77
\$\begingroup\$

After spending a while on Stack Exchange, I can recognise most sites in the Hot Network Questions by their little icon (which is also their favicon), but certainly not all of them. Let's write a program which can! You're to write code which determines the site, given one of the (currently) 132 favicons, including Stack Overflow in Japanese (which is still in private beta):

Academia Android Enthusiasts Anime & Manga Ask Different Arduino Ask Ubuntu Astronomy Aviation Video Production Beer Bicycles Biology Bitcoin Blender Board & Card Games Stack Overflow in Portuguese LEGO Answers Buddhism Chemistry Chess Chinese Language Christianity Programming Puzzles & Code Golf Code Review Cognitive Sciences Community Building Seasoned Advice Craft CMS Cryptography Computer Science Theoretical Computer Science Data Science Database Administrators
Home Improvement Drupal Answers Signal Processing Earth Science Ebooks Economics Electrical Engineering English Language Learners Emacs English Language & Usage Expatriates ExpressionEngine Answers Physical Fitness Freelancing French Language Game Development Arqade Gardening & Landscaping Genealogy & Family History German Language Geographic Information Systems Graphic Design Amateur Radio Biblical Hermeneutics Hinduism History Homebrewing History of Science and Math Islam Italian Language Stack Overflow in Japanese Japanese Language Joomla
Mi Yodeya Linguistics Magento Martial Arts Mathematics Mathematics Educators Mathematica MathOverflow Motor Vehicle Maintenance & Repair Meta Stack Exchange Personal Finance & Money Movies & TV Music, Practice & Theory Network Engineering Open Data The Great Outdoors Parenting Ask Patents Pets Philosophy Photography Physics Project Management Poker Politics Personal Productivity Programmers Puzzling Quantitative Finance Raspberry Pi Reverse Engineering Robotics Role-playing Games
Russian Language Salesforce Computational Science Science Fiction & Fantasy Information Security Server Fault SharePoint Skeptics Software Recommendations Sound Design Space Exploration Spanish Language Sports Software Quality Assurance & Testing Stack Apps Stack Overflow Startups Cross Validated Super User Sustainable Living TeX - LaTeX Tor Travel Tridion Unix & Linux User Experience Web Applications Webmasters Windows Phone WordPress Development The Workplace Worldbuilding Writers

I've uploaded a ZIP file with all these images as PNGs on GitHub. Click the "Raw" button to download it. The order of the above icons is the alphabetical order of the file names in the zip.

The corresponding site names (in this order) are:

Academia
Android Enthusiasts
Anime & Manga
Ask Different
Arduino
Ask Ubuntu
Astronomy
Aviation
Video Production
Beer
Bicycles
Biology
Bitcoin
Blender
Board & Card Games
Stack Overflow in Portuguese
LEGO Answers
Buddhism
Chemistry
Chess
Chinese Language
Christianity
Programming Puzzles & Code Golf
Code Review
Cognitive Sciences
Community Building
Seasoned Advice
Craft CMS
Cryptography
Computer Science
Theoretical Computer Science
Data Science
Database Administrators
Home Improvement
Drupal Answers
Signal Processing
Earth Science
Ebooks
Economics
Electrical Engineering
English Language Learners
Emacs
English Language & Usage
Expatriates
ExpressionEngine Answers
Physical Fitness
Freelancing
French Language
Game Development
Arqade
Gardening & Landscaping
Genealogy & Family History
German Language
Geographic Information Systems
Graphic Design
Amateur Radio
Biblical Hermeneutics
Hinduism
History
Homebrewing
History of Science and Math
Islam
Italian Language
Stack Overflow in Japanese
Japanese Language
Joomla
Mi Yodeya
Linguistics
Magento
Martial Arts
Mathematics
Mathematics Educators
Mathematica
MathOverflow
Motor Vehicle Maintenance & Repair
Meta Stack Exchange
Personal Finance & Money
Movies & TV
Music, Practice & Theory
Network Engineering
Open Data
The Great Outdoors
Parenting
Ask Patents
Pets
Philosophy
Photography
Physics
Project Management
Poker
Politics
Personal Productivity
Programmers
Puzzling
Quantitative Finance
Raspberry Pi
Reverse Engineering
Robotics
Role-playing Games
Russian Language
Salesforce
Computational Science
Science Fiction & Fantasy
Information Security
Server Fault
SharePoint
Skeptics
Software Recommendations
Sound Design
Space Exploration
Spanish Language
Sports
Software Quality Assurance & Testing
Stack Apps
Stack Overflow
Startups
Cross Validated
Super User
Sustainable Living
TeX - LaTeX
Tor
Travel
Tridion
Unix & Linux
User Experience
Web Applications
Webmasters
Windows Phone
WordPress Development
The Workplace
Worldbuilding
Writers

Notes:

  • I've removed the ® from LEGO® Answers and ExpressionEngine® Answers, so you don't have to worry about Unicode.
  • I've used English names for the Stack Overflows in Japanese and Portuguese for the same reason.
  • The icons of Earth Science and Spanish Language are indistinguishable. Therefore, given one of these icons, your code may return either of those sites (of your choosing). The same goes for Magento and Martial Arts.

Rules

You may write a program or function which

  • receives the (local) file name of the image via STDIN, command-line argument or function argument or receives the image file's contents via STDIN
  • returns or prints to STDOUT the name of the site, as listed above.

Your code must recognise all 132 sites correctly (with the exception mentioned above).

You may not make any assumptions about the file name (like that it's called codegolf.png). You may assume that the image has dimensions 16x16, and that it will indeed be one of the 132 images above. The above images are all PNGs, but you may use any other convenient raster graphics format, but you'll have to convert the images yourself. You should not make any assumptions about the actual byte stream of the image file, other than that it's a valid image in whatever format you chose. In particular, if there are multiple ways to encode the same image in your format (e.g. by adding irrelevant fields to the header section), your code must work for all of them. In short, your code should only rely on the pixel values themselves, no details of the file encoding it.

As usual, you may not fetch any data from the internet. You have to determine the site from the image alone.

You may use built-in or third-party functions to read the image file and get a list of colour values, but you must not use any other existing image processing functions.

This is code golf, so the shortest answer (in bytes) wins.

\$\endgroup\$
15
  • 5
    \$\begingroup\$ CRC32 is shorter than MD5. Anyone who knows an even shorter checksum algorithm will probably win. \$\endgroup\$
    – manatwork
    Commented Dec 8, 2014 at 13:14
  • 1
    \$\begingroup\$ @squeamishossifrage You may generate a hash from the pixel values. I don't know how flexible the PNG format is, but whatever format you choose, if there are two different files that encode the exact same image, your program has to work for both of them. So you might not be able to make any assumptions about the precise byte stream from the image file, other than it's a valid image in whatever format you chose. I should clarify that. \$\endgroup\$ Commented Dec 8, 2014 at 16:57
  • 1
    \$\begingroup\$ @MartinBüttner Yes there can be two different PNGs that encode the same image, otherwise pngcrush will not exist. \$\endgroup\$
    – kennytm
    Commented Dec 8, 2014 at 17:14
  • 1
    \$\begingroup\$ I feel that this could be done amazingly with a decision tree that analyses pixels at specific co-ordinates in the image... I'm sure there's some simple ML technique to identify the "key features" that can be used to distinguish the images... Time to dig out my ML notes! \$\endgroup\$ Commented Dec 8, 2014 at 21:51
  • 1
    \$\begingroup\$ @MarkKCowan Look at my solution, I basically do that. For every pixel, I make a list of the different color values it has in all the images. Then I pick the median value, and split the images on two depending on their color in that pixel. Then do the same recursively to the two halves until you're down to a single image. \$\endgroup\$ Commented Dec 9, 2014 at 13:39

