|
// Licensed to the .NET Foundation under one or more agreements.
// The .NET Foundation licenses this file to you under the MIT license.
// See the LICENSE file in the project root for more information.
//
//
// Description: FontFamilyIdentifier type
//
//
using System;
using System.Diagnostics;
using System.Security;
using System.Text;
using System.Windows;
using System.Windows.Media;
using System.Windows.Media.TextFormatting;
using MS.Utility;
using MS.Internal;
using MS.Internal.Shaping;
using MS.Internal.FontCache;
using MS.Internal.TextFormatting;
namespace MS.Internal.FontFace
{
/// <summary>
/// The FontFamilyIdentifier value type encapsulates a friendly name and base Uri
/// and provides access to the corresponding canonical name(s) via an indexer.
/// </summary>
//
// The friendly name is a string in the format passed to the FontFamily constructor and the
// corresponding type converter. It comprises one or more font family references separated by
// commas (with literal commas escaped by doubling). Each font family reference comprises a
// family name and an optional location.
//
// Pseudo-BNF:
//
// friendlyName = ESCAPED(fontFamilyReference) *( "," ESCAPED(fontFamilyReference) )
// fontFamilyReference = [ location "#" ] escapedFamilyName
// location = "" | relativeUri | absoluteUri
//
// where ESCAPED(fontFamilyReference) denotes a fontFamilyReference in which comma characters (",")
// have been replaced by doubled commas (",,").
//
// Both the location (if present) and the escapedFamilyName may contain hexadecimal escape
// sequences in the form %XX as in URI references.
//
// The location may be an empty string so "#ARIAL" is a valid font family reference;
// in fact it is the normalized form of "Arial".
//
// Canonicalization
//
// Canonicalization is the process of converting a font family reference to an absolute URI
// which specifies the font family location, plus a fragment which specifies the family name.
// The family name is converted to uppercase (since family names are not case-sensitive) but
// the rest of the URI is not.
//
// The process for canonicalizing the entire friendly name is as follows:
//
// 1. Split the friendly name into font family references.
// Treat single commas (",") as delimiters, and unescape double commas (",,").
//
// 2. Convert each font family reference to a normalized form.
// See Util.NormalizeFontFamilyReference.
//
// 3. Canonicalize the normalized font family reference.
// See Util.GetCanonicalUriForFamily.
//
// This is essentially what the Canonicalize method does, with addition of caching the resulting
// canonical names in the TypefaceMetricsCache. To save working set, the result of canonicalization
// is stored as a single null-delimited string rather than an array of strings.
//
// Since canonicalization is potentially expensive, we do not always canonicalize the entire
// friendly name. In this case, the indexer canonicalizes the requested font family reference
// on demand. However, we still cache the result in the TypefaceMetricsCache.
//
internal struct FontFamilyIdentifier
{
/// <summary>
/// FontFamilyIdentifier constructor
/// </summary>
/// <param name="friendlyName">friendly name in the format passed to
/// FontFamily constructor and type converter.</param>
/// <param name="baseUri">Base Uri used to resolve the location part of a font family reference
/// if one exists and is relative</param>
internal FontFamilyIdentifier(string friendlyName, Uri baseUri)
{
_friendlyName = friendlyName;
_baseUri = baseUri;
_tokenCount = (friendlyName != null) ? -1 : 0;
_canonicalReferences = null;
}
/// <summary>
/// Create a FontFamilyIdentifier by concatenating two existing identifiers.
