Most unusable technology award (my nomination - regular expressions)
-
Every now and then I need to solve a problem for which regular expressions looks like it is the "perfect" answer. Today that happens to be validating and extracting UK postal codes from addresses. BUT Every time I try to use regular expressions I find that no-one (especially me) has a clue how to use them and that all "posted" solutions can be demonstrated as flawed and therefore dangerous to use. The sheer complexity of the expressions makes them virtually impossible to read and therefore understand. I post the wikipedia solution to demonstrate my case (GIR 0AA)|(((A[BL]|B[ABDHLNRSTX]?|C[ABFHMORTVW]|D[ADEGHLNTY]|E[HNX]?|F[KY]|G[LUY]?|H[ADGPRSUX]|I[GMPV]|JE|K [ATWY]|L[ADELNSU]?|M[EKL]?|N[EGNPRW]?|O[LX]|P[AEHLOR]|R[GHM]|S[AEGKLMNOPRSTY]?|T[ADFNQRSW]|UB|W[ADFNRSV] |YO|ZE)[1-9]?[0-9]|((E|N|NW|SE|SW|W)1|EC[1-4]|WC[12])[A-HJKMNPR-Y]|(SW|W)([2-9]|[1-9][0-9])|EC[1-9] [0-9]) [0-9][ABD-HJLNP-UW-Z]{2}) Can anyone think of a less usable technology?
www.it-workplace.com
"If a man speaks in a forest where there is no woman to hear him, is he still wrong?"You're not supposed to understand it, just go to a "regular expression website", read the description and the comments and copy/paste what you need. Never try to debug it yourself, just complain to the person who wrote it. :laugh: There's a community of "special people" for everything.
Giraffes are not real.
-
Manfred R. Bihy wrote:
If the only tool you've got is a hammer, every problem looks like a nail!
I like that. :thumbsup:
Just along for the ride. "the meat from that butcher is just the dogs danglies, absolutely amazing cuts of beef." - DaveAuld (2011)
"No, that is just the earthly manifestation of the Great God Retardon." - Nagy Vilmos (2011) "It is the celestial scrotum of good luck!" - Nagy Vilmos (2011)Slacker007 wrote:
Manfred R. Bihy wrote:
If the only tool you've got is a hammer, every problem looks like a nail!
I like that. :thumbsup:
I believe it was a quote from Jesus, in an early draft of the Bible. It was discarded the final edit, along with many other gems, like "Ow, my #$%&ing thumb!"
I wanna be a eunuchs developer! Pass me a bread knife!
-
Nearly, but doesn't cover the case where the user has not entered the space (i.e. 'LS1 9EL' vs 'LS19EL'). We also have some other occasional but common variations such as 'LS1_9EL' that we can try to parse for. My understanding is also that whilst this expression will validate the general format of the postcode there are specific exceptions that it does not cover. Unfortunately the task is not one of validating data at point of entry but matching data that has not been properly validated in the first place (>5m records), so refering to a web service such as the BING api is ruled out for performance reasons.
www.it-workplace.com
"If a man speaks in a forest where there is no woman to hear him, is he still wrong?"Then test for basic, then retest the exceptions. Beats the Hell out of a 300-character regex.
I wanna be a eunuchs developer! Pass me a bread knife!
-
Every now and then I need to solve a problem for which regular expressions looks like it is the "perfect" answer. Today that happens to be validating and extracting UK postal codes from addresses. BUT Every time I try to use regular expressions I find that no-one (especially me) has a clue how to use them and that all "posted" solutions can be demonstrated as flawed and therefore dangerous to use. The sheer complexity of the expressions makes them virtually impossible to read and therefore understand. I post the wikipedia solution to demonstrate my case (GIR 0AA)|(((A[BL]|B[ABDHLNRSTX]?|C[ABFHMORTVW]|D[ADEGHLNTY]|E[HNX]?|F[KY]|G[LUY]?|H[ADGPRSUX]|I[GMPV]|JE|K [ATWY]|L[ADELNSU]?|M[EKL]?|N[EGNPRW]?|O[LX]|P[AEHLOR]|R[GHM]|S[AEGKLMNOPRSTY]?|T[ADFNQRSW]|UB|W[ADFNRSV] |YO|ZE)[1-9]?[0-9]|((E|N|NW|SE|SW|W)1|EC[1-4]|WC[12])[A-HJKMNPR-Y]|(SW|W)([2-9]|[1-9][0-9])|EC[1-9] [0-9]) [0-9][ABD-HJLNP-UW-Z]{2}) Can anyone think of a less usable technology?
