lsm/PdfPig
1
0
Fork 0
mirror of https://github.com/UglyToad/PdfPig.git synced 2025-08-20 09:21:57 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

normalize some line endings

Browse Source
This commit is contained in:
Eliot Jones 2023-05-21 19:17:14 +01:00
parent bb4c6f2f1e
commit c3dd69388d
2 changed files with 390 additions and 394 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

252
src/UglyToad.PdfPig.Fonts/Encodings/Encoding.cs
View File

@ -1,65 +1,65 @@
namespace UglyToad.PdfPig.Fonts.Encodings
{
using System.Collections.Generic;
using Tokens;
/// <summary>
/// Maps character codes to glyph names from a PostScript encoding.
/// </summary>
public abstract class Encoding
{
/// <summary>
/// Mutable code to name map.
/// </summary>
protected readonly Dictionary<int, string> CodeToName = new Dictionary<int, string>(250);
/// <summary>
/// Maps from character codes to names.
/// </summary>
public IReadOnlyDictionary<int, string> CodeToNameMap => CodeToName;
/// <summary>
/// Mutable name to code map.
/// </summary>
protected readonly Dictionary<string, int> NameToCode = new Dictionary<string, int>(250);
/// <summary>
/// Maps from names to character cocdes.
/// </summary>
public IReadOnlyDictionary<string, int> NameToCodeMap => NameToCode;
/// <summary>
/// The name of this encoding.
/// </summary>
public abstract string EncodingName { get; }
/// <summary>
/// Whether this encoding contains a code for the name.
/// </summary>
public bool ContainsName(string name)
{
return NameToCode.ContainsKey(name);
}
/// <summary>
/// Whether this encoding contains a name for the code.
/// </summary>
public bool ContainsCode(int code)
{
return CodeToName.ContainsKey(code);
}
/// <summary>
/// Get the character name corresponding to the given code.
/// </summary>
public virtual string GetName(int code)
{
if (!CodeToName.TryGetValue(code, out var name))
{
return ".notdef";
}
return name;
namespace UglyToad.PdfPig.Fonts.Encodings
{
using System.Collections.Generic;
using Tokens;
/// <summary>
/// Maps character codes to glyph names from a PostScript encoding.
/// </summary>
public abstract class Encoding
{
/// <summary>
/// Mutable code to name map.
/// </summary>
protected readonly Dictionary<int, string> CodeToName = new Dictionary<int, string>(250);
/// <summary>
/// Maps from character codes to names.
/// </summary>
public IReadOnlyDictionary<int, string> CodeToNameMap => CodeToName;
/// <summary>
/// Mutable name to code map.
/// </summary>
protected readonly Dictionary<string, int> NameToCode = new Dictionary<string, int>(250);
/// <summary>
/// Maps from names to character cocdes.
/// </summary>
public IReadOnlyDictionary<string, int> NameToCodeMap => NameToCode;
/// <summary>
/// The name of this encoding.
/// </summary>
public abstract string EncodingName { get; }
/// <summary>
/// Whether this encoding contains a code for the name.
/// </summary>
public bool ContainsName(string name)
{
return NameToCode.ContainsKey(name);
}
/// <summary>
/// Whether this encoding contains a name for the code.
/// </summary>
public bool ContainsCode(int code)
{
return CodeToName.ContainsKey(code);
}
/// <summary>
/// Get the character name corresponding to the given code.
/// </summary>
public virtual string GetName(int code)
{
if (!CodeToName.TryGetValue(code, out var name))
{
return ".notdef";
}
return name;
}
@ -67,67 +67,67 @@
/// Get the character code from name
/// </summary>
/// <param name="name">Character name (eg. euro, ampersand, A, space)</param>
/// <returns>-1 if not found otherwise the character code</returns>
public virtual int GetCode(string name)
{
if (!NameToCode.TryGetValue(name, out var code))
{
return -1;
}
return code;
}
/// <summary>
/// Add a character code and name pair.
/// </summary>
protected void Add(int code, string name)
{
CodeToName[code] = name;
if (!NameToCode.ContainsKey(name))
{
NameToCode[name] = code;
}
}
/// <summary>
/// Get a known encoding instance with the given name.
