I'm trying to query a database of Book titles based on the first letter of the title. However, I want to ignore common words such as "The" and "A".
So when searching for books that start with the letter "T"
"The Adventures of Huck Finn" - would NOT be matched
"Transformation of a Runner" - would be matched
I'm not very experienced with REGEX, but this is what I have so far (where $first_letter could equal 't')
... WHERE title = '^[(a )(the )]*[$first_letter]' ...
This successfully matches book titles that start with a particular letter even after the words "A" or "The", but doesn't ignore those words. So if $first_letter='t', it would match BOTH books mentioned above.
I've tried googling it, but haven't found any solutions. Any help would be greatly appreciated.
Thanks in advance.
Kevin
Read about MySQL full text search
The regular expression you've written isn't valid. []s are used to denote what is called a character class. Everything you enter between the brackets (with some characters potentially needing to be escaped, such as the literal characters [ and ]) is treated as standing-in for a single character.
edit After re-reading my answer, I realized lookaround wasn't a good way to approach this.
The functionality you're groping for is called negative lookahead, negative lookbehind, or some similar variant. I'm unsure whether MySQL's regex flavor supports it, but I don't think it would be a good fit for this problem.
Alternatively, you could do a regex that looks like this:
^((a|the|of|and) )?[letter of interest]
The breakdown:
There are two groups
The inner-most group looks for instances of words you want to ignore
The outer-most group just adds a space to the end of that
The ? asserts that there could be 0 or 1 instances of this group
You'll have to do the legwork of translating this into MySQL regex syntax yourself. My apologies.
Related
We have thousands of structured filenames stored in our database, and unfortunately many hundreds have been manually altered to names that do not follow our naming convention. Using regex, I'm trying to match the correct file names in order to identify all the misnamed ones.
The files are all relative to a meeting agenda, and use the date, meeting type, Agenda Item#, and description in the name.
Our naming convention is yyyymmdd_aa[_bbb]_ccccc.pdf where:
yyyymmdd is a date (and may optionally use underscores such as yyyy_mm_dd)
aa is a 2-3 character Meeting Type code
bbb is an optional Agenda Item
ccccc is a freeform variable length description of the file (alphanumeric only)
Example filenames:
20200225_RM_agenda.pdf
20200225_RM_2_memo.pdf
20200225_SS1_3c_presenTATION.pdf
20200225_CA_4d_SiGnEd.pdf
20200225_RM_5_Order1234.pdf
2021_02_25_EV_Notice.pdf
The regex I'm using to match these files is below (regex demo):
/^(\d{4}[_]?\d{2}[_]?\d{2})_(\w{2,3})_([a-z0-9]{1,3})_?(.+)?.pdf/i
The Problem:
In general, it's working fine, BUT if the Agenda Number ("bbb") is NOT in the filename, the regex captures and returns the first 3 characters of the description. It seems to me that the 3rd capture group _([a-z0-9]{1,3})_ is saying 1-3 alphanumeric characters between underscores, but I don't know how to "force the underscore delimiters", or otherwise tell it that the group may not be there, and that it's now looking at the descriptive text. This can be seen in the demo code where the first and last filenames do not use an Agenda Number.
Any assistance is appreciated.
The optional identifier ? is for the last thing, either a characters or group. So the expression ([a-z0-9]{1,3})_? makes the underscore optional, but not the preceding group. The solution is to move the underscore into the parenthesis.
^(\d{4}[_]?\d{2}[_]?\d{2})_(\w{2,3})_([a-z0-9]{1,3}_)?(.+)?.pdf
Additionally, the [_]? can be simplified to just _?, file name periods should be escaped (otherwise they are a wildcard), and I personally like to name my groups using (?<name>) syntax. Putting that all together you get:
^(?<date>\d{4}_?\d{2}_?\d{2})_(?<meeting_type>\w{2,3})_(?<agenda>[a-z0-9]{1,3}_)?(?<description>.+)?\.pdf$
Demo here: https://regex101.com/r/BUKCih/1
Updated:
I've made some updates based on the comments. I added $ to the end to force "end of filename" as #Chris Maurer said. This stops file.pdf.txt from getting through. I also made a sub-group and moved the name into that group, which allows the trailing underscore to not be included in the named-group. I'm going to leave Chris's other comment about tightening the last matching group alone, although I do agree with it, and the OP might find a couple of non-conforming files if they use [a-z0-9]+ or similar. I don't remember off-hand if PHP supports POSIX but if so [:alnum:] could be used too.
