4

It is a well-known fact that spammers crawl web pages and automatically extract email addresses not only from HTML files, but also from other formats, e.g., PDF.

It is therefore desirable to have an obfuscation method for email addresses in TeX that works with commonly used formats (e.g., PDF, PostScript, DVI), produces results that look identical to what one would get without obfuscation, and does not allow any commonly used software (e.g., pdftotext) to extract the unobfuscated email address. (A use case I have a mind is uploading a paper on arXiv with an email address in it.)

Thus the desired properties of obfuscation can be summarized as follows.

  1. Portability: the end result must look exactly like the unobfuscated version when compiled with any commonly used software: tex, tex+dvips, tex+dvipdfm(x), pdftex, tex+dvips+ghostscript (directly or via a script such as ps2pdf). In particular, when compiled by arXiv, all three resulting files (DVI, PostScript, and PDF) must have an email address in them that looks exactly like an unobfuscated email address, i.e., the reader of the paper should not be able to guess that obfuscation was used by merely looking at the paper (though copy-paste might fail).

  2. Robustness: none of the commonly used software, such as

    • catdvi, dvi2tty, crudetype for DVI,
    • ghostscript (directly or via a script such as ps2ascii), ps2txt, ps2a, pstotext for PostScript,
    • pdftotext, pdfbox, pdf2txt.py, or ghostscript for PDF,

    should be able to extract (after a possible removal of whitespace) the unobfuscated email from any of the files produced in 1.

  3. Self-containedness: the method must not require any external files and must only use the current font. (Otherwise one could just use a font with randomly remapped characters, which would be a very reliable obfuscation method, but not a very practical one if one wants to publish a paper in journal. For the same reason using other fonts (even in the same family) is also not acceptable because it easily breaks if the fonts are changed.)

Methods that are known not to work:

  1. Typesetting the email address in a nontrivial order, e.g., randomly permuting letters. Fails with catdvi, dvipdfm+pdftotext, pdftex+pdftotext.

  2. Inserting in the middle of the email address some additional text typeset in a microscopic font (cmr10 at 1sp, for example). Fails with tex+dvipdfm+pdftotext. It also exposes bugs in catdvi and pdftex 1.40.15, i.e., the latter is a noncompliant implementation of TeX. Even though one can work around these bugs by using 8sp instead of 1sp, the method also fails with pdftex+pdftotext.

  3. Inserting horizontal microspaces such as 1sp or -1sp in the middle of the email address. Same as 2.

  4. Raising or lowering individual letters by a microscopic amount (e.g., 1sp). Works with catdvi, but fails with pdftotext.

  5. Replacing some period with the dot accent lowered by its height minus the height of the period. Although this method does satisfy all of the above criteria, and, in a certain sense, is a solution (which I currently use in my papers together with typesetting in a different order), for pdftotext it merely replaces periods with spaces, which is still vulnerable to automatic collection (spam software is known to deobfuscate even more sophisticated variants such as “user at example dot com” and periods replaced by spaces certainly do not present a problem), so it violates condition 2, if not in letter, then in spirit. It also relies on rather subtle properties of Computer Modern fonts, which means it might break condition 1 if the publisher uses a more poorly designed font.

  6. Inserting some additional white-colored text that is typeset before the email address and overlaps with it. Unlike the previous methods this one is somewhat nonportable, but it works fine with most software using color specials, and some special hacks for pdftex (it does break some less commonly used software such as dvilj). Unfortunately, this method totally fails when one prints the document on a nonwhite paper or simply displays it on a nonwhite background (which is not that uncommon, some people prefer light gray or similar colors as background). Ostensibly, the problem could be solved by using background-colored text, but I don't know any reasonably portable way to do it, though some programs might allow such hacks.

