Bug#928472: pdftohtml litters the directory with all sort of binary named files

[積丹尼 Dan Jacobson]

Package: poppler-utils
Version: 0.71.0-3
Severity: important
File: /usr/bin/pdftohtml

pdftohtml is crazy, littering the directory with all kinds of raw binary
names for the *.png files it creates.
And then it can't find them anyway:
<img src="BR186585-"/><br/>
<img src="BR186585-"/><br/>
<img src="BR186585-"/><br/>
And also the internal links are not ASCII either and all the same by the way.
<a href="BR186585s.html#^P<B3>">
<a href="BR186585s.html#^P<B3>">
Also should run it through HTML TIDY (-e) to find the rest of the
problems. And no, this has nothing to do with -enc.
Tested on https://www.nlsc.gov.tw/uploadfile/BR186585.pdf