On Mon, Jul 02, 2012 at 05:14:34PM +0100, Andy Whitcroft wrote:
> On Mon, Jul 02, 2012 at 11:56:56AM +0200, Jakub Wilk wrote:
> > tags 679891 + fixed-upstream
> > thanks
> >
> > * Andy Whitcroft <a...@canonical.com>, 2012-07-02, 10:37:
> > >* Disable tests/test-nfkc.sh which tests ligature handling but
> > >libpoppler25 now always decomposes such ligatures.
> >
> > Well, this test shows that pdf2djvu with your patch doesn't work
> > like it's shouldâ¦
>
> The issue is that the semantics of the underlying library has changed.
> The call used no longer returns ligatures they are always decomposed in
> libpoppler before being returned. This is a deliberate change in the
> libpoppler code, new code introduced to ensure that ligatures are not
> exposed. The new functionality did not appear to be optional.
To be specific this is the code delta in poppler which triggers the test
to fail. As the test is specifically seeing if the pdf2djvu ligature
removal is working by confirming that the output changes when it removes
ligatures it is going to fail:
@@ -852,6 +1253,17 @@
// construct the char code -> Unicode mapping object
ctu = CharCodeToUnicode::make8BitToUnicode(toUnicode);
+ // pass 1a: Expand ligatures in the Alphabetic Presentation Form
+ // block (eg "fi", "ffi") to normal form
+ for (code = 0; code < 256; ++code) {
+ if (unicodeIsAlphabeticPresentationForm(toUnicode[code])) {
+ Unicode *normalized = unicodeNormalizeNFKC(&toUnicode[code], 1,
&len, NULL);
+ if (len > 1)
+ ctu->setMapping((CharCode)code, normalized, len);
+ gfree(normalized);
+ }
+ }
+
// pass 2: try to fill in the missing chars, looking for ligatures,
// numeric
// references and variants
if (missing) {
Hopefully you'll have a better idea as to how to proceed.
-apw
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
