font-preview/FontPreview/FontCmap.cs

using System;
using System.Collections.Generic;
using System.IO;
using System.Text;

namespace FontPreview
{
	/// <summary>
	/// 轻量 cmap 表解析器：纯字节读取字体文件，判断某 Unicode 码点是否有字形。
	/// 支持 TTF/OTF/TTC，覆盖 cmap 格式 4（BMP）与格式 12（全平面）。
	/// 不依赖 GDI+/DirectWrite，跨平台可编译。
	/// </summary>
	public class FontCmap
	{
		private List<int[]> ranges = new List<int[]>(); // 排序的 [start, end] 闭区间

		public static FontCmap Load(string path)
		{
			try
			{
				byte[] d = File.ReadAllBytes(path);
				FontCmap c = new FontCmap();
				c.Parse(d);
				return c.ranges.Count > 0 ? c : null;
			}
			catch
			{
				return null;
			}
		}

		private static ushort RU16(byte[] d, uint o)
		{
			return (ushort)(d[o] * 256u + d[o + 1]);
		}

		private static uint RU32(byte[] d, uint o)
		{
			return (uint)(d[o] * 16777216u + d[o + 1] * 65536u + d[o + 2] * 256u + d[o + 3]);
		}

		private static string Tag(byte[] d, uint o)
		{
			try
			{
				return Encoding.ASCII.GetString(d, (int)o, 4);
			}
			catch { return ""; }
		}

		private void Parse(byte[] d)
		{
			uint sfntOff = 0;
			if (Tag(d, 0) == "ttcf")
			{
				sfntOff = RU32(d, 8); // TTC：第一个字体的偏移
			}
			uint numTables = RU16(d, sfntOff + 4);
			uint cmapOff = 0;
			for (uint i = 0; i < numTables; i++)
			{
				uint rec = sfntOff + 12 + i * 16;
				if (Tag(d, rec) == "cmap")
				{
					cmapOff = RU32(d, rec + 8);
					break;
				}
			}
			if (cmapOff == 0) return;

			uint numSub = RU16(d, cmapOff + 2);
			uint best = 0;
			bool gotFull = false;
			for (uint i = 0; i < numSub; i++)
			{
				uint rec = cmapOff + 4 + i * 8;
				ushort platform = RU16(d, rec);
				ushort enc = RU16(d, rec + 2);
				uint off = cmapOff + RU32(d, rec + 4);
				// Windows: platformID=3；enc=10 为全 Unicode 平面，enc=1 为 BMP
				if (platform == 3 && (enc == 10 || enc == 1))
				{
					if (enc == 10 && !gotFull) { best = off; gotFull = true; }
					else if (best == 0) { best = off; }
				}
			}
			if (best == 0)
			{
				// 退而求其次：任意子表
				if (numSub > 0) best = cmapOff + RU32(d, cmapOff + 8);
			}
			if (best == 0) return;

			ushort fmt = RU16(d, best);
			if (fmt == 4) ParseFormat4(d, best);
			else if (fmt == 12) ParseFormat12(d, best);
			else if (fmt == 0) ParseFormat0(d, best);

			ranges.Sort((a, b) => a[0].CompareTo(b[0]));
			Merge();
		}

		private void ParseFormat4(byte[] d, uint o)
		{
			int segCount = RU16(d, o + 6) / 2;
			uint endOff = o + 14;
			uint startOff = endOff + (uint)segCount * 2 + 2; // 跳过 endCode 与 reservedPad
			for (int i = 0; i < segCount; i++)
			{
				int s = RU16(d, startOff + (uint)i * 2);
				int e = RU16(d, endOff + (uint)i * 2);
				if (s != 0xFFFF && e >= s && s != 0)
				{
					ranges.Add(new int[] { s, e });
				}
			}
		}

		private void ParseFormat12(byte[] d, uint o)
		{
			uint numGroups = RU32(d, o + 12);
			uint g = o + 16;
			for (uint i = 0; i < numGroups; i++)
			{
				int s = (int)RU32(d, g + i * 12);
				int e = (int)RU32(d, g + i * 12 + 4);
				if (e >= s) ranges.Add(new int[] { s, e });
			}
		}

		private void ParseFormat0(byte[] d, uint o)
		{
			// 256 字节 glyph index 表，非 0 即有字形
			int start = -1;
			for (int i = 0; i < 256; i++)
			{
				bool has = d[o + 6 + i] != 0;
				if (has && start < 0) start = i;
				if (!has && start >= 0) { ranges.Add(new int[] { start, i - 1 }); start = -1; }
			}
			if (start >= 0) ranges.Add(new int[] { start, 255 });
		}

		private void Merge()
		{
			List<int[]> merged = new List<int[]>();
			foreach (int[] r in ranges)
			{
				if (merged.Count > 0 && r[0] <= merged[merged.Count - 1][1] + 1)
				{
					if (r[1] > merged[merged.Count - 1][1]) merged[merged.Count - 1][1] = r[1];
				}
				else merged.Add(new int[] { r[0], r[1] });
			}
			ranges = merged;
		}

		public bool HasGlyph(int codepoint)
		{
			int lo = 0, hi = ranges.Count - 1;
			while (lo <= hi)
			{
				int mid = (lo + hi) / 2;
				if (codepoint < ranges[mid][0]) hi = mid - 1;
				else if (codepoint > ranges[mid][1]) lo = mid + 1;
				else return true;
			}
			return false;
		}
	}
}