www.it-workplace.com
"If a man speaks in a forest where there is no woman to hear him, is he still wrong?"Can I nominate Lotus Notes? "I am rarely happier than when spending entire day programming my computer to perform automatically a task that it would otherwise take me a good ten seconds to do by hand." - Douglas Adams
-
Every now and then I need to solve a problem for which regular expressions looks like it is the "perfect" answer. Today that happens to be validating and extracting UK postal codes from addresses. BUT Every time I try to use regular expressions I find that no-one (especially me) has a clue how to use them and that all "posted" solutions can be demonstrated as flawed and therefore dangerous to use. The sheer complexity of the expressions makes them virtually impossible to read and therefore understand. I post the wikipedia solution to demonstrate my case (GIR 0AA)|(((A[BL]|B[ABDHLNRSTX]?|C[ABFHMORTVW]|D[ADEGHLNTY]|E[HNX]?|F[KY]|G[LUY]?|H[ADGPRSUX]|I[GMPV]|JE|K [ATWY]|L[ADELNSU]?|M[EKL]?|N[EGNPRW]?|O[LX]|P[AEHLOR]|R[GHM]|S[AEGKLMNOPRSTY]?|T[ADFNQRSW]|UB|W[ADFNRSV] |YO|ZE)[1-9]?[0-9]|((E|N|NW|SE|SW|W)1|EC[1-4]|WC[12])[A-HJKMNPR-Y]|(SW|W)([2-9]|[1-9][0-9])|EC[1-9] [0-9]) [0-9][ABD-HJLNP-UW-Z]{2}) Can anyone think of a less usable technology?
www.it-workplace.com
"If a man speaks in a forest where there is no woman to hear him, is he still wrong?"I would like to nominate my iPhone 3G for the award. Via iOS updates, Apple managed to turn my phone into a completely useless brick. At times, it could take up to 10 seconds for one of the little "folders" to open up after tapping on it. And that was just one of the features that failed to perform. Voicemail problems, WiFi problems, frequent crashing, etc. Fortunately, my wife dropped the phone into the swimming pool and now I am a very much happier Windows Phone user.
-
That would be my approach too, and I spent a long time working with the Postal Address File in the UK. To attempt to write one regex for the whole thing leads to monstrosities like the one demonstrated. That said, I too dislike R.E.'s. They seem to me to be easy and quick to write, but tough to read. Good in an editor's search box, or tool like grep, bad in code that must be viewed by other developers. I don't object to regular languages, just the form of reg ex's that has come into use over the years. There a few examples of regular languages better done, notably in the MGrammar parser technology Microsoft tech previewed a while ago (which I guess has, unfortunately, floundered in their dev labs as they seem to have gone silent).
I have to say, my tendency would be to have a list of the known postal codes and put that into an array. Then use a string compare to just look for matches to all known postal codes first. Then after that start doing evals to look for things that fit the known formats but aren't on the known list, if possible, presuming each entry is a separate line or item in an array, only putting the lines or items that haven't already matched into a separate array and checking those. Not sure if my solution is more or less complicated. It's probably colored by my experience as a LAMP dev though.
CaptJosh There are only 10 kinds of people in the world; those who understand binary and those who don't.
-
Can I nominate Lotus Notes? "I am rarely happier than when spending entire day programming my computer to perform automatically a task that it would otherwise take me a good ten seconds to do by hand." - Douglas Adams
-
Every now and then I need to solve a problem for which regular expressions looks like it is the "perfect" answer. Today that happens to be validating and extracting UK postal codes from addresses. BUT Every time I try to use regular expressions I find that no-one (especially me) has a clue how to use them and that all "posted" solutions can be demonstrated as flawed and therefore dangerous to use. The sheer complexity of the expressions makes them virtually impossible to read and therefore understand. I post the wikipedia solution to demonstrate my case (GIR 0AA)|(((A[BL]|B[ABDHLNRSTX]?|C[ABFHMORTVW]|D[ADEGHLNTY]|E[HNX]?|F[KY]|G[LUY]?|H[ADGPRSUX]|I[GMPV]|JE|K [ATWY]|L[ADELNSU]?|M[EKL]?|N[EGNPRW]?|O[LX]|P[AEHLOR]|R[GHM]|S[AEGKLMNOPRSTY]?|T[ADFNQRSW]|UB|W[ADFNRSV] |YO|ZE)[1-9]?[0-9]|((E|N|NW|SE|SW|W)1|EC[1-4]|WC[12])[A-HJKMNPR-Y]|(SW|W)([2-9]|[1-9][0-9])|EC[1-9] [0-9]) [0-9][ABD-HJLNP-UW-Z]{2}) Can anyone think of a less usable technology?