Dmitri Pavlov
  • 153
  • 1
    I'm flabbergasted by your scope of operation for spam mails. But maybe you can explain what fail means in this context? I can stick anything in the email address. And from the look of it the best thing you can do is to put your website address and let people find you there. However you do realize I hope that if somebody who has your mail in her/his addressbook gets compromised to spam indexing all your efforts goes down the drain right? – percusse Apr 05 '15 at 20:49
  • @percusse: Failing means that the indicated software recovers the unobfuscated email address. As for your suggestion to put the website address, some of my coauthors simply don't have one. – Dmitri Pavlov Apr 05 '15 at 20:52
  • If you want to avoid this for DVI files, you're out of luck: an @ will correspond to @ in the DVI file, so dvi2tty will most certainly display the address. You could use a remapped font, but this would violate several of your constraints. – egreg Apr 05 '15 at 21:00
  • @egreg: Not at all, method 2 already works for dvi2tty. And the possibility of a remapped font is discussed in the post. – Dmitri Pavlov Apr 05 '15 at 21:11
  • @JohnKormylo: This is method 1 in my list and catdvi, dvipdfm+pdftotext, pdftex+pdftotext all recover the unobfuscated email. – Dmitri Pavlov Apr 05 '15 at 21:32
  • seeing as arXiv accept images in documents why not just include an image? – David Carlisle Apr 05 '15 at 21:48
  • @DavidCarlisle: It violates condition 3. (You can't make an image that uses the current TeX font.) – Dmitri Pavlov Apr 05 '15 at 21:54
  • I know, that's the question, why condition 3 – David Carlisle Apr 05 '15 at 21:55
  • @DavidCarlisle: Many journals like to typeset papers in the fonts of their choice. This includes emails. – Dmitri Pavlov Apr 05 '15 at 22:00
  • 2
    many journals wouldn't accept any of the suggested obfuscation techniques so i don't think that's much of an objection:-) – David Carlisle Apr 05 '15 at 22:21
  • Related: What can cause generated PDF document whose text are not correctly copyable? – Werner Apr 05 '15 at 22:41
  • Very much related: http://tex.stackexchange.com/questions/150375/crawler-resistance-email-address. However, it is not wholly clear from your question what you seek. Do you want something that visually looks like the email address in the PDF, but does not copy/paste as an email address? Or do you want something that visually obscures the email address as it appears in the PDF? – Steven B. Segletes Apr 06 '15 at 00:35
  • Also related: http://tex.stackexchange.com/a/160367/8528 -- uses the current font as an image, too. – jon Apr 06 '15 at 05:39
  • @Werner: My question is not specific to PDF; arXiv also generates PostScript and DVI and the method should also work for them. – Dmitri Pavlov Apr 06 '15 at 12:11
  • @StevenB.Segletes: As the original post says, I want something that looks exactly like the unobfuscated email would look like. So visually obscuring anything is out of the question. – Dmitri Pavlov Apr 06 '15 at 12:12
  • @jon: The answer in the link is not at all related to my question: it has massive portability problems (\write18 is disabled by default on most distributions, not to mention that the software it uses need not be available). – Dmitri Pavlov Apr 06 '15 at 12:14
  • @DavidCarlisle: Do you have any concrete examples of journals that are known not to accept methods 1–4 in my list? – Dmitri Pavlov Apr 06 '15 at 12:16
  • 1
    @DmitriPavlov Any journal that publishes an HTML version of articles is going to simply retype your address. – Joseph Wright Apr 06 '15 at 13:17
  • @JosephWright: I don't know many math journals that produce HTML versions, but those that do (e.g., Advances in Mathematics) obfuscate their email addresses properly, for example, see http://www.sciencedirect.com/science/article/pii/S0001870815000936. So your claim is simply wrong. – Dmitri Pavlov Apr 06 '15 at 13:40