/// </summary>
internal FontFamilyIdentifier(FontFamilyIdentifier first, FontFamilyIdentifier second)
{
first.Canonicalize();
second.Canonicalize();
_friendlyName = null;
_tokenCount = first._tokenCount + second._tokenCount;
_baseUri = null;
if (first._tokenCount == 0)
{
_canonicalReferences = second._canonicalReferences;
}
else if (second._tokenCount == 0)
{
_canonicalReferences = first._canonicalReferences;
}
else
{
_canonicalReferences = new CanonicalFontFamilyReference[_tokenCount];
int i = 0;
foreach (CanonicalFontFamilyReference family in first._canonicalReferences)
{
_canonicalReferences[i++] = family;
}
foreach (CanonicalFontFamilyReference family in second._canonicalReferences)
{
_canonicalReferences[i++] = family;
}
}
}
internal string Source
{
get { return _friendlyName; }
}
internal Uri BaseUri
{
get { return _baseUri; }
}
public bool Equals(FontFamilyIdentifier other)
{
if (_friendlyName == other._friendlyName && _baseUri == other._baseUri)
return true;
int c = Count;
if (other.Count != c)
return false;
if (c != 0)
{
Canonicalize();
other.Canonicalize();
for (int i = 0; i < c; ++i)
{
if (!_canonicalReferences[i].Equals(other._canonicalReferences[i]))
return false;
}
}
return true;
}
public override bool Equals(object obj)
{
return obj is FontFamilyIdentifier && Equals((FontFamilyIdentifier)obj);
}
public override int GetHashCode()
{
int hash = 1;
if (Count != 0)
{
Canonicalize();
foreach (CanonicalFontFamilyReference family in _canonicalReferences)
{
hash = HashFn.HashMultiply(hash) + family.GetHashCode();
}
}
return HashFn.HashScramble(hash);
}
internal int Count
{
get
{
// negative value represents uninitialized count
if (_tokenCount < 0)
{
_tokenCount = CountTokens(_friendlyName);
}
return _tokenCount;
}
}
internal CanonicalFontFamilyReference this[int tokenIndex]
{
get
{
ArgumentOutOfRangeException.ThrowIfNegative(tokenIndex);
ArgumentOutOfRangeException.ThrowIfGreaterThanOrEqual(tokenIndex, Count);
// Have we already been canonicalized?
if (_canonicalReferences != null)
{
// We have already canonicalized. This is typically the case for longer-lived
// identifiers such as belong to FontFamily objects.
return _canonicalReferences[tokenIndex];
}
else
{
// We have not already canonicalized. This is probably a short-lived object so
// it's not worthwhile to canonicalize all the names up front and cache them for
// later use. Just canonicalize each name as we need it.
// find the Nth font family reference.
int i, length;
int j = FindToken(_friendlyName, 0, out i, out length);
for (int k = 0; k < tokenIndex; ++k)
{
j = FindToken(_friendlyName, j, out i, out length);
}
// canonicalize just this font family reference
return GetCanonicalReference(i, length);
}
}
}
internal void Canonicalize()
{
if (_canonicalReferences != null)
return;
int count = this.Count;
if (count == 0)
return;
// First look up the entire friendly name in the cache; this may enable us to
// save working set by sharing the same array of may equal FontFamilyIdentifier.
BasedFriendlyName hashKey = new BasedFriendlyName(_baseUri, _friendlyName);
CanonicalFontFamilyReference[] canonicalReferences = TypefaceMetricsCache.ReadonlyLookup(hashKey) as CanonicalFontFamilyReference[];
if (canonicalReferences == null)
{
// We need to construct a new array.
canonicalReferences = new CanonicalFontFamilyReference[count];
// Add the first canonical family reference.
int i, length;
int j = FindToken(_friendlyName, 0, out i, out length);
canonicalReferences[0] = GetCanonicalReference(i, length);
// Add subsequent family references.
for (int k = 1; k < count; ++k)
{
j = FindToken(_friendlyName, j, out i, out length);
canonicalReferences[k] = GetCanonicalReference(i, length);
}
// Add the array to the cache.
TypefaceMetricsCache.Add(hashKey, canonicalReferences);
}
// for thread safety, we assign to the field only after the array is fully initialized
_canonicalReferences = canonicalReferences;
}
#region Friendly name parsing methods
private static int CountTokens(string friendlyName)
{
int count = 0;
int i, length;
int j = FindToken(friendlyName, 0, out i, out length);
while (j >= 0)
{
// Limit the number of family names in a single string.
if (++count == MaxFamilyNamePerFamilyMapTarget)
break;
j = FindToken(friendlyName, j, out i, out length);
}
return count;
}
/// <summary>
/// Scans the specified friendly name starting at the specified index and gets the index and
/// length of the first token it finds.
