public void FixCasing(List <string> nameList, bool changeNameCases, bool makeUppercaseAfterBreak, bool checkLastLine, string lastLine)
{
var replaceIds = new List <string>();
var replaceNames = new List <string>();
var originalNames = new List <string>();
ReplaceNames1Remove(nameList, 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("..."))
{
if (!StrippedText.StartsWith("www.", StringComparison.OrdinalIgnoreCase) &&
!StrippedText.StartsWith("http", StringComparison.OrdinalIgnoreCase))
{
StrippedText = char.ToUpper(StrippedText[0]) + StrippedText.Substring(1);
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] = char.ToUpper(originalNames[i][0]) + originalNames[i].Remove(0, 1);
}
break;
}
}
}
}
}
}
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 (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 = 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(new[] { '.', '!', '?', ':', ';', ')', ']', '}', '(', '[', '{' }))
{
const string breakAfterChars = @".!?:;)]}([{";
var sb = new StringBuilder();
bool lastWasBreak = false;
for (int i = 0; i < StrippedText.Length; i++)
{
var s = StrippedText[i];
if (lastWasBreak)
{
if (("\"`´'()<>!?.- " + Environment.NewLine).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);
}
}
private static List <Paragraph> MakeHorizontalParagraphs(Paragraph p, int width, int height)
{
var lines = p.Text.SplitToLines();
var adjustment = 34;
var startY = height - lines.Count * 2 * adjustment + 30;
if (p.Text.StartsWith("{\\an8", StringComparison.Ordinal))
{
startY = 40;
}
var list = new List <Paragraph>();
var furiganaList = new List <Paragraph>();
var rubyOn = false;
var italinOn = false;
int startX;
using (var g = Graphics.FromHwnd(IntPtr.Zero))
{
var actualText = NetflixImsc11Japanese.RemoveBoutens(HtmlUtil.RemoveHtmlTags(p.Text, true));
var actualTextSize = g.MeasureString(actualText, new Font(SystemFonts.DefaultFont.FontFamily, 20));
startX = (int)(width / 2.0 - actualTextSize.Width / 2.0);
if (p.Text.StartsWith("{\\an5", StringComparison.Ordinal))
{
startY = (int)(height / 2.0 - actualTextSize.Height / 2.0);
}
}
for (var index = 0; index < lines.Count; index++)
{
var line = lines[index];
if (italinOn)
{
line = "<i>" + line;
}
var actual = new StringBuilder();
int i = 0;
while (i < line.Length)
{
if (line.Substring(i).StartsWith("{\\"))
{
var end = line.IndexOf('}', i);
if (end < 0)
{
break;
}
i = end + 1;
}
else if (line.Substring(i).StartsWith("<i>", StringComparison.Ordinal) || line.Substring(i).StartsWith("<u>", StringComparison.Ordinal) || line.Substring(i).StartsWith("<b>", StringComparison.Ordinal))
{
actual.Append("{\\i1}");
i += 3;
italinOn = true;
}
else if (line.Substring(i).StartsWith("</i>", StringComparison.Ordinal) || line.Substring(i).StartsWith("</u>", StringComparison.Ordinal) || line.Substring(i).StartsWith("</b>", StringComparison.Ordinal))
{
actual.Append("{\\i0}");
i += 4;
italinOn = false;
}
else if (line.Substring(i).StartsWith("<bouten-", StringComparison.Ordinal))
{
var end = line.IndexOf('>', i);
if (end < 0)
{
break;
}
if (end + 1 >= line.Length)
{
break;
}
var endTagStart = line.IndexOf("</", end, StringComparison.Ordinal);
if (endTagStart < 0)
{
break;
}
var tag = line.Substring(i + 1, end - i - 1);
var text = line.Substring(end + 1, endTagStart - end - 1);
foreach (var ch in text)
{
var furiganaChar = BoutenTagToUnicode(tag);
if (!string.IsNullOrWhiteSpace(furiganaChar))
{
furiganaList.Add(new Paragraph($"{{\\alpha&FF&}}{actual}{{\\alpha&0&}}{furiganaChar}", p.StartTime.TotalMilliseconds, p.EndTime.TotalMilliseconds));
}
actual.Append(ch);
}
var endTagEnd = line.IndexOf('>', endTagStart);
if (endTagEnd < 0)
{
break;
}
i = endTagEnd + 1;
}
else if (line.Substring(i).StartsWith("<ruby-container>", StringComparison.Ordinal))
{
var baseTextStart = line.IndexOf("<ruby-base>", i, StringComparison.Ordinal);
var baseTextEnd = line.IndexOf("</ruby-base>", i, StringComparison.Ordinal);
if (baseTextStart < 0 || baseTextEnd < 0)
{
baseTextStart = line.IndexOf("<ruby-base-italic>", i, StringComparison.Ordinal);
baseTextEnd = line.IndexOf("</ruby-base-italic>", i, StringComparison.Ordinal);
if (baseTextStart < 0 || baseTextEnd < 0)
{
break;
}
}
baseTextStart += "<ruby-base>".Length;