  • If I inspect the HTML source which email harvesters also do I clearly see all the emails embedded in the source. There is no obfuscation whatsoever which is what @JosephWright is mentioning. – percusse Apr 06 '15 at 14:34
  • @percusse: To which source code are you referring? TeX? Or is this a response to the previous comment about HTML? – Dmitri Pavlov Apr 06 '15 at 14:36
  • @DmitriPavlov Edited. And note that I don't have a subscription to that journal which makes things much much easier. – percusse Apr 06 '15 at 14:37
  • @percusse: Not true, try for yourself: curl -s http://www.sciencedirect.com/science/article/pii/S0001870815000936 | grep dinasso (this searches for the first author's email, and yields no results whether you have a subscription or not). – Dmitri Pavlov Apr 06 '15 at 14:40
  • You are using very primitive tools. It is true --> http://i.imgur.com/T8gqx5P.png Trust me this battle has been lost and you are trying something many tried for decades. Spam detection software is your best bet. – percusse Apr 06 '15 at 15:00
  • Note that I didn't download or anything the simple Firefox extension is enough to discover these. – percusse Apr 06 '15 at 15:03
  • @percusse: I see, the “source” you claim to have inspected has been modified by JavaScript. Automatic crawlers used by spammers don't run JavaScript. – Dmitri Pavlov Apr 06 '15 at 15:05
  • If that's your preception.. – percusse Apr 06 '15 at 15:09
  • @percusse: If you look into the (real) source, you will see that the email addresses are constructed using JavaScript. Try to disable JavaScript in your browser and see for yourself if any email addresses are still visible. – Dmitri Pavlov Apr 06 '15 at 15:12
  • The question is why would I disable it if I'm a webcrawler? They tried email engines they tried many other things. You can too... – percusse Apr 06 '15 at 15:16
  • @percusse: I don't know why, but they just don't do it (even though it's possible, one simply has to run a command-line JavaScript interpreter, e.g., phantomjs, after downloading the page): http://superuser.com/questions/235937/does-e-mail-address-obfuscation-actually-work – Dmitri Pavlov Apr 06 '15 at 15:33
  • Re portability in my comment: using the ifplatform package can get you (maybe) other similar solutions specific to those OSes (which I don't know or care to know). That is why it was a comment, and not an answer: think of it as a suggestion to explore, not a complete answer that deserves only a 'OK'/'Not OK' vote. (Or not....) – jon Apr 06 '15 at 15:37
  • @jon: No, the problem is that most distributions simply do not have any form of external command execution enabled by default. There is nothing one can do about it, and packages won't help here. – Dmitri Pavlov Apr 06 '15 at 15:42
  • I'm very surprised: I've seen questions here about how to enable shell-escape on other platforms, but I don't remember hearing ever that it is simply impossible to do so. (Mind you, I've never really paid attention to this issue, either.) Anyway, good luck. I have nothing useful to add.... – jon Apr 06 '15 at 19:21
  • @jon: Even if there was a portable way to enable shell escapes, what kind of commands would you run? You can't assume that everyone has POSIX, let alone GhostScript or other image-processing software. Also, think about somebody who will try to compile the file in 20 years… – Dmitri Pavlov Apr 07 '15 at 10:41
  • The point would be to run commands based on the result of the ifplatform test. As for your worries about "the future", that's another matter. Your quest is academically interesting to me, but I don't necessarily agree with all the constraints you've imposed. I'm inclined to think your demands are too stringent, but the ingenuity of some people here is truly remarkable. (Not me, however.) – jon Apr 07 '15 at 12:31

