public static string Numerize(string value)
{
var result = value;
// preprocess
result = @" +|([^\d])-([^\d])".Compile().Replace(result, "$1 $2");
// will mutilate hyphenated-words but shouldn't matter for date extraction
result = result.Replace("a half", "haAlf");
// take the 'a' out so it doesn't turn into a 1, save the half for the end
// easy/direct replacements
DIRECT_NUMS.ForEach <string, string>(
(p, r) =>
result =
Regex.Replace(
result,
p,
"<num>" + r));
ORDINALS.ForEach <string, string>(
(p, r) =>
result =
Regex.Replace(
result,
p,
"<num>" + r +
p.LastCharacters
(2)));
// ten, twenty, etc.
TEN_PREFIXES.ForEach <string, int>(
(p, r) =>
result =
Regex.Replace(
result,
"(?:" + p + @") *<num>(\d(?=[^\d]|$))*",
match => "<num>" + (r + int.Parse(match.Groups[1].Value))));
TEN_PREFIXES.ForEach <string, int>(
(p, r) => result = Regex.Replace(result, p, "<num>" + r.ToString()));
// hundreds, thousands, millions, etc.
BIG_PREFIXES.ForEach <string, long>(
(p, r) =>
{
result = Regex.Replace(result, @"(?:<num>)?(\d*) *" + p, match => "<num>" + (r * int.Parse(match.Groups[1].Value)).ToString());
result = Andition(result);
});
// fractional addition
// I'm not combining this with the previous block as using float addition complicates the strings
// (with extraneous .0"s and such )
result = Regex.Replace(result, @"(\d +)(?: |and | -)*haAlf", match => (float.Parse(match.Groups[1].Value) + 0.5).ToString());
result = result.Replace("<num>", "");
return(result);
}
/// <summary>
/// limitation à 69
/// </summary>
/// <param name="result"></param>
/// <returns></returns>
private static string NumerizeFrench(string result)
{
result = @" +|([^\d])-([^\d])".Compile().Replace(result, "$1 $2");
result = result.Replace("et demi", "etDemi");
FRENCH_DIRECT_NUMS.ForEach <string, string>(
(p, r) =>
result =
Regex.Replace(
result,
p,
"<num>" + r));
FRENCH_ORDINALS.ForEach <string, string>(
(p, r) =>
result =
Regex.Replace(
result,
p,
"<num>" + r +
p.
LastCharacters
(2)));
// ten, twenty, etc.
FRENCH_TEN_PREFIXES.ForEach <string, int>(
(p, r) =>
result =
Regex.Replace(
result,
"(?:" + p + @")(?: et)? *<num>(\d(?=[^\d]|$))*",
match => "<num>" + (r + int.Parse(match.Groups[1].Value))));
FRENCH_TEN_PREFIXES.ForEach <string, int>(
(p, r) => result = Regex.Replace(result, p, "<num>" + r.ToString()));
// hundreds, thousands, millions, etc.
FRENCH_BIG_PREFIXES.ForEach <string, long>(
(p, r) =>
{
result = Regex.Replace(result, @"(?:<num>)?(\d*) *" + p, match => "<num>" + (r * int.Parse(match.Groups[1].Value == string.Empty ? "1" : match.Groups[1].Value)).ToString());
result = Andition(result);
});
// fractional addition
// I'm not combining this with the previous block as using float addition complicates the strings
// (with extraneous .0"s and such )
result = Regex.Replace(result, @"(\d +)(?: |et | -)*etDemi", match => (float.Parse(match.Groups[1].Value) + 0.5).ToString());
result = result.Replace("<num>", "");
return(result);
}
public static string Numerize(string value, bool intendTime)
{
var result = value;
string pattern;
MatchCollection matches;
// preprocess
result = @" +|([^\d])-([^\d])".Compile().Replace(result, "$1 $2");
// will mutilate hyphenated-words but shouldn't matter for date extraction
result = result.Replace("a half", "haAlf");
// take the 'a' out so it doesn't turn into a 1, save the half for the end
// easy/direct replacements
DIRECT_NUMS.ForEach <string, string>(
(p, r) =>
result =
Regex.Replace(
result,
p,
"<num>" + r));
ORDINALS.ForEach <string, string>(
(p, r) =>
result =
Regex.Replace(
result,
p,
"<num>" + r +
p.
LastCharacters
(2)));
// ten, twenty, etc.
TEN_PREFIXES.ForEach <string, int>(
(p, r) =>
result =
Regex.Replace(
result,
"(?:" + p + @") *<num>(\d(?=[^\d]|$))*",
match => "<num>" + (r + int.Parse(match.Groups[1].Value))));
var prefix = intendTime ? "<ten><num>" : "<num>";
TEN_PREFIXES.ForEach <string, int>(
(p, r) => result = Regex.Replace(result, p, prefix + r.ToString()));
if (intendTime)
{
var tens = result.Split(new string[] { "<ten>" }, StringSplitOptions.None);
var newResult = tens[0];
for (int i = 1; i < tens.Length; i++)
{
var clean = tens[i].Replace("- <num>", "<num>");
newResult += Andition(clean);
}
result = newResult;
pattern = @"<num>\b(0*(?:[0-9][0-9]?))\b";
matches = Regex.Matches(result, pattern);
if (matches.Count > 0)
{
for (int i = matches.Count - 1; i > 0; i--)
{
dynamic match = matches[i];
var index = match.Groups[1].Index;
if (match.Groups[1].Value.Length == 1)
{
result = result.Insert(index, "0");
}
result = result.Insert(index, ":");
var length = " <num>".Length;
result = result.Remove(index - length, length);
}
}
}
else
{
result = result.Replace("<ten>", "");
}
// hundreds, thousands, millions, etc.
BIG_PREFIXES.ForEach <string, long>(
(p, r) =>
{
result = Regex.Replace(result, @"(?:<num>)?(\d*) *" + p, match => "<num>" + (r * int.Parse(match.Groups[1].Value)).ToString());
result = Andition(result);
});
// fractional addition
// I'm not combining this with the previous block as using float addition complicates the strings
// (with extraneous .0"s and such )
result = Regex.Replace(result, @"(\d +)(?: |and | -)*haAlf", match => (float.Parse(match.Groups[1].Value) + 0.5).ToString());
result = result.Replace("<num>", "");
return(result);
}