5 Answers 5

53
\$\begingroup\$

C#, 2760 bytes

namespace System{class P{static void Main(){var c="us>g System;us>g System.Draw>g;class p{publ' stZ' void c(){var b=(Bitmap)Bitmap.FromFile(Console.ReadL>e());Console.Write(!1,1[3,3#2?!4,4#1?!5,5#2?!@#1?!+$SupK UsK*Salesforce%]^$< F't~_Fantasy%8,8#6?$Arqade*Ask Ubuntu%@#6?$WordPress Development%8,8#4?!9,9#2?$Phys's*UsK ExpKience%+$Travel*Craft CMS%5,5^!@#6?!]#1?$TeX - LaTeX%+$Mi Yodeya*Meta Stack Exchange%]#6?$ProgrammK|2,2#5?$Christianity*MZhemZ'a%@[]^$MZhemZ'|0,0#1?$Cross ValidZed*Theoret'al ComputK <%2,2^!]#5?$Seasoned Adv'e*Drupal AnswK|]^$Graph' Design*WebmastK|2,2^!4,4#5?!5,5#2?!@^$Ask PZents*Home Improvement%@#6?$Android Enthusiasts*Skept'|5,5#4?!@#5?$Web Appl'Z~s*English/_Usage%@#4?!]#8?$SharePo>t*B'ycle|]#1?$Electr'al Eng>eK>g*DZabase Adm>istrZor|4,4^!5,5#2?!+$Game Development*Photography%@:SKvK Fault%+$Geograph' InformZ~ Systems*Unix_L>ux%5,5^!0`Stack Apps*The Workplace%+!@[]#4?$InformZ~ Security*Ask DiffKent%]:Role-play>g Games*PKsonal F>ance_Money%3,9#8?$Academia%5,15^$& > Japanese%0,13:&*& > Portuguese%2,2:MZhOvKflow%3,3:AviZ~%4,4^$LEGO AnswK|5,5[@[][0,1:PokK*Anime_Manga%+$BlendK*H>duism%][8,8:WritK|0,1^!2`RaspbKry Pi*W>dows Phone%+$Biology*Worldbuild>g%9,9:Movies_TV%8,8[6`Trid~*Islam%0,1:Mus', Pract'e_Theory*L>guist'|][@[9,9:Ches|+!4,6:Garden>g_Landscap>g*Susta>able Liv>g%4,6:Buddhism*Startup|8,8:English/ LearnK|3,4[6,7:Network Eng>eK>g%9,5:Magento%+$Motor Veh'le Ma>tenance_Repair*MZhemZ's EducZor|4,8:GKman/%7`Joomla%7,6[+$QuantitZive F>ance*Ardu>o%9`Chemistry*Cryptography%8,8[0,1^!@[3`Tor%4,5:French/*The GreZ Outdoor|3`Project Management%3,6:Ch>ese/*Space ExplorZ~%+!2,6:Programm>g Puzzles_Code Golf%@:Cognitive <s*Signal Process>g%1,0:Ebook|@[4,5:Earth <*BeK%8,5:Emacs*ExpZriZe|1`History of < and MZh%3,4[@[+$Homebrew>g%3,6:DZa <*Video Product~%5,7:Russian/%9,4[4,8:Bibl'al HKmeneut's*Philosophy%11,6:Board_Card Games*Bitco>%3,5[@[3,8:AmZeur Radio%5,6:Genealogy_Family History%6,9:Software Quality Assurance_Test>g*Open DZa%3,6[11,7:Code Review%+$ComputK <*Community Build>g%10,6[+$Sports*Software RecommendZ~|9`Sound Design*ComputZ~al <%3,7:Japanese/%5,8[7,5:Puzzl>g%8,5:Econom's*Express~Eng>e AnswK|6,5[9,4[7`Freelanc>g*History%10,6:Italian/%6`Parent>g*Astronomy%6,7[9`RevKse Eng>eK>g*Robot'|10,8[+$Polit's*Pet|12,5:PKsonal Productivity*Phys'al Fitness$);}}";foreach(var s in "~ion;|s%;`,4:;&Stack OvKflow;_ & ;^#3?;]7,7;'ic;['!;Zat;Ker;@6,6;>in;<Science;:'$;/ Language;+0,0#3?;*$:$;'#7?;%$:!;$\";#).B%9<;!b.GetPixel(".Split(';'))c=c.Replace(s[0]+"",s.Substring(1));new Microsoft.CSharp.CSharpCodeProvider().CompileAssemblyFromSource(new CodeDom.Compiler.CompilerParameters("mscorlib.dll System.Core.dll System.Drawing.dll".Split()),c).CompiledAssembly.CreateInstance("p").GetType().GetMethod("c").Invoke(c,null);}}}

My solution doesn't use hashing, it actually examines the individual pixels of the image and decides based on that. I discovered that it was enough to examine the blue component of the image modulo 9. The idea is to repeatedly split based on the value of pixel.B%9, like so:

b.GetPixel(3,4).B % 9 < 7 ? "Russian Language" : (b.GetPixel(8,4).B % 9 < 6 ? "Project Management" : "Philosophy")

Using a script, I generated the following monstrous program (5197 bytes) that solves the problem using a binary decision tree:

using System;using System.Drawing;class p {public static void c(){var b=(Bitmap)Bitmap.FromFile(Console.ReadLine());Console.Write(b.GetPixel(4,5).B%9<6?b.GetPixel(1,1).B%9<5?b.GetPixel(1,10).B%9<1?b.GetPixel(3,4).B%9<1?b.GetPixel(2,8).B%9<2?b.GetPixel(0,0).B%9<3?b.GetPixel(3,10).B%9<8?"Super User":"Travel":b.GetPixel(0,2).B%9<3?"Salesforce":"Craft CMS":b.GetPixel(2,4).B%9<2?b.GetPixel(2,2).B%9<6?"Ask Ubuntu":"Ask Different":b.GetPixel(1,3).B%9<6?"Mathematics":"Programmers":b.GetPixel(2,7).B%9<3?b.GetPixel(2,4).B%9<1?b.GetPixel(2,8).B%9<2?"Database Administrators":"Christianity":b.GetPixel(0,1).B%9<3?"TeX - LaTeX":"Game Development":b.GetPixel(1,6).B%9<3?b.GetPixel(2,3).B%9<7?"English Language & Usage":"Skeptics":b.GetPixel(0,0).B%9<3?"StackApps":"Unix & Linux":b.GetPixel(0,6).B%9<1?b.GetPixel(1,4).B%9<2?b.GetPixel(1,5).B%9<3?b.GetPixel(0,4).B%9<1?"Arqade":"Mi Yodeya":b.GetPixel(1,6).B%9<6?"Drupal Answers":"User Experience":b.GetPixel(1,5).B%9<4?b.GetPixel(1,6).B%9<3?"SharePoint":"Bicycles":b.GetPixel(0,0).B%9<3?"Home Improvement":"Photography":b.GetPixel(0,0).B%9<3?b.GetPixel(0,7).B%9<2?b.GetPixel(0,1).B%9<1?"Science Fiction & Fantasy":"Theoretical Computer Science":b.GetPixel(0,4).B%9<2?"Android Enthusiasts":"Ask Patents":b.GetPixel(5,15).B%9<3?b.GetPixel(2,0).B%9<3?"Meta Stack Exchange":"Stack Overflow inJapanese":b.GetPixel(0,13).B%9<7?"Stack Overflow":"Stack Overflow in Portuguese":b.GetPixel(5,8).B%9<7?b.GetPixel(4,4).B%9<7?b.GetPixel(0,0).B%9<3?b.GetPixel(7,5).B%9<7?b.GetPixel(0,1).B%9<5?"German Language":"Server Fault":b.GetPixel(8,5).B%9<7?"Economics":"ExpressionEngine Answers":b.GetPixel(1,13).B%9<7?b.GetPixel(0,1).B%9<7?"Ebooks":"MathOverflow":b.GetPixel(7,5).B%9<7?"Buddhism":"Earth Science":b.GetPixel(6,6).B%9<7?b.GetPixel(4,6).B%9<7?b.GetPixel(3,6).B%9<7?"Poker":"Gardening & Landscaping":b.GetPixel(0,0).B%9<3?"Sustainable Living":"Startups":b.GetPixel(7,6).B%9<7?b.GetPixel(0,0).B%9<3?"Quantitative Finance":"Arduino":b.GetPixel(9,4).B%9<7?"Chemistry":"Cryptography":b.GetPixel(5,4).B%9<7?b.GetPixel(9,5).B%9<7?b.GetPixel(10,8).B%9<7?b.GetPixel(0,0).B%9<3?"Reverse Engineering":"Pets":b.GetPixel(12,5).B%9<7?"Personal Productivity":"Physical Fitness":b.GetPixel(6,7).B%9<7?b.GetPixel(3,4).B%9<7?"Tor":"Robotics":b.GetPixel(7,5).B%9<7?"Politics":"French Language":b.GetPixel(0,1).B%9<3?b.GetPixel(9,5).B%9<7?b.GetPixel(3,4).B%9<3?"LEGO Answers":"Magento":b.GetPixel(6,5).B%9<7?"Motor Vehicle Maintenance & Repair":"Network Engineering":b.GetPixel(2,6).B%9<7?b.GetPixel(0,0).B%9<7?"Programming Puzzles & Code Golf":"History of Science and Math":b.GetPixel(0,2).B%9<7?"Mathematics Educators":b.GetPixel(1,4).B%9<7?"Signal Processing":"Cognitive Sciences":b.GetPixel(3,6).B%9<7?b.GetPixel(4,6).B%9<5?b.GetPixel(5,8).B%9<5?b.GetPixel(6,5).B%9<7?b.GetPixel(9,8).B%9<7?b.GetPixel(1,1).B%9<7?"Geographic Information Systems":"Biblical Hermeneutics":b.GetPixel(11,6).B%9<7?"Board & Card Games":"Bitcoin":b.GetPixel(0,0).B%9<3?b.GetPixel(2,4).B%9<7?"RaspberryPi":"English Language Learners":b.GetPixel(8,5).B%9<7?"Emacs":"Expatriates":b.GetPixel(5,4).B%9<6?b.GetPixel(6,8).B%9<7?b.GetPixel(3,4).B%9<7?"Russian Language":"Amateur Radio":b.GetPixel(8,4).B%9<7?"Project Management":"Philosophy":b.GetPixel(0,7).B%9<7?b.GetPixel(1,1).B%9<7?"Cross Validated":"Homebrewing":b.GetPixel(0,1).B%9<7?"Web Applications":"Biology":b.GetPixel(3,4).B%9<7?b.GetPixel(0,2).B%9<3?b.GetPixel(1,7).B%9<6?b.GetPixel(1,6).B%9<6?"WordPress Development":"Electrical Engineering":b.GetPixel(1,1).B%9<7?"Physics":"Windows Phone":b.GetPixel(1,0).B%9<7?b.GetPixel(0,0).B%9<3?"Webmasters":"Academia":b.GetPixel(3,8).B%9<7?"Data Science":"Worldbuilding":b.GetPixel(6,6).B%9<7?b.GetPixel(5,6).B%9<7?b.GetPixel(0,1).B%9<7?"Genealogy & Family History":"Anime & Manga":b.GetPixel(6,9).B%9<7?"Software Quality Assurance & Testing":"Open Data":b.GetPixel(9,4).B%9<7?b.GetPixel(0,0).B%9<3?"Code Review":"Community Building":b.GetPixel(8,4).B%9<7?"Chinese Language":"Computer Science":b.GetPixel(5,6).B%9<6?b.GetPixel(4,4).B%9<7?b.GetPixel(9,8).B%9<7?b.GetPixel(10,6).B%9<7?b.GetPixel(0,0).B%9<3?"Sports":"Software Recommendations":b.GetPixel(1,1).B%9<7?"Information Security":"Sound Design":b.GetPixel(3,5).B%9<7?b.GetPixel(8,4).B%9<7?"Space Exploration":"Computational Science":b.GetPixel(0,0).B%9<3?"Tridion":"Puzzling":b.GetPixel(0,5).B%9<3?b.GetPixel(6,8).B%9<7?b.GetPixel(5,8).B%9<7?"Blender":"Writers":b.GetPixel(6,5).B%9<7?"Music, Practice & Theory":"Islam":b.GetPixel(0,1).B%9<3?b.GetPixel(0,4).B%9<7?"Mathematica":"The Workplace":b.GetPixel(0,0).B%9<3?"Linguistics":"Hinduism":b.GetPixel(6,4).B%9<5?b.GetPixel(9,8).B%9<7?b.GetPixel(3,7).B%9<7?b.GetPixel(0,0).B%9<3?"Japanese Language":"Aviation":b.GetPixel(7,6).B%9<7?"History":"Italian Language":b.GetPixel(9,4).B%9<7?b.GetPixel(5,3).B%9<7?"Chess":"Freelancing":b.GetPixel(0,0).B%9<3?"Parenting":"Beer":b.GetPixel(0,5).B%9<1?b.GetPixel(4,8).B%9<7?b.GetPixel(1,1).B%9<7?"Role-playing Games":"Movies & TV":b.GetPixel(6,8).B%9<7?"Astronomy":"The Great Outdoors":b.GetPixel(0,4).B%9<3?b.GetPixel(0,6).B%9<7?"SeasonedAdvice":"Personal Finance & Money":b.GetPixel(0,0).B%9<3?"Graphic Design":b.GetPixel(3,4).B%9<7?"Video Production":"Joomla");}}

Some people have used built-in compression functions in their solutions. I made my own, writing a script that identifies common substrings and replaces them by single-character shorthands. The code is compressed to the following 2502 bye string:

us@g System;us@g System.Draw@g;class p {publ| st^| void c(){var b=(Bitmap)Bitmap.FromFile(Console.ReadL@e());Console.Write(!4,5#6?!1,1#5?!1,10#1?!3`#1?!2,8#2?*!3,10#8?$Sup] Us]+Travel~0,2_$Salesforce+Craft CMS~2`#2?!2,2#6?$Ask Ubuntu+Ask Diff]ent~1,3#6?$M^hem^|s+Programm]s~2,7_!2`#1?!2,8#2?$D^abase Adm@istr^ors+Christianity~0,1_$TeX - LaTeX+Game Development~1,6_!2,3[English/:Usage+Skept|s'*$StackApps+Unix:L@ux~0,6#1?!1`#2?!1,5_!0`#1?$Arqade+Mi Yodeya~1,6#6?$Drupal Answ]s+Us] Exp]ience~1,5#4?!1,6_$SharePo@t+B|ycles'*$Home Improvement+Photography'*!0,7#2?!0,1#1?$> F|tion:Fantasy+Theoret|al Comput] >~0`#2?$Android Enthusiasts+Ask P^ents~5,15_!2,0_$Meta Stack Exchange+Stack Ov]flow @Japanese~0,13[Stack Ov]flow+Stack Ov]flow @ Portuguese~5,8<4`%*!7,5<0,1#5?$G]man/+S]v] Fault~8,5[Econom|s+ExpressionEng@e Answ]s~1,13<0,1[Ebooks+M^hOv]flow~7,5[Buddhism+Earth >~6,6<4,6<3,6[Pok]+Garden@g:Landscap@g'*$Susta@able Liv@g+Startups~7,6%*$Quantit^ive F@ance+Ardu@o~9`[Chemistry+Cryptography~5`<9,5<10,8%*$Rev]se Eng@e]@g+Pets~12,5[P]sonal Productivity+Phys|al Fitness~6,7<3`[Tor+Robot|s~7,5[Polit|s+French/~0,1_!9,5<3`_$LEGO Answ]s+Magento~6,5[Motor Veh|le Ma@tenance:Repair+Network Eng@e]@g~2,6<0,0[Programm@g Puzzles:Code Golf+History of > and M^h~0,2[M^hem^|s Educ^ors~1`[Signal Process@g+Cognitive >s~3,6<4,6#5?!5,8#5?!6,5<9,8<1,1[Geograph| Inform^ion Systems+Bibl|al H]meneut|s~11,6[Board:Card Games+Bitco@'*!2`[Raspb]ryPi+English/ Learn]s~8,5[Emacs+Exp^ri^es~5`#6?!6,8<3`[Russian/+Am^eur Radio~8`[Project Management+Philosophy~0,7<1,1[Cross Valid^ed+Homebrew@g~0,1[Web Appl|^ions+Biology~3`<0,2_!1,7#6?!1,6#6?$WordPress Development+Electr|al Eng@e]@g~1,1[Phys|s+W@dows Phone~1,0%*$Webmast]s+Academia~3,8[D^a >+Worldbuild@g~6,6<5,6<0,1[Genealogy:Family History+Anime:Manga~6,9[Software Quality Assurance:Test@g+Open D^a~9`%*$Code Review+Community Build@g~8`[Ch@ese/+Comput] >~5,6#6?!4`<9,8<10,6%*$Sports+Software Recommend^ions~1,1[Inform^ion Security+Sound Design~3,5<8`[Space Explor^ion+Comput^ional >'*$Tridion+Puzzl@g~0,5_!6,8<5,8[Blend]+Writ]s~6,5[Mus|, Pract|e:Theory+Islam~0,1_!0`[M^hem^|a+The Workplace'*$L@guist|s+H@duism~6`#5?!9,8<3,7%*$Japanese/+Avi^ion~7,6[History+Italian/~9`<5,3[Chess+Freelanc@g'*$Parent@g+Be]~0,5#1?!4,8<1,1[Role-play@g Games+Movies:TV~6,8[Astronomy+The Gre^ Outdoors~0`_!0,6[SeasonedAdv|e+P]sonal F@ance:Money'*$Graph| Design~3`[Video Production+Joomla$);}}

The dictionary needed for decompression is just 108 bytes:

: & ;<%!;|ic;`,4;_#3?;^at;]er;[%$;@in;>Science;~'!;/Language;+'$;*!0,0#3?;'$:;%#7?;$\";#).B%9<;!b.GetPixel(

The dictionary uses semicolons as delimiter, and contains single characters followed by their decompression. So to decompress, ":" would first be replaced by " & ", then "<" by "%!", "|" by "ic", and so on. Decompression of a string c can be expressed in quite a concise way:

foreach (var s in "[dictionary]".Split(';')) c = c.Replace(s[0] + "", s.Substring(1));

Then, after decompression I use some reflection black-magic to compile the code on the fly and run it:

namespace System
{
    using Collections.Generic;
    using CodeDom.Compiler;
    using Microsoft.CSharp;
    using Linq;
    using Reflection;
    class P
    {
        static void Main()
        {
            var c = "us@g System;us@g System.Draw@g;class p {publ| st^| void c(){var b=(Bitmap)Bitmap.FromFile(Console.ReadL@e());Console.Write(!4,5#6?!1,1#5?!1,10#1?!3`#1?!2,8#2?*!3,10#8?$Sup] Us]+Travel~0,2_$Salesforce+Craft CMS~2`#2?!2,2#6?$Ask Ubuntu+Ask Diff]ent~1,3#6?$M^hem^|s+Programm]s~2,7_!2`#1?!2,8#2?$D^abase Adm@istr^ors+Christianity~0,1_$TeX - LaTeX+Game Development~1,6_!2,3[English/:Usage+Skept|s'*$StackApps+Unix:L@ux~0,6#1?!1`#2?!1,5_!0`#1?$Arqade+Mi Yodeya~1,6#6?$Drupal Answ]s+Us] Exp]ience~1,5#4?!1,6_$SharePo@t+B|ycles'*$Home Improvement+Photography'*!0,7#2?!0,1#1?$> F|tion:Fantasy+Theoret|al Comput] >~0`#2?$Android Enthusiasts+Ask P^ents~5,15_!2,0_$Meta Stack Exchange+Stack Ov]flow @Japanese~0,13[Stack Ov]flow+Stack Ov]flow @ Portuguese~5,8<4`%*!7,5<0,1#5?$G]man/+S]v] Fault~8,5[Econom|s+ExpressionEng@e Answ]s~1,13<0,1[Ebooks+M^hOv]flow~7,5[Buddhism+Earth >~6,6<4,6<3,6[Pok]+Garden@g:Landscap@g'*$Susta@able Liv@g+Startups~7,6%*$Quantit^ive F@ance+Ardu@o~9`[Chemistry+Cryptography~5`<9,5<10,8%*$Rev]se Eng@e]@g+Pets~12,5[P]sonal Productivity+Phys|al Fitness~6,7<3`[Tor+Robot|s~7,5[Polit|s+French/~0,1_!9,5<3`_$LEGO Answ]s+Magento~6,5[Motor Veh|le Ma@tenance:Repair+Network Eng@e]@g~2,6<0,0[Programm@g Puzzles:Code Golf+History of > and M^h~0,2[M^hem^|s Educ^ors~1`[Signal Process@g+Cognitive >s~3,6<4,6#5?!5,8#5?!6,5<9,8<1,1[Geograph| Inform^ion Systems+Bibl|al H]meneut|s~11,6[Board:Card Games+Bitco@'*!2`[Raspb]ryPi+English/ Learn]s~8,5[Emacs+Exp^ri^es~5`#6?!6,8<3`[Russian/+Am^eur Radio~8`[Project Management+Philosophy~0,7<1,1[Cross Valid^ed+Homebrew@g~0,1[Web Appl|^ions+Biology~3`<0,2_!1,7#6?!1,6#6?$WordPress Development+Electr|al Eng@e]@g~1,1[Phys|s+W@dows Phone~1,0%*$Webmast]s+Academia~3,8[D^a >+Worldbuild@g~6,6<5,6<0,1[Genealogy:Family History+Anime:Manga~6,9[Software Quality Assurance:Test@g+Open D^a~9`%*$Code Review+Community Build@g~8`[Ch@ese/+Comput] >~5,6#6?!4`<9,8<10,6%*$Sports+Software Recommend^ions~1,1[Inform^ion Security+Sound Design~3,5<8`[Space Explor^ion+Comput^ional >'*$Tridion+Puzzl@g~0,5_!6,8<5,8[Blend]+Writ]s~6,5[Mus|, Pract|e:Theory+Islam~0,1_!0`[M^hem^|a+The Workplace'*$L@guist|s+H@duism~6`#5?!9,8<3,7%*$Japanese/+Avi^ion~7,6[History+Italian/~9`<5,3[Chess+Freelanc@g'*$Parent@g+Be]~0,5#1?!4,8<1,1[Role-play@g Games+Movies:TV~6,8[Astronomy+The Gre^ Outdoors~0`_!0,6[SeasonedAdv|e+P]sonal F@ance:Money'*$Graph| Design~3`[Video Production+Joomla$);}}";
            foreach (var s in ": & ;<%!;|ic;`,4;_#3?;^at;]er;[%$;@in;>Science;~'!;/ Language;+'$;*!0,0#3?;'$:;%#7?;$\";#).B%9<;!b.GetPixel(".Split(';')) c = c.Replace(s[0] + "", s.Substring(1));

            var o = new CSharpCodeProvider().CompileAssemblyFromSource(new CompilerParameters("mscorlib.dll System.Core.dll System.Drawing.dll".Split()), c).CompiledAssembly.CreateInstance("p").GetType().GetMethod("c");
            o.Invoke(o, null);
        }
    }
}

Note that the examples used here for explanation are slightly different from the one used in the 2876 byte solution.

