private void ReplaceNames1Remove(List <string> namesEtc, List <string> replaceIds, List <string> replaceNames, List <string> originalNames)
{
if (Post.StartsWith('.'))
{
StrippedText += ".";
Post = Post.Remove(0, 1);
}
string lower = StrippedText.ToLower();
foreach (string name in namesEtc)
{
int start = lower.IndexOf(name, StringComparison.OrdinalIgnoreCase);
while (start >= 0 && start < lower.Length)
{
bool startOk = (start == 0) || (lower[start - 1] == ' ') || (lower[start - 1] == '-') ||
(lower[start - 1] == '"') || (lower[start - 1] == '\'') || (lower[start - 1] == '>') ||
Environment.NewLine.EndsWith(lower[start - 1]);
if (startOk && string.CompareOrdinal(name, "Don") == 0 && lower.Substring(start).StartsWith("don't"))
{
startOk = false;
}
if (startOk)
{
int end = start + name.Length;
bool endOk = end <= lower.Length;
if (endOk)
{
endOk = end == lower.Length || (@" ,.!?:;')- <""" + Environment.NewLine).Contains(lower[end]);
}
if (endOk && StrippedText.Length >= start + name.Length)
{
string originalName = StrippedText.Substring(start, name.Length);
originalNames.Add(originalName);
StrippedText = StrippedText.Remove(start, name.Length);
StrippedText = StrippedText.Insert(start, GetAndInsertNextId(replaceIds, replaceNames, name));
lower = StrippedText.ToLower();
}
}
if (start + 3 > lower.Length)
{
start = lower.Length + 1;
}
else
{
start = lower.IndexOf(name.ToLower(), start + 3, StringComparison.Ordinal);
}
}
}
if (StrippedText.EndsWith('.'))
{
Post = "." + Post;
StrippedText = StrippedText.TrimEnd('.');
}
}
private bool ShouldStartWithUpperCase(string lastLine, double millisecondsgaps)
{
// do not capitalize url
if (StrippedText.StartsWith("www.", StringComparison.OrdinalIgnoreCase) || StrippedText.StartsWith("http", StringComparison.OrdinalIgnoreCase))
{
return(false);
}
// do not capitalize word like iPhone
if (StrippedText.Length > 1 && StrippedText[0] == 'i' && char.IsUpper(StrippedText[1]))
{
return(false);
}
// shouldn't capitalize current line not closed
if (Pre.Contains("...") || Pre.Contains("…"))
{
return(false);
}
// too much gaps between lines, so should be considered as closed
if (millisecondsgaps > 5000)
{
return(true);
}
var preLine = HtmlUtil.RemoveHtmlTags(lastLine).TrimEnd().TrimEnd('\"', '”').TrimEnd();
// check if previous line was fully closed
if (string.IsNullOrEmpty(preLine))
{
return(true);
}
char lastChar = preLine[preLine.Length - 1];
if (lastChar == '♪')
{
string tempPreLine = preLine.Substring(0, preLine.Length - 1).TrimEnd();
// update last char
if (tempPreLine.Length > 0)
{
lastChar = tempPreLine[tempPreLine.Length - 1];
}
}
if (lastChar == '.' || lastChar == '!' || lastChar == '?' || lastChar == ']' || lastChar == ')' || lastChar == ':' || lastChar == '_')
{
return(true);
}
// previous line ends with music symbol but current line doesn't contains any music symbol
if ((preLine.EndsWith('♪') || preLine.EndsWith('♫')) && !Pre.Contains(new[] { '♪', '♫' }))
{
return(true);
}
// do not capitalize
return(false);
}
public void FixCasing(List <string> namesEtc, bool changeNameCases, bool makeUppercaseAfterBreak, bool checkLastLine, string lastLine)
{
var replaceIds = new List <string>();
var replaceNames = new List <string>();
var originalNames = new List <string>();
ReplaceNames1Remove(namesEtc, replaceIds, replaceNames, originalNames);
if (checkLastLine)
{
string s = HtmlUtil.RemoveHtmlTags(lastLine).TrimEnd().TrimEnd('\"').TrimEnd();
bool startWithUppercase = string.IsNullOrEmpty(s) ||
s.EndsWith('.') ||
s.EndsWith('!') ||
s.EndsWith('?') ||
s.EndsWith(". ♪", StringComparison.Ordinal) ||
s.EndsWith("! ♪", StringComparison.Ordinal) ||
s.EndsWith("? ♪", StringComparison.Ordinal) ||
s.EndsWith(']') ||
s.EndsWith(')') ||
s.EndsWith(':');
// start with uppercase after music symbol - but only if next line does not start with music symbol
if (!startWithUppercase && (s.EndsWith('♪') || s.EndsWith('♫')))
{
if (!Pre.Contains(new[] { '♪', '♫' }))
{
startWithUppercase = true;
}
}
if (startWithUppercase && StrippedText.Length > 0 && !Pre.Contains("..."))
