mirror of
https://github.com/MindWorkAI/AI-Studio.git
synced 2026-11-13 13:07:21 +00:00
485 lines
23 KiB
C#
485 lines
23 KiB
C#
|
using System.Globalization;
|
|||
|
|
using System.Text;
|
|||
|
|
using System.Text.RegularExpressions;
|
|||
|
|
|
|||
|
|
namespace AIStudio.Assistants.BeyondFactsLab;
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Removes the personal details it recognizes from a session of the BeyondFacts Lab before the person shares it: links,
|
|||
|
|
/// e-mail addresses, paths, phone numbers, and the names and terms the person lists.
|
|||
|
|
/// </summary>
|
|||
|
|
/// <remarks>
|
|||
|
|
/// <para>
|
|||
|
|
/// The anonymizer follows rules and asks no model, so it recognizes only what its rules describe. Names count only when
|
|||
|
|
/// the person lists them; the lab suggests the ones it knows, see BflShareNames. A name is replaced in full, with its
|
|||
|
|
/// titles, in the genitive, with umlauts written out or left out, and in the forms of web addresses, such as
|
|||
|
|
/// "anna-mueller"; its first and its last name are replaced on their own as well, but only as they are written with a
|
|||
|
|
/// capital letter, since many names are common words. A part which several people share becomes a placeholder
|
|||
|
|
/// without a number.
|
|||
|
|
/// </para>
|
|||
|
|
/// <para>
|
|||
|
|
/// Running the anonymizer again over its own result changes nothing: it leaves its placeholders alone, and the same
|
|||
|
|
/// lists give the same numbers. That is why the person may edit the result and have it anonymized once more.
|
|||
|
|
/// </para>
|
|||
|
|
/// </remarks>
|
|||
|
|
public static partial class BflAnonymizer
|
|||
|
|
{
|
|||
|
|
private static string TB(string fallbackEN) => Tools.PluginSystem.I18N.I.T(fallbackEN, typeof(BflAnonymizer).Namespace, nameof(BflAnonymizer));
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// The fewest digits a phone number has; a shorter number may be anything.
|
|||
|
|
/// </summary>
|
|||
|
|
private const int MIN_PHONE_DIGITS = 7;
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// The fewest letters a part of a name needs to be replaced on its own, so that an initial stays.
|
|||
|
|
/// </summary>
|
|||
|
|
private const int MIN_PART_LETTERS = 2;
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// The titles which may stand before a name, longer ones first.
|
|||
|
|
/// </summary>
|
|||
|
|
private const string TITLE = @"(?:Professorin|Professor|Prof|Dr\.?-Ing|Dipl\.?-Ing|Dr|PD|Herrn|Herr|Frau|Mrs|Mr|Ms|Mx|Sir|Dame)";
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Any number of titles before a name, each followed by a period or a space, e.g., "Prof. Dr. ".
|
|||
|
|
/// </summary>
|
|||
|
|
private const string TITLES = $@"(?:{TITLE}(?:\.[ \t]*|[ \t]+))*";
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// What may stand between the parts of a name: spaces, or the characters with which web addresses join words.
|
|||
|
|
/// </summary>
|
|||
|
|
private const string PART_SEPARATOR = @"(?:\s+|[._-])?";
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Where a name or a term may begin: not inside a word.
|
|||
|
|
/// </summary>
|
|||
|
|
private const string BEFORE = @"(?<![\p{L}\p{M}\p{N}_])";
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Where a name or a term may end: not inside a word.
|
|||
|
|
/// </summary>
|
|||
|
|
private const string AFTER = @"(?![\p{L}\p{M}\p{N}_])";
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Where a part of a name on its own may begin: neither inside a word nor after the hyphen of a double name.
|
|||
|
|
/// </summary>
|
|||
|
|
private const string BEFORE_PART = @"(?<![\p{L}\p{M}\p{N}_])(?<!\p{L}[-‐‑])";
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Where a part of a name on its own may end: neither inside a word nor before the hyphen of a double name.
|
|||
|
|
/// </summary>
|
|||
|
|
private const string AFTER_PART = @"(?![\p{L}\p{M}\p{N}_]|[-‐‑]\p{L})";
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// The small words with which many last names begin, e.g., "von der Leyen".
