Esempio n. 1
0
void
pdf_load_to_unicode(pdf_document *doc, pdf_font_desc *font,
	char **strings, char *collection, pdf_obj *cmapstm)
{
	pdf_cmap *cmap;
	int cid;
	int ucsbuf[8];
	int ucslen;
	int i;
	fz_context *ctx = doc->ctx;

	if (pdf_is_stream(doc, pdf_to_num(cmapstm), pdf_to_gen(cmapstm)))
	{
		cmap = pdf_load_embedded_cmap(doc, cmapstm);

		font->to_unicode = pdf_new_cmap(ctx);

		for (i = 0; i < (strings ? 256 : 65536); i++)
		{
			cid = pdf_lookup_cmap(font->encoding, i);
			if (cid >= 0)
			{
				ucslen = pdf_lookup_cmap_full(cmap, i, ucsbuf);
				if (ucslen == 1)
					pdf_map_range_to_range(ctx, font->to_unicode, cid, cid, ucsbuf[0]);
				if (ucslen > 1)
					pdf_map_one_to_many(ctx, font->to_unicode, cid, ucsbuf, ucslen);
			}
		}

		pdf_sort_cmap(ctx, font->to_unicode);

		pdf_drop_cmap(ctx, cmap);
		font->size += pdf_cmap_size(ctx, font->to_unicode);
	}

	else if (collection)
	{
		if (!strcmp(collection, "Adobe-CNS1"))
			font->to_unicode = pdf_load_system_cmap(ctx, "Adobe-CNS1-UCS2");
		else if (!strcmp(collection, "Adobe-GB1"))
			font->to_unicode = pdf_load_system_cmap(ctx, "Adobe-GB1-UCS2");
		else if (!strcmp(collection, "Adobe-Japan1"))
			font->to_unicode = pdf_load_system_cmap(ctx, "Adobe-Japan1-UCS2");
		else if (!strcmp(collection, "Adobe-Korea1"))
			font->to_unicode = pdf_load_system_cmap(ctx, "Adobe-Korea1-UCS2");

		return;
	}

	if (strings)
	{
		/* TODO one-to-many mappings */

		font->cid_to_ucs_len = 256;
		font->cid_to_ucs = fz_malloc_array(ctx, 256, sizeof(unsigned short));
		font->size += 256 * sizeof(unsigned short);

		for (i = 0; i < 256; i++)
		{
			if (strings[i])
				font->cid_to_ucs[i] = pdf_lookup_agl(strings[i]);
			else
				font->cid_to_ucs[i] = '?';
		}
	}

	if (!font->to_unicode && !font->cid_to_ucs)
	{
		/* TODO: synthesize a ToUnicode if it's a freetype font with
		 * cmap and/or post tables or if it has glyph names. */
	}
}
static pdf_font_desc *
pdf_load_simple_font(pdf_document *xref, pdf_obj *dict)
{
	pdf_obj *descriptor;
	pdf_obj *encoding;
	pdf_obj *widths;
	unsigned short *etable = NULL;
	pdf_font_desc *fontdesc = NULL;
	char *subtype;
	FT_Face face;
	FT_CharMap cmap;
	int symbolic;
	int kind;

	char *basefont;
	char *estrings[256];
	char ebuffer[256][32];
	int i, k, n;
	int fterr;
	fz_context *ctx = xref->ctx;

	fz_var(fontdesc);
	fz_var(etable);

	basefont = pdf_to_name(pdf_dict_gets(dict, "BaseFont"));

	/* Load font file */
	fz_try(ctx)
	{
		fontdesc = pdf_new_font_desc(ctx);

		descriptor = pdf_dict_gets(dict, "FontDescriptor");
		if (descriptor)
			pdf_load_font_descriptor(fontdesc, xref, descriptor, NULL, basefont);
		else
			pdf_load_builtin_font(ctx, fontdesc, basefont);