\$\endgroup\$
10
  • 2
    \$\begingroup\$ I've edited your post into our usual format, I hope you don't mind! :) \$\endgroup\$ Commented Dec 8, 2014 at 12:17
  • 2
    \$\begingroup\$ I think, you can remove namespace to save some bytes. \$\endgroup\$
    – Qwertiy
    Commented Dec 8, 2014 at 12:52
  • 31
    \$\begingroup\$ Mildly interesting: i.imgur.com/MNnndMd.png - picture showing which pixels (highlighted in red) in each icon are used to identify it in my program. \$\endgroup\$ Commented Dec 8, 2014 at 16:58
  • 2
    \$\begingroup\$ @MartinBüttner No, that wouldn't make a difference. The non-beta sites are recognized by a pixel or two, but those pixels are only examined after recognizing a bunch of other sites first. Changing the order of comparisons wouldn't affect the code length. The code uses a total of 129 comparisons which optimal given that there are 130 images. \$\endgroup\$ Commented Dec 8, 2014 at 18:48
  • 8
    \$\begingroup\$ Good lord. The world needs more people like you. \$\endgroup\$
    – ldam
    Commented Dec 9, 2014 at 17:22
30
\$\begingroup\$

Python 3.x + Pillow, 2301 1894 1878 bytes

import sys,re,zlib as z,base64,PIL.Image as i
print(re.search(str(z.adler32(i.open(sys.argv[1]).tobytes()[2::6])%2003)+'(\\D+)',z.decompress(base64.b85decode(b'c${Tc%XZ>86#NyfdNqBH<%b?iLI_EpFkwO-GfQC$tcfjoB$+h*^;0s?o$g*?uzgjxs%~BKif!C3zALZEU!&?wJd#-mqgS@p<P&+)AsV|c*52*C>U&(1_4w^urvvWJU8Bj;bvv4ia;eiJLu+bsXLru`DjH|W<`|;xLq++kv5k8VWZgQe*`_u+opEF8>A6qjM%k#skzbiuJ7dL_*><K@9l6!M*H(|w)PZJ2`RYy7K2VVrv+zc)&2Fc?wvh|^#ksyymE?D!Q$3!LbFX%h%vT%1#lr486Iybw?0!`H)ZlZ#XIao2uXTr}(IBl@wRWHa#})bgV);O3HoW7+T*!5M3=^mu6Ri$`7L;uUWj~gsqR6gQR9h7^nKdBIgy>ats36%|M|8p1x;32(ZfK9R5_xK5ts=Sz70c%SIK&g3N1E|sg{n3>JSbKwwhssGcb$8uMNw=Nv;f_;rj?}kz(3-&#FQw@O;q(8nGHjbrTC(NQv=ckXT|eHr|amahCq(7T6<LBTr5Gj>fHV~3DcYINN!Qr`6Dl?jT>!4u5~bb%Zi6VTax;%vg*m325J_$nEC%yqbbYeEp!K_K7k)W-XCP`6+}N7tktMVX+{@c2L&kLAc&IvspGrzZ$F`uOBg}fA3W2neAU}tg~<+iCN=<sR^Zke+9WNhnCyy6!8nfa>33&OWiPt2s=OMTrUgilSYQxrG!gziB!Aj83zX&6S|yjIQkA?iF0n%?<+B=%_D11aoCv?@Q&eei{sqNkr7RG`#*xyzSUE6BUNnZK)3|3Wls}K$qaKv;jOO#!{3Ms)4CQ?shafnAF^zVq(`K9o2xL5JoQoGea+W1$+xmw&8OMOtN1Pid6UAcd+*_bB6IY$KjrO=ac^tPv3(8O`&(agOBbT{Va6s}Rm9&`I#yivC;MhW8=n+^^&Reho;~!~W<xgH6bVq3^=H7)sUXV2nv~DP4`CY`f55i&qH)DvPMQy>)y*Z=|68w@v=#oo5^RSh5R2F%1-vt>h=|+1n0pE^Yq-72!)1+o{hD_3<Cyxpyr6r%cei-AFQ*h1seM$MP>ovLS2k#Ci$to#do$s3MXu2lZsw~9HkbjVWM@1!{w>}LWVj#j2qC(+*Okc>=<qcV}*}+U_QK+(ZZ(x>ZMdDDWtgHiBK+7p6aHHI`u-h6z7|Xm`>D1t4uEp271y7oCI^{1^H~O7QRxTA_U&92Q8z$J2O|5Lbeg_0umd^pSU<%-ki7LtDvo&AgIt0h~6(y@Hr+kyLJ$<B#07KdGob0EW$7@f{GszaUv#tkpSyk759Khmvq-aiyM-_&x_VCTx$Vy<;$<k{T$-@{Mhc<<%KD1AHK^MKkg%pYIW}SxhV?&a{o9=rs@*+hqyoioE=TF>9S<KxY+^69^OA0jkS#0!w@kw4}KTcP9#+PVW+qB|7B`Y?eRs+7}8RSAs?i?-Cz#Wp(Iihd`nI!Q_i%R^o_;fvDQ5|{Cr;yWeirL>_l#(wn)$QenJkK6Sgw)5NleBzvo&G#@>X`7Ib*aS4LHx9epI%BXXV&y7LJ&54g}D25-Q8>DZQ5hGKyIeJkQQu08%j|<x~)q(AOVw9P~Z*c0Z@VM6n#<}tBT6+T%4mZ9f*y76xlEFM~o$xr9*SZHd^6D-C!pkkp3m(-~W3&>KY@106L+;I><MJ^Dz`f{(I3i?u`2pWjUjlwZx`B1JB#h#u3xTH*b1K?TG40|4lxF1$_Pw+t?h2')).decode()).group(1))

The idea is hash the image and find the corresponding string from a dictionary (just like other answers).

The key code is this:

adler32(open(sys.argv[1]).tobytes()[2::6])%2003

We open the file, and then convert it into a 1024-byte string of RGBA values. With a bit of experimentation, we find that the ADLER-32 checksum of every sixth byte of this byte string is unique for these 132 images. And then further testing shows that, taking modulus of 2003 of the checksum gives the smallest dictionary.

The original dictionary looks like:

{
    3: 'Programming Puzzles & Code Golf',
    6: 'Science Fiction & Fantasy',
    7: 'Bitcoin',
    8: 'Biblical Hermeneutics',
    …,
    1969: 'Unix & Linux',
    1989: 'WordPress Development',
    1993: 'Cognitive Sciences',
    2001: 'Personal Productivity',
}

We noticed that all site names do not contain numbers. Therefore we could concatenate the whole dictionary into a single string:

"256Software Quality Assurance & Testing3Programming Puzzles & Code Golf1284Geographic Info…"

And then use the regex e.g. 1969(\D+) to extract the site name. This huge string is then compressed to save space (let the compression engine notice the multiple occurrences of "Language"), and finally base-85 encoded.