var baseText = line.Substring(baseTextStart, baseTextEnd - baseTextStart);
var extraText = string.Empty;
var extraTextStart = line.IndexOf("<ruby-text>", i, StringComparison.Ordinal);
var extraTextEnd = line.IndexOf("</ruby-text>", i, StringComparison.Ordinal);
if (extraTextStart >= 0 || extraTextEnd >= 0 && extraTextStart < extraTextEnd)
{
extraTextStart += "<ruby-text>".Length;
extraText = line.Substring(extraTextStart, extraTextEnd - extraTextStart);
}
if (string.IsNullOrEmpty(extraText))
{
extraTextStart = line.IndexOf("<ruby-text-italic>", i, StringComparison.Ordinal);
extraTextEnd = line.IndexOf("</ruby-text-italic>", i, StringComparison.Ordinal);
if (extraTextStart >= 0 || extraTextEnd >= 0 && extraTextStart < extraTextEnd)
{
extraTextStart += "<ruby-text-italic>".Length;
extraText = line.Substring(extraTextStart, extraTextEnd - extraTextStart);
}
}
var extraTextAfter = string.Empty;
var extraTextStartAfter = line.IndexOf("<ruby-text-after>", i, StringComparison.Ordinal);
var extraTextEndAfter = line.IndexOf("</ruby-text-after>", i, StringComparison.Ordinal);
if (extraTextStartAfter >= 0 || extraTextEndAfter >= 0 && extraTextStartAfter < extraTextEndAfter)
{
extraTextStartAfter += "<ruby-text-after>".Length;
extraText = line.Substring(extraTextStartAfter, extraTextEndAfter - extraTextStartAfter);
}
var preFurigana = string.Empty;
if (actual.Length > 0)
{
preFurigana = $"{{\\alpha&FF&}}{actual.ToString().TrimEnd()}{{\\alpha&0&}}";
}
if (!string.IsNullOrWhiteSpace(extraText))
{
furiganaList.Add(new Paragraph($"{preFurigana}{{\\fs20}}{extraText}", p.StartTime.TotalMilliseconds, p.EndTime.TotalMilliseconds));
}
if (!string.IsNullOrWhiteSpace(extraTextAfter))
{
furiganaList.Add(new Paragraph($"{preFurigana}{{\\fs20}} {extraTextAfter}", p.StartTime.TotalMilliseconds, p.EndTime.TotalMilliseconds));
}
actual.Append(baseText);
var endTagEnd = line.IndexOf("</ruby-container>", i, StringComparison.Ordinal);
if (endTagEnd < 0)
{
break;
}
i = endTagEnd + "</ruby-container>".Length;
rubyOn = true;
}
else
{
actual.Append(line.Substring(i, 1));
i++;
}
}
var actualText = actual.ToString().TrimEnd();
bool displayBefore = lines.Count == 2 && index == 0 || lines.Count == 1;
if (displayBefore && furiganaList.Count > 0)
{
foreach (var fp in furiganaList)
{
var beforeText = "{\\an1}{\\pos(" + startX + "," + startY + ")}" + fp.Text;
list.Add(new Paragraph(beforeText, p.StartTime.TotalMilliseconds, p.EndTime.TotalMilliseconds));
}
startY += adjustment;
if (rubyOn && index == 0 && lines.Count == 2)
{
startY += 3;
}
}
actualText = "{\\an1}{\\pos(" + startX + "," + startY + ")}" + actualText;
list.Add(new Paragraph(actualText, p.StartTime.TotalMilliseconds, p.EndTime.TotalMilliseconds));
startY += adjustment;
if (!displayBefore && furiganaList.Count > 0)
{
if (rubyOn && index == 1 && lines.Count == 2)
{
startY = (int)(startY - adjustment * 0.4);
}
foreach (var fp in furiganaList)
{
var beforeText = "{\\an1}{\\pos(" + startX + "," + startY + ")}" + fp.Text;
list.Add(new Paragraph(beforeText, p.StartTime.TotalMilliseconds, p.EndTime.TotalMilliseconds));
}
startY += adjustment;
}
furiganaList.Clear();
}
return(list);
}
public void FixCasing(List <string> nameList, bool changeNameCases, bool makeUppercaseAfterBreak, bool checkLastLine, string lastLine)
{
var replaceIds = new List <string>();
var replaceNames = new List <string>();
var originalNames = new List <string>();
ReplaceNames1Remove(nameList, 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("..."))
{
if (!StrippedText.StartsWith("www.", StringComparison.OrdinalIgnoreCase) &&
!StrippedText.StartsWith("http", StringComparison.OrdinalIgnoreCase))
{
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 && !char.IsLower(temp[temp.Length - 1]))
{
lastWasBreak = true;
}
}
else
{
idx = sb.ToString().LastIndexOf(' ');
if (idx >= 0 && idx < sb.Length - 2 && !IsInMiddleOfUrl(i - idx, StrippedText.Substring(idx + 1)))
{
lastWasBreak = true;
}
}
}
}
}
StrippedText = sb.ToString();
}
ReplaceNames2Fix(replaceIds, changeNameCases ? replaceNames : originalNames);
}
public static int CountWords(this string source)
{
return(HtmlUtil.RemoveHtmlTags(source, true).Split(new[] { ' ', '\n', '\r' }, StringSplitOptions.RemoveEmptyEntries).Length);
}