|
|||
|
|
/// </summary>
|
|||
|
|
internal static readonly IReadOnlySet<string> NAME_PARTICLES = new HashSet<string>(StringComparer.OrdinalIgnoreCase)
|
|||
|
|
{
|
|||
|
|
"von", "vom", "van", "de", "der", "den", "del", "della", "di", "da", "du", "la", "le", "zu", "zum", "zur", "ten", "ter",
|
|||
|
|
};
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Replaces the personal details the anonymizer recognizes.
|
|||
|
|
/// </summary>
|
|||
|
|
/// <param name="text">The text, e.g., a session of the lab in Markdown.</param>
|
|||
|
|
/// <param name="people">The names of people to replace, one person each, e.g., "Prof. Dr. Anna Müller".</param>
|
|||
|
|
/// <param name="terms">Further terms to replace, e.g., an institution or a place.</param>
|
|||
|
|
/// <returns>The text with placeholders, and what was replaced.</returns>
|
|||
|
|
public static BflAnonymization Anonymize(string text, IReadOnlyList<string> people, IReadOnlyList<string> terms)
|
|||
|
|
{
|
|||
|
|
var report = new List<BflReplacement>();
|
|||
|
|
var link = Placeholder(TB("Link"));
|
|||
|
|
var email = Placeholder(TB("E-mail address"));
|
|||
|
|
var path = Placeholder(TB("Path"));
|
|||
|
|
var phone = Placeholder(TB("Phone number"));
|
|||
|
|
var result = text.Normalize(NormalizationForm.FormC);
|
|||
|
|
|
|||
|
|
// Links first, since an address may hold an e-mail address, a path, or a name:
|
|||
|
|
result = MarkdownLinkPattern().Replace(result, match =>
|
|||
|
|
{
|
|||
|
|
Record(report, link, match.Groups["url"].Value);
|
|||
|
|
var label = match.Groups["label"].Value.Trim();
|
|||
|
|
return label.Length > 0 ? $"{label} {link}" : link;
|
|||
|
|
});
|
|||
|
|
|
|||
|
|
result = AutoLinkPattern().Replace(result, match => Record(report, link, match.Groups["url"].Value));
|
|||
|
|
result = UrlPattern().Replace(result, match => ReplaceWithoutEnd(report, link, match.Value));
|
|||
|
|
result = EmailPattern().Replace(result, match => Record(report, email, match.Value));
|
|||
|
|
result = PathPattern().Replace(result, match => ReplaceWithoutEnd(report, path, match.Value));
|
|||
|
|
result = PhonePattern().Replace(result, match => IsPhoneNumber(match.Value) ? Record(report, phone, match.Value) : match.Value);
|
|||
|
|
result = ReplaceNamesAndTerms(result, people, terms, report, [link, email, path, phone]);
|
|||
|
|
return new BflAnonymization(result, report);
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Gets the parts of a name as the person listed it: without titles and remarks in parentheses, and in the order
|
|||
|
|
/// first name, last name, also when the person wrote "Müller, Anna".
|
|||
|
|
/// </summary>
|
|||
|
|
/// <param name="name">The name.</param>
|
|||
|
|
/// <returns>The parts, empty when nothing like a name remains.</returns>
|
|||
|
|
internal static IReadOnlyList<string> PartsOfName(string name)
|
|||
|
|
{
|
|||
|
|
var cleaned = ParenthesesPattern().Replace(name.Normalize(NormalizationForm.FormC), " ").Replace("**", " ").Trim().Trim(',', ';', ':', '-', '–', '*', '_').Trim();
|
|||
|
|
var commaParts = cleaned.Split(',', StringSplitOptions.TrimEntries | StringSplitOptions.RemoveEmptyEntries);
|
|||
|
|
if (commaParts.Length > 1)
|
|||
|
|
{
|
|||
|
|
// "Müller, Anna" names the last name first; "Anna Müller, TUM" adds a remark:
|
|||
|
|
cleaned = commaParts.Length is 2 && !commaParts[0].Contains(' ') && !commaParts[1].Contains(' ') ? $"{commaParts[1]} {commaParts[0]}" : commaParts[0];
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
var parts = cleaned.Split((char[]?)null, StringSplitOptions.RemoveEmptyEntries).Select(part => part.Trim(',', ';', ':')).Where(part => part.Any(char.IsLetter)).ToList();
|
|||
|
|
while (parts.Count > 0 && TitlePattern().IsMatch(parts[0]))
|
|||
|
|
parts.RemoveAt(0);
|
|||
|
|
|
|||
|
|
return parts;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Tells whether a word is a title which may stand before a name, e.g., "Prof." or "Dr.".