www.it-workplace.com
"If a man speaks in a forest where there is no woman to hear him, is he still wrong?"I think that you can use regular expressions for small issues but when it comes to a great problem you should use parsing instead, i am an informatic enginneer and i had to make a programm that received call records from a PBX, a friend of mine that is a cibernetic made me a regular expresssion to parse the line. but that was sort of a blackbox , because when it crashed for any reason you couldnt tell why, later I replaced it by a parser an it was much better
-
I find the carriage return key to be next to useless and I see that you agree. :-D
Henry Minute Do not read medical books! You could die of a misprint. - Mark Twain Girl: (staring) "Why do you need an icy cucumber?" “I want to report a fraud. The government is lying to us all.” I wouldn't let CG touch my Abacus! When you're wrestling a gorilla, you don't stop when you're tired, you stop when the gorilla is.
-
Wait.. you have a useless key on your keyboard? I only have a 104 key one, now I gotta go upgrade :)
We can program with only 1's, but if all you've got are zeros, you've got nothing.
Yup. Right next to the 'Any' key.
Henry Minute Do not read medical books! You could die of a misprint. - Mark Twain Girl: (staring) "Why do you need an icy cucumber?" “I want to report a fraud. The government is lying to us all.” I wouldn't let CG touch my Abacus! When you're wrestling a gorilla, you don't stop when you're tired, you stop when the gorilla is.
-
I find the carriage return key to be next to useless and I see that you agree. :-D
Henry Minute Do not read medical books! You could die of a misprint. - Mark Twain Girl: (staring) "Why do you need an icy cucumber?" “I want to report a fraud. The government is lying to us all.” I wouldn't let CG touch my Abacus! When you're wrestling a gorilla, you don't stop when you're tired, you stop when the gorilla is.
The carriage-return key isn't on my keyboard. Ancient typewriters had to return the carriage in order to start a new line, thus the name. Of course, you already know this, just explicating for those that don't. ;P Steve
Just think of it as evolution in action.
-
The carriage-return key isn't on my keyboard. Ancient typewriters had to return the carriage in order to start a new line, thus the name. Of course, you already know this, just explicating for those that don't. ;P Steve
Just think of it as evolution in action.
Steve Burchett wrote:
just explicating for those that don't
Explicating them from what situation? Ahh, those were the days, you could cut your paper in half with your printer. (Used impact devices to strike the ribbon on the paper. Then all you had to do is forget that carriage return just moves it back to the front of the line. If you didn't tell it to line feed the next line would overwrite the same location, that one line getting blacker and blacker until it was "paper" thin and wear out.
-
I recall reading that line at least some 10 years ago, and I wondered where it came from. Apparently it's much older: at least from 1964
-
C# is my hammer! RegEx is one of those tools that was bought a long time ago for a very specific job and now lurks at the back of the toolbox. It only comes out very infrequently so: 1. I have to read the manual each time I use it because I have forgotten the instructions since the last time. 2. I am unwilling to invest the amount of time required to master its use because (refer to 1) I only use it very infrequently. 3. Each time I use it I have a lingering concern that I have left a problem for the next poor soul who has to maintain the solution.
www.it-workplace.com
"If a man speaks in a forest where there is no woman to hear him, is he still wrong?" -
Every now and then I need to solve a problem for which regular expressions looks like it is the "perfect" answer. Today that happens to be validating and extracting UK postal codes from addresses. BUT Every time I try to use regular expressions I find that no-one (especially me) has a clue how to use them and that all "posted" solutions can be demonstrated as flawed and therefore dangerous to use. The sheer complexity of the expressions makes them virtually impossible to read and therefore understand. I post the wikipedia solution to demonstrate my case (GIR 0AA)|(((A[BL]|B[ABDHLNRSTX]?|C[ABFHMORTVW]|D[ADEGHLNTY]|E[HNX]?|F[KY]|G[LUY]?|H[ADGPRSUX]|I[GMPV]|JE|K [ATWY]|L[ADELNSU]?|M[EKL]?|N[EGNPRW]?|O[LX]|P[AEHLOR]|R[GHM]|S[AEGKLMNOPRSTY]?|T[ADFNQRSW]|UB|W[ADFNRSV] |YO|ZE)[1-9]?[0-9]|((E|N|NW|SE|SW|W)1|EC[1-4]|WC[12])[A-HJKMNPR-Y]|(SW|W)([2-9]|[1-9][0-9])|EC[1-9] [0-9]) [0-9][ABD-HJLNP-UW-Z]{2}) Can anyone think of a less usable technology?
