SubtitleEdit/libse/StripableText.cs

285 lines
12 KiB
C#
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

using System;
using System.Collections.Generic;
using System.Text;
namespace Nikse.SubtitleEdit.Core
{
public class StripableText
{
public string Pre { get; set; }
public string Post { get; set; }
public string StrippedText { get; set; }
public string OriginalText { get; private set; }
public string MergedString
{
get { return Pre + StrippedText + Post; }
}
public StripableText(string text)
: this(text, " >-\"”“['`´¶(♪¿¡.…—", " -\"”“]'`´¶)♪.!?:…—")
{
}
public StripableText(string text, string stripStartCharacters, string stripEndCharacters)
{
OriginalText = text;
Pre = string.Empty;
if (text.Length > 0 && ("<{" + stripStartCharacters).Contains(text[0]))
{
int beginLength;
do
{
beginLength = text.Length;
while (text.Length > 0 && stripStartCharacters.Contains(text[0]))
{
Pre += text[0];
text = text.Remove(0, 1);
}
// ASS/SSA codes like {\an9}
int endIndex = text.IndexOf('}');
if (endIndex > 0 && text.StartsWith("{\\", StringComparison.Ordinal))
{
int nextStartIndex = text.IndexOf('{', 2);
if (nextStartIndex == -1 || nextStartIndex > endIndex)
{
endIndex++;
Pre += text.Substring(0, endIndex);
text = text.Remove(0, endIndex);
}
}
// tags like <i> or <font face="Segoe Print" color="#ff0000">
endIndex = text.IndexOf('>');
if (text.StartsWith('<') && endIndex >= 2)
{
endIndex++;
Pre += text.Substring(0, endIndex);
text = text.Remove(0, endIndex);
}
}
while (text.Length < beginLength);
}
Post = string.Empty;
if (text.Length > 0 && (">" + stripEndCharacters).Contains(text[text.Length - 1]))
{
int beginLength;
do
{
beginLength = text.Length;
while (text.Length > 0 && stripEndCharacters.Contains(text[text.Length - 1]))
{
Post = text[text.Length - 1] + Post;
text = text.Substring(0, text.Length - 1);
}
if (text.EndsWith('>'))
{
// tags </i> </b> </u>
if (text.EndsWith("</i>", StringComparison.OrdinalIgnoreCase) ||
text.EndsWith("</b>", StringComparison.OrdinalIgnoreCase) ||
text.EndsWith("</u>", StringComparison.OrdinalIgnoreCase))
{
Post = text.Substring(text.Length - 4) + Post;
text = text.Substring(0, text.Length - 4);
}
// tag </font>
if (text.EndsWith("</font>", StringComparison.OrdinalIgnoreCase))
{
Post = text.Substring(text.Length - 7) + Post;
text = text.Substring(0, text.Length - 7);
}
}
}
while (text.Length < beginLength);
}
StrippedText = text;
}
private static string GetAndInsertNextId(List<string> replaceIds, List<string> replaceNames, string name)
{
int i = 0;
string id = string.Format("_@{0}_", i);
while (replaceIds.Contains(id))
{
i++;
id = string.Format("_@{0}_", i);
}
replaceIds.Add(id);
replaceNames.Add(name);
return id;
}
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 void ReplaceNames2Fix(List<string> replaceIds, List<string> replaceNames)
{
for (int i = 0; i < replaceIds.Count; i++)
{
StrippedText = StrippedText.Replace(replaceIds[i], replaceNames[i]);
}
}
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);
}
}
}