/// </summary>
/// <param name="friendlyName">friendly name containing zero or more comma-delimited tokens</param>
/// <param name="i">character index to scan from</param>
/// <param name="tokenIndex">receives the index of the token (or zero if none)</param>
/// <param name="tokenLength">receives the length of the token (or zero if none)</param>
/// <returns>If a token was found, the return value is a positive integer specifying where to begin
/// scanning for the next token; if no token was found, the return value is -1.</returns>
private static int FindToken(string friendlyName, int i, out int tokenIndex, out int tokenLength)
{
int length = friendlyName.Length;
while (i < length)
{
// skip leading whitespace
while (i < length && char.IsWhiteSpace(friendlyName[i]))
++i;
int begin = i;
// find delimiter or end of string
while (i < length)
{
if (friendlyName[i] == FamilyNameDelimiter)
{
if (i + 1 < length && friendlyName[i + 1] == FamilyNameDelimiter)
{
// Don't treat double commas as the family name delimiter.
i += 2;
}
else
{
break; // single comma delimiter
}
}
else if (friendlyName[i] == '\0')
{
break; // might as well treat null as a delimiter too
}
else
{
++i;
}
}
// exclude trailing whitespace
int end = i;
while (end > begin && char.IsWhiteSpace(friendlyName[end - 1]))
--end;
// make sure it's not an empty string
if (begin < end)
{
tokenIndex = begin;
tokenLength = end - begin;
return i + 1;
}
// continue after delimiter
++i;
}
// no token
tokenIndex = length;
tokenLength = 0;
return -1;
}
private CanonicalFontFamilyReference GetCanonicalReference(int startIndex, int length)
{
string normalizedString = Util.GetNormalizedFontFamilyReference(_friendlyName, startIndex, length);
// For caching normalized names, we use a different type of key which does not compare equal
// to the keys we used for caching friendly names.
BasedNormalizedName hashKey = new BasedNormalizedName(_baseUri, normalizedString);
// Look up the normalized string and base URI in the cache?
CanonicalFontFamilyReference canonicalReference = TypefaceMetricsCache.ReadonlyLookup(hashKey) as CanonicalFontFamilyReference;
// Do we already have a cached font family reference?
if (canonicalReference == null)
{
// Not in cache. Construct a new font family reference.
canonicalReference = CanonicalFontFamilyReference.Create(_baseUri, normalizedString);
// Add it to the cache.
TypefaceMetricsCache.Add(hashKey, canonicalReference);
}
return canonicalReference;
}
#endregion
#region BasedFriendlyName and BasedNormalizedName
/// <summary>
/// BasedFriendlyName represents a friendly name with an associated URI or use as a hash key.
/// </summary>
private sealed class BasedFriendlyName : BasedName
{
public BasedFriendlyName(Uri baseUri, string name)
: base(baseUri, name)
{
}
public override int GetHashCode()
{
// Specify a different seed than BasedNormalizedName
return InternalGetHashCode(1);
}
public override bool Equals(object obj)
{
// Only compare equal to other BasedFriendlyName objects
return InternalEquals(obj as BasedFriendlyName);
}
}
/// <summary>
/// BasedNormalizedName represents a normalized name with an associated URI or use as a hash key.
/// </summary>
private sealed class BasedNormalizedName : BasedName
{
public BasedNormalizedName(Uri baseUri, string name)
: base(baseUri, name)
{
}
public override int GetHashCode()
{
// Specify a different seed than BasedFriendlyName
return InternalGetHashCode(int.MaxValue);
}
public override bool Equals(object obj)
{
// Only compare equal to other BasedNormalizedName objects
return InternalEquals(obj as BasedNormalizedName);
}
}
/// <summary>
/// BasedName implements shared functionality of BasedFriendlyName and BasedNormalizedName.
/// The reason for the two derived classes (and for not just using Pair) is that we don't
/// want these two different types of keys to compare equal.
/// </summary>
private abstract class BasedName
{
private Uri _baseUri;
private string _name;
protected BasedName(Uri baseUri, string name)
{
_baseUri = baseUri;
_name = name;
}
public abstract override int GetHashCode();
public abstract override bool Equals(object obj);
protected int InternalGetHashCode(int seed)
{
int hash = seed;
if (_baseUri != null)
hash += HashFn.HashMultiply(_baseUri.GetHashCode());
if (_name != null)
hash = HashFn.HashMultiply(hash) + _name.GetHashCode();
return HashFn.HashScramble(hash);
}
protected bool InternalEquals(BasedName other)
{
return other != null &&
other._baseUri == _baseUri &&
other._name == _name;
}
}
#endregion
private string _friendlyName;
private Uri _baseUri;
private int _tokenCount;
private CanonicalFontFamilyReference[] _canonicalReferences;
internal const char FamilyNameDelimiter = ',';
internal const int MaxFamilyNamePerFamilyMapTarget = 32;
}
}
|