/// </summary>
public static bool TryGetNamedEncoding(NameToken name, out Encoding encoding)
{
encoding = null;
if (name == null)
{
return false;
}
if (name.Equals(NameToken.StandardEncoding))
{
encoding = StandardEncoding.Instance;
return true;
}
if (name.Equals(NameToken.WinAnsiEncoding))
{
encoding = WinAnsiEncoding.Instance;
return true;
}
if (name.Equals(NameToken.MacExpertEncoding))
{
encoding = MacExpertEncoding.Instance;
return true;
}
if (name.Equals(NameToken.MacRomanEncoding))
{
encoding = MacRomanEncoding.Instance;
return true;
}
return false;
}
}
}
/// <returns>-1 if not found otherwise the character code</returns>
public virtual int GetCode(string name)
{
if (!NameToCode.TryGetValue(name, out var code))
{
return -1;
}
return code;
}
/// <summary>
/// Add a character code and name pair.
/// </summary>
protected void Add(int code, string name)
{
CodeToName[code] = name;
if (!NameToCode.ContainsKey(name))
{
NameToCode[name] = code;
}
}
/// <summary>
/// Get a known encoding instance with the given name.
/// </summary>
public static bool TryGetNamedEncoding(NameToken name, out Encoding encoding)
{
encoding = null;
if (name == null)
{
return false;
}
if (name.Equals(NameToken.StandardEncoding))
{
encoding = StandardEncoding.Instance;
return true;
}
if (name.Equals(NameToken.WinAnsiEncoding))
{
encoding = WinAnsiEncoding.Instance;
return true;
}
if (name.Equals(NameToken.MacExpertEncoding))
{
encoding = MacExpertEncoding.Instance;
return true;
}
if (name.Equals(NameToken.MacRomanEncoding))
{
encoding = MacRomanEncoding.Instance;
return true;
}
return false;
}
}
}

532
src/UglyToad.PdfPig.Fonts/Encodings/WinAnsiEncoding.cs
View File

@ -1,271 +1,267 @@
namespace UglyToad.PdfPig.Fonts.Encodings
{
namespace UglyToad.PdfPig.Fonts.Encodings
{
/// <summary>
/// Windows ANSI encoding.
/// </summary>
public class WinAnsiEncoding : Encoding
{
/// <summary>
/// The encoding table is taken from the Appendix of the specification.
/// These codes are in octal.
/// </summary>
private static readonly (int, string)[] EncodingTable =
{
(0101, "A"),
(0306, "AE"),
(0301, "Aacute"),
(0302, "Acircumflex"),
(0304, "Adieresis"),
(0300, "Agrave"),
(0305, "Aring"),
(0303, "Atilde"),
(0102, "B"),
(0103, "C"),
(0307, "Ccedilla"),
(0104, "D"),
(0105, "E"),
(0311, "Eacute"),
(0312, "Ecircumflex"),
(0313, "Edieresis"),
(0310, "Egrave"),
(0320, "Eth"),
(0200, "Euro"),
(0106, "F"),
(0107, "G"),
(0110, "H"),
(0111, "I"),
(0315, "Iacute"),
(0316, "Icircumflex"),
(0317, "Idieresis"),
(0314, "Igrave"),
(0112, "J"),
(0113, "K"),
(0114, "L"),
(0115, "M"),
(0116, "N"),
(0321, "Ntilde"),
(0117, "O"),
(0214, "OE"),
(0323, "Oacute"),
(0324, "Ocircumflex"),
(0326, "Odieresis"),
(0322, "Ograve"),
(0330, "Oslash"),
(0325, "Otilde"),
(0120, "P"),
(0121, "Q"),
(0122, "R"),
(0123, "S"),
(0212, "Scaron"),
(0124, "T"),
(0336, "Thorn"),
(0125, "U"),
(0332, "Uacute"),
(0333, "Ucircumflex"),
(0334, "Udieresis"),
(0331, "Ugrave"),
(0126, "V"),
(0127, "W"),
(0130, "X"),
(0131, "Y"),
(0335, "Yacute"),
(0237, "Ydieresis"),
(0132, "Z"),
