Log Message:
xpdf: update to 4.04.

4.04 (2022-apr-18)
------------------
When closing a file, save the current page number in ~/.xpdf.pages,
  and restore it next time the file is opened.  This can be disabled
  with "savePageNumbers no" in your xpdfrc file.
Allow the tab list to be reordered, using drag-and-drop.
Added a document information dialog, with metadata and fonts, to xpdf.
Pdftohtml now generates HTML links for URI links anchored on text.
Added the useTrueTypeUnicodeMapping xpdfrc setting.
Added the 'enableXFA' xpdfrc setting back: if set to 'no', xpdf will
  not read an XFA form, meaning all form info comes from the AcroForm.
Add support for long paths on Windows 10.
Added the "-formfields" option to pdftohtml.
Added the "-embedbackground" and "-embedfonts" options to \ 
pdftohtml.
Added the "-vstretch" option to pdftohtml.
Added the "-verbose" flag to pdfimages, pdftohtml, pdftopng, pdftoppm,
  pdftops, and pdftotext.
Added the showAttachmentsPane, showLayersPane, and showOutlinePane
  commands.
Added the 'openFile2' command to xpdf.
Added the showMenuBar, hideMenuBar, and toggleMenuBar commands to
  xpdf.
Rearrange file names in the tab list to "foo.pdf [/full/path/]", to
  improve the usability of the (narrow) tab list.
Added the 'imageMaskAntialias' xpdfrc setting.
Added the '-table' switch to pdftohtml.
Fixed a couple of progression order bugs in the JPEG 2000 decoder.
  [Thanks to shellway for the bug reports.]
Fixed a problem with subsampled pixel index computations in the JPEG
  2000 decoder.  [Thanks to shellway for the bug report.]
Fixed a couple of memory leaks triggered by fuzzed PDF files.  [Thanks
  to shellway for the bug report.]
Changed pdftohtml to generate spans with 'class=' instead of 'id=', so
  there aren't duplicate ids.  Also removed the unused id in the
  background img element.
Optimized SplashOutputDev::drawTilingPattern for the case where the
  tile is much larger than the area being drawn (i.e., the clipping
  bbox).  In this case, we can render just the needed portion of the
  tile.
Added a missing zero check for precinct size in the JPEG 2000 decoder.
  [Thanks to shellway for the bug report.]
Added a missing null check in the XFA form scanner.  [Thanks to Taolaw
  for the bug report.]
Deal with xref streams that include a free entry with gen=0xffffffff.
Fixed a problematic corner case in looking up XObject resources.
  [Thanks to shellway for the bug report.]
The unicodeRemapping config command wasn't correctly handling unsorted
  remapping files.
Improved handling of RTL text in the text extractor.
Extended the saveTabState/loadTabState commands to save/restore
  display mode, zoom, rotation, and scroll position for each tab.
Added support for Qt6.
Added a missing bounds check on stream DecodeParms arrays.  [Thanks to
  minipython for the bug report.]
Fixed an integer overflow check in XRef::readXRefTable.  [Thanks to
  yangshufan for the bug report.]
The builtin Latin1 encoding was mapping 'endash' instead of
  'softhyphen' to 0xad.  [Thanks to Jach Fong for the bug report.]
Fixed a bug in PSOutputDev::checkPageSlice() when built without
  SPLASH_CMYK.  [Thanks to irfanariq for the bug report.]
Added missing array length and type checks in Gfx::doForm().  [Thanks
  to shaohua for the bug report.]
Fixed an integer overflow security hole in the JBIG2 decoder.
Substitute Helvetica for undefined fonts and bad font objects.
Added an integer overflow check in JPXStream.  (JPXStream issue)
  [Thanks to Shin Ando @ Ricera Security for the bug report.]
The DCT (JPEG) decoder was allowing the 'interleaved' flag to be
  changed after the first scan of the image.  (CVE-2022-24106) [Thanks
  to Shin Ando @ Ricera Security for the bug report.]