Forcing linebreaks in \url

Question

This question led to a new package:
xurl

I wish to typeset some relatively long URLs in a piece of text, and when I use \url{..}, the resulting text does not respect the margin boundaries that govern the main text body, instead going all the way to the edge of the paper before wrapping around.

\documentclass{article}
\usepackage{hyperref}
\begin{document}
\url{very-long-url-very-long-url-very-long-url-very-long-url-very-long-url-very-long-url-very-long-url-very-long-url}
\end{document}

I'm compiling using pdflatex, although I doubt this makes a difference.

Answer 1

(converting a previous comment to an answer)

A quick google search (alas, too quick) reveals one solution. Use the [hyphens] option with the url package:

\PassOptionsToPackage{hyphens}{url}\usepackage{hyperref}

Some recommendations also include inserting \sloppy in case latex is trying too hard to align things.

Answer 2

Herbert Voß created (December 2017) the package xurl.

This package allows urls to break everywhere, can be compiled simply with pdflatex, and has the same options as url (indeed, it loads url package).

Here an MWE:

% arara: pdflatex
\documentclass{article}
\parindent=0pt

\usepackage{xurl}

\begin{document}
\rule{\linewidth}{1pt}

An example of everywhere breaking url: 
\url{very-long-url-very-long-url-very-long-url-very-long-url-very-long-url}

Another example of everywhere breaking url: 
\url{very-long-url-very-long-url-very-long-url-very-long-url-very-long-url}

A third example: 
\url{very-long-url-very-long-url-very-long-url-very-long-url-very-long-url}

\end{document}

Here the output:

Update: since version 0.05 (December 2018) the package author have added support for biblatex which has its own url handling.

If you load xurl after biblatex this is done by default.

If you want to avoid it, load xurl before biblatex or use the option nobiblatex:

\usepackage[nobiblatex]{xurl}

For more info, see package documentation.

Answer 3

The package url defines a command \def\UrlOrds{\do\*\do\-\do\~\do\'\do\"\do\-}% which can be added to the default url break characters at which a line can be broken. Below, the line \g@addto@macro appends the list of characters defined in \UrlOrds to the \UrlBreaks macro.

\documentclass{article}
\textwidth=8cm
\parindent=0pt

\usepackage{url}
\makeatletter
\g@addto@macro{\UrlBreaks}{\UrlOrds}
\makeatother
\begin{document}
\rule{\linewidth}{1pt}

foo bar baz \url{very-long-url-very-long-url-very-long-url-very-long-url-very-long-url-}

\end{document}

The same effect could be obtained by using the more usual \renewcommand on \UrlBreaks, but this would remove the characters that were already defined in \UrlBreaks.

Edit:
If you load package etoolbox you can reduce these 3 lines:

\makeatletter
\g@addto@macro{\UrlBreaks}{\UrlOrds}
\makeatother

to this oneliner:

\gappto{\UrlBreaks}{\UrlOrds}

Answer 4

I carefully read the documentation of the url package (I'm using PDFLaTeX) and found the following to work quite nice to let LaTeX add line breaks after every normal alphabetic character:

\expandafter\def\expandafter\UrlBreaks\expandafter{\UrlBreaks%  save the current one
  \do\a\do\b\do\c\do\d\do\e\do\f\do\g\do\h\do\i\do\j%
  \do\k\do\l\do\m\do\n\do\o\do\p\do\q\do\r\do\s\do\t%
  \do\u\do\v\do\w\do\x\do\y\do\z\do\A\do\B\do\C\do\D%
  \do\E\do\F\do\G\do\H\do\I\do\J\do\K\do\L\do\M\do\N%
  \do\O\do\P\do\Q\do\R\do\S\do\T\do\U\do\V\do\W\do\X%
  \do\Y\do\Z}

Update:
Herbert Voß (who provided the edited version above with all the \expandafters) created a package xurl which makes it much simpler (see also answer of CarLaTeX). This package takes also care of biblatex and does also include numbers and the characters that are defined as \UrlOrds in url, cf. this other answer (which is also from Herbert, though you can not recognize this anymore.)

Answer 5

Using the sloppypar environment as in

\begin{sloppypar}
  Figure \ref{fig:example_instant_lumi}:
  Public result available from \url{https://twiki.cern.ch/twiki/bin/view/AtlasPublic/LuminosityPublicResults}, version of 16.12.2011.
\end{sloppypar}

helps in my case. Without sloppypar the AtlasPublic part runs over the right margin, with sloppypar LaTeX breaks after view/ and everything's fine. (Doesn't need the url package.)