(0216, "Zcaron"),
(0141, "a"),
(0341, "aacute"),
(0342, "acircumflex"),
(0264, "acute"),
(0344, "adieresis"),
(0346, "ae"),
(0340, "agrave"),
(046, "ampersand"),
(0345, "aring"),
(0136, "asciicircum"),
(0176, "asciitilde"),
(052, "asterisk"),
(0100, "at"),
(0343, "atilde"),
(0142, "b"),
(0134, "backslash"),
(0174, "bar"),
(0173, "braceleft"),
(0175, "braceright"),
(0133, "bracketleft"),
(0135, "bracketright"),
(0246, "brokenbar"),
(0225, "bullet"),
(0143, "c"),
(0347, "ccedilla"),
(0270, "cedilla"),
(0242, "cent"),
(0210, "circumflex"),
(072, "colon"),
(054, "comma"),
(0251, "copyright"),
(0244, "currency"),
(0144, "d"),
(0206, "dagger"),
(0207, "daggerdbl"),
(0260, "degree"),
(0250, "dieresis"),
(0367, "divide"),
(044, "dollar"),
(0145, "e"),
(0351, "eacute"),
(0352, "ecircumflex"),
(0353, "edieresis"),
(0350, "egrave"),
(070, "eight"),
(0205, "ellipsis"),
(0227, "emdash"),
(0226, "endash"),
(075, "equal"),
(0360, "eth"),
(041, "exclam"),
(0241, "exclamdown"),
(0146, "f"),
(065, "five"),
(0203, "florin"),
(064, "four"),
(0147, "g"),
(0337, "germandbls"),
(0140, "grave"),
(076, "greater"),
(0253, "guillemotleft"),
(0273, "guillemotright"),
(0213, "guilsinglleft"),
(0233, "guilsinglright"),
(0150, "h"),
(055, "hyphen"),
(0151, "i"),
(0355, "iacute"),
(0356, "icircumflex"),
(0357, "idieresis"),
(0354, "igrave"),
(0152, "j"),
(0153, "k"),
(0154, "l"),
(074, "less"),
(0254, "logicalnot"),
(0155, "m"),
(0257, "macron"),
(0265, "mu"),
(0327, "multiply"),
(0156, "n"),
(071, "nine"),
(0361, "ntilde"),
(043, "numbersign"),
(0157, "o"),
(0363, "oacute"),
(0364, "ocircumflex"),
(0366, "odieresis"),
(0234, "oe"),
(0362, "ograve"),
(061, "one"),
(0275, "onehalf"),
(0274, "onequarter"),
(0271, "onesuperior"),
(0252, "ordfeminine"),
(0272, "ordmasculine"),
(0370, "oslash"),
(0365, "otilde"),
(0160, "p"),
(0266, "paragraph"),
(050, "parenleft"),
(051, "parenright"),
(045, "percent"),
(056, "period"),
(0267, "periodcentered"),
(0211, "perthousand"),
(053, "plus"),
(0261, "plusminus"),
(0161, "q"),
(077, "question"),
(0277, "questiondown"),
(042, "quotedbl"),
(0204, "quotedblbase"),
(0223, "quotedblleft"),
(0224, "quotedblright"),
(0221, "quoteleft"),
(0222, "quoteright"),
(0202, "quotesinglbase"),
(047, "quotesingle"),
(0162, "r"),
(0256, "registered"),
(0163, "s"),
(0232, "scaron"),
(0247, "section"),
(073, "semicolon"),
(067, "seven"),
(066, "six"),
(057, "slash"),
(040, "space"),
(0243, "sterling"),
(0164, "t"),
(0376, "thorn"),
(063, "three"),
(0276, "threequarters"),
(0263, "threesuperior"),
(0230, "tilde"),
(0231, "trademark"),
(062, "two"),
(0262, "twosuperior"),
(0165, "u"),
(0372, "uacute"),
(0373, "ucircumflex"),
(0374, "udieresis"),
(0371, "ugrave"),
(0137, "underscore"),
(0166, "v"),
(0167, "w"),
(0170, "x"),
(0171, "y"),
(0375, "yacute"),
(0377, "ydieresis"),
(0245, "yen"),
(0172, "z"),
(0236, "zcaron"),
(060, "zero"),
// adding some additional mappings as defined in Appendix D of the pdf spec
(0240, "space"),
(0255, "hyphen")
};
/// <summary>
/// Single instance of this encoding.
