Mirrors/SubtitleEdit
1
0
Fork 0
mirror of https://github.com/SubtitleEdit/subtitleedit.git synced 2024-11-25 12:44:46 +01:00
Code Issues Projects Releases Wiki Activity

New method to normalize Unicode characters

Browse Source
This commit is contained in:
ivandrofly 2015-03-06 21:34:33 +00:00
parent d5866958d7
commit 72029ebdb7
1 changed files with 32 additions and 30 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

62
src/Forms/Main.cs
View File

@ -3017,31 +3017,7 @@ namespace Nikse.SubtitleEdit.Forms
if (!isUnicode)
{
const string defHyphen = "-"; // - Hyphen-minus (\u002D) (Basic Latin)
const string defColon = ":"; // : Colon (\uu003A) (Basic Latin)
// Hyphens
allText = allText.Replace("\u2043", defHyphen); // ⁃ Hyphen bullet (\u2043)
allText = allText.Replace("\u2010", defHyphen); // ‐ Hyphen (\u2010)
allText = allText.Replace("\u2012", defHyphen); // ‒ Figure dash (\u2012)
allText = allText.Replace("\u2013", defHyphen); // – En dash (\u2013)
allText = allText.Replace("\u2014", defHyphen); // — Em dash (\u2014)
allText = allText.Replace("\u2015", defHyphen); // ― Horizontal bar (\u2015)
// Colons:
allText = allText.Replace("\u02F8", defColon); // ˸ Modifier Letter Raised Colon (\u02F8)
allText = allText.Replace("\uFF1A", defColon); // ： Fullwidth Colon (\uFF1A)
allText = allText.Replace("\uF313", defColon); // ︓ Presentation Form for Vertical Colon (\uF313)
// Others
allText = allText.Replace("…", "...");
allText = allText.Replace("♪", "#");
allText = allText.Replace("♫", "#");
// Spaces
allText = allText.Replace("\u00A0", " "); // No-Break Space
allText = allText.Replace("\u200B", string.Empty); // Zero Width Space
allText = allText.Replace("\uFEFF", string.Empty); // Zero Width No-Break Space
allText = NormalizeUnicode(allText);
}
bool containsNegativeTime = false;
@ -3147,11 +3123,7 @@ namespace Nikse.SubtitleEdit.Forms
if (!isUnicode)
{
allText = allText.Replace("—", "-"); // mdash, code 8212
allText = allText.Replace("―", "-"); // mdash, code 8213
allText = allText.Replace("…", "...");
allText = allText.Replace("♪", "#");
allText = allText.Replace("♫", "#");
allText = NormalizeUnicode(allText);
}
bool containsNegativeTime = false;
@ -3181,6 +3153,36 @@ namespace Nikse.SubtitleEdit.Forms
}
}
public string NormalizeUnicode(string text)
{
const string defHyphen = "-"; // - Hyphen-minus (\u002D) (Basic Latin)
const string defColon = ":"; // : Colon (\uu003A) (Basic Latin)
// Hyphens
text = text.Replace("\u2043", defHyphen); // ⁃ Hyphen bullet (\u2043)
text = text.Replace("\u2010", defHyphen); // ‐ Hyphen (\u2010)
text = text.Replace("\u2012", defHyphen); // ‒ Figure dash (\u2012)
text = text.Replace("\u2013", defHyphen); // – En dash (\u2013)
text = text.Replace("\u2014", defHyphen); // — Em dash (\u2014)
text = text.Replace("\u2015", defHyphen); // ― Horizontal bar (\u2015)
// Colons:
text = text.Replace("\u02F8", defColon); // ˸ Modifier Letter Raised Colon (\u02F8)
text = text.Replace("\uFF1A", defColon); // ： Fullwidth Colon (\uFF1A)
text = text.Replace("\uF313", defColon); // ︓ Presentation Form for Vertical Colon (\uF313)
// Others
text = text.Replace("…", "...");
text = text.Replace("♪", "#");
text = text.Replace("♫", "#");
// Spaces
text = text.Replace("\u00A0", " "); // No-Break Space
text = text.Replace("\u200B", string.Empty); // Zero Width Space
text = text.Replace("\uFEFF", string.Empty); // Zero Width No-Break Space
return text;
}
private void NewToolStripMenuItemClick(object sender, EventArgs e)
{
ReloadFromSourceView();