|
|||
|
|
/// </summary>
|
|||
|
|
/// <param name="word">The word.</param>
|
|||
|
|
internal static bool IsTitle(string word) => TitlePattern().IsMatch(word);
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Replaces the names of people and the further terms.
|
|||
|
|
/// </summary>
|
|||
|
|
/// <param name="text">The text.</param>
|
|||
|
|
/// <param name="people">The names of people, one person each.</param>
|
|||
|
|
/// <param name="terms">The further terms.</param>
|
|||
|
|
/// <param name="report">What was replaced so far.</param>
|
|||
|
|
/// <param name="fixedPlaceholders">The placeholders of links, e-mail addresses, paths, and phone numbers, which stay as they are.</param>
|
|||
|
|
/// <returns>The text with placeholders.</returns>
|
|||
|
|
private static string ReplaceNamesAndTerms(string text, IReadOnlyList<string> people, IReadOnlyList<string> terms, List<BflReplacement> report, IReadOnlyList<string> fixedPlaceholders)
|
|||
|
|
{
|
|||
|
|
var personLabel = TB("Person");
|
|||
|
|
var termLabel = TB("Term");
|
|||
|
|
var patterns = new List<BflNamePattern>();
|
|||
|
|
var singleParts = new Dictionary<string, string>(StringComparer.Ordinal);
|
|||
|
|
var ambiguous = Placeholder(personLabel);
|
|||
|
|
|
|||
|
|
var knownPeople = new HashSet<string>(StringComparer.OrdinalIgnoreCase);
|
|||
|
|
var number = 0;
|
|||
|
|
foreach (var name in people)
|
|||
|
|
{
|
|||
|
|
var parts = PartsOfName(name);
|
|||
|
|
if (parts.Count is 0 || !knownPeople.Add(string.Join(' ', parts)))
|
|||
|
|
continue;
|
|||
|
|
|
|||
|
|
var placeholder = Placeholder($"{personLabel} {++number}");
|
|||
|
|
if (parts.Count is 1)
|
|||
|
|
{
|
|||
|
|
AddSinglePart(singleParts, parts[0], placeholder, ambiguous);
|
|||
|
|
continue;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
var first = parts[0];
|
|||
|
|
var lastName = LastNameOf(parts);
|
|||
|
|
patterns.Add(new(TITLES + string.Join(PART_SEPARATOR, parts.Select(PartPattern)), string.Join(' ', parts).Length, placeholder, true));
|
|||
|
|
patterns.Add(new(PhrasePattern(lastName) + @"(?:,\s*|\s+|[._-])?" + PartPattern(first), lastName.Length + first.Length + 1, placeholder, true));
|
|||
|
|
if (parts.Count > 2)
|
|||
|
|
patterns.Add(new(TITLES + PartPattern(first) + PART_SEPARATOR + PhrasePattern(lastName), first.Length + lastName.Length + 1, placeholder, true));
|
|||
|
|
|
|||
|
|
AddSinglePart(singleParts, first, placeholder, ambiguous);
|
|||
|
|
AddSinglePart(singleParts, lastName, placeholder, ambiguous);
|
|||
|
|
if (lastName.Contains(' '))
|
|||
|
|
AddSinglePart(singleParts, parts[^1], placeholder, ambiguous);
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
foreach (var (part, placeholder) in singleParts)
|
|||
|
|
patterns.Add(new(TITLES + PhrasePattern(part), part.Length, placeholder, false));
|
|||
|
|
|
|||
|
|
var knownTerms = new HashSet<string>(StringComparer.OrdinalIgnoreCase);
|
|||
|
|
number = 0;
|
|||
|
|
foreach (var term in terms)
|
|||
|
|
{
|
|||
|
|
var parts = term.Normalize(NormalizationForm.FormC).Split((char[]?)null, StringSplitOptions.RemoveEmptyEntries).Where(part => part.Any(char.IsLetterOrDigit)).ToList();
|
|||
|
|
if (parts.Count is 0 || !knownTerms.Add(string.Join(' ', parts)))
|
|||
|
|
continue;
|
|||
|
|
|
|||
|
|
patterns.Add(new(string.Join(PART_SEPARATOR, parts.Select(PartPattern)), string.Join(' ', parts).Length, Placeholder($"{termLabel} {++number}"), true));