2 Answers2

5

If you really want to hide the text you could extract the paths and draw them, the following's pretty poor, mostly as I ignored the distinction between move and line segments but it could be improved to render essentially the same as the original text.

First use a bit of postscript to convert the text to a path;

/Times-Roman findfont 10 scalefont setfont

newpath
100 100 moveto
(someone@example.org)
true
charpath
flattenpath
{== == (move) ==}
{== == (line) ==}
{== == == ==(curve) ==}
{(close) ==}
pathforall
showpage

that produces

a path of the form

104.429482
103.061745
(line)
104.429482
102.952805
(line)
104.320549
102.795456
(line)

then a bit of editing you can render that path with tikz:

enter image description here

with a bit more care the individual path segments could be closed and filled to produce essentially the original font rendering, but here I just stroke the entire path.

\documentclass{article}
\usepackage{tikz}

\begin{document}


aaa

\fbox{%
\begin{tikzpicture}[xscale=-.03,yscale=.03,rotate=90]
\draw

(103.085953,103.098053)
--
(104.429482,103.061745)
--
(104.429482,102.952805)
--
(104.320549,102.795456)
--
(104.393173,102.529175)
--
(104.526314,101.839249)
--
(104.178329,100.876991)
--
(103.303818,100.495712)
--
(102.565483,100.762)
--
(101.887665,101.657692)
--
(101.56086,102.22657)
--
(100.822517,102.73494)
--
(100.308105,102.514038)
--
(100.120491,101.923973)
--
(100.432167,101.122093)
--
(101.488235,100.665169)
--
(101.488235,100.50782)
--
(99.9631424,100.50782)
--
(99.9631424,100.640961)
--
(100.072075,100.870934)
--
(99.9994583,101.318779)
--
(99.902626,102.032913)
--
(100.274818,103.019379)
--
(101.16143,103.424858)
--
(101.793861,103.234222)
--
(102.33551,102.601791)
--
(102.964912,101.536652)
--
(103.63063,101.113014)
--
(104.114784,101.321808)
--
(104.296341,101.875557)
--
(104.024,102.535225)
--
(103.085953,102.952805)
--
(103.085953,103.098053)
--
(104.526314,106.344925)
--
(103.884811,104.780495)
--
(102.22657,104.178329)
--
(100.562286,104.786552)
--
(99.902626,106.332825)
--
(100.589523,107.885147)
--
(102.299194,108.511528)
--
(103.89994,107.900284)
--
(104.526314,106.344925)
--
(104.247925,106.223892)
--
(103.609444,107.234566)
--
(101.960289,107.627945)
--
(100.65004,107.310211)
--
(100.181015,106.441757)
--
(100.810417,105.497658)
--
(102.698624,105.061913)
--
(103.827316,105.373589)
--
(104.247925,106.223892)
--
(99.9994583,109.044098)
--
(99.9994583,111.234909)
--
(100.144699,111.234909)
--
(100.268768,110.696281)
--
(100.653069,110.557091)
--
(103.424858,110.557091)
--
(103.594315,110.690231)
--
(104.005852,111.719063)
--
(103.754692,112.209274)
--
(102.977013,112.372673)
--
(100.846725,112.372673)
--
(100.302055,112.248611)
--
(100.144699,111.706963)
--
(99.9994583,111.706963)
--
(99.9994583,113.909866)
--
(100.144699,113.909866)
--
(100.317184,113.337959)
--
(100.931458,113.19574)
--
(103.412758,113.19574)
--
(104.005852,114.272987)
--
(103.77285,114.850952)
--
(102.928604,115.011322)
--
(100.858833,115.011322)
--
(100.317184,114.896339)
--
(100.144699,114.357712)
--
(99.9994583,114.357712)
--
(99.9994583,116.512207)
--
(100.144699,116.512207)
--
(100.168907,116.258026)
--
(100.749901,115.834389)
--
(102.771248,115.834389)
--
(104.084526,115.546921)
--
(104.526314,114.696625)
--
(103.691147,113.086807)
--
(104.526314,112.045868)
--
(104.381065,111.422516)
--
(103.763771,110.520775)
--
(104.502106,110.520775)
--
(104.526314,110.448151)
--
(104.078468,109.068306)
--
(103.909019,109.068306)
--
(103.957436,109.383011)
--
(103.328026,109.734024)
--
(100.834625,109.734024)
--
(100.296,109.597855)
--
(100.144699,109.044098)
--
(99.9994583,109.044098)
--
(101.609276,120.685036)
--
(100.813446,119.97998)
--
(100.580444,119.159943)
--
(101.367195,117.913239)
--
(102.722832,117.622742)
--
(102.722832,120.64872)
--
(103.464195,120.503471)
--
(103.957436,120.225082)
--
(104.526314,118.966278)
--
(103.851524,117.477493)
--
(102.105537,116.908615)
--
(100.495712,117.404877)
