# What does this input encoding error mean? [closed]

! Package inputenc Error: Keyboard character used is undefined (inputenc) in inputencoding utf8'.

See the inputenc package documentation for explanation. Type H for immediate help. ...

l.78 \end {document} You need to provide a definition with \DeclareInputText or \DeclareInputMath before using this key.

Line 78 of the document is simply \end{document}. The documentation for the package has been of no use to me, in the log however, it appears on the next line, with some indentation.

I have changed every character in all of the relevant files (chapters as well as the .bib) to the format \'{x} as needed for german, french and spanish (the languages represented) with no success.

This error appears twice, same line, same description. I have one warning as follows:

Data encoding is 'utf8'.

I also have about 30 overfull \hbox bad boxes, although I do not think that is related, and clearing the auxiliary files does not seem to solve the problem.

The document compiles, and I get the associated pdf, but still with this message:

Process started: pdflatex.exe -synctex=1 -interaction=nonstopmode "tesis v2.0".tex

Process exited with error(s)

I am on a Windows 7 machine, using Miktex 2.9 64-bit, working on TeXStudio 2.8.8.

Edit: I've been taking all of the suggestions in the comments to try them. I cannot see any particular invisible character in my editor (switched to Notepad++), but it seems to be created by the bibliography. The relevant block of the preamble looks like this:

\usepackage[backend=bibtex,
style=authoryear,natbib=true,url=false,doi=true,
language=spanish,mergedate=false
]{biblatex}

Currently, removing that block from the preamble eliminates the error, but creates a bunch more (all related to the .aux file). Removing parts of the .bib file does not solve the problem. I am working on a MWE to update.
• Somewhere, you have a character which is not recognised. Bear in mind that the character may not be visible in your editor. You may be able to copy it from the error message somewhere and see what it is, or you may not be able to. Try deleting and entirely retyping te line it says the character occurs on. Otherwise, if the error is really somewhere else and there are no clues in the .log file, you can use brute force. Comment out almost the entire contents of the document environment, compile. If that is OK, uncomment half etc. If not, comment half the preamble. Etc. – cfr Jan 29 '15 at 3:35
• Since you are using utf8 with inputenc you can type á etc. directly - there's no need to change these to the macro versions. (They aren't the problem - at least when in the main .tex. You might need to change them in the .bib if you are using bibtex.) – cfr Jan 29 '15 at 3:37
• as it is at \end{document} it is probably reading the aux file back, if you showed more of the log we could see which file was being read. Use {}` to mark as a code section rather than use the quotation markup, so linebreaks in the log are preserved. – David Carlisle Jan 29 '15 at 9:54