/// </summary>
public static WinAnsiEncoding Instance { get; } = new WinAnsiEncoding();
/// <summary>
/// Windows ANSI encoding.
/// </summary>
public class WinAnsiEncoding : Encoding
{
/// <summary>
/// The encoding table is taken from the Appendix of the specification.
/// These codes are in octal.
/// </summary>
private static readonly (int, string)[] EncodingTable =
{
(0101, "A"),
(0306, "AE"),
(0301, "Aacute"),
(0302, "Acircumflex"),
(0304, "Adieresis"),
(0300, "Agrave"),
(0305, "Aring"),
(0303, "Atilde"),
(0102, "B"),
(0103, "C"),
(0307, "Ccedilla"),
(0104, "D"),
(0105, "E"),
(0311, "Eacute"),
(0312, "Ecircumflex"),
(0313, "Edieresis"),
(0310, "Egrave"),
(0320, "Eth"),
(0200, "Euro"),
(0106, "F"),
(0107, "G"),
(0110, "H"),
(0111, "I"),
(0315, "Iacute"),
(0316, "Icircumflex"),
(0317, "Idieresis"),
(0314, "Igrave"),
(0112, "J"),
(0113, "K"),
(0114, "L"),
(0115, "M"),
(0116, "N"),
(0321, "Ntilde"),
(0117, "O"),
(0214, "OE"),
(0323, "Oacute"),
(0324, "Ocircumflex"),
(0326, "Odieresis"),
(0322, "Ograve"),
(0330, "Oslash"),
(0325, "Otilde"),
(0120, "P"),
(0121, "Q"),
(0122, "R"),
(0123, "S"),
(0212, "Scaron"),
(0124, "T"),
(0336, "Thorn"),
(0125, "U"),
(0332, "Uacute"),
(0333, "Ucircumflex"),
(0334, "Udieresis"),
(0331, "Ugrave"),
(0126, "V"),
(0127, "W"),
(0130, "X"),
(0131, "Y"),
(0335, "Yacute"),
(0237, "Ydieresis"),
(0132, "Z"),
(0216, "Zcaron"),
(0141, "a"),
(0341, "aacute"),
(0342, "acircumflex"),
(0264, "acute"),
(0344, "adieresis"),
(0346, "ae"),
(0340, "agrave"),
(046, "ampersand"),
(0345, "aring"),
(0136, "asciicircum"),
(0176, "asciitilde"),
(052, "asterisk"),
(0100, "at"),
(0343, "atilde"),
(0142, "b"),
(0134, "backslash"),
(0174, "bar"),
(0173, "braceleft"),
(0175, "braceright"),
(0133, "bracketleft"),
(0135, "bracketright"),
(0246, "brokenbar"),
(0225, "bullet"),
(0143, "c"),
(0347, "ccedilla"),
(0270, "cedilla"),
(0242, "cent"),
(0210, "circumflex"),
(072, "colon"),
(054, "comma"),
(0251, "copyright"),
(0244, "currency"),
(0144, "d"),
(0206, "dagger"),
(0207, "daggerdbl"),
(0260, "degree"),
(0250, "dieresis"),
(0367, "divide"),
(044, "dollar"),
(0145, "e"),
(0351, "eacute"),
(0352, "ecircumflex"),
(0353, "edieresis"),
(0350, "egrave"),
(070, "eight"),
(0205, "ellipsis"),
(0227, "emdash"),
(0226, "endash"),
(075, "equal"),
(0360, "eth"),
(041, "exclam"),
(0241, "exclamdown"),
(0146, "f"),
(065, "five"),
(0203, "florin"),
(064, "four"),
(0147, "g"),
(0337, "germandbls"),
(0140, "grave"),
(076, "greater"),
(0253, "guillemotleft"),
(0273, "guillemotright"),
(0213, "guilsinglleft"),
(0233, "guilsinglright"),
(0150, "h"),
(055, "hyphen"),
(0151, "i"),
(0355, "iacute"),
(0356, "icircumflex"),
(0357, "idieresis"),
(0354, "igrave"),
(0152, "j"),
(0153, "k"),
(0154, "l"),
(074, "less"),
(0254, "logicalnot"),
(0155, "m"),
(0257, "macron"),
(0265, "mu"),
(0327, "multiply"),
(0156, "n"),
(071, "nine"),
(0361, "ntilde"),
(043, "numbersign"),
(0157, "o"),
(0363, "oacute"),
(0364, "ocircumflex"),
(0366, "odieresis"),
(0234, "oe"),
(0362, "ograve"),
(061, "one"),
(0275, "onehalf"),
(0274, "onequarter"),
(0271, "onesuperior"),
(0252, "ordfeminine"),
(0272, "ordmasculine"),
(0370, "oslash"),
(0365, "otilde"),
(0160, "p"),