Answer 6

1. \usepackage[hyphens]{url} should be ok. If it does not work, then go to (2).

2. \RequirePackage[hyphens]{url} before \documentclass.

3. \expandafter\def\expandafter\UrlBreaks\expandafter{\UrlBreaks% save the current one \do\a\do\b\do\c\do\d\do\e\do\f\do\g\do\h\do\i\do\j% \do\k\do\l\do\m\do\n\do\o\do\p\do\q\do\r\do\s\do\t% \do\u\do\v\do\w\do\x\do\y\do\z\do\A\do\B\do\C\do\D% \do\E\do\F\do\G\do\H\do\I\do\J\do\K\do\L\do\M\do\N% \do\O\do\P\do\Q\do\R\do\S\do\T\do\U\do\V\do\W\do\X% \do\Y\do\Z}

   It does work, but it is not a good way because it breaks the url address at an arbitrary place, which looks ugly.

4. \sloppy sometimes does work, sometimes does not. It is not a thorough solution.

I recommend Solution (1) or (2).

Answer 7

If it's not necessary for the url to appear verbatim, you could use \href{url}{text}. That way you could give the url a human-readable form. If your document is used electronically you just use it as an hyperlink and if the document is printed you could (or should) question the relevance of long (and probably cryptic) url's.

Answer 8

You may also want to look at the breakurl package.

Answer 9

According to the documenation, BigBreaks is the way to go otherwise it breaks at the first character.

\def\UrlBigBreaks{\do\/\do-\do:}

See: http://texdoc.net/texmf-dist/doc/latex/url/url.pdf I had the problem that a line break was inserted always after http:, even when using \UrlBreaks, the documentation clarified that this is the intended behavior.

Answer 10

Not using \url, but \href, the URL is not visible anyway, so the suggestion in regard to employing tinyurl is relevant as it does avoid the ugliness of the URL in the text.

Much better, I find, to make an appropriate part of the text active, as in \href{URL}{active in context text}.

This way, the URL is not visible in the print out either.

Answer 11

I troubles I had with url breaking in an itemize environment could be solved by adding \Urlmuskip = 0mu plus 1mu in the preamble.

Answer 12

The top-voted answer did not completely work for me and I really won't use the most ridicule way of requesting a line break for each letter of the alphabet(!) as some answers suggest here. (I'd wonder what happens if you then have a lot of emojis in your URLs or so, which is – theoretically at least – a thing in 2019.)

Instead though, I found this nice solutions, which really seems to force breaking URLs at any position:

% always break URLs anywhere!
\setcounter{biburllcpenalty}{9000} % lower letters
\setcounter{biburlucpenalty}{9000} % upper letters

Source is this forum post, which actually refers to this German forum post.

In my testing it worked, and, okay, I admit, I did not test it with emojis.

Answer 13

I have a similar problem with itemize environment. When a word (or an url) is too long to respect the margin it does not starts a newline. Using \sloppy works but is very ugly.

My LaTeX code is generated by rst2latex.py and I use pdflatex on it.

I found that much better solution on the latex-community.org forum :

\tolerance 1414
\hbadness 1414
\emergencystretch 1.5em
\hfuzz 0.3pt
\widowpenalty=10000
\vfuzz \hfuzz
\raggedbottom

So I add this snippet in my docutils.tex and everything is just fine !

Answer 14

I used the .bib file with these packages.

\usepackage{breakurl}
\usepackage{url}

And that made the work.

Answer 15

I would like to link to the following question here on TeX-SE: https://tex.stackexchange.com/a/207712/128658

Followed by \Urlmuskip=0mu plus 1mu\relax this does result in URLs that are properly broken over multiple lines without causing any overfull boxes. Plus this does also work for URLs in bibliographys (opposed to @xamde's answer to this question).

