[poppler] Utils - Fix UTF-16 file name on Windows environment

[William Bader]

I haven't used VMS for about two decades, but I remember that the extra options to fopen() are important. Without them, I think that you end up with a record oriented file instead of a stream file, and stdio might not work as you expect, for example, writing '\n' might close the current record instead of writing a 0x0A byte. The full list of parameters is in the entry for creat() at http://h41379.www4.hpe.com/commercial/c/docs/5763p021.html#index_x_600

HP sold VMS to a company that is porting it to Intel x86-64, so VMS might soon be easier to run. http://www.openvms.org/node/107
Regards, William

________________________________
From: poppler <poppler-bounces at lists.freedesktop.org> on behalf of Martin (gzlist) <gzlist at googlemail.com>
Sent: Sunday, June 24, 2018 8:05 AM
To: poppler at lists.freedesktop.org
Subject: Re: [poppler] Utils - Fix UTF-16 file name on Windows environment

On 19/06/2018, Albert Astals Cid <aacid at kde.org> wrote:
>
> It does make some sense given that function exists in the first place but my
> experience with windows is veeeeeeeeeeeeery limited so I'd like for someone
> else to vouch for this before landing it.

Overall the change might be, but there are some tricky aspects.

On windows fopen takes a narrow string of variable codepage, and
openFile takes (sort-of*) utf-8 but the types are not distinguished in
the codebase so it's tricky to see what callers are providing.

In cases where the string being used comes straight from say, the
command line or environment block, it will not be utf-8 so non-ascii
characters will be mangled. That's probably best fixed by ensuring on
GooString construction that it's converted to utf-8 but at present
that's entirely unvalidated.

Also:

 #ifdef VMS
-    f = fopen(fileName->getCString(), "rb", "ctx=stm");
+    f = openFile(fileName->getCString(), "rb", "ctx=stm");
 #else
-    f = fopen(fileName->getCString(), "rb");
+    f = openFile(fileName->getCString(), "rb");
 #endif

Breaks compilation on VMS (if that's still a platform that matters) as
openFile takes two args only. Oddly, GooFile::open already includes
this logic but openFile does not. Can just drop the first branch
change for now.

Martin

*sort-of utf-8: gfile.cc has a pretty half-assed utf-8 to utf-16
conversion algorithm in several places that only correctly handles a
subset of inputs.
_______________________________________________
poppler mailing list
poppler at lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/poppler
poppler Info Page - freedesktop.org<https://lists.freedesktop.org/mailman/listinfo/poppler>
lists.freedesktop.org
Subscribing to poppler: Subscribe to poppler by filling out the following form. Use of all freedesktop.org lists is subject to our Code of Conduct.



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/poppler/attachments/20180624/1affa187/attachment.html>