How to count words frequency by removing non-letters of a string?

问题

I have a string:

var text = @"
I have a long string with a load of words,
and it includes new lines and non-letter characters.
I want to remove all of them and split this text to have one word per line, then I can count how many of each word exist."

What is the best way to go about removing all non-letter characters, then splitting each word onto a new line so I can store and count how many of each word there are?

var words = text.Split(' ');

foreach(var word in words)
{
    word.Trim(',','.','-');
}

I have tried various things such as text.Replace(characters) with whitespace then split. I have tried Regex (which I would rather not use).

I have also tried to use the StringBuilder class to take the characters from the text (string) and only appending the character if it is a letter a-z / A-Z.

Also tried calling sb.Replace or sb.Remove the characters I don't want before storing them in a Dictionary. But I still seem to end up with characters I don't want?

Everything I try, I seem to have at least one character I don't want in there and can't quite figure out why it isn't working.

Thanks!

回答1:

Using an extension method without RegEx nor Linq

static public class StringHelper
{
  static public Dictionary<string, int> CountDistinctWords(this string text)
  {
    string str = text.Replace(Environment.NewLine, " ");
    var words = new Dictionary<string, int>();
    var builder = new StringBuilder();
    char charCurrent;
    Action processBuilder = () =>
    {
      var word = builder.ToString();
      if ( !string.IsNullOrEmpty(word) )
        if ( !words.ContainsKey(word) )
          words.Add(word, 1);
        else
          words[word]++;
    };
    for ( int index = 0; index < str.Length; index++ )
    {
      charCurrent = str[index];
      if ( char.IsLetter(charCurrent) )
        builder.Append(charCurrent);
      else
      if ( !char.IsNumber(charCurrent) )
        charCurrent = ' ';
      if ( char.IsWhiteSpace(charCurrent) )
      {
        processBuilder();
        builder.Clear();
      }
    }
    processBuilder();
    return words;
  }
}

It parses all chars rejecting all non letters while creating a dictionary of each words having the number of occurrences counted.

Test

var result = text.CountDistinctWords();
Console.WriteLine($"Found {result.Count()} distinct words:");
Console.WriteLine();
foreach ( var item in result )
  Console.WriteLine($"{item.Key}: {item.Value}");

Result on your sample

Found 36 distinct words:

I: 3
have: 2
a: 2
long: 1
string: 1
with: 1
load: 1
of: 3
words: 1
and: 3
it: 1
includes: 1
new: 1
lines: 1
non: 1
letter: 1
characters: 1
want: 1
to: 2
remove: 1
all: 1
them: 1
split: 1
this: 1
text: 1
one: 1
word: 2
per: 1
line: 1
then: 1
can: 1
count: 1
how: 1
many: 1
each: 1
exist: 1

回答2:

I do believe the solution with dictionary for counting frequencies is a best practice in terms of performance and clarity. Here's my version which is slightly different from the accepted answer (I use String.Split() instead of iterating through string's chars):

var text = @"
    I have a long string with a load of words,
    and it includes new lines and non-letter characters.
    I want to remove all of them and split this text to have one word per line, then I       can count how many of each word exist.";

var words = text.Split(new [] {',', '.', '-', ' ', '\n', '\r' }, StringSplitOptions.RemoveEmptyEntries);

var freqByWord = new Dictionary<string, int>();

foreach (var word in words)
{
    if (freqByWord.ContainsKey(word))
    {
        freqByWord[word]++; // we found the same word
    }
    else
    {
        freqByWord.Add(word, 1); // we don't have this one yet
    }
}

foreach (var word in freqByWord.Keys)
{
    Console.WriteLine($"{word}: {freqByWord[word]}");
}

And the results are pretty much the same:

I: 3
have: 2
a: 2
long: 1
string: 1
with: 1
load: 1
of: 3
words: 1
and: 3
it: 1
includes: 1
new: 1
lines: 1
non: 1
letter: 1
characters: 1
want: 1
to: 2
remove: 1
all: 1
them: 1
split: 1
this: 1
text: 1
one: 1
word: 2
per: 1
line: 1
then: 1
can: 1
count: 1
how: 1
many: 1
each: 1
exist: 1

回答3:

Using regex excluding non letter characters. This will also give you a collection of all the words.

var text = @"
I have a long string with a load of words,
and it includes new lines and non-letter characters.
I want to remove all of them and split this text to have one word per line, then I can count how many of each word exist.";

var words = Regex.Matches(text, @"[A-Za-z ]+").Cast<Match>().SelectMany(n => n.Value.Trim().Split(' '));
int wordCount = words.Count();

来源：https://stackoverflow.com/questions/58734248/how-to-count-words-frequency-by-removing-non-letters-of-a-string