www.it-workplace.com
"If a man speaks in a forest where there is no woman to hear him, is he still wrong?"Right, regular expressions easily become cryptic. But don't they look childish compared to the marvels you can so easily produce with most Unix scripting languages ?
lsof "$lsof_args" | grep $connection_type | grep -v "$no_match" |
awk '{print $9}' | cut -d : -f $field | sort | uniq |
sed s/"^$router"//if [ "$(eval "echo \${$(echo get$(echo -ne $arch |
sed 's/^\(.\).*/\1/g' | tr 'a-z' 'A-Z'; echo $arch |
sed 's/^.\(.*\)/\1/g')):-false}")" = true ]
then
return 0;
else
return 1;
fi;find . -name '*[+{;"\\=?~()<>&*|$ ]*' -maxdepth 0 \
-exec rm -f '{}' \; -
Andrew Wiles wrote:
The sheer complexity of the expressions makes them virtually impossible to read and therefore understand
It is at this point that you need the tool support. Better said, it is not because the thing seems complicated and unreadable that I won't use it if it solves my problem best.
Whilst I agree that RegExps are like Java and C, namely WORN (Write Once, Read Never), I would say that it is amazing how much you can accomplish with them with a fairly short definition. If I were writing the Post Code validator, I'd sacrifice a bit of (possible?) efficiency to aid readability, e.g. I'd change
A[BL]|B[ABDHLNRSTX]?|C[ABFHMORTVW]| etc
to
AB|AL|B|BA|BB|BD|BH|BL|BN|BR|BS|BT|BX|CA|CB|CF|CH|CM|CO|CR|CT|CV|CW|
Try looking for, for example, the code for Belfast (BT) in the first extract above and compare it with looking for the BT in the second extract. Just to make it simpler, I have highlighted them. [aside] What is BX the code for? [/aside] When codi ng it, you would not write it all as a long string, it could easily be broken up into manageable pieces Actually, I would add one efficiency: The Bootle code G1R 0AA is such a rare special case that I would put it at the end so it only rarely got touched rather than putting it at the begining. When coding it, you would not write it all as a long string, it could easily be broken up into manageable pieces. E.g. The following is a RegExp for TLDs (as I am too lazy to rewrite the PostCode one) which shows the technique in action:
clsForm_TLDs = _
"aero|biz|cat|com|coop|edu|gov|info|int|mil|mobi|museum|name|net|org|pro|travel" _
& "|ac|ad|ae|af|ag|ai|al|am|an|ao|aq|ar|as|at|au|aw|ax|az" _
& "|ba|bb|bd|be|bf|bg|bh|bi|bj|bm|bn|bo|br|bs|bt|bv|bw|by|bz" _
& "|ca|cc|cd|cf|cg|ch|ci|ck|cl|cm|cn|co|cr|cs|cu|cv|cx|cy|cz" _
& "|dd|de|dj|dk|dm|do|dz" _
& "|ec|ee|eg|eh|er|es|et|eu" _
& "|fi|fj|fk|fm|fo|fr|fx" _
& "|ga|gb|gd|ge|gf|gg|gh|gi|gl|gm|gn|gp|gq|gr|gs|gt|gu|gw|gy" _
& "|hk|hm|hn|hr|ht|hu" _
& "|id|ie|il|im|in|io|iq|ir|is|it" _
& "|je|jm|jo|jp" _
& "|ke|kg|kh|ki|km|kn|kp|kr|kw|ky|kz" _
& "|la|lb|lc|li|lk|lr|ls|lt|lu|lv|ly" _
& "|ma|mc|md|mg|mh|mk|ml|mm|mn|mo|mp|mq|mr|ms|mt|mu|mv|mw|mx|my|mz" _
& "|na|nc|ne|nf|ng|ni|nl|no|np|nr|nu|nz" _
& "|om" _
& "|pa|pc|pe|pf|pg|ph|pk|pl|pm|pn|pr|ps|pt|pw|py" _
& "|qa" _
& "|re|ro|ru|rw" _
& "|sa|sb|sc|sd|se|sg|sh|si|sj|sk|sl|sm|sn|so|sr|st|su|sv|sy|sz" _
& "|tc|td|tf|tg|th|tj|tk|tm|tn|to|tp|tr|tt|tv|tw|tz" _
& "|ua|ug|uk|um|us|uy|uz" _
& "|va|vc|ve|vg|vi|vn|vu" _
& "|wf|ws" _
& "|ye|yt|yu" _
& "|za|zm|zr|zw" -
Whilst I agree that RegExps are like Java and C, namely WORN (Write Once, Read Never), I would say that it is amazing how much you can accomplish with them with a fairly short definition. If I were writing the Post Code validator, I'd sacrifice a bit of (possible?) efficiency to aid readability, e.g. I'd change