^(?<date>\d{4}_?\d{2}_?\d{2})_(?<meeting_type>\w{2,3})_((?<agenda>[a-z0-9]{1,3})_)?(?<description>.+)?\.pdf$
Updated demo here: https://regex101.com/r/ebmxkF/1
Have read the following which have some overlap (pun intended!) with the issue I am facing:
preg_match_all how to get *all* combinations? Even overlapping ones
Overlapping matches with preg_match_all and pattern ending with repeated character
However, I don’t really know how to apply their answers to my issue which is a little more complicated.
My regex that I use with preg_match_all():
/.{240}[^\[]Order[^ ][^\(].{9}/u
With the following string:
56A. Subject to the provisions of this Act, any decision of the Court or the Appeal Board shall be final and conclusive, and no decision or order of the Court or the Appeal Board shall be challenged, appealed against, reviewed, quashed or called into question in any court and shall not be subject to any Quashing Order, Prohibiting Order, Mandatory Order or injunction in any court on any account.[20/99; 42/2005]
I intended it to match exactly 3 times. The first match has “Quashing Order” 9 characters before the end. The second match has “Prohibiting Order” 9 characters before the end. The third match has “Mandatory Order” 9 characters before the end.
However, as expected it’s only matching the first one, as the expected matches are overlapping.
I applied what I read in the other posts, I tried this:
(?=(.{240}[^\[]Order[^ ][^\(].{9}))
I still don’t get what I need.
How do I solve this?
You can use
\w+\s+Order\b
See the regex demo.
Regex details
\w+ - one or more word chars
\s+ - 1 or more whitespaces
Order\b - a whole word Order, as \b is a word boundary.
You will need to use a positive look-behind assertion for .{240}, just like the answer you found suggests using a positive look-ahead assertion for .{9}:
/(?<=.{240})[^\[]Order[^ ][^\(](?=.{9})/u
This RE matches your string only twice because of [^ ], as #bobblebubble said. Adjust that part as necessary.
I would like to use regular expression to find certain combination of words from a phrase in php. I can't even get the regex expression part to work.
The sentence should match any phrase that has the words (proficient/proficiency/fluent) in (chinese/mandarin/cantonese) in the same sentence. So it would match "She is fluent in Chinese." and "His proficiency in Mandarin is excellent"
regex = (fluent)|(proficient)|(proficiency).*(chinese)|(mandarin)|(cantonese)
I can get it to match the word fluent but how to make it match both words in the same sentence before it is considered a match?
Your grouping is wrong, it should be rather
(fluent|proficient|proficiency)[^.]*(chinese|mandarin|cantonese)
[^.] ensures (naively) that the words occur within the same sentence. Also, don't forget the i flag to match title-cased words like Chinese.
((fluent)|(proficient)|(proficiency)).*((chinese)|(mandarin)|(cantonese))
You need to put aditional brackets, if you also want to match the whole sentence you need to do something like this
[.!?].*((fluent)|(proficient)|(proficiency)).*((chinese)|(mandarin)|(cantonese)).*[.!?]
If the order doesn't matter, you could use two regexp, the first for the first group and a second to match the second group. Than you match two times and if both hit, you got it.
In case you're dealing with a fluent text, I would try to split it in sentences.