Since it is now also tagged as , here a 2394-byte solution that does not use compression (zlib is still imported for adler32).

import sys,re,zlib,PIL.Image as i
l=' Language'
s='Science'
print(re.search(str(zlib.adler32(i.open(sys.argv[1]).tobytes()[2::6])%2003)+'(\\D+)','256Software Quality Assurance & Testing3Programming Puzzles & Code Golf1284Geographic Information Systems517Windows Phone6'+s+' Fiction & Fantasy7Bitcoin8Biblical Hermeneutics1034Writers1803Ask Different781Joomla597Islam1809Craft CMS878English'+l+' & Usage431Freelancing1565Politics32LEGO Answers289Physical Fitness1316Spanish'+l+'1830Database Administrators296Pets1066Philosophy1582Ask Patents560Cryptography1073Magento1588MathOverflow1333Salesforce1591Linguistics1844Stack Apps948Japanese'+l+'572Electrical Engineering778Genealogy & Family History735Sound Design63Open Data505Tridion328Arqade1609Hinduism588Music, Practice & Theory1101Expatriates846Network Engineering79German'+l+'1104Webmasters1208Signal Processing118Puzzling867ExpressionEngine Answers654Buddhism1593Quantitative Finance1625Parenting859Biology1117Academia1118Stack Overflow353Project Management1123Motor Vehicle Maintenance & Repair612Chinese'+l+'876Startups877Video Production110Reverse Engineering701French'+l+'677Bicycles1683Ebooks1908Blender190Physics1654Beer119History of '+s+' and Math1913Android Enthusiasts634Chemistry1252Travel1148Cross Validated1662Italian'+l+'565Sustainable Living387Programmers797Arduino1932Data '+s+'141Server Fault1422Chess109Amateur Radio147Computational '+s+'917Home Improvement1599Worldbuilding1438Mi Yodeya197Christianity1185Community Building1956Aviation1701Poker1703Mathematics Educators1704SharePoint425Meta Stack Exchange171Code Review1964Game Development1967Homebrewing944User Experience1969Unix & Linux1715Board & Card Games1716Graphic Design969Tor696Economics955Drupal Answers1213Raspberry Pi958The Great Outdoors885Mathematics1731Emacs1989WordPress Development910Photography1736Space Exploration1993Cognitive '+s+'s1227Computer '+s+'1230Mathematica207Gardening & Landscaping2001Personal Productivity1235Super User1358Information Security727Theoretical Computer '+s+'1497Earth '+s+'220Russian'+l+'1914Role-playing Games1758Movies & TV479Anime & Manga67English'+l+' Learners228Web Applications485Robotics593The Workplace232Astronomy1519Personal Finance & Money808Stack Overflow in Japanese1266Seasoned Advice467Software Recommendations1012Sports332Stack Overflow in Portuguese761History763Ask Ubuntu1957Skeptics85TeX - LaTeX').group(1))
\$\endgroup\$
5
  • \$\begingroup\$ @MartinBüttner: .tobytes() gets the pixel data (16×16×4 = 1024), so it relies on the image. The file needs to be in RGBA though. \$\endgroup\$
    – kennytm
    Commented Dec 8, 2014 at 17:12
  • \$\begingroup\$ Ah, great. I thought it would return the actual byte stream of the file. As far as I'm aware the PNG images I uploaded don't have an alpha channel, though? \$\endgroup\$ Commented Dec 8, 2014 at 17:15
  • \$\begingroup\$ Oh looks like they do... I guess I messed that up. \$\endgroup\$ Commented Dec 8, 2014 at 17:17
  • \$\begingroup\$ Pillow? isn't it PIL? \$\endgroup\$
    – TheDoctor
    Commented Dec 10, 2014 at 17:26
  • 1
    \$\begingroup\$ @TheDoctor: Pillow is a PIL fork. PIL is unmaintained and doesn't work on Python 3. \$\endgroup\$
    – kennytm
    Commented Dec 10, 2014 at 19:05
15
\$\begingroup\$

Node.js, 3178 3130 2667 2608 Bytes

Computes the SHA1 hash of each file's image data and using bytes 16 through 19 of the hex digest, indexes the site names.

Using bytes 12 through 16 of the hex digest of each file's SHA1 hash, indexes the site names. There may be a shorter combination using only 3 bytes of the hex digest.

function $(e){r=require;r("pngparse").parseFile(e,function(e,t){console.log("Academia;Android Enthusiasts;Anime & Manga;Ask Different;Arduino;Ask Ubuntu;Astronomy;Aviation;Video Production;Beer;Bicycles;Biology;Bitcoin;Blender;Board & Card Games;Stack Overflow in Portuguese;LEGO Answers;Buddhism;Chemistry;Chess;Chinese Language;Christianity;Programming Puzzles & Code Golf;Code Review;Cognitive Sciences;Community Building;Seasoned Advice;Craft CMS;Cryptography;Computer Science;Theoretical Computer Science;Data Science;Database Administrators;Home Improvement;Drupal Answers;Signal Processing;Earth Science;Ebooks;Economics;Electrical Engineering;English Language Learners;Emacs;English Language & Usage;Expatriates;ExpressionEngine Answers;Physical Fitness;Freelancing;French Language;Game Development;Arqade;Gardening & Landscaping;Genealogy & Family History;German Language;Geographic Information Systems;Graphic Design;Amateur Radio;Biblical Hermeneutics;Hinduism;History;Homebrewing;History of Science and Math;Islam;Italian Language;Stack Overflow in Japanese;Japanese Language;Joomla;Mi Yodeya;Linguistics;Magento;Martial Arts;Mathematics;Mathematics Educators;Mathematica;MathOverflow;Motor Vehicle Maintenance & Repair;Meta Stack Exchange;Personal Finance & Money;Movies & TV;Music, Practice & Theory;Network Engineering;Open Data;The Great Outdoors;Parenting;Ask Patents;Pets;Philosophy;Photography;Physics;Project Management;Poker;Politics;Personal Productivity;Programmers;Puzzling;Quantitative Finance;Raspberry Pi;Reverse Engineering;Robotics;Role-playing Games;Russian Language;Salesforce;Computational Science;Science Fiction & Fantasy;Information Security;Server Fault;SharePoint;Skeptics;Software Recommendations;Sound Design;Space Exploration;Spanish Language;Sports;Software Quality Assurance & Testing;Stack Apps;Stack Overflow;Startups;Cross Validated;Super User;Sustainable Living;TeX - LaTeX;Tor;Travel;Tridion;Unix & Linux;User Experience;Web Applications;Webmasters;Windows Phone;WordPress Development;The Workplace;Worldbuilding;Writers".split(";")[x=r("crypto").createHash("sha1").update(t.data).digest("hex").slice(16,20),x=="c7da"?27:"f38c4a899f50c7d09a0f9dde6e05bbaebb274fc59f1ba9c4b8c315490f3d887b55f27c526f6f79c7db63f12feccafebf80fde98b13f6f6dbb8e59d6b97a09fe8cb0101f6abc8c963004f2f1aafbe216457ef8210ea104d841381d529cdd2697cf809f3d5cb0f1d91d970a981c41fe617d62c01284afdbbf78588490de40a56a0c7e613480354d13a738a84fc0bf2433d2351d585039c9795f8167997d15c1b2c8144ca40223794e66864608e305d5271faddc88177ef9d44bbd4c752b9d938ec92af827550fb".match(/.{3}/g).indexOf(x.slice(0,3))])})}
\$\endgroup\$
6
  • \$\begingroup\$ The word “Language” is used 9 times. You could use a variable for it. \$\endgroup\$
    – manatwork
    Commented Dec 8, 2014 at 13:41
  • \$\begingroup\$ Didn't observe the site names. Thanks. \$\endgroup\$
    – c.P.u1
    Commented Dec 8, 2014 at 14:01
  • 6
    \$\begingroup\$ You can reduce your table's size by 446 bytes. Switch to a CSV format: h="17352368".match(/.{4}/g) (the 4 character hashes, split into an array), s="MathOverflow;StackOverflow in Portuguese".split(";") (names separated by a ;) and then put it all together again: t={}h.forEach(function(k,i){t[k]=s[i]}) (results in the same object as your code). There are 132 semicolons, so even if you switch to a 2-byte character (OP says there should be no unicode in the site names), you'll save space. Also, you can add @manatwork 's optimisation on top of that additional saving. \$\endgroup\$
    – RickN
    Commented Dec 8, 2014 at 14:36
  • \$\begingroup\$ You can shorten it further by sparing the split()s: us a single string like "1234Site;5678Other". Then, supposing there will be no clash between hash fragments and site names, a single match() will do it: function $(e){r=require;return"8d4fAcademia;3a6dAndroid Enthusiasts;5caeAnime & Manga;804cAsk Different;bef3Arduino".match(r("crypto").createHash("sha1").update(r("fs").readFileSync(e)).digest("hex").slice(12,16)+"([^;]+)")[1]} \$\endgroup\$
    – manatwork
    Commented Dec 8, 2014 at 15:46
  • \$\begingroup\$ @MartinBüttner, the initial post did not mention that file hashing isn't allowed. I was hashing the file but have now updated my post to hash the image's data instead. \$\endgroup\$
    – c.P.u1
    Commented Dec 9, 2014 at 7:53
5
\$\begingroup\$