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
if (patterns.Count is 0)
|
|||
|
|
return text;
|
|||
|
|
|
|||
|
|
// The longest names first, so that "Anna Schmidt-Weber" is not taken for "Anna Schmidt":
|
|||
|
|
patterns = [.. patterns.OrderByDescending(pattern => pattern.Length)];
|
|||
|
|
var expression = new StringBuilder();
|
|||
|
|
expression.Append("(?<keep>");
|
|||
|
|
foreach (var placeholder in fixedPlaceholders)
|
|||
|
|
expression.Append(Regex.Escape(placeholder)).Append('|');
|
|||
|
|
|
|||
|
|
expression.Append(Regex.Escape($"[{personLabel}")).Append(@"(?: \d+)?\]|").Append(Regex.Escape($"[{termLabel}")).Append(@"(?: \d+)?\])");
|
|||
|
|
for (var index = 0; index < patterns.Count; index++)
|
|||
|
|
{
|
|||
|
|
//
|
|||
|
|
// A part of a name on its own must not be a part of a double name, e.g., "Schmidt" in "Müller-Schmidt",
|
|||
|
|
// which may be somebody else. A full name or a term may stand in a longer word with a hyphen, e.g., in
|
|||
|
|
// the address "anna-mueller-lab":
|
|||
|
|
//
|
|||
|
|
var pattern = patterns[index];
|
|||
|
|
var groupIndex = index.ToString(CultureInfo.InvariantCulture);
|
|||
|
|
expression.Append('|').Append(pattern.IgnoresCase ? BEFORE : BEFORE_PART);
|
|||
|
|
expression.Append("(?<v").Append(groupIndex).Append('>').Append(pattern.IgnoresCase ? "(?i:" : "(?:").Append(pattern.Pattern).Append("))");
|
|||
|
|
expression.Append("(?<s").Append(groupIndex).Append(">['’]s|['’]|s)?");
|
|||
|
|
expression.Append(pattern.IgnoresCase ? AFTER : AFTER_PART);
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
var regex = new Regex(expression.ToString(), RegexOptions.CultureInvariant);
|
|||
|
|
var nameGroups = Enumerable.Range(0, patterns.Count).Select(index => regex.GroupNumberFromName($"v{index.ToString(CultureInfo.InvariantCulture)}")).ToArray();
|
|||
|
|
var suffixGroups = Enumerable.Range(0, patterns.Count).Select(index => regex.GroupNumberFromName($"s{index.ToString(CultureInfo.InvariantCulture)}")).ToArray();
|
|||
|
|
return regex.Replace(text, match =>
|
|||
|
|
{
|
|||
|
|
if (match.Groups["keep"].Success)
|
|||
|
|
return match.Value;
|
|||
|
|
|
|||
|
|
for (var index = 0; index < patterns.Count; index++)
|
|||
|
|
{
|
|||
|
|
// The genitive stays after the placeholder, e.g., "[Person 1]s":
|
|||
|
|
var group = match.Groups[nameGroups[index]];
|
|||
|
|
if (group.Success)
|
|||
|
|
return Record(report, patterns[index].Placeholder, group.Value) + match.Groups[suffixGroups[index]].Value;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
return match.Value;
|
|||
|
|
});
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Gets the last name of a name with several parts: from its first particle on, e.g., "von der Leyen", otherwise
|
|||
|
|
/// its last part.
|
|||
|
|
/// </summary>
|
|||
|
|
/// <param name="parts">The parts of the name, at least two.</param>
|
|||
|
|
private static string LastNameOf(IReadOnlyList<string> parts)
|
|||
|
|
{
|
|||
|
|
for (var index = 1; index < parts.Count - 1; index++)
|
|||
|
|
{
|
|||
|
|
if (NAME_PARTICLES.Contains(parts[index]))
|
|||
|
|
return string.Join(' ', parts.Skip(index));
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
return parts[^1];
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Adds a part of a name which the anonymizer replaces on its own; a part several people share gets the
|
|||
|
|
/// placeholder without a number.
