How to search a string for values and convert values

Question

I have an API that accepts a string that needs to be properly formatted before going into the server.

The format for going into the server is the following

"{Country ABR} {Day/Hour} {State ABR} {Title} {hrs.} ({Month Year}.)"

Several Possibilities the client may send in :

"US Construction 7/70 hrs."

"IA Private hrs US.

"OIL US 8/70 hrs (Dec 2014).

Several valid examples after converting user input are:

"US 7/70 MI Construction hrs."

"US IA Private hrs."

"US OIL 8/70 hrs. (Dec 2014)"

the converter arranges the input into the correct order. hrs always ends with a period and rearranges ({Month Year}) outside the sentence as shown.

so far I have

       [TestMethod]
    public void TestMethod1()
    {
        var toConvert = "USA Construction 70/700 (Dec 2014) hrs";
        var converted = ConvertHOSRules(toConvert);

        Assert.AreEqual(converted, "USA 70/700 Construction hrs.(Dec 2014)");
    }

    private string ConvertHOSRules(string input)
    {
        //todo refactor
        string output = "";

        string country = Regex.Match(input, @"\b(USA|CAN|MEX)\b").Value +" ";
        string dateHours =  Regex.Match(input,@"\d{1,2}\/\d{1,3}").Value + " ";
        string hrs = Regex.Match(input, @"\b(hrs)\b").Value ;
        var date = Regex.Match(input, @"\(([a-zA-Z]+\s{1}[0-9]{4})\)").Value + " ";
        string title = input.Replace(country, "").Replace(date, "").Replace(dateHours, "").Replace(hrs, "");
        output = $"{country} {dateHours} {title} {hrs}.{date}";
        return output;

    }

This is passing i need to refactor.. the + " " is like a null guard by lazy programmer

Make tests out of these examples. Then try to write regexp(s) to pass tests. — Renat, May 30 '19 at 14:20
This sounds like a very challenging problem, largely because it looks like users are entering plain text. That means that they don't know about the rules by which you parse the text, and so you can't expect them to enter anything in a consistent way. Then any state or country abbreviation could also be part of a title. And lots of other things no one would think of. Can you change the API so that the data is already separated into fields? — Scott Hannen, May 30 '19 at 14:51
there is documentation on examples so the possibilities clients may send in are the only values I'm considering. If it is way off then I am sending a bad request back with an error code explaining the text is invalid see the documentation for help — Dylan Godfrey, May 30 '19 at 14:54
The matches also report their locations. Try to match all the subparts whose format you know. Using interval arithmetic on the original string, subtract the matched subparts. Then, remove any unmatched interval that consists only of white spaces. After that, there should be at most one interval left. This is your title. — Nico Schertler, May 30 '19 at 15:27
@NicoSchertler exactly what im doing I just have an issue with the (Dec 2014). Any suggestions? — Dylan Godfrey, May 30 '19 at 15:40
Suggestion for the regex? Something like `\((Jan|Feb|Mar)\s+\d{4}\)` should work. — Nico Schertler, May 30 '19 at 15:44

Emma · Accepted Answer · 2019-06-01T20:02:51.820

This question is quite interesting, especially if we would want to design algorithms for it, because my guess is that our regular expressions would be rather unnecessary.

If we wish to do that with expressions, I would start with a simple expression such as listing possible countries and states in two capturing groups:

(US|UK|FR)
(CA|WA|IA|MO|MI)

then our hours are well-structured:

(\d+\/\d+)

so is the month (.+?) and year ([0-9]+):

\(((.+?)\s+([0-9]+))\)

and here is where we would be facing problem with other keywords such as Construction and OIL, we could add a min 3 chars not to possibly conflict with states and countries:

([A-Z][a-z]{2,}|[A-Z]{3,})

and last we would clean our string by collecting all spaces and other chars left, such as hrs. which is just repeating and we might not want to match or capture that.

(.*?)

Finally, we would combine using alternation:

(US|UK|FR)|(CA|NY|IA|TX|MI)|(\d+\/\d+)|\(((.+?)\s+([0-9]+))\)|([A-Z][a-z]{2,}|[A-Z]{3,})|(.*?)

DEMO

Test

using System;
using System.Text.RegularExpressions;

public class Example
{
    public static void Main()
    {
        string pattern = @"(US|UK|FR)|(CA|NY|IA|TX|MI)|(\d+\/\d+)|\(((.+?)\s+([0-9]+))\)|([A-Z][a-z]{2,}|[A-Z]{3,})|(.*?)";
        string input = @"US 7/70 MI Construction hrs.
US IA Private hrs.
US OIL 8/70 hrs. (Dec 2014)
UK 7/70 MI Construction hrs.
UK IA Private hrs.
UK OIL 8/70 hrs. (Dec 2014)
FR 7/70 MI Construction hrs.
FR IA Private hrs.
FR OIL 8/70 hrs. (Dec 2014)";
        RegexOptions options = RegexOptions.Multiline;

        foreach (Match m in Regex.Matches(input, pattern, options))
        {
            Console.WriteLine("'{0}' found at index {1}.", m.Value, m.Index);
        }
    }
}

DEMO

RegEx

If this expression wasn't desired, it can be modified/changed in regex101.com.

RegEx Circuit

jex.im visualizes regular expressions:

this is interesting and useful.. thanks for the response – Dylan Godfrey May 30 '19 at 16:15 — Dylan Godfrey, May 30 '19 at 16:15

How to search a string for values and convert values

1 Answers1

DEMO

Test

DEMO

RegEx

RegEx Circuit