I am new to regular expression and trying to match the following pattern using regular expression:
Groups of numbers, each looks like either a single number like 12, or a number range like 19-39
Groups are separated by semicolon(;)
All numbers are within range 1-48 (but we don't need to verify this in regular expression)
So an example match would be 12;13;19-39;43
For a single group, I can think of using
\b[1-9]{1}|[1-9]{1}[0-9]{1}\b
for single number, and
\b[1-9]{1}|[1-9]{1}[0-9]{1}-[1-9]{1}|[1-9]{1}[0-9]{1}\b
for number range.
The question is how to take the semicolon(;) into consideration also: any number of the above groups of number(s) connected by ; can be matched.
This should exactly match your requirement:
\d*[0-9](|-\d*[0-9]|;\d*[0-9])*$
Explanation:
Match any digit multiple times.
Next, check for a - or ; followed by another series of digits.
Repeat this till matches are found.
Try it out here:
http://gskinner.com/RegExr/
You can paste sample text in the big text area and see the exp in action. Cheers!
Try this:
/^\d*[0-9](|.\d*[0-9]|;\d*[0-9])*$/;
Its matches your requirement.
One trick to learning these is to try and break it into parts and write brutal ones to start:
1-48 alone ending in ; you can be as complicated as: ((\d)|([1-3]\d)|(4[0-8]));
for dashed groups just the same components repeated with a dash: ((\d)|([1-3]\d)|(4[0-8]))-((\d)|([1-3]\d)|(4[0-8]));
Now Combine to get either / or and repeat the whole group: ((((\d)|([1-3]\d)|(4[0-8]));)|(((\d)|([1-3]\d)|(4[0-8]))-((\d)|([1-3]\d)|(4[0-8]));))*
Now we have this gross, brute force, regex with a ridiculous number of groupings above, but it works. Next we can think about simplifying and you have an even better place (sort of) to start asking for help from.
Was going to start simplifying, but you have a other answers here already.
Simplifying a little and just noting your final number does not end with a semicolon you can start with merging with something like #Sunny has:
^((\d)|([1-3]\d)|(4[0-8]))(|-((\d)|([1-3]\d)|(4[0-8]))|;((\d)|([1-3]\d)|(4[0-8])))*$
I want to highlight a group of words, they can appear single or in a row. I'd like them to be highlighted together if they appear one after the other, and if they don't, they should also be highlighted, like the normal behavior. For instance, if I want to highlight the words:
results as
And the subject is:
real time results: shows results as you type
I'd like the result to be:
real time results: shows <span class="highlighted"> results as </span> you type
The whitespaces are also a headache, because I tried using an or expression:
( results )|( as )
with whitespaces to prevent highlighting words like bass, crash, and so on. But since the whitespace after results is the same as the whitespace before as, the regexp ignores it and only highlights results.
It can be used to highlighted many words so combinations of
( (one) (two) )|( (two) (one) )|( one )|( two )
are not an option :(
Then I thought that there may be an operator that worked like | that could be use to match both if possible, else one, or the other.
Using spaces to ensure you match full words is the wrong approach. That's what word boundaries are for: \b matches a position between a word and a non-word character (where word characters usually are letters, digits and underscores). To match combinations of your desired words, you can simply put them all in an alternation (like you already do), and repeat as often as possible. Like so:
(?:\bresults\b\s*|\bas\b\s*)+
This assumes that you want to highlight the first and separate results in your example as well (which would satisfy your description of the problem).
Perhaps you do not need to match a string of words next to each other. Why not just apply your highlighting like so:
real time results: shows <span class="highlighted">results</span> <span class="highlighted">as</span> you type
The only realy difference is that the space between the words is not highlighted, but it's a clean and easy compromise which will save you hours of work and doesn't seem to hurt the UX in the least (in my opinion).
In that case, you could just use alternation:
\b(results|as)\b
(\b being the word boundary anchor)
If you really don't like the space between words not being highlight, you could write a jQuery function to find "highlighted" spans separated by only white space and then combine them (a "second stage" to achieve your UX design goals).
Update
(OK... so merging spans is actually kind of difficult via jQuery. See Find text between two tags/nodes)