		/* Some chinese documents mistakenly consider WinAnsiEncoding to be codepage 936 */
		if (descriptor && pdf_is_string(pdf_dict_gets(descriptor, "FontName")) &&
			!pdf_dict_gets(dict, "ToUnicode") &&
			!strcmp(pdf_to_name(pdf_dict_gets(dict, "Encoding")), "WinAnsiEncoding") &&
			pdf_to_int(pdf_dict_gets(descriptor, "Flags")) == 4)
		{
			char *cp936fonts[] = {
				"\xCB\xCE\xCC\xE5", "SimSun,Regular",
				"\xBA\xDA\xCC\xE5", "SimHei,Regular",
				"\xBF\xAC\xCC\xE5_GB2312", "SimKai,Regular",
				"\xB7\xC2\xCB\xCE_GB2312", "SimFang,Regular",
				"\xC1\xA5\xCA\xE9", "SimLi,Regular",
				NULL
			};
			for (i = 0; cp936fonts[i]; i += 2)
				if (!strcmp(basefont, cp936fonts[i]))
					break;
			if (cp936fonts[i])
			{
				fz_warn(ctx, "workaround for S22PDF lying about chinese font encodings");
				pdf_drop_font(ctx, fontdesc);
				fontdesc = pdf_new_font_desc(ctx);
				pdf_load_font_descriptor(fontdesc, xref, descriptor, "Adobe-GB1", cp936fonts[i+1]);
				fontdesc->encoding = pdf_load_system_cmap(ctx, "GBK-EUC-H");
				fontdesc->to_unicode = pdf_load_system_cmap(ctx, "Adobe-GB1-UCS2");
				fontdesc->to_ttf_cmap = pdf_load_system_cmap(ctx, "Adobe-GB1-UCS2");

				face = fontdesc->font->ft_face;
				kind = ft_kind(face);
				goto skip_encoding;
			}
		}

		face = fontdesc->font->ft_face;
		kind = ft_kind(face);

		/* Encoding */

		symbolic = fontdesc->flags & 4;

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

		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 (symbolic && test->platform_id == 3 && test->encoding_id == 0)
					cmap = test;
			}
		}

		if (cmap)
		{
			fterr = FT_Set_Charmap(face, cmap);
			if (fterr)
				fz_warn(ctx, "freetype could not set cmap: %s", ft_error_string(fterr));
		}
		else
			fz_warn(ctx, "freetype could not find any cmaps");

		etable = fz_malloc_array(ctx, 256, sizeof(unsigned short));
		fontdesc->size += 256 * sizeof(unsigned short);
		for (i = 0; i < 256; i++)
		{
			estrings[i] = NULL;
			etable[i] = 0;
		}

		encoding = pdf_dict_gets(dict, "Encoding");
		if (encoding)
		{
			if (pdf_is_name(encoding))
				pdf_load_encoding(estrings, pdf_to_name(encoding));

			if (pdf_is_dict(encoding))
			{
				pdf_obj *base, *diff, *item;

				base = pdf_dict_gets(encoding, "BaseEncoding");
				if (pdf_is_name(base))
					pdf_load_encoding(estrings, pdf_to_name(base));
				else if (!fontdesc->is_embedded && !symbolic)
					pdf_load_encoding(estrings, "StandardEncoding");

				diff = pdf_dict_gets(encoding, "Differences");
				if (pdf_is_array(diff))
				{
					n = pdf_array_len(diff);
					k = 0;
					for (i = 0; i < n; i++)
					{
						item = pdf_array_get(diff, i);
						if (pdf_is_int(item))
							k = pdf_to_int(item);
						if (pdf_is_name(item) && k >= 0 && k < nelem(estrings))
							estrings[k++] = pdf_to_name(item);
					}
				}
			}
		}

		/* start with the builtin encoding */
		for (i = 0; i < 256; i++)
			etable[i] = ft_char_index(face, i);

		fz_lock(ctx, FZ_LOCK_FREETYPE);

		/* built-in and substitute fonts may be a different type than what the document expects */
		subtype = pdf_to_name(pdf_dict_gets(dict, "Subtype"));
		if (!strcmp(subtype, "Type1"))
			kind = TYPE1;
		else if (!strcmp(subtype, "MMType1"))
			kind = TYPE1;
		else if (!strcmp(subtype, "TrueType"))
			kind = TRUETYPE;
		else if (!strcmp(subtype, "CIDFontType0"))
			kind = TYPE1;
		else if (!strcmp(subtype, "CIDFontType2"))
			kind = TRUETYPE;