--
(99.902626,118.748405)
--
(100.326263,119.99511)
--
(101.548752,120.842384)
--
(101.609276,120.685036)
--
(103.037537,117.64695)
--
(103.890862,117.976784)
--
(104.1632,118.687889)
--
(103.924149,119.341499)
--
(103.037537,119.644096)
--
(103.037537,117.64695)
--
(104.526314,123.562737)
--
(103.884811,121.998306)
--
(102.22657,121.396133)
--
(100.562286,122.004356)
--
(99.902626,123.550629)
--
(100.589523,125.102959)
--
(102.299194,125.729332)
--
(103.89994,125.118088)
--
(104.526314,123.562737)
--
(104.247925,123.441696)
--
(103.609444,124.45237)
--
(101.960289,124.845749)
--
(100.65004,124.528023)
--
(100.181015,123.659569)
--
(100.810417,122.715462)
--
(102.698624,122.279724)
--
(103.827316,122.5914)
--
(104.247925,123.441696)
--
(99.9994583,126.286118)
--
(99.9994583,128.367981)
--
(100.144699,128.367981)
--
(100.280869,127.853569)
--
(100.653069,127.714378)
--
(103.424858,127.714378)
--
(103.866653,128.277206)
--
(103.981644,128.73111)
--
(103.748642,129.278809)
--
(103.025429,129.457336)
--
(100.967766,129.457336)
--
(100.338364,129.327225)
--
(100.144699,128.827942)
--
(99.9994583,128.827942)
--
(99.9994583,130.873489)
--
(100.144699,130.873489)
--
(100.308105,130.398422)
--
(100.798309,130.280411)
--
(103.049637,130.280411)
--
(104.132942,129.971756)
--
(104.526314,129.118423)
--
(104.362915,128.486008)
--
(103.727455,127.69017)
--
(104.502106,127.69017)
--
(104.526314,127.617546)
--
(104.078468,126.261909)
--
(103.909019,126.261909)
--
(103.957436,126.528191)
--
(103.824287,126.818687)
--
(103.328026,126.891312)
--
(100.883041,126.891312)
--
(100.320206,126.7733)
--
(100.144699,126.286118)
--
(99.9994583,126.286118)
--
(101.609276,135.121964)
--
(100.813446,134.416916)
--
(100.580444,133.596878)
--
(101.367195,132.350174)
--
(102.722832,132.059677)
--
(102.722832,135.085663)
--
(103.464195,134.940414)
--
(103.957436,134.662018)
--
(104.526314,133.403214)
--
(103.851524,131.914429)
--
(102.105537,131.345551)
--
(100.495712,131.841812)
--
(99.902626,133.185349)
--
(100.326263,134.432037)
--
(101.548752,135.279312)
--
(101.609276,135.121964)
--
(103.037537,132.083893)
--
(103.890862,132.413712)
--
(104.1632,133.124817)
--
(103.924149,133.778427)
--
(103.037537,134.081024)
--
(103.037537,132.083893)
--
(100.713585,142.32077)
--
(100.241531,140.38414)
--
(100.459404,139.225189)
--
(101.061569,138.317398)
--
(101.990547,137.724304)
--
(103.170677,137.515518)
--
(104.514214,137.754562)
--
(105.591461,138.423309)
--
(106.341904,140.287308)
--
(106.130081,141.379684)
--
(105.555145,142.278397)
--
(104.704849,142.883591)
--
(103.666939,143.107513)
--
(102.326431,142.717163)
--
(101.730316,141.848709)
--
(102.178162,141.461395)
--
(102.359718,141.473495)
--
(104.853119,142.115)
--
(104.853119,141.43718)
--
(104.477898,141.340347)
--
(104.998367,140.626221)
--
(104.550522,139.573181)
--
(103.63063,138.937729)
--
(102.517067,138.701691)
--
(101.736366,138.949829)
--
(101.415611,139.573181)
--
(102.117638,140.79567)
--
(101.603226,141.089188)
--
(101.403511,141.69136)
--
(102.108559,142.965302)
--
(103.763771,143.506943)
--
(104.898514,143.25882)
--
(105.812355,142.574951)
--
(106.423607,141.549149)
--
(106.644501,140.275208)
--
(106.378212,138.868134)
--
(105.642899,137.727341)
--
(104.553551,136.95874)
--
(103.206993,136.680344)
--
(101.890694,136.955704)
--
(100.828575,137.724304)
--
(100.12352,138.874176)
--
(99.8663101,140.311523)
--
(99.9813,141.270752)
--
(100.423088,142.429703)
--
(100.713585,142.32077)
--
(104.56263,140.771469)
--
(103.933228,141.170898)
--
(102.408134,140.735153)
--
(101.839249,139.996811)
--
(102.063171,139.564102)
--
(102.674416,139.403717)
--
(104.017952,139.875778)
--
(104.56263,140.771469)
--
(101.609276,148.769119)
--
(100.813446,148.064072)
--
(100.580444,147.244034)
--
(101.367195,145.99733)
--
(102.722832,145.706833)
--
(102.722832,148.732803)
--
(103.464195,148.58757)
--
(103.957436,148.309174)
--
(104.526314,147.050369)
--
(103.851524,145.561584)
--
(102.105537,144.992706)
--