A[BL]|B[ABDHLNRSTX]?|C[ABFHMORTVW]| etc
to
AB|AL|B|BA|BB|BD|BH|BL|BN|BR|BS|BT|BX|CA|CB|CF|CH|CM|CO|CR|CT|CV|CW|
Try looking for, for example, the code for Belfast (BT) in the first extract above and compare it with looking for the BT in the second extract. Just to make it simpler, I have highlighted them. [aside] What is BX the code for? [/aside] When codi ng it, you would not write it all as a long string, it could easily be broken up into manageable pieces Actually, I would add one efficiency: The Bootle code G1R 0AA is such a rare special case that I would put it at the end so it only rarely got touched rather than putting it at the begining. When coding it, you would not write it all as a long string, it could easily be broken up into manageable pieces. E.g. The following is a RegExp for TLDs (as I am too lazy to rewrite the PostCode one) which shows the technique in action:
clsForm_TLDs = _
"aero|biz|cat|com|coop|edu|gov|info|int|mil|mobi|museum|name|net|org|pro|travel" _
& "|ac|ad|ae|af|ag|ai|al|am|an|ao|aq|ar|as|at|au|aw|ax|az" _
& "|ba|bb|bd|be|bf|bg|bh|bi|bj|bm|bn|bo|br|bs|bt|bv|bw|by|bz" _
& "|ca|cc|cd|cf|cg|ch|ci|ck|cl|cm|cn|co|cr|cs|cu|cv|cx|cy|cz" _
& "|dd|de|dj|dk|dm|do|dz" _
& "|ec|ee|eg|eh|er|es|et|eu" _
& "|fi|fj|fk|fm|fo|fr|fx" _
& "|ga|gb|gd|ge|gf|gg|gh|gi|gl|gm|gn|gp|gq|gr|gs|gt|gu|gw|gy" _
& "|hk|hm|hn|hr|ht|hu" _
& "|id|ie|il|im|in|io|iq|ir|is|it" _
& "|je|jm|jo|jp" _
& "|ke|kg|kh|ki|km|kn|kp|kr|kw|ky|kz" _
& "|la|lb|lc|li|lk|lr|ls|lt|lu|lv|ly" _
& "|ma|mc|md|mg|mh|mk|ml|mm|mn|mo|mp|mq|mr|ms|mt|mu|mv|mw|mx|my|mz" _
& "|na|nc|ne|nf|ng|ni|nl|no|np|nr|nu|nz" _
& "|om" _
& "|pa|pc|pe|pf|pg|ph|pk|pl|pm|pn|pr|ps|pt|pw|py" _
& "|qa" _
& "|re|ro|ru|rw" _
& "|sa|sb|sc|sd|se|sg|sh|si|sj|sk|sl|sm|sn|so|sr|st|su|sv|sy|sz" _
& "|tc|td|tf|tg|th|tj|tk|tm|tn|to|tp|tr|tt|tv|tw|tz" _
& "|ua|ug|uk|um|us|uy|uz" _
& "|va|vc|ve|vg|vi|vn|vu" _
& "|wf|ws" _
& "|ye|yt|yu" _
& "|za|zm|zr|zw" -
I have to say, my tendency would be to have a list of the known postal codes and put that into an array. Then use a string compare to just look for matches to all known postal codes first. Then after that start doing evals to look for things that fit the known formats but aren't on the known list, if possible, presuming each entry is a separate line or item in an array, only putting the lines or items that haven't already matched into a separate array and checking those. Not sure if my solution is more or less complicated. It's probably colored by my experience as a LAMP dev though.
CaptJosh There are only 10 kinds of people in the world; those who understand binary and those who don't.
Use a trie instead of a list. The complete list of postal codes is long, you know. Or at least have it ordered, and employ a somewhat optimized search algorithm.
-
Use a trie instead of a list. The complete list of postal codes is long, you know. Or at least have it ordered, and employ a somewhat optimized search algorithm.
-
Yes, I've considered that too - and most of the XML-related stuff. But at least you have the option to make it halfway readable. It can be a real pain though if you have to delve in to other people's code and the original programmer took no care of naming conventions and formatting...
I second that, XML looks like it was designed by a committee. Schema ( huh why use latin words, why not use english words like plan or design or layout) DTD, XSD, XPath oh my god what a mess. Some guy from the Vatican whose first language was Latin must have gone beserk when designing XML.