{
StrippedText = char.ToUpper(StrippedText[0]) + StrippedText.Substring(1);
}
}
if (makeUppercaseAfterBreak && StrippedText.Contains(ExpectedCharsArray))
{
const string breakAfterChars = @".!?:;)]}([{";
const string ExpectedChars = "\"`´'()<>!?.- \r\n";
var sb = new StringBuilder();
bool lastWasBreak = false;
for (int i = 0; i < StrippedText.Length; i++)
{
var s = StrippedText[i];
if (lastWasBreak)
{
if (ExpectedChars.Contains(s))
{
sb.Append(s);
}
else if ((sb.EndsWith('<') || sb.ToString().EndsWith("</", StringComparison.Ordinal)) && i + 1 < StrippedText.Length && StrippedText[i + 1] == '>')
{ // tags
sb.Append(s);
}
else if (sb.EndsWith('<') && s == '/' && i + 2 < StrippedText.Length && StrippedText[i + 2] == '>')
{ // tags
sb.Append(s);
}
else if (sb.ToString().EndsWith("... ", StringComparison.Ordinal))
{
sb.Append(s);
lastWasBreak = false;
}
else
{
if (breakAfterChars.Contains(s))
{
sb.Append(s);
}
else
{
lastWasBreak = false;
sb.Append(char.ToUpper(s));
}
}
}
else
{
sb.Append(s);
if (breakAfterChars.Contains(s))
{
var idx = sb.ToString().IndexOf('[');
if (s == ']' && idx > 1)
{ // I [Motor roaring] love you!
string temp = sb.ToString(0, idx - 1).Trim();
if (temp.Length > 0 && !Utilities.LowercaseLetters.Contains(temp[temp.Length - 1]))
{
lastWasBreak = true;
}
}
else
{
lastWasBreak = true;
}
}
}
}
StrippedText = sb.ToString();
}
if (changeNameCases)
{
ReplaceNames2Fix(replaceIds, replaceNames);
}
else
{
ReplaceNames2Fix(replaceIds, originalNames);
}
}
public void FixCasing(List <string> nameList, bool changeNameCases, bool makeUppercaseAfterBreak, bool checkLastLine, string lastLine, double millisecondsFromLast = 0)
{
var replaceIds = new List <string>();
var replaceNames = new List <string>();
var originalNames = new List <string>();
ReplaceNames1Remove(nameList, replaceIds, replaceNames, originalNames);
if (checkLastLine && ShouldStartWithUpperCase(lastLine, millisecondsFromLast))
{
if (StrippedText.StartsWith("_@", StringComparison.Ordinal))
{
for (int i = 0; i < replaceIds.Count; i++)
{
string id = $"_@{i}_";
if (StrippedText.StartsWith(id, StringComparison.Ordinal))
{
if (!string.IsNullOrEmpty(originalNames[i]))
{
originalNames[i] = originalNames[i].CapitalizeFirstLetter();
}
break;
}
}
}
else
{
StrippedText = StrippedText.CapitalizeFirstLetter();
}
}
if (makeUppercaseAfterBreak && StrippedText.Contains(ExpectedCharsArray))
{
const string breakAfterChars = @".!?:;)]}([{";
const string expectedChars = "\"“`´'()<>!?.- \r\n";
var sb = new StringBuilder(StrippedText.Length);
bool lastWasBreak = false;
for (int i = 0; i < StrippedText.Length; i++)
{
var s = StrippedText[i];
