Broken search in various pdf readers for "Theorem 7.1"

bellecp commented 4 years ago

Using the Evince pdf reader, searching for the text string "Theorem 7.1" in the sample pdf file https://github.com/vtex-soft/texsupport.ims-aos/blob/master/aos-sample.pdf does not find Theorem 7.1. However, searching for "heorem 7.1" without the capital T finds the theorem. This suggests that some invisible character or other issue is present between T and "heorem 7.1".

bellecp commented 4 years ago

It appears that most published version of AOS papers in recent years suffer from this. It should be an easy-fix and very useful to quickly locate theorems in published papers. I am happy to test a possible fix in some PDF readers.

bellecp commented 4 years ago

Annals of Probability suffers from the same issue with the sample pdf in https://github.com/vtex-soft/texsupport.ims-aop/blob/master/aop-sample.pdf

edgaras-sakuras commented 4 years ago

This problem is related with two things:

1) LaTeX engine in general adds some kerning (space) between letters for so called visual tuning. For small caps too. And this "fix-up" is even font dependent, e.g. compiling this file:

```LaTeX
\documentclass{article}
\begin{document}
\textsc{Theorem 7.1}
\end{document}
```

would result PDF with searchable "theo" (by Evince). In this case default Computer Modern font is used.
But adding these two line before `\begin{document}`:

```LaTeX
\usepackage[T1]{fontenc}
\usepackage{times}
```

would result PDF with **non**-searchable "theo" (by Evince). In this case default Nimbus Roman is used.

2) Not all PDF viewers "understand" such pseudo spaces (bug or feature in heuristics?). Usually these are PDF viewers with Poppler based PDF rendering - Evince, Zathura, Qpdfview, etc.

vtex-soft / texsupport.ims-aos

Broken search in various pdf readers for "Theorem 7.1" #1