Fix Type1 (PFA and PFB) font embedding

Issue 4989/3: Add a guard against invalid Type 1 (PFB) font

  This commit makes LilyPond can detect invalid Type1 (PFB) font.

Issue 4989/2: Fix Type 1 (PFB) font embedding

  Type 1 PFA data which is converted from PFB files can contain '\0'.

  In order to avoid problems with '\0',
  this commit makes to use the file length
  instead of zero-terminated string.

Issue 4989/1: Fix Type 1 (PFA) font embedding

  Type 1 (PFA) fonts can contain '\0'.
  e.g. URW++ June 2016 (Ghostscript 9.20)

  In order to avoid problems with '\0',
  this commit makes to use the file length
  instead of zero-terminated string.

[lemzwerg, 2016-10-26 13:07:16]

LGTM.

https://codereview.appspot.com/313930043/diff/1/lily/pfb.cc
File lily/pfb.cc (right):

https://codereview.appspot.com/313930043/diff/1/lily/pfb.cc#newcode49
lily/pfb.cc:49: seglen |= (static_cast<Byte>(*p++) << 24);
I suggest to add a guard against invalid, possibly extremely large `seglen'
values...

https://codereview.appspot.com/313930043/diff/1/lily/pfb.cc#newcode62
lily/pfb.cc:62: static_cast<int>(static_cast<Byte>(*p++)));
As above, we should probably add a guard against invalid data...

I also wonder whether `sprintf' is the most efficient tool for converting two
hex digits into a byte value.

[trueroad, 2016-10-26 14:09:57]

Thank you for your reviewing.

> https://codereview.appspot.com/313930043/diff/1/lily/pfb.cc
> File lily/pfb.cc (right):
> 
> https://codereview.appspot.com/313930043/diff/1/lily/pfb.cc#newcode49
> lily/pfb.cc:49: seglen |= (static_cast<Byte>(*p++) << 24);
> I suggest to add a guard against invalid, possibly extremely large `seglen'
> values...
> 
> https://codereview.appspot.com/313930043/diff/1/lily/pfb.cc#newcode62
> lily/pfb.cc:62: static_cast<int>(static_cast<Byte>(*p++)));
> As above, we should probably add a guard against invalid data...

I'll consider the detection of out-of-range access.

> I also wonder whether `sprintf' is the most efficient tool for converting two
> hex digits into a byte value.

How about stringstream?

std::stringstream ss;
ss << std::hex << std::setw (2) << std::setfill ('0')
   << static_cast<int>(static_cast<Byte>(*p++));
out.push_back (ss.str[0]);
out.push_back (ss.str[1]);

[lemzwerg, 2016-10-26 15:08:50]

> How about stringstream?
> 
> std::stringstream ss;
> ss << std::hex << std::setw (2) << std::setfill ('0')
>    << static_cast<int>(static_cast<Byte>(*p++));
> out.push_back (ss.str[0]);
> out.push_back (ss.str[1]);

Looks better to me, but I have no experience with the `stringstream' class.

[trueroad, 2016-10-27 15:02:03]

Use stringstream and add PFB error check

[lemzwerg, 2016-10-27 15:35:36]

LGTM now, thanks!

[trueroad, 2016-11-04 11:29:18]

I've pushed.

Issue 4989/3: Add a guard against invalid Type 1 (PFB) font
commit 3ec0872f35d8ab79d07127c3e7be5f08977c2394

Issue 4989/2: Fix Type 1 (PFB) font embedding
commit 65a0059b69385985896a24f407b3a791d33aef0f

Issue 4989/1: Fix Type1 (PFA) font embedding
commit 1e0bc03d9e50ea4d69e08fdbb0a85e56203299f2