pdf_font.c

SumatraPDF是一款小型开源的pdf阅读工具。虽然玲珑小巧（只有800多KB）

#include "fitz.h"
#include "mupdf.h"

#define noHINT

#include <ft2build.h>
#include FT_FREETYPE_H
#include FT_XFREE86_H

static char *basefontnames[14][7] =
{
	{ "Courier", "CourierNew", "CourierNewPSMT", 0 },
	{ "Courier-Bold", "CourierNew,Bold", "Courier,Bold",
		"CourierNewPS-BoldMT", "CourierNew-Bold", 0 },
	{ "Courier-Oblique", "CourierNew,Italic", "Courier,Italic",
		"CourierNewPS-ItalicMT", "CourierNew-Italic", 0 },
	{ "Courier-BoldOblique", "CourierNew,BoldItalic", "Courier,BoldItalic",
		"CourierNewPS-BoldItalicMT", "CourierNew-BoldItalic", 0 },
	{ "Helvetica", "ArialMT", "Arial", 0 },
	{ "Helvetica-Bold", "Arial-BoldMT", "Arial,Bold", "Arial-Bold",
		"Helvetica,Bold", 0 },
	{ "Helvetica-Oblique", "Arial-ItalicMT", "Arial,Italic", "Arial-Italic",
		"Helvetica,Italic", "Helvetica-Italic", 0 },
	{ "Helvetica-BoldOblique", "Arial-BoldItalicMT",
		"Arial,BoldItalic", "Arial-BoldItalic",
		"Helvetica,BoldItalic", "Helvetica-BoldItalic", 0 },
	{ "Times-Roman", "TimesNewRomanPSMT", "TimesNewRoman",
		"TimesNewRomanPS", 0 },
	{ "Times-Bold", "TimesNewRomanPS-BoldMT", "TimesNewRoman,Bold",
		"TimesNewRomanPS-Bold", "TimesNewRoman-Bold", 0 },
	{ "Times-Italic", "TimesNewRomanPS-ItalicMT", "TimesNewRoman,Italic",
		"TimesNewRomanPS-Italic", "TimesNewRoman-Italic", 0 },
	{ "Times-BoldItalic", "TimesNewRomanPS-BoldItalicMT",
		"TimesNewRoman,BoldItalic", "TimesNewRomanPS-BoldItalic",
		"TimesNewRoman-BoldItalic", 0 },
	{ "Symbol", 0 },
	{ "ZapfDingbats", 0 }
};

/*
 * FreeType and Rendering glue
 */

enum { UNKNOWN, TYPE1, TRUETYPE, CID };

#undef __FTERRORS_H__
#define FT_ERRORDEF(e, v, s)	{ (e), (s) },
#define FT_ERROR_START_LIST
#define FT_ERROR_END_LIST	{ 0, NULL }

struct ft_error
{
	int err;
	char *str;
};

const struct ft_error ft_errors[] =
{
#include FT_ERRORS_H
};

char *ft_errstr(int err)
{
	const struct ft_error *e;

	for (e = ft_errors; e->str != NULL; e++)
		if (e->err == err)
			return e->str;

	return "Unknown error";
}

static int ftkind(FT_Face face)
{
	const char *kind = FT_Get_X11_Font_Format(face);
	pdf_logfont("ft font format %s\n", kind);
	if (!strcmp(kind, "TrueType"))
		return TRUETYPE;
	if (!strcmp(kind, "Type 1"))
		return TYPE1;
	if (!strcmp(kind, "CFF"))
		return TYPE1;
	if (!strcmp(kind, "CID Type 1"))
		return TYPE1;
	return UNKNOWN;
}

static int ftcharindex(FT_Face face, int cid)
{
	int gid = FT_Get_Char_Index(face, cid);
	if (gid == 0)
		gid = FT_Get_Char_Index(face, 0xf000 + cid);
	return gid;
}

static inline int ftcidtogid(pdf_font *font, int cid)
{
	if (font->tottfcmap)
	{
		cid = pdf_lookupcmap(font->tottfcmap, cid);
		return ftcharindex(font->ftface, cid);
	}

	if (font->cidtogid)
		return font->cidtogid[cid];

	return cid;
}

static int ftwidth(pdf_font *font, int cid)
{
	int e;

	cid = ftcidtogid(font, cid);

	e = FT_Load_Glyph(font->ftface, cid,
			FT_LOAD_NO_HINTING | FT_LOAD_NO_BITMAP | FT_LOAD_IGNORE_TRANSFORM);
	if (e)
		return 0;
	return ((FT_Face)font->ftface)->glyph->advance.x;
}

static fz_error *
ftrender(fz_glyph *glyph, fz_font *fzfont, int cid, fz_matrix trm)
{
	pdf_font *font = (pdf_font*)fzfont;
	FT_Face face = font->ftface;
	FT_Matrix m;
	FT_Vector v;
	FT_Error fterr;
	float scale;
	int gid;
	int x, y;
	int hint = font->hint;

	gid = ftcidtogid(font, cid);

	if (font->substitute && fzfont->wmode == 0)
	{
		fz_hmtx subw;
		int realw;