(100.495712,145.488968)
--
(99.902626,146.832504)
--
(100.326263,148.079193)
--
(101.548752,148.926468)
--
(101.609276,148.769119)
--
(103.037537,145.731049)
--
(103.890862,146.060867)
--
(104.1632,146.771973)
--
(103.924149,147.425583)
--
(103.037537,147.72818)
--
(103.037537,145.731049)
--
(99.9994583,151.925217)
--
(99.9994583,153.910263)
--
(100.144699,153.910263)
--
(100.737793,153.099289)
--
(102.662315,151.840485)
--
(103.848495,152.651443)
--
(104.272133,153.450302)
--
(104.429482,153.450302)
--
(104.429482,151.901)
--
(104.272133,151.901)
--
(104.05426,152.28833)
--
(103.424858,151.925217)
--
(102.98912,151.63472)
--
(103.182785,151.489471)
--
(104.05426,151.041626)
--
(104.272133,151.465271)
--
(104.429482,151.465271)
--
(104.429482,149.431808)
--
(104.272133,149.431808)
--
(104.272133,149.516541)
--
(103.691147,150.266983)
--
(102.274986,151.198975)
--
(100.653069,150.073318)
--
(100.259689,149.728363)
--
(100.144699,149.359192)
--
(99.9994583,149.359192)
--
(99.9994583,150.787445)
--
(100.144699,150.787445)
--
(100.326263,150.388016)
--
(100.725693,150.58168)
--
(101.936081,151.368439)
--
(100.556236,152.26413)
--
(100.38678,152.324646)
--
(100.144699,151.925217)
--
(99.9994583,151.925217)
--
(100.653069,158.533951)
--
(100.459404,158.110306)
--
(101.02829,157.807709)
--
(102.952805,157.807709)
--
(103.957436,157.62616)
--
(104.526314,156.391556)
--
(104.1632,155.144852)
--
(103.424858,154.745422)
--
(103.001221,155.169067)
--
(103.412758,155.604797)
--
(103.570107,155.580597)
--
(103.800079,155.556381)
--
(104.142014,155.762161)
--
(104.284241,156.270523)
--
(104.078468,156.815201)
--
(103.473274,157.00885)
--
(102.86808,157.00885)
--
(102.253807,155.547318)
--
(101.80294,154.902771)
--
(100.955666,154.551758)
--
(100.199173,154.839233)
--
(99.902626,155.580597)
--
(100.616753,157.020966)
--
(100.066025,157.202515)
--
(99.902626,157.65036)
--
(100.38678,158.533951)
--
(100.653069,158.533951)
--
(101.209846,157.00885)
--
(100.689377,156.754669)
--
(100.471504,156.040543)
--
(100.695427,155.598755)
--
(101.234055,155.423248)
--
(101.270363,155.423248)
--
(102.008705,155.774261)
--
(102.638107,157.00885)
--
(101.209846,157.00885)
--
(99.9994583,158.78511)
--
(99.9994583,160.975906)
--
(100.144699,160.975906)
--
(100.268768,160.437286)
--
(100.653069,160.298096)
--
(103.424858,160.298096)
--
(103.594315,160.431244)
--
(104.005852,161.460068)
--
(103.754692,161.950272)
--
(102.977013,162.113678)
--
(100.846725,162.113678)
--
(100.302055,161.989609)
--
(100.144699,161.447968)
--
(99.9994583,161.447968)
--
(99.9994583,163.650879)
--
(100.144699,163.650879)
--
(100.317184,163.078964)
--
(100.931458,162.936752)
--
(103.412758,162.936752)
--
(104.005852,164.013992)
--
(103.77285,164.591949)
--
(102.928604,164.752335)
--
(100.858833,164.752335)
--
(100.317184,164.637344)
--
(100.144699,164.098724)
--
(99.9994583,164.098724)
--
(99.9994583,166.25322)
--
(100.144699,166.25322)
--
(100.168907,165.999039)
--
(100.749901,165.575394)
--
(102.771248,165.575394)
--
(104.084526,165.287933)
--
(104.526314,164.437637)
--
(103.691147,162.827805)
--
(104.526314,161.78688)
--
(104.381065,161.163528)
--
(103.763771,160.26178)
--
(104.502106,160.26178)
--
(104.526314,160.189163)
--
(104.078468,158.809311)
--
(103.909019,158.809311)
--
(103.957436,159.124008)
--
(103.328026,159.475021)
--
(100.834625,159.475021)
--
(100.296,159.338852)
--
(100.144699,158.78511)
--
(99.9994583,158.78511)
--
(97.8691635,166.455963)
--
(97.8691635,168.840424)
--
(98.03862,168.840424)
--
(98.1808395,168.126297)
--
(98.7769623,167.968948)
--
(100.326263,167.968948)
--
(99.902626,168.961472)
--
(100.631882,170.43512)
--
(102.432343,171.031235)
--
(103.933228,170.565231)
--
(104.526314,169.385101)
--
(103.739563,167.968948)
--
(104.502106,167.968948)
--
(104.526314,167.908432)
--
(104.017952,166.492264)
--
(103.860603,166.492264)
--
(103.872704,166.746445)
--
(103.766792,167.0672)
--
(103.315926,167.145874)
--
(98.7164383,167.145874)
--
(98.2020264,167.018784)