Python 2.7, 1906 1889 bytes

This solution uses CRC32 on the pixel data to make a unique 2 digit base 95 identifier. The index of the identifier is then used to look up the answer string.

The tricky bit was finding a combination of hash-style functions that would result in 132 (or 131) small but unique labels. I tried a few options before settling on this one. It seems quite compact.

The program uses Python PIL to read the pixel data from the file.

import sys,re,zlib as Z,base64 as B,PIL.Image as P
g=lambda p:chr(32+Z.crc32(p)%95)
D=P.open(sys.argv[1]).tostring()
print Z.decompress(B.b64decode('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')).split('\n')[re.findall('..','''<3 JlXAa}J>33:7XotUSZDHC }=c4G]wdaIwhi)Xa(h6B"jdfcQ_0^\wLo{Iyn7;XaRco&w7Ht(/]v^$u~?0:r,!?*y``7>5j%\SAq`G{v@&'#LuZn7)['#JTY d4(6mR%ljnE@/s6s6A-[&-\kR*[C$3*qgY25!vy&~>0H<TEGV.N$K_y|m5O#)nkhwg=E+CYmW<c /BQgFHgujMCm`793uA)T.&'f$vFRh zXin[3q:y\CGH.!hAw=k5v!gCd4g0]mU/'O''').index(g(D)+g(D*3))]

Python 2.7 2150 bytes

This is a version without using compression or encoding libraries. The Stack Exchange List is compressed with a simple exchange method. The characters that are not used in the text:

`~!@#$%^*()_+=[]{}:;"<>?/0123456789

are used to hold common string segments. The text is uncompressed with the for k,v in [(v[0],v[1:]) for v in K.split('|')]:T=T.replace(k,v) line. The two character indexing table is the same as in the above program.

import sys,re,zlib as Z,PIL.Image as P
g=lambda p:chr(32+Z.crc32(p)%95)
D=P.open(sys.argv[1]).tostring()
K='~ Engineer?|%]ematic|^*{|)Computer @|[ Develop3|" Answ7|! Language|@Science|#Software|$ Design|*Stack |(Japanese|_History|+Programm|=English|` & |<tion|>raph|?ing|/Ask |]Math|{Overflow|}tics|:logy|;ian|0Data|1Game|2ance|3ment|4Home|5ard|6ity|7ers|8Exp|9The'
T='Academia|Android Enthusiasts|Anime`Manga|/Different|Arduino|/Ubuntu|Astronomy|Avia<|Video Produc<|Beer|Bicycles|Bio:|Bitcoin|Blender|Bo5`C5 1s|^ in Portuguese|LEGO"|Buddhism|Chemistry|Chess|Chinese!|Christ;6|+? Puzzles`Code Golf|Code Review|Cognitive @s|Commun6 Build?|Seasoned Advice|Craft CMS|Cryptog>y|)|9oretical )|0 @|0base Administrators|4 Improve3|Drupal"|Signal Process?|Earth @|Ebooks|Economics|Electrical~|=! Learn7|Emacs|=!`Usage|8atriates|8ressionEngine"|Physical Fitness|Freelanc?|French!|1[|Arqade|G5en?`Landscap?|Genea:`Family _|German!|Geog>ic Informa< Systems|G>ic$|Amateur Radio|Biblical Hermeneu}|Hinduism|_|4brew?|_ of @ and ]|Islam|Ital;!|^ in (|(!|Joomla|Mi Yodeya|L?uis}|Magento|Martial Arts|%s|%s Educators|%a|]{|Motor Vehicle Mainten2`Repair|Meta *Exchange|P7onal Fin2`Money|Movies`TV|Music, Practice`9ory|Network~|Open 0|9 Great Outdoors|Parent?|/Patents|Pets|Philosophy|Photog>y|Physics|Project Manage3|Poker|Poli}|P7onal Productiv6|+7|Puzzl?|Quantitative Fin2|Raspberry Pi|Rev7e~|Robo}|Role-play? 1s|Russ;!|Salesforce|Computa<al @|@ Fic<`Fantasy|Informa< Secur6|Server Fault|SharePoint|Skep}|# Recommenda<s|Sound$|Space 8lora<|Spanish!|Sports|# Qual6 Assur2`Test?|*Apps|^|Startups|Cross Validated|Super User|Sustainable Liv?|TeX - LaTeX|Tor|Travel|Tridion|Unix`Linux|User 8erience|Web Applica<s|Webmast7|Windows Phone|WordPress[|9 Workplace|Worldbuild?|Writ7|'
for k,v in [(v[0],v[1:]) for v in K.split('|')]:T=T.replace(k,v)
print T.split('|')[re.findall('..','''<3 JlXAa}J>33:7XotUSZDHC }=c4G]wdaIwhi)Xa(h6B"jdfcQ_0^\wLo{Iyn7;XaRco&w7Ht(/]v^$u~?0:r,!?*y``7>5j%\SAq`G{v@&'#LuZn7)['#JTY d4(6mR%ljnE@/s6s6A-[&-\kR*[C$3*qgY25!vy&~>0H<TEGV.N$K_y|m5O#)nkhwg=E+CYmW<c /BQgFHgujMCm`793uA)T.&'f$vFRh zXin[3q:y\CGH.!hAw=k5v!gCd4g0]mU/'O''').index(g(D)+g(D*3))]
\$\endgroup\$
0
\$\begingroup\$

