossier-legislatif/pjl16-537.html =-
  [] download SENAT version
  [] parse SENAT version
  [] download AN version
  [] parse AN version
        title: projet de loi ordonnances relatives à la modification des règles applicables à l'évaluation environnementale des projets et portant réforme des procédures destinées à assurer l'information du public à l'élaboration de décisions ayant une incidence sur l'environnement
  [] parse the texts
** parsing texts
    ^ text:  http://www.assemblee-nationale.fr/15/projets/pl0011.asp
    ^ text:  http://www.assemblee-nationale.fr/15/ta-commission/r0091-a0.asp
    ^ text:  http://www.assemblee-nationale.fr/15/ta/ta0004.asp
    ^ text:  https://www.senat.fr/leg/pjl16-666.html
    ^ text:  https://www.senat.fr/leg/pjl17-009.html
    ^ text:  https://www.senat.fr/leg/tas17-001.html
    ^ text:  https://www.senat.fr/leg/pjl17-199.html
Segmentation fault: 11

Since it's parsing HTML this could very well be a lower-level lxml or Python bug but I thought you'd want to know.

[mdamien]

Sorry but I can't reproduce it on my Ubuntu machine, looks like it's a MacOS-specific segfault, maybe you can also try without PyPy. Also I'm transfering the issue to the main repo since we don't yet where the error is coming from.