If you are doing this sorcery inside a package instead of your actual document you have to encapsulate the counter-changes in a \AtBeginDocument as they aren't defined otherwise. It could therefore look something like this:

% set special behaviour for hyperlinks in pdfs
\RequirePackage[breaklinks=true, pdfencoding=auto]{hyperref}
\g@addto@macro\UrlBreaks
{%
    \do\a\do\b\do\c\do\d\do\e\do\f\do\g\do\h\do\i\do\j%
    \do\k\do\l\do\m\do\n\do\o\do\p\do\q\do\r\do\s\do\t%
    \do\u\do\v\do\w\do\x\do\y\do\z\do\&\do\1\do\2\do\3%
    \do\4\do\5\do\6\do\7\do\8\do\9\do\0\do\/\do\.%
}
\g@addto@macro\UrlSpecials
{%
    \do\/{\mbox{\UrlFont/}\hskip 0pt plus 10pt}%
}

\Urlmuskip=0mu plus 1mu\relax

\AtBeginDocument{%
    \setcounter{biburlucpenalty}{1}  %break URL after uppercase character
    \setcounter{biburlnumpenalty}{1} %break URL after number
    \setcounter{biburllcpenalty}{1}  %break URL after lowercase character
} 

Answer 16

The solution that works perfect for me both for bibtex and biblatex is the combination of sloppypar that @fuenfundachtzig mentioned with the \usepackage[hyphens]{url} that proposed by @Suresh.

The next code is used to be able to use both ways (bibtex and biblatex) to compile your file and works for me in cases of natbibs agsm style and bibers authoryear. Actually you don't have to use both ways and the method will work (I think) anyway if you just keep the code of each case (bibtex or biblatex) [except of the trick of @moewe with the url that solved the problem I had in my post here that will be possibly need to be adjusted according to the bibtex style we will use].

\documentclass{article}
\usepackage{filecontents}
% \UseBibLaTex is a variable that if is less than 1 the document 
% will have to compiled with `bibTeX`... else with biber
\def\UseBibLaTeX{1}
% ``url'' package is used in order to make LaTeX to break the urls 
% in places like the bibliography combined with the next (upcoming) 
% command
%\usepackage{breakurl}
\usepackage[hyphens]{url}
\Urlmuskip=0mu plus 1mu minus 5mu


\ifnum\UseBibLaTeX>0
\usepackage[backend=biber,style=authoryear]{biblatex}
\addbibresource{myBib.bib}
\else
\usepackage{natbib}
\def\harvardurl#1{{\bfseries URL:}\space\url{#1}}
\fi


\begin{filecontents*}{myBib.bib}
@Article{cite1,
title={One Article with a url that has to break both in bibtex and biblatex},
author={Someone NotMe and Someother Me},
year={2018},
month={10},
url={http://A-long-URL-that-has-to-break-through-lines-and-biber-has-to-ignore-the-url-command},
doi={http://dx.doi.org/10.1038/nphys1170}
}
@Article{cite2,
title={One Article with another url that has to break both in bibtex and biblatex},
author={Someone NotMe and Myself Me},
year={2017},
month={10},
url={http://AlongUrlThatHasToBreak.ThroughlinesAndBoth.BiberAndBiblatex.HasToAcceptTheCitation.com},
doi={http://dx.doi.org/10.1038/nphys1170}
}
\end{filecontents*}

\title{Double Functionality about Bibliography}
\author{Konstantinos Leledakis}
\date{October 2018}

\begin{document}

\maketitle

\nocite{*}
\ifnum\UseBibLaTeX>0
\begin{sloppypar}
\printbibliography
\end{sloppypar}
\else
\bibliographystyle{agsm}
\begin{sloppypar}
  \bibliography{myBib}
\end{sloppypar}
\fi

\end{document}

Output with bibtex:

Output with biblatex (biber):

Answer 17

There is another possibility to force a linebreak by using a user-defined character inside the url. This is documented in the url package under 5.2.1 Arbitrary character actions:

You can do arbitrarily complex things with characters by specifying their definition(s) in \UrlSpecials. This makes them ‘active’ in math mode (mathcode "8000). The format for setting each special character c is: \do\c{hdefinitioni}, but other definitions not following this style can also be included. Here is an example to make “!” inside \url force a line break instead of being treated verbatim (it uses LATEX’s \g@addto@macro):

\makeatletter
\g@addto@macro\UrlSpecials{\do\!{\newline}}

I think there are some downsides like using only ascii characters as the user-defined character (maybe, I'm not sure), and to have the user-defined character in the original url text, of course. I successfully tried the character | which I haven't found in any of my urls.

So you can write:

\url{https://what-an-example.org/subfolder/view.!php?id=12345}

which will give you:

https://what-an-example.org/subfolder/view.
php?id=12345