|
|||
|
|
/// </summary>
|
|||
|
|
/// <param name="singleParts">The parts so far, with their placeholders.</param>
|
|||
|
|
/// <param name="part">The part, e.g., a first name.</param>
|
|||
|
|
/// <param name="placeholder">The placeholder of the person.</param>
|
|||
|
|
/// <param name="ambiguous">The placeholder without a number.</param>
|
|||
|
|
private static void AddSinglePart(Dictionary<string, string> singleParts, string part, string placeholder, string ambiguous)
|
|||
|
|
{
|
|||
|
|
if (part.Count(char.IsLetter) < MIN_PART_LETTERS || NAME_PARTICLES.Contains(part))
|
|||
|
|
return;
|
|||
|
|
|
|||
|
|
singleParts[part] = singleParts.TryGetValue(part, out var known) && known != placeholder ? ambiguous : placeholder;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Gets the pattern of several parts of a name which stand one after the other, e.g., "von der Leyen".
|
|||
|
|
/// </summary>
|
|||
|
|
/// <param name="phrase">The parts, separated by spaces.</param>
|
|||
|
|
private static string PhrasePattern(string phrase) => string.Join(@"\s+", phrase.Split(' ').Select(PartPattern));
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Gets the pattern of one part of a name: umlauts match their spellings with and without them, letters with
|
|||
|
|
/// accents also without them, and the period of an initial may be missing.
|
|||
|
|
/// </summary>
|
|||
|
|
/// <param name="part">The part of the name.</param>
|
|||
|
|
private static string PartPattern(string part)
|
|||
|
|
{
|
|||
|
|
var pattern = new StringBuilder();
|
|||
|
|
for (var index = 0; index < part.Length; index++)
|
|||
|
|
{
|
|||
|
|
var character = part[index];
|
|||
|
|
var next = index + 1 < part.Length ? part[index + 1] : '\0';
|
|||
|
|
if (next is 'e' or 'E' && character is 'a' or 'o' or 'u' or 'A' or 'O' or 'U')
|
|||
|
|
{
|
|||
|
|
// A name written as "Mueller" also stands for "Müller":
|
|||
|
|
pattern.Append("(?:").Append(Regex.Escape(part.Substring(index, 2))).Append('|').Append(UmlautOf(character)).Append(')');
|
|||
|
|
index++;
|
|||
|
|
continue;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
if (character is 's' && next is 's')
|
|||
|
|
{
|
|||
|
|
pattern.Append("(?:ss|ß)");
|
|||
|
|
index++;
|
|||
|
|
continue;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
pattern.Append(character switch
|
|||
|
|
{
|
|||
|
|
'ä' => "(?:ä|ae|a)",
|
|||
|
|
'ö' => "(?:ö|oe|o)",
|
|||
|
|
'ü' => "(?:ü|ue|u)",
|
|||
|
|
'Ä' => "(?:Ä|Ae|AE|A)",
|
|||
|
|
'Ö' => "(?:Ö|Oe|OE|O)",
|
|||
|
|
'Ü' => "(?:Ü|Ue|UE|U)",
|
|||
|
|
'ß' => "(?:ß|ss)",
|
|||
|
|
// A double name may lose its hyphen in an address, e.g., "MuellerLuedenscheidt":
|
|||
|
|
'-' or '‐' or '‑' => "[-‐‑._]?",
|
|||
|
|
'\'' or '’' => "['’]",
|
|||
|
|
'.' => @"\.?",
|
|||
|
|
|
|||
|
|
_ => LetterPattern(character),
|
|||
|
|
});
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
return pattern.ToString();
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Gets the umlaut of a vowel, e.g., "ü" for "u".
|
|||
|
|
/// </summary>
|
|||
|
|
/// <param name="vowel">The vowel a, o, or u, in either case.</param>
|
|||
|
|
private static string UmlautOf(char vowel) => vowel switch
|
|||
|
|
{
|
|||
|
|
'a' => "ä",
|
|||
|
|
'o' => "ö",
|
|||
|
|
'u' => "ü",
|
|||
|
|
'A' => "Ä",
|
|||
|
|
'O' => "Ö",
|
|||
|
|
|
|||
|
|
_ => "Ü",
|
|||
|
|
};
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Gets the pattern of a letter, which matches a letter with an accent also without it.