--
(98.03862,166.455963)
--
(97.8691635,166.455963)
--
(103.279617,167.968948)
--
(103.721405,168.298782)
--
(103.933228,168.973572)
--
(103.415779,169.854126)
--
(102.032913,170.18396)
--
(100.716614,169.854126)
--
(100.217323,168.997772)
--
(100.429146,168.310883)
--
(100.858833,167.968948)
--
(103.279617,167.968948)
--
(99.9994583,171.612228)
--
(99.9994583,173.936172)
--
(100.144699,173.936172)
--
(100.299026,173.330978)
--
(100.822517,173.19783)
--
(106.692917,173.19783)
--
(106.717125,173.161514)
--
(106.28138,171.588013)
--
(106.124031,171.588013)
--
(106.124031,171.648529)
--
(106.148239,171.963242)
--
(106.01812,172.287018)
--
(105.543045,172.374771)
--
(100.858833,172.374771)
--
(100.320206,172.223465)
--
(100.144699,171.612228)
--
(99.9994583,171.612228)
--
(101.609276,178.202805)
--
(100.813446,177.497742)
--
(100.580444,176.677704)
--
(101.367195,175.431)
--
(102.722832,175.140503)
--
(102.722832,178.166489)
--
(103.464195,178.02124)
--
(103.957436,177.742844)
--
(104.526314,176.484039)
--
(103.851524,174.99527)
--
(102.105537,174.426376)
--
(100.495712,174.922638)
--
(99.902626,176.266174)
--
(100.326263,177.512878)
--
(101.548752,178.360153)
--
(101.609276,178.202805)
--
(103.037537,175.164719)
--
(103.890862,175.494553)
--
(104.1632,176.205658)
--
(103.924149,176.859268)
--
(103.037537,177.161865)
--
(103.037537,175.164719)
--
(100.979874,179.858)
--
(100.423088,179.313339)
--
(99.8905182,179.845901)
--
(100.423088,180.402679)
--
(100.979874,179.858)
--
(104.526314,183.579956)
--
(103.884811,182.015533)
--
(102.22657,181.413361)
--
(100.562286,182.021576)
--
(99.902626,183.567856)
--
(100.589523,185.120178)
--
(102.299194,185.746552)
--
(103.89994,185.135315)
--
(104.526314,183.579956)
--
(104.247925,183.458923)
--
(103.609444,184.469589)
--
(101.960289,184.862976)
--
(100.65004,184.545242)
--
(100.181015,183.676788)
--
(100.810417,182.732681)
--
(102.698624,182.296951)
--
(103.827316,182.608627)
--
(104.247925,183.458923)
--
(99.9994583,186.170197)
--
(99.9994583,188.530457)
--
(100.144699,188.530457)
--
(100.320206,187.86171)
--
(100.883041,187.695282)
--
(103.098053,187.695282)
--
(103.615494,187.934341)
--
(103.896912,188.385208)
--
(103.751663,188.639389)
--
(103.558006,189.038818)
--
(104.005852,189.414047)
--
(104.526314,188.88147)
--
(104.329628,188.318634)
--
(103.594315,187.695282)
--
(104.502106,187.695282)
--
(104.526314,187.646866)
--
(103.993744,186.194397)
--
(103.836395,186.194397)
--
(103.872704,186.53331)
--
(103.279617,186.872223)
--
(100.822517,186.872223)
--
(100.347443,186.757233)
--
(100.144699,186.170197)
--
(99.9994583,186.170197)
--
(103.812187,194.074051)
--
(104.199509,194.074051)
--
(104.199509,193.323608)
--
(104.296341,192.778931)
--
(104.368965,192.561066)
--
(104.526314,191.774307)
--
(104.063339,190.60022)
--
(102.916496,190.128174)
--
(102.147896,190.33696)
--
(101.597168,191.048065)
--
(101.209846,190.624435)
--
(100.532028,190.16449)
--
(100.011559,190.684952)
--
(99.3579483,189.943588)
--
(98.8132706,189.728745)
--
(98.1354523,190.221985)
--
(97.8570633,191.423294)
--
(98.4864655,193.396225)
--
(99.5153,193.989319)
--
(100.259689,193.668564)
--
(100.568336,192.803131)
--
(100.62886,191.532227)
--
(100.895142,190.757584)
--
(101.249184,190.954269)
--
(101.512444,191.350662)
--
(101.488235,191.532227)
--
(101.464027,191.90744)
--
(101.851349,192.936279)
--
(102.98912,193.444641)
--
(103.812187,193.263092)
--
(103.812187,194.074051)
--
(99.9752502,190.902817)
--
(99.8542099,192.488434)
--
(99.7452698,193.435562)
--
(99.3700485,193.710922)
--
(98.68013,193.196518)
--
(98.4138412,191.846924)
--
(98.6105347,190.805984)
--
(99.1400757,190.418671)
--
(99.4547806,190.500366)
--
(99.9752502,190.902817)
--
(103.328026,190.951233)
--
(104.002823,191.141876)
--
(104.247925,191.677475)
--
(103.787979,192.391602)
--
(102.601791,192.694199)
--
(101.95121,192.497513)
--
(101.706108,191.955856)
--
(102.144875,191.226608)
--
(103.291718,190.951233)
--
(103.328026,190.951233)
;

