# What features of PDF produced by TeX and Co determine its interpretation by other programs

This is a revision of my previously posted question that was put on hold. In short, I would like to know how different fonts are embedded in PDFs and interpreted by other programs. It is my naïve interpretation that any PDF document produced by TeX, LaTeX and alike contains a table with prescriptions how each symbol of the font must be rendered, and this prescription may vary depending on the type of font. I also assume that any textual information is represented as a sequence of coordinates---position of letters and their size. Thus, I was assuming until today, that any piece of text (be it a letter of Latin alphabet, a Unicode symbol, or a part of a formula) can be selected with a mouse in PDF viewer, copied and pasted. Analogically I assumed that text in PDF documents is editable by Inkscape, Adobe Illustrator, Coral Draw provided that fonts are embedded. There are many questions here concerning the font embedding, but I do not have this problem. The generated files do contain font metrics as I readily can check with Adobe Acrobat. Yet, I observed today, that not all letters are interpreted in the same way by Adobe Illustrator. I realized today that under some circumstances a letter or a symbol attains additional attributes that prevents their handling as textual information in other applications. I would like to know what are these attributes. Below is my original post.

I've created some nice looking diagrams using tikz and lualatex. I am using Cambria for text and Cambria Math for equations. I have several diagrams on one page and they are hard to position properly. However, I am sure I can do this solely with tex provided enough time. Unfortunately I do not have much of it, but I am experienced Adobe Illustrator user. I open the produced page and text gets immediately converted to paths. This is not necessary a bad thing. However, the text is very strangely grouped. This means a lot of extra work. How can I open lualatex-created file preserving text as text without converting to paths? Cambria and Cambria Math are system-wide fonts on my mac. Minimal not-working example:

\RequirePackage{luatex85}
\documentclass{standalone}
\usepackage[T1]{fontenc}
\usepackage[]{amsmath}
\usepackage{fontspec}
\setmainfont[Ligatures=TeX]{Cambria}
\usepackage[math-style=TeX]{unicode-math}
\setmathfont{Cambria Math}
\begin{document}
$4 F_{1, 4} F_{2, 3} \Delta^{1,2}_{4,3}$
\end{document}


Here, only $\Delta$ is a text. The rest of the symbols are paths in AI.

• maybe graphicdesign.stackexchange.com is a better choice for an adobe question. – user36296 Dec 22 '17 at 11:38
• @samcarter I guess the problem is on TeX side, why otherwise Delta is shown as text, but super/subscripts and associated symbols are shown as paths. Clearly, there is some inconsistency in the way how math is represented. – yarchik Dec 22 '17 at 11:42
• @samcarter I have edited my question to bring it to the scope of this site. I am kindly asking you to change your verdict. – yarchik Dec 22 '17 at 17:50
• @RobtAll I guess 3000 rep are required to vote to reopen tex.stackexchange.com/help/privileges/close-questions – user36296 Dec 22 '17 at 19:16
• From time to time, I have noticed that if I copy some example code from a package documentation PDF, some of the characters are not as expected in plain text. Math mode does not seem to be involved, and certainly not LuaLaTeX or Cambria Math. This leads me to believe that the old-fashioned OT1 (or similar) font encoding is confusing the PDF reader. That is, the character looks OK in the PDF, but it is not mapped to a standard character code. I ask the OP: With LuaLaTeX and fontspec, but without fontenc, and with utf-8 document, and with Libertinus Math as math font, what happens? – user139954 Dec 23 '17 at 0:48