|
|||
|
|
/// </summary>
|
|||
|
|
/// <param name="character">The letter.</param>
|
|||
|
|
private static string LetterPattern(char character)
|
|||
|
|
{
|
|||
|
|
var decomposed = character.ToString().Normalize(NormalizationForm.FormD);
|
|||
|
|
if (decomposed.Length > 1 && decomposed.Skip(1).All(mark => CharUnicodeInfo.GetUnicodeCategory(mark) is UnicodeCategory.NonSpacingMark))
|
|||
|
|
return $"(?:{Regex.Escape(character.ToString())}|{Regex.Escape(decomposed[..1])})";
|
|||
|
|
|
|||
|
|
return Regex.Escape(character.ToString());
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Tells whether a run of digits is a phone number rather than a date, a time, or a decimal number.
|
|||
|
|
/// </summary>
|
|||
|
|
/// <param name="candidate">The run of digits, as the pattern found it.</param>
|
|||
|
|
private static bool IsPhoneNumber(string candidate) => candidate.Count(char.IsDigit) >= MIN_PHONE_DIGITS && !NotAPhoneNumberPattern().IsMatch(candidate);
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Replaces an address or a path, but leaves the punctuation after it, which belongs to the sentence.
|
|||
|
|
/// </summary>
|
|||
|
|
/// <param name="report">What was replaced so far.</param>
|
|||
|
|
/// <param name="placeholder">The placeholder.</param>
|
|||
|
|
/// <param name="value">What the pattern found.</param>
|
|||
|
|
/// <returns>The placeholder and the punctuation after it.</returns>
|
|||
|
|
private static string ReplaceWithoutEnd(List<BflReplacement> report, string placeholder, string value)
|
|||
|
|
{
|
|||
|
|
var length = value.Length;
|
|||
|
|
while (length > 0)
|
|||
|
|
{
|
|||
|
|
var last = value[length - 1];
|
|||
|
|
var opened = value[..length].Count(character => character == OpeningOf(last));
|
|||
|
|
var closed = value[..length].Count(character => character == last);
|
|||
|
|
if (last is '.' or ',' or ';' or ':' or '!' or '?' or '*' or '_' or '\'' or '"' || last is ')' or ']' or '}' && closed > opened)
|
|||
|
|
length--;
|
|||
|
|
else
|
|||
|
|
break;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
return length is 0 ? value : Record(report, placeholder, value[..length]) + value[length..];
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Gets the opening bracket of a closing one, or a character no text holds.
|
|||
|
|
/// </summary>
|
|||
|
|
/// <param name="closing">The closing bracket.</param>
|
|||
|
|
private static char OpeningOf(char closing) => closing switch
|
|||
|
|
{
|
|||
|
|
')' => '(',
|
|||
|
|
']' => '[',
|
|||
|
|
'}' => '{',
|
|||
|
|
|
|||
|
|
_ => '\0',
|
|||
|
|
};
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Notes a replacement in the report.
|
|||
|
|
/// </summary>
|
|||
|
|
/// <param name="report">What was replaced so far.</param>
|
|||
|
|
/// <param name="placeholder">The placeholder.</param>
|
|||
|
|
/// <param name="original">What it replaces.</param>
|
|||
|
|
/// <returns>The placeholder.</returns>
|
|||
|
|
private static string Record(List<BflReplacement> report, string placeholder, string original)
|
|||
|
|
{
|
|||
|
|
var index = report.FindIndex(replacement => replacement.Placeholder == placeholder && replacement.Original == original);
|
|||
|
|
if (index < 0)
|
|||
|
|
report.Add(new BflReplacement(placeholder, original, 1));
|
|||
|
|
else
|
|||
|
|
report[index] = report[index] with { Count = report[index].Count + 1 };
|
|||
|
|
|
|||
|
|
return placeholder;
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// Gets a placeholder, e.g., "[Person 1]".
|
|||
|
|
/// </summary>
|
|||
|
|
/// <param name="label">What it stands for.</param>
|
|||
|
|
private static string Placeholder(string label) => $"[{label}]";
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// A link in Markdown, whose text stays and whose address goes.