\end{tikzpicture}}

bbb

\end{document}
David Carlisle
  • 757,742
  • Sorry, this is not at all related to my question. I asked for a method to produce output that looks exactly like a normal email address, not some sort of TeX CAPTCHA. – Dmitri Pavlov Apr 06 '15 at 12:17
  • Since this point seems to be unclear, I added a clarification (in bold) to the original post. – Dmitri Pavlov Apr 06 '15 at 13:06
  • @DmitriPavlov yes that point was clear as I said in the answer, you can use the same technique but using the same font, I just used Times Roman for simplicity. – David Carlisle Apr 06 '15 at 19:26
2

Here's an even more "self-contained" way to do what David Carlisle suggested, made by following links from Can we extract the points making the character from the font file? and playing with the code found there. This compiled with pdflatex.

enter image description here

\immediate\write18{tex pgflibraryshapes.letters.dtx}
%\documentclass[border=10]{standalone}
\documentclass[landscape]{article}
\pagestyle{empty}
\thispagestyle{empty}
\usepackage{tikz}
\usetikzlibrary{
  shapes.letters,
  decorations.pathmorphing,
  decorations.pathreplacing,
  decorations.text,
  decorations.footprints,
  fadings,
}


\pgfmathsetmacro{\emcm}{1em/1cm}
\pgfkeys{
  /pgf/letter/.cd,
%  load font={stikz}{italic},
  load font={stikz}{normal},
  size=1,
  load encoding=char,
  every letter/.append style={
    fill,
  },
  sentence width=.9\textwidth,
}


\begin{document}
\begin{tikzpicture}[every letter/.append style={path fading=west},scale=.1]
\node[letter=h] (1) {};
\node[letter=o,right of previous letter,at=(1.east)] (2) {};
\node[letter=l,right of previous letter,at=(2.east)] (3) {};
\node[letter=t,right of previous letter,at=(3.east)] (4) {};
\node[letter=z,right of previous letter,at=(4.east)] (5) {};
\node[letter=e,right of previous letter,at=(5.east)] (6) {};
\node[letter=r,right of previous letter,at=(6.east)] (7) {};
\node[letter=m,right of previous letter,at=(7.east)] (8) {};
\node[letter=a,right of previous letter,at=(8.east)] (9) {};
\node[letter=n,right of previous letter,at=(9.east)] (10) {};
\node[letter=n,right of previous letter,at=(10.east)] (11) {};
\node[letter=1,right of previous letter,at=(11.east)] (12) {};
\node[letter=7,right of previous letter,at=(12.east)] (13) {};
\node[letter=@,right of previous letter,at=(13.east)] (14) {};
\node[letter=g,right of previous letter,at=(14.east)] (15) {};
\node[letter=m,right of previous letter,at=(15.east)] (16) {};
\node[letter=a,right of previous letter,at=(16.east)] (17) {};
\node[letter=i,right of previous letter,at=(17.east)] (18) {};
\node[letter=l,right of previous letter,at=(18.east)] (19) {};
\node[letter=.,right of previous letter,at=(19.east)] (20) {};
\node[letter=c,right of previous letter,at=(20.east)] (21) {};
\node[letter=o,right of previous letter,at=(21.east)] (22) {};
\node[letter=m,right of previous letter,at=(22.east)] (23) {};
\node[above=23] (joe) {Hi, I'm Joe};
\node[below=23] (joe) {Corneli, Phd};
\end{tikzpicture}

\end{document}
Community
  • 1
Joe Corneli
  • 4,340
  • \write18 is disabled by default on most distributions. This is as far from being portable (criterion 1 in my list, by the way) as possible. Also this totally fails with DVI, and arXiv produces DVI, so it is out of the question. – Dmitri Pavlov Apr 06 '15 at 12:20
  • What's worse, this solution requires an external file pgflibraryshapes.letters.dtx that is not even present in my distribution of TeX Live. The original post specifically requests that no external files can be used. – Dmitri Pavlov Apr 06 '15 at 12:50