C#, 2672 bytes

using System.IO;using System.IO.Compression;using System.Security.Cryptography;using System.Text;namespace l
{internal class Program
{private static void Main(string[]args)
{string m="H4sIAAAAAAAEAGVWy3bbNhT8Fayyas8hxTd2IAk66bEa1XKSdgmRsISaBFgAtKN8fQeU5dgnKxHAxcV9zMxVmXasF4OclKBpkzE9 WKMGwrU/LU4J5x1t8phpNUnygWyFPgrKupS5R9KqhwdppfY0iltmh0VpQ+OyDWdfDov2C43h3XlrtJnONOo69qSEV0bTbBN/VYM0 ZGfNsPSXPdbVUloaNUWt+nM/SkfLNK+VGc3xTOu2qpXvjdK02hT1KPUA46ZsaiPsgOCa8HMjJlyra7b3on8kn5+kfRjNM1Ga7Iz1 y3GRTtIky275zWfCtHuW1tEs6eplGE7KTbSo4+aEeiDuM23jBgvnaJSXzUlpXCa3KMIijhJJ4NDCUAmt/JkmmxjpHK2YJqWPZLf8 +IEcQmRmkOTGjA+0ZZt1cSeflHymXVk35ojL6kmSfa+k7hF9lWSNmaYlOCX1osYB7mgeJXspnNFyIGx4Ur2kXdE0Vjx40mz3cB01 9jz7EMB8OlO4hpd58dJeXdO8iu5P0ljpVS9G8st5G8et8OJ1nURFWB8E0mYDsgpVEd6gZG1efTQAxadptuZJTgEHLE5au8zwfC1s xLO9OmrsoDLIzYVMSs65sP70+kxVb/jBmEdHGWO8D3BRPboSx3yUvbdrsFwfUX9pg4d0w7AclTu9doPcSmF1eJNXJZ8E7hdZ9ovV B/LFhd6lPOffZwHfwgectRxLGwI0+vLSaw5dl+xOZ7cG0SmvAxx4XXdWylHoPsSTtQWWuv/5EIAZByySVj7J0cxrfaKyZvY/sI12 PLoBXKUOQPkQbg2uF3PwFSfljdRSBNDjqBOTGs/kIypvAEjO+I20k9A/X0pKbF26rnryST8YnAdGkf3ZeTmhC5vo5uW4lQ79oG3X MBjJxZI7MShDU1bW6jCuSX7EA4hg8aEJRcc+Kg1ygxoo2jWOKstC+w9WPq8NafjLCTEP174SZAXJ8KeQ+Cc3iolGrPzkxajexl9v 4l/J+oeYRWAb5WV7/X5zJU/+MGYaoVlRtFXkH3DqLGjeFbeIZgmU7INw1VtYa28unxZEBTQtNC2L8hCYDIWCZVpUb5aEQ5IuKM9Y /PNA0Kwsw/IaKM052xoYkq8SxR0lslXaSy1C9h/A8lkoi8eSrQysWrPk3/sT8kDfWLMDvoxecXW9swXBoZVVszWQiKAe919Bo24L Oe5/A40E1HK1XJl8pqzc/Cn9s7GP7yjS8fjzLDUJ/KUNq2BNbqwUnnxe/GBCclXNdyIIeLBP8iwI9w6g0ChQ3eU7GQpVxbuTGo0z QVMYr3Yn8yoxLIovzIBdmoHh/4KtYUig6ivgNyXfmUfINE/znRnVS7Hj17yv8v8U9LPM2qt+Bt61cbZK6Cp+dffXIhCpF6tWvtSL Vml2J9x8kBbI2yla5DWkFbflu2LEXXZnDmZ9PmniO4NW/U7mUZwD/15mRlHcLeD/W2jmbbwX0HBQalXH5KKYK7sQ/VW/eFxdId+p dZitxNVeuDC32neclP1iQ7Y17/bSIlZYLqOnaZrvT2jHDhPO06qr9o9yXiOuKr43D/4Zh4BUj9GA2bd6Q5Gidm8W8OyF2AXL97NA IJCz0djLtG0ajk39Vggp26T72QQucJ6+ukeRxzB1mHOLfUHkvXQrQuq4vSCYzTMeTpL3rKVdyrGDGYvTkrPGGufIV/hDrHKgeVLv lxnpfnEAxCZL94vzoIs4oBm3QMCqffxe/o3W3Ar8QpWre2MhVu29FZBRWhTFvVVDyKmrmi9afQ/iqfTyPSAhOA55o+drWzLefpOH EG7QtUu92CbG3oQ/NuuoaJtvEDfz7AhwrcNIz74ZO+zCJHgn3W2dBAbh8BGwgfMy5liMw+E6n7Ok+obGwu3/ajKqV1QJAAA=";string f=BitConverter.ToString(new SHA1Managed().ComputeHash(File.OpenRead(Console.ReadLine()))).Replace("-","").Substring(0,3);foreach(string n in d(m).Split(':'))
{if(f==n.Substring(0,3))
{Console.WriteLine(n.Substring(3,n.Length-3));break;}}}
public static String d(String c)
{byte[]d=Convert.FromBase64String(c);using(MemoryStream i=new MemoryStream(d))
using(GZipStream z=new GZipStream(i,CompressionMode.Decompress))
using(MemoryStream o=new MemoryStream())
{z.CopyTo(o);StringBuilder sb=new StringBuilder();return Encoding.UTF8.GetString(o.ToArray());}}}}

The table (string) of labels and partial SHA hashes is compressed to save a few bytes. The original dictionary looks like:

84FAcademia:4C5Android Enthusiasts:C61Anime & Manga:AF4Ask Different:01DArduino:18DAsk Ubuntu:14FAstronomy:0FFAviation:521Video Production:5AFBeer:0C7Bicycles:846Biology:BD9Bitcoin:927Blender:C8CBoard & Card Games:BBAStack Overflow in Portuguese:355LEGO Answers:53FBuddhism:7B1Chemistry:D1CChess:068Chinese Language:52CChristianity:321Programming Puzzles & Code Golf:DA2Code Review:F8BCognitive Sciences:935Community Building:603Seasoned Advice:F7CCraft CMS:DA0Cryptography:8BCComputer Science:690Theoretical Computer Science:D11Data Science:307Database Administrators:D69Home Improvement:A13Drupal Answers:0E5Signal Processing:8EEEarth Science:9B2Ebooks:AAAEconomics:511Electrical Engineering:42AEnglish Language Learners:E98Emacs:755English Language & Usage:4E6Expatriates:8DEExpressionEngine Answers:FF3Physical Fitness:EBBFreelancing:5D7French Language:C81Game Development:08BArqade:FE0Gardening & Landscaping:138Genealogy & Family History:EAEGerman Language:38EGeographic Information Systems:020Graphic Design:DFCAmateur Radio:4A8Biblical Hermeneutics:7FAHinduism:4E6History:955Homebrewing:4CEHistory of Science and Math:08BIslam:0A8Italian Language:B21Stack Overflow in Japanese:E8DJapanese Language:B63Joomla:400Mi Yodeya:6F7Linguistics:C6BMagento:C6BMartial Arts:506Mathematics:479Mathematics Educators:5A1Mathematica:588MathOverflow:6EAMotor Vehicle Maintenance & Repair:503Meta Stack Exchange:3ACPersonal Finance & Money:09CMovies & TV:0EFMusic, Practice & Theory:A82Network Engineering:FE1Open Data:CA9The Great Outdoors:9BEParenting:365Ask Patents:BF6Pets:591Philosophy:AE9Photography:A01Physics:545Project Management:28EPoker:E46Politics:471Personal Productivity:85DProgrammers:D15Puzzling:6BFQuantitative Finance:945Raspberry Pi:76BReverse Engineering:1F5Robotics:3C1Role - playing Games:B77Russian Language:6D1Salesforce:D13Computational Science:E19Science Fiction & Fantasy:BDDInformation Security:BEFServer Fault:446SharePoint:9F9Skeptics:99ESoftware Recommendations:D0DSound Design:7A6Space Exploration:CCESpanish Language:A24Sports:EE4Software Quality Assurance & Testing:B1DStack Apps:D33Stack Overflow:F4EStartups:8EACross Validated:63BSuper User:254Sustainable Living:13ETeX - LaTeX:8D9Tor:DFDTravel:777Tridion:F9CUnix & Linux:945User Experience:5EDWeb Applications:A21Webmasters:EDCWindows Phone:525WordPress Development:DB3The Workplace:81EWorldbuilding:539Writers
\$\endgroup\$
2
  • 15
    \$\begingroup\$ It looks like you're taking the hash of the file not of the actual image data. So if I added some meta-data to the file without altering the image, wouldn't that break your code? \$\endgroup\$ Commented Dec 8, 2014 at 20:11
  • 1
    \$\begingroup\$ 1) var should safe a couple of bytes. 2) What's with the stringbuilder? I don't see it used. 3) StreamReader.ReadToEnd might help a bit as well. \$\endgroup\$ Commented Dec 8, 2014 at 20:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.