|
|||
|
|
/// </summary>
|
|||
|
|
[GeneratedRegex("""\[(?<label>[^\]\n]*)\]\(\s*<?(?<url>(?:[a-z][a-z0-9+.-]*:|www\.)[^\s<>()]*(?:\([^\s<>()]*\)[^\s<>()]*)*)>?(?:\s+"[^"\n]*")?\s*\)""", RegexOptions.IgnoreCase | RegexOptions.CultureInvariant)]
|
|||
|
|
private static partial Regex MarkdownLinkPattern();
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// An address in angle brackets, e.g., "<https://example.org>".
|
|||
|
|
/// </summary>
|
|||
|
|
[GeneratedRegex(@"<(?<url>(?:https?|ftp|file|mailto):[^\s<>]+)>", RegexOptions.IgnoreCase | RegexOptions.CultureInvariant)]
|
|||
|
|
private static partial Regex AutoLinkPattern();
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// An address of a web page or a file, with its scheme or beginning with "www.".
|
|||
|
|
/// </summary>
|
|||
|
|
[GeneratedRegex(@"(?<![\p{L}\p{N}@.])(?:(?:https?|ftp|file)://|www\.)[^\s<>""'`]+", RegexOptions.IgnoreCase | RegexOptions.CultureInvariant)]
|
|||
|
|
private static partial Regex UrlPattern();
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// An e-mail address.
|
|||
|
|
/// </summary>
|
|||
|
|
[GeneratedRegex(@"(?<![\p{L}\p{N}._%+-])[\p{L}\p{N}._%+-]+@[\p{L}\p{N}-]+(?:\.[\p{L}\p{N}-]+)*\.\p{L}{2,}", RegexOptions.CultureInvariant)]
|
|||
|
|
private static partial Regex EmailPattern();
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// A path of a file or a folder: on Windows with its drive or its server, otherwise beginning with "/" or "~" and
|
|||
|
|
/// with at least two folders.
|
|||
|
|
/// </summary>
|
|||
|
|
[GeneratedRegex(@"(?<![\p{L}\p{N}])[A-Za-z]:[\\/][^\s""<>|*?]*|\\\\[^\s\\""<>|*?]+(?:\\[^\s""<>|*?]*)+|%[A-Za-z_]+%\\[^\s""<>|*?]*|(?<![\p{L}\p{M}\p{N}/.:~\\\])_-])(?:(?:~|\$HOME)(?:/[\p{L}\p{M}\p{N}._@+-]+)+|(?:/[\p{L}\p{M}\p{N}._@+-]+){2,})/?", RegexOptions.CultureInvariant)]
|
|||
|
|
private static partial Regex PathPattern();
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// A run which may be a phone number: beginning with "+" or "0", with spaces, periods, slashes, hyphens, or
|
|||
|
|
/// parentheses between the digits; IsPhoneNumber decides.
|
|||
|
|
/// </summary>
|
|||
|
|
[GeneratedRegex(@"(?<![\p{L}\p{N}+(])\(?(?:\+|0)\d[\d \t./()-]{5,}\d(?![\p{L}\p{N}])", RegexOptions.CultureInvariant)]
|
|||
|
|
private static partial Regex PhonePattern();
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// What looks like a phone number but is not: a date, a span of time, or a decimal number.
|
|||
|
|
/// </summary>
|
|||
|
|
[GeneratedRegex(@"^\(?\d{1,2}\.\d{1,2}\.(?:\d{4}|\d{2})$|^\d{1,2}[.:]\d{2}\s*-\s*\d{1,2}[.:]\d{2}$|^0[.,]", RegexOptions.CultureInvariant)]
|
|||
|
|
private static partial Regex NotAPhoneNumberPattern();
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// A remark in parentheses after a name, e.g., "(TUM)".
|
|||
|
|
/// </summary>
|
|||
|
|
[GeneratedRegex(@"\([^)]*\)", RegexOptions.CultureInvariant)]
|
|||
|
|
private static partial Regex ParenthesesPattern();
|
|||
|
|
|
|||
|
|
/// <summary>
|
|||
|
|
/// A title on its own, e.g., "Prof." or "Dr.".
|
|||
|
|
/// </summary>
|
|||
|
|
[GeneratedRegex($@"^{TITLE}\.?$", RegexOptions.CultureInvariant)]
|
|||
|
|
private static partial Regex TitlePattern();
|
|||
|
|
}
|