		FT_Set_Char_Size(face, 1000, 1000, 72, 72);

		fterr = FT_Load_Glyph(font->ftface, gid,
				FT_LOAD_NO_HINTING | FT_LOAD_NO_BITMAP | FT_LOAD_IGNORE_TRANSFORM);
		if (fterr)
			return fz_throw("freetype failed to load glyph: %s", ft_errstr(fterr));

		realw = ((FT_Face)font->ftface)->glyph->advance.x;
		subw = fz_gethmtx(fzfont, cid);
		if (realw)
			scale = (float) subw.w / realw;
		else
			scale = 1.0;

		trm = fz_concat(fz_scale(scale, 1.0), trm);
	}

	glyph->w = 0;
	glyph->h = 0;
	glyph->x = 0;
	glyph->y = 0;
	glyph->samples = nil;

	/* freetype mutilates complex glyphs if they are loaded
	 * with FT_Set_Char_Size 1.0. it rounds the coordinates
	 * before applying transformation. to get more precision in
	 * freetype, we shift part of the scale in the matrix
	 * into FT_Set_Char_Size instead
	 */

#ifdef HINT
	hint = 1;
#endif

	if (hint)
	{
		scale = fz_matrixexpansion(trm);
		m.xx = trm.a * 65536 / scale;
		m.yx = trm.b * 65536 / scale;
		m.xy = trm.c * 65536 / scale;
		m.yy = trm.d * 65536 / scale;
		v.x = 0;
		v.y = 0;

		FT_Set_Char_Size(face, 64 * scale, 64 * scale, 72, 72);
		FT_Set_Transform(face, &m, &v);

		fterr = FT_Load_Glyph(face, gid, FT_LOAD_NO_BITMAP);
		if (fterr)
			fz_warn("freetype load glyph: %s", ft_errstr(fterr));
	}
	else
	{
		m.xx = trm.a * 64;	/* should be 65536 */
		m.yx = trm.b * 64;
		m.xy = trm.c * 64;
		m.yy = trm.d * 64;
		v.x = trm.e * 64;
		v.y = trm.f * 64;

		FT_Set_Char_Size(face, 65536, 65536, 72, 72); /* should be 64, 64 */
		FT_Set_Transform(face, &m, &v);

		fterr = FT_Load_Glyph(face, gid, FT_LOAD_NO_BITMAP | FT_LOAD_NO_HINTING);
		if (fterr)
			fz_warn("freetype load glyph: %s", ft_errstr(fterr));
	}

	fterr = FT_Render_Glyph(face->glyph, ft_render_mode_normal);
	if (fterr)
		fz_warn("freetype render glyph: %s", ft_errstr(fterr));

	glyph->w = face->glyph->bitmap.width;
	glyph->h = face->glyph->bitmap.rows;
	glyph->x = face->glyph->bitmap_left;
	glyph->y = face->glyph->bitmap_top - glyph->h;
	glyph->samples = face->glyph->bitmap.buffer;

	for (y = 0; y < glyph->h / 2; y++)
	{
		for (x = 0; x < glyph->w; x++)
		{
			unsigned char a = glyph->samples[y * glyph->w + x ];
			unsigned char b = glyph->samples[(glyph->h - y - 1) * glyph->w + x];
			glyph->samples[y * glyph->w + x ] = b;
			glyph->samples[(glyph->h - y - 1) * glyph->w + x] = a;
		}
	}

	return fz_okay;
}

/*
 * Basic encoding tables
 */

static char *cleanfontname(char *fontname)
{
	int i, k;
	for (i = 0; i < 14; i++)
		for (k = 0; basefontnames[i][k]; k++)
			if (!strcmp(basefontnames[i][k], fontname))
				return basefontnames[i][0];
	return fontname;
}

static int mrecode(char *name)
{
	int i;
	for (i = 0; i < 256; i++)
		if (pdf_macroman[i] && !strcmp(name, pdf_macroman[i]))
			return i;
	return -1;
}