(0266, "paragraph"),
(050, "parenleft"),
(051, "parenright"),
(045, "percent"),
(056, "period"),
(0267, "periodcentered"),
(0211, "perthousand"),
(053, "plus"),
(0261, "plusminus"),
(0161, "q"),
(077, "question"),
(0277, "questiondown"),
(042, "quotedbl"),
(0204, "quotedblbase"),
(0223, "quotedblleft"),
(0224, "quotedblright"),
(0221, "quoteleft"),
(0222, "quoteright"),
(0202, "quotesinglbase"),
(047, "quotesingle"),
(0162, "r"),
(0256, "registered"),
(0163, "s"),
(0232, "scaron"),
(0247, "section"),
(073, "semicolon"),
(067, "seven"),
(066, "six"),
(057, "slash"),
(040, "space"),
(0243, "sterling"),
(0164, "t"),
(0376, "thorn"),
(063, "three"),
(0276, "threequarters"),
(0263, "threesuperior"),
(0230, "tilde"),
(0231, "trademark"),
(062, "two"),
(0262, "twosuperior"),
(0165, "u"),
(0372, "uacute"),
(0373, "ucircumflex"),
(0374, "udieresis"),
(0371, "ugrave"),
(0137, "underscore"),
(0166, "v"),
(0167, "w"),
(0170, "x"),
(0171, "y"),
(0375, "yacute"),
(0377, "ydieresis"),
(0245, "yen"),
(0172, "z"),
(0236, "zcaron"),
(060, "zero"),
// adding some additional mappings as defined in Appendix D of the pdf spec
(0240, "space"),
(0255, "hyphen")
};
/// <summary>
/// Single instance of this encoding.
/// </summary>
public static WinAnsiEncoding Instance { get; } = new WinAnsiEncoding();
/// <inheritdoc />
public override string EncodingName => "WinAnsiEncoding";
private WinAnsiEncoding()
{
foreach ((var codeToBeConverted, var name) in EncodingTable)
{
// In source code an int literal with a leading zero ('0')
// in other languages ('C' and 'Java') would be interpreted
// as octal (base 8) and converted but C# does not support and
// so arrives here as a different value parsed as base10.
// Convert 'codeToBeConverted' to intended value as if it was an octal literal before using.
// For example 040 converts to string "40" then convert string to int again but using base 8 (octal) so result is 32 (base 10).
var code = System.Convert.ToInt32($"{codeToBeConverted}", 8); // alternative is OctalHelpers.FromOctalInt()
Add(code, name);
}
// In WinAnsiEncoding, all unused codes greater than 40 map to the bullet character.
for (var i = 041; i <= 255; i++)
{
if (!CodeToName.ContainsKey(i))
{
Add(i, "bullet");
}
}
}
}
/// <inheritdoc />
public override string EncodingName => "WinAnsiEncoding";
private WinAnsiEncoding()
{
foreach ((var codeToBeConverted, var name) in EncodingTable)
{
// In source code an int literal with a leading zero ('0')
// in other languages ('C' and 'Java') would be interpreted
// as octal (base 8) and converted but C# does not support and
// so arrives here as a different value parsed as base10.
// Convert 'codeToBeConverted' to intended value as if it was an octal literal before using.
// For example 040 converts to string "40" then convert string to int again but using base 8 (octal) so result is 32 (base 10).
var code = System.Convert.ToInt32($"{codeToBeConverted}", 8); // alternative is OctalHelpers.FromOctalInt()
Add(code, name);
}
// In WinAnsiEncoding, all unused codes greater than 40 map to the bullet character.
for (var i = 041; i <= 255; i++)
{
if (!CodeToName.ContainsKey(i))
{
Add(i, "bullet");
}
}
}
}
}