if (lastWasBreak)
{
if (expectedChars.Contains(s))
{
sb.Append(s);
}
else if ((sb.EndsWith('<') || sb.ToString().EndsWith("</", StringComparison.Ordinal)) && i + 1 < StrippedText.Length && StrippedText[i + 1] == '>')
{ // tags
sb.Append(s);
}
else if (sb.EndsWith('<') && s == '/' && i + 2 < StrippedText.Length && StrippedText[i + 2] == '>')
{ // tags
sb.Append(s);
}
else if (sb.ToString().EndsWith("... ", StringComparison.Ordinal))
{
sb.Append(s);
lastWasBreak = false;
}
else
{
if (breakAfterChars.Contains(s))
{
sb.Append(s);
}
else
{
lastWasBreak = false;
sb.Append(char.ToUpper(s));
if (StrippedText.Substring(i).StartsWith("_@", StringComparison.Ordinal))
{
var ks = StrippedText.Substring(i);
for (int k = 0; k < replaceIds.Count; k++)
{
string id = $"_@{k}_";
if (ks.StartsWith(id, StringComparison.Ordinal))
{
if (!string.IsNullOrEmpty(originalNames[k]))
{
originalNames[k] = char.ToUpper(originalNames[k][0]) + originalNames[k].Remove(0, 1);
}
break;
}
}
}
}
}
}
else
{
sb.Append(s);
if (breakAfterChars.Contains(s))
{
var idx = sb.ToString().IndexOf('[');
if (s == ']' && idx > 1)
{ // I [Motor roaring] love you!
string temp = sb.ToString(0, idx - 1).Trim();
if (temp.Length > 0 && !char.IsLetterOrDigit(temp[temp.Length - 1]))
{
lastWasBreak = true;
}
}
else if (s == ']' && idx == -1 && Pre.Contains('['))
{ // [ Motor roaring ] Hallo!
lastWasBreak = true;
}
else if (s == ':') // seems to be the rule (in subtitles) to nearly always capitalize first letter efter semicolon
{
lastWasBreak = true;
}
else
{
idx = sb.ToString().LastIndexOf(' ');
if (idx >= 0 && idx < sb.Length - 2 && !IsInMiddleOfUrl(i - idx, StrippedText.Substring(idx + 1)))
{
lastWasBreak = true;
}
else if (StrippedText.Length > i + 1 && " \r\n".Contains(StrippedText[i + 1]))
{
lastWasBreak = true;
}
}
}
else if (s == '-' && Pre.Contains('-'))
{
if (sb.ToString().EndsWith(Environment.NewLine + "-"))
{
var prevLine = HtmlUtil.RemoveHtmlTags(sb.ToString().Substring(0, sb.Length - 2).TrimEnd());
if (prevLine.EndsWith('.') ||
prevLine.EndsWith('!') ||
prevLine.EndsWith('?') ||
prevLine.EndsWith(". ♪", StringComparison.Ordinal) ||
prevLine.EndsWith("! ♪", StringComparison.Ordinal) ||
prevLine.EndsWith("? ♪", StringComparison.Ordinal) ||
prevLine.EndsWith(']') ||
prevLine.EndsWith(')') ||
prevLine.EndsWith(':'))
{
lastWasBreak = true;
}
}
}
}
}
StrippedText = sb.ToString();
}
ReplaceNames2Fix(replaceIds, changeNameCases ? replaceNames : originalNames);
}
public void FixCasing(List <string> namesEtc, bool changeNameCases, bool makeUppercaseAfterBreak, bool checkLastLine, string lastLine)
{
var replaceIds = new List <string>();