/*
 * Create and destroy
 */

static void ftdropfont(fz_font *font)
{
	pdf_font *pfont = (pdf_font*)font;
	if (pfont->encoding)
		pdf_dropcmap(pfont->encoding);
	if (pfont->tottfcmap)
		pdf_dropcmap(pfont->tottfcmap);
	if (pfont->tounicode)
		pdf_dropcmap(pfont->tounicode);
	fz_free(pfont->cidtogid);
	fz_free(pfont->cidtoucs);
	if (pfont->ftface)
		FT_Done_Face((FT_Face)pfont->ftface);
	if (pfont->fontdata)
		fz_dropbuffer(pfont->fontdata);
}

/* List of fonts that don't render properly without hinting. This list comes
   freetype */
static const char* const force_hinting[] =
{
  "DFKaiSho-SB",	 /* dfkaisb.ttf */
  "DFKai-SB",		 /* kaiu.ttf */
  "HuaTianSongTi?",  /* htst3.ttf */
  "MingLiU",		 /* mingliu.ttf & mingliu.ttc */
  "PMingLiU",		 /* mingliu.ttc */
  "MingLi43",		 /* mingli.ttf */
  NULL
};

pdf_font *
pdf_newfont(char *name)
{
	pdf_font *font;
	int i;

	font = fz_malloc(sizeof (pdf_font));
	if (!font)
		return nil;

	fz_initfont((fz_font*)font, name);
	font->super.render = ftrender;
	font->super.drop = (void(*)(fz_font*)) ftdropfont;

	font->ftface = nil;
	font->substitute = 0;

	font->flags = 0;
	font->italicangle = 0;
	font->ascent = 0;
	font->descent = 0;
	font->capheight = 0;
	font->xheight = 0;
	font->missingwidth = 0;

	font->encoding = nil;
	font->tottfcmap = 0;
	font->ncidtogid = 0;
	font->cidtogid = nil;

	font->tounicode = nil;
	font->ncidtoucs = 0;
	font->cidtoucs = nil;

	font->filename = nil;
	font->fontdata = nil;

	for (i = 0; i < 256; i++)
		font->charprocs[i] = nil;

	font->hint = 0;
	for (i = 0; force_hinting[i]; i++)
	{
		char *pos = strstr(name, force_hinting[i]);
		if (pos && (pos == name || pos[-1] == '+'))
		{
			font->hint = 1;
			break;
		}
	}
	return font;
}

/*
 * Simple fonts (Type1 and TrueType)
 */

static fz_error *
loadsimplefont(pdf_font **fontp, pdf_xref *xref, fz_obj *dict, fz_obj *ref)
{
	fz_error *error;
	fz_obj *descriptor = nil;
	fz_obj *encoding = nil;
	fz_obj *widths = nil;
	fz_obj *basefontobj = nil;
	unsigned short *etable = nil;
	pdf_font *font;
	fz_irect bbox;
	FT_Face face;
	FT_CharMap cmap;
	int kind;
	int symbolic;

	char *basefont;
	char *fontname;
	char *estrings[256];
	char ebuffer[256][32];
	int i, k, n;
	int fterr;

	basefontobj = fz_dictgets(dict, "BaseFont");
	error = pdf_resolve(&basefontobj, xref);
	if (error)
		return fz_rethrow(error, "cannot load simple font");
	basefont = fz_toname(basefontobj);
	fontname = cleanfontname(basefont);

	/*
	 * Load font file
	 */

	font = pdf_newfont(fontname);
	if (!font)
		return fz_outofmem;

	pdf_logfont("load simple font (%d %d R) ptr=%p {\n", fz_tonum(ref), fz_togen(ref), font);
	pdf_logfont("basefont0 %s\n", basefont);
	pdf_logfont("basefont1 %s\n", fontname);

	descriptor = fz_dictgets(dict, "FontDescriptor");
	if (descriptor && basefont == fontname)
		error = pdf_loadfontdescriptor(font, xref, descriptor, nil);
	else
		error = pdf_loadbuiltinfont(font, fontname);
	if (error)
		goto cleanup;

	face = font->ftface;
	kind = ftkind(face);

	pdf_logfont("ft name '%s' '%s'\n", face->family_name, face->style_name);

	bbox.x0 = (face->bbox.xMin * 1000) / face->units_per_EM;
	bbox.y0 = (face->bbox.yMin * 1000) / face->units_per_EM;
	bbox.x1 = (face->bbox.xMax * 1000) / face->units_per_EM;
	bbox.y1 = (face->bbox.yMax * 1000) / face->units_per_EM;

	pdf_logfont("ft bbox [%d %d %d %d]\n", bbox.x0, bbox.y0, bbox.x1, bbox.y1);

	if (bbox.x0 == bbox.x1)
		fz_setfontbbox((fz_font*)font, -1000, -1000, 2000, 2000);
	else
		fz_setfontbbox((fz_font*)font, bbox.x0, bbox.y0, bbox.x1, bbox.y1);

