Why ImportString[ExportString[, "PDF"], "PDF"] sometimes gives an extra point

Question

Consider this code:

fc = ImportString[ExportString[Style["Ma"], "PDF"], "PDF"][[1, 1, 2, 1, 1]];
lines = GeometricFunctions`DecodeFilledCurve[fc][[1]];
{lines // Length,Graphics[JoinedCurve[#]] & /@ lines}

will give what is expected:

enter image description here

but this code:

fc = ImportString[ExportString[Style["Ma", FontFamily -> "Times"], "PDF"], "PDF"][[1, 1, 2, 1, 1]];
lines = GeometricFunctions`DecodeFilledCurve[fc][[1]];
{lines // Length, Graphics[JoinedCurve[#]] & /@ lines}

gives four things, with the fourth element is just a point:

enter image description here

lines[[-1]]
(*{Line[{{10.5607,9.40406},{10.5607,9.40406}}]}*)

Why there is an extra point in the second case? I'm using version 9 on Mac.

Update

screenshots of the problem on version 8 and 9 on mac, and version 8 on windows 7, the results seems different:

enter image description here

BTW The first picture is not created by the first block of code. — Sjoerd C. de Vries, May 10 '13 at 23:42
Ok, so it is OS dependent then. Unfortunately, I can't help with that as I don't have OSX. I assume that it is caused by its PDF exporter. There is a remote possibility that it may be a problem of the font itself. — Sjoerd C. de Vries, May 11 '13 at 06:29
Not sure why Times looks different on Mac, but you could work around it by replacing Times with TimesNewRoman. — Jens, May 12 '13 at 00:30
@Jens So another way of distinguishing between Times and Times New Roman...! :) The problem also appears in some other fonts, eg a Helvetica "g" (but not Helvetica Neue). — cormullion, May 12 '13 at 10:52

cormullion · Accepted Answer · 2013-05-12T07:23:18.980

I've encountered this problem, coming from a slightly different direction. I think the compound paths contain an extra pair of coordinates that you're noticing when you split up the results of a FilledCurve.

Here's how I first came across this:

glyph[code_, font_, size_] := 
  First[First[
    ImportString[
     ExportString[
      Style[FromCharacterCode[code], FontSize -> size, 
       FontFamily -> font], "PDF"], "TextMode" -> "Outlines"]]]; 

a = glyph[97, "Times", 24]

{Thickness[0.130073],
  {FilledCurve[
    {
      {{1,4,3},{1,3,3},{0,1,0},{1,3,3},{1,3,3},{1,3,3},{1,3,3}},
      {{1,4,3},{1,3,3},{0,1,0},{1,3,3},{1,3,3},{1,3,3}, ...
      {{0,2,0}}},
     {
      {{3.74836,5.265},{4.01063,5.265},{4.25426,5.35078}, ...
      {{0.405,5.96742},{0.405,6.74086},{0.77502,7.35469}, ...
      {{4.1175,12.4875},{4.1175,12.4875}}}]}}

The {{4.1175,12.4875},{4.1175,12.4875}} is a short zero-length line that I think might be causing the problems.

I noticed it when I selected the graphic. I could see that there was something odd there:

Graphics[{Gray, a}]

the letter a selected

The short line registers oddly as a large dot.

Looking at all the letter forms, it appears that you definitely get this extra line for many letters, not just for "compound paths".

Length[#[[2, 1, 1, 2]]] & /@ (glyph[#, "Helvetica", 24] & /@ 
   ToCharacterCode /@ Characters["abcdefghijklmnopqrstuvwxyz"]);
BarChart[%, ChartLabels -> Characters["abcdefghijklmnopqrstuvwxyz"]]

chart

There's "a", "e", and "o" which have three elements in the relevant FilledCurve. "b" has two, like it should have, but "c" should only have one. "n", "s", and "u" also have more elements than they need. And so on.

You could probably fix this by adding a rule that makes zero-length lines get deleted from the FilledCurve, but that's proved beyond my expertise...

tl;dr I think it's a bug, but who knows whether in Mathematica or the Mac OS system.

Why ImportString[ExportString[, "PDF"], "PDF"] sometimes gives an extra point

1 Answers1

Linked