var replaceNames = new List <string>();
var originalNames = new List <string>();
ReplaceNames1Remove(namesEtc, replaceIds, replaceNames, originalNames);
if (checkLastLine)
{
string s = Utilities.RemoveHtmlTags(lastLine).TrimEnd().TrimEnd('\"').TrimEnd();
bool startWithUppercase = string.IsNullOrEmpty(s) ||
s.EndsWith('.') ||
s.EndsWith('!') ||
s.EndsWith('?') ||
s.EndsWith(". ♪", StringComparison.Ordinal) ||
s.EndsWith("! ♪", StringComparison.Ordinal) ||
s.EndsWith("? ♪", StringComparison.Ordinal) ||
s.EndsWith(']') ||
s.EndsWith(')') ||
s.EndsWith(':');
// start with uppercase after music symbol - but only if next line does not start with music symbol
if (!startWithUppercase && (s.EndsWith('♪') || s.EndsWith('♫')))
{
if (!Pre.Contains("♪") && !Pre.Contains("♫"))
{
startWithUppercase = true;
}
}
if (startWithUppercase && StrippedText.Length > 0 && !Pre.Contains("..."))
{
StrippedText = StrippedText.Remove(0, 1).Insert(0, StrippedText[0].ToString().ToUpper());
}
}
if (makeUppercaseAfterBreak &&
(StrippedText.Contains(".") ||
StrippedText.Contains("!") ||
StrippedText.Contains("?") ||
StrippedText.Contains(":") ||
StrippedText.Contains(";") ||
StrippedText.Contains(")") ||
StrippedText.Contains("]") ||
StrippedText.Contains("}") ||
StrippedText.Contains("(") ||
StrippedText.Contains("[") ||
StrippedText.Contains("{")))
{
var sb = new StringBuilder();
bool lastWasBreak = false;
for (int i = 0; i < StrippedText.Length; i++)
{
string s = StrippedText[i].ToString();
if (lastWasBreak)
{
if (("\"`´'()<>!?.- " + Environment.NewLine).Contains(s))
{
sb.Append(s);
}
else if ((sb.ToString().EndsWith('<') || sb.ToString().EndsWith("</", StringComparison.Ordinal)) && i + 1 < StrippedText.Length && StrippedText[i + 1] == '>')
{ // tags
sb.Append(s);
}
else if (sb.ToString().EndsWith('<') && s == "/" && i + 2 < StrippedText.Length && StrippedText[i + 2] == '>')
{ // tags
sb.Append(s);
}
else if (sb.ToString().EndsWith("... ", StringComparison.Ordinal))
{
sb.Append(s);
lastWasBreak = false;
}
else
{
if (".!?:;)]}([{".Contains(s))
{
sb.Append(s);
}
else
{
lastWasBreak = false;
sb.Append(s.ToUpper());
}
}
}
else
{
sb.Append(s);
if (".!?:;)]}([{".Contains(s))
{
if (s == "]" && sb.ToString().IndexOf('[') > 1)
{ // I [Motor roaring] love you!
string temp = sb.ToString().Substring(0, sb.ToString().IndexOf('[') - 1).Trim();
if (temp.Length > 0 && !Utilities.LowercaseLetters.Contains(temp[temp.Length - 1].ToString()))
{
lastWasBreak = true;
}
}
else
{
lastWasBreak = true;
}
}
}
}
StrippedText = sb.ToString();
}
if (changeNameCases)
{
ReplaceNames2Fix(replaceIds, replaceNames);
}
else
{
ReplaceNames2Fix(replaceIds, originalNames);
}
}