	/*
	 * Encoding
	 */

	symbolic = font->flags & 4;

	if (face->num_charmaps > 0)
		cmap = face->charmaps[0];
	else
		cmap = nil;

	for (i = 0; i < face->num_charmaps; i++)
	{
		FT_CharMap test = face->charmaps[i];

		if (kind == TYPE1)
		{
			if (test->platform_id == 7)
				cmap = test;
		}

		if (kind == TRUETYPE)
		{
			if (test->platform_id == 1 && test->encoding_id == 0)
				cmap = test;
			if (test->platform_id == 3 && test->encoding_id == 1)
				cmap = test;
		}
	}

	if (cmap)
	{
		fterr = FT_Set_Charmap(face, cmap);
		if (fterr)
		{
			error = fz_throw("freetype could not set cmap: %s", ft_errstr(fterr));
			goto cleanup;
		}
	}
	else
		fz_warn("freetype could not find any cmaps");

	etable = fz_malloc(sizeof(unsigned short) * 256);
	if (!etable)
		goto cleanup;

	for (i = 0; i < 256; i++)
	{
		estrings[i] = nil;
		etable[i] = 0;
	}

	encoding = fz_dictgets(dict, "Encoding");
	if (encoding && !(kind == TRUETYPE && symbolic))
	{
		error = pdf_resolve(&encoding, xref);
		if (error)
			goto cleanup;

		if (fz_isname(encoding))
			pdf_loadencoding(estrings, fz_toname(encoding));

		if (fz_isdict(encoding))
		{
			fz_obj *base, *diff, *item;

			base = fz_dictgets(encoding, "BaseEncoding");
			if (fz_isname(base))
				pdf_loadencoding(estrings, fz_toname(base));

			diff = fz_dictgets(encoding, "Differences");
			if (fz_isarray(diff))
			{
				n = fz_arraylen(diff);
				k = 0;
				for (i = 0; i < n; i++)
				{
					item = fz_arrayget(diff, i);
					if (fz_isint(item))
						k = fz_toint(item);
					if (fz_isname(item))
						estrings[k++] = fz_toname(item);
					if (k < 0) k = 0;
					if (k > 255) k = 255;
				}
			}
		}

		if (kind == TYPE1)
		{
			pdf_logfont("encode type1/cff by strings\n");
			for (i = 0; i < 256; i++)
				if (estrings[i])
					etable[i] = FT_Get_Name_Index(face, estrings[i]);
				else
					etable[i] = ftcharindex(face, i);
		}

		if (kind == TRUETYPE)
		{
			/* Unicode cmap */
			if (face->charmap && face->charmap->platform_id == 3)
			{
				pdf_logfont("encode truetype via unicode\n");
				for (i = 0; i < 256; i++)
					if (estrings[i])
					{
						int aglbuf[256];
						int aglnum;
						aglnum = pdf_lookupagl(estrings[i], aglbuf, nelem(aglbuf));
						if (aglnum != 1)
							etable[i] = FT_Get_Name_Index(face, estrings[i]);
						else
							etable[i] = ftcharindex(face, aglbuf[0]);
					}
					else
						etable[i] = ftcharindex(face, i);
			}

			/* MacRoman cmap */
			else if (face->charmap && face->charmap->platform_id == 1)
			{
				pdf_logfont("encode truetype via macroman\n");
				for (i = 0; i < 256; i++)
					if (estrings[i])
					{
						k = mrecode(estrings[i]);
						if (k <= 0)
							etable[i] = FT_Get_Name_Index(face, estrings[i]);
						else
							etable[i] = ftcharindex(face, k);
					}
					else
						etable[i] = ftcharindex(face, i);
			}

			/* Symbolic cmap */
			else
			{
				pdf_logfont("encode truetype symbolic\n");
				for (i = 0; i < 256; i++)
				{
					etable[i] = ftcharindex(face, i);
					if (etable[i])
						FT_Get_Glyph_Name(face, etable[i], ebuffer[i], 32);
					else
						FT_Get_Glyph_Name(face, i, ebuffer[i], 32);
				}

				/* map encoding to gid via glyph names */
				for (i = 0; i < 256; i++)
				{
					char *s = estrings[i];
					if (!etable[i] && s)