		/* encode by glyph name where we can */
		if (kind == TYPE1)
		{
			for (i = 0; i < 256; i++)
			{
				if (estrings[i])
				{
					etable[i] = FT_Get_Name_Index(face, estrings[i]);
					if (etable[i] == 0)
					{
						int aglcode = pdf_lookup_agl(estrings[i]);
						const char **dupnames = pdf_lookup_agl_duplicates(aglcode);
						while (*dupnames)
						{
							etable[i] = FT_Get_Name_Index(face, (char*)*dupnames);
							if (etable[i])
								break;
							dupnames++;
						}
					}
				}
			}
		}

		/* encode by glyph name where we can */
		if (kind == TRUETYPE)
		{
			/* Unicode cmap */
			if (!symbolic && face->charmap && face->charmap->platform_id == 3)
			{
				for (i = 0; i < 256; i++)
				{
					if (estrings[i])
					{
						int aglcode = pdf_lookup_agl(estrings[i]);
						if (!aglcode)
							etable[i] = FT_Get_Name_Index(face, estrings[i]);
						else
							etable[i] = ft_char_index(face, aglcode);
					}
				}
			}

			/* MacRoman cmap */
			else if (!symbolic && face->charmap && face->charmap->platform_id == 1)
			{
				for (i = 0; i < 256; i++)
				{
					if (estrings[i])
					{
						k = lookup_mre_code(estrings[i]);
						if (k <= 0)
							etable[i] = FT_Get_Name_Index(face, estrings[i]);
						else
							etable[i] = ft_char_index(face, k);
					}
				}
			}

			/* Symbolic cmap */
			else if (!face->charmap || face->charmap->encoding != FT_ENCODING_MS_SYMBOL)
			{
				for (i = 0; i < 256; i++)
				{
					if (estrings[i])
					{
						etable[i] = FT_Get_Name_Index(face, estrings[i]);
						if (etable[i] == 0)
							etable[i] = ft_char_index(face, i);
					}
				}
			}
		}

		/* try to reverse the glyph names from the builtin encoding */
		for (i = 0; i < 256; i++)
		{
			if (etable[i] && !estrings[i])
			{
				if (FT_HAS_GLYPH_NAMES(face))
				{
					fterr = FT_Get_Glyph_Name(face, etable[i], ebuffer[i], 32);
					if (fterr)
						fz_warn(ctx, "freetype get glyph name (gid %d): %s", etable[i], ft_error_string(fterr));
					if (ebuffer[i][0])
						estrings[i] = ebuffer[i];
				}
				else
				{
					estrings[i] = (char*) pdf_win_ansi[i]; /* discard const */
				}
			}
		}

		/* symbolic Type 1 fonts with an implicit encoding and non-standard glyph names */
		if (kind == TYPE1 && symbolic)
		{
			for (i = 0; i < 256; i++)
				if (etable[i] && estrings[i] && !pdf_lookup_agl(estrings[i]))
					estrings[i] = (char*) pdf_standard[i];
		}

		fz_unlock(ctx, FZ_LOCK_FREETYPE);

		fontdesc->encoding = pdf_new_identity_cmap(ctx, 0, 1);
		fontdesc->size += pdf_cmap_size(ctx, fontdesc->encoding);
		fontdesc->cid_to_gid_len = 256;
		fontdesc->cid_to_gid = etable;

		fz_try(ctx)
		{
			pdf_load_to_unicode(xref, fontdesc, estrings, NULL, pdf_dict_gets(dict, "ToUnicode"));
		}
		fz_catch(ctx)
		{
			fz_warn(ctx, "cannot load ToUnicode CMap");
		}

	skip_encoding:

		/* Widths */

		pdf_set_default_hmtx(ctx, fontdesc, fontdesc->missing_width);

		widths = pdf_dict_gets(dict, "Widths");
		if (widths)
		{
			int first, last;

			first = pdf_to_int(pdf_dict_gets(dict, "FirstChar"));
			last = pdf_to_int(pdf_dict_gets(dict, "LastChar"));

			if (first < 0 || last > 255 || first > last)
				first = last = 0;

			for (i = 0; i < last - first + 1; i++)
			{
				int wid = pdf_to_int(pdf_array_get(widths, i));
				pdf_add_hmtx(ctx, fontdesc, i + first, i + first, wid);
			}
		}
		else
		{
			fz_lock(ctx, FZ_LOCK_FREETYPE);
			fterr = FT_Set_Char_Size(face, 1000, 1000, 72, 72);
			if (fterr)
				fz_warn(ctx, "freetype set character size: %s", ft_error_string(fterr));
			for (i = 0; i < 256; i++)
			{
				pdf_add_hmtx(ctx, fontdesc, i, i, ft_width(ctx, fontdesc, i));
			}
			fz_unlock(ctx, FZ_LOCK_FREETYPE);
		}

		pdf_end_hmtx(ctx, fontdesc);
	}
	fz_catch(ctx)
	{
		if (fontdesc && etable != fontdesc->cid_to_gid)
			fz_free(ctx, etable);
		pdf_drop_font(ctx, fontdesc);
		fz_throw(ctx, "cannot load simple font (%d %d R)", pdf_to_num(dict), pdf_to_gen(dict));
	}
	return fontdesc;
}
fz_error
pdf_load_to_unicode(pdf_font_desc *font, pdf_xref *xref,
	char **strings, char *collection, fz_obj *cmapstm)
{
	fz_error error = fz_okay;
	pdf_cmap *cmap;
	int cid;
	int ucsbuf[8];
	int ucslen;
	int i;

	if (pdf_is_stream(xref, fz_to_num(cmapstm), fz_to_gen(cmapstm)))
	{
		error = pdf_load_embedded_cmap(&cmap, xref, cmapstm);
		if (error)
			return fz_rethrow(error, "cannot load embedded cmap (%d %d R)", fz_to_num(cmapstm), fz_to_gen(cmapstm));

		font->to_unicode = pdf_new_cmap();

		for (i = 0; i < (strings ? 256 : 65536); i++)
		{
			cid = pdf_lookup_cmap(font->encoding, i);
			if (cid >= 0)
			{
				ucslen = pdf_lookup_cmap_full(cmap, i, ucsbuf);
				if (ucslen == 1)
					pdf_map_range_to_range(font->to_unicode, cid, cid, ucsbuf[0]);
				if (ucslen > 1)
					pdf_map_one_to_many(font->to_unicode, cid, ucsbuf, ucslen);
			}
		}

		pdf_sort_cmap(font->to_unicode);

		pdf_drop_cmap(cmap);
	}

	else if (collection)
	{
		error = fz_okay;

		if (!strcmp(collection, "Adobe-CNS1"))
			error = pdf_load_system_cmap(&font->to_unicode, "Adobe-CNS1-UCS2");
		else if (!strcmp(collection, "Adobe-GB1"))
			error = pdf_load_system_cmap(&font->to_unicode, "Adobe-GB1-UCS2");
		else if (!strcmp(collection, "Adobe-Japan1"))
			error = pdf_load_system_cmap(&font->to_unicode, "Adobe-Japan1-UCS2");
		else if (!strcmp(collection, "Adobe-Korea1"))
			error = pdf_load_system_cmap(&font->to_unicode, "Adobe-Korea1-UCS2");

		if (error)
			return fz_rethrow(error, "cannot load ToUnicode system cmap %s-UCS2", collection);
	}

	if (strings)
	{
		/* TODO one-to-many mappings */

		font->cid_to_ucs_len = 256;
		font->cid_to_ucs = fz_calloc(256, sizeof(unsigned short));

		for (i = 0; i < 256; i++)
		{
			if (strings[i])
				font->cid_to_ucs[i] = pdf_lookup_agl(strings[i]);
			else
				font->cid_to_ucs[i] = '?';
		}
	}

	if (!font->to_unicode && !font->cid_to_ucs)
	{
		/* TODO: synthesize a ToUnicode if it's a freetype font with
		 * cmap and/or post tables or if it has glyph names. */
	}

	return fz_okay;
}