Add simple fonts with 8-bit greek and cyrillic encodings.

Use KOI8-U for Cyrillic, and ISO 8859-7 for Greek.

Use with 'mutool create' using an extra argument to the %%Font directive:

    %%Font TmRmC Times-Roman Cyrillic
    BT
    /TmRmC 16 Tf 10 10 Td
    <fa c4 d2 c1 d7 d3 d4 d7 d5 ca d4 c5 21> Tj
    ET

The alternatives are "Latin", "Greek", and "Cyrillic".

6 files changed, 212 insertions, 12 deletions

diff --git a/source/pdf/pdf-encoding.c b/source/pdf/pdf-encoding.c
index 0b2cdfca..04874457 100644
--- a/source/pdf/pdf-encoding.c
+++ b/source/pdf/pdf-encoding.c
@@ -85,3 +85,41 @@ pdf_lookup_agl_duplicates(int ucs)
 	}
 	return empty_dup_list;
 }
+
+int pdf_cyrillic_from_unicode(int u)
+{
+	int l = 0;
+	int r = nelem(koi8u_from_unicode) - 1;
+	if (u < 128)
+		return u;
+	while (l <= r)
+	{
+		int m = (l + r) >> 1;
+		if (u < koi8u_from_unicode[m].u)
+			r = m - 1;
+		else if (u > koi8u_from_unicode[m].u)
+			l = m + 1;
+		else
+			return koi8u_from_unicode[m].c;
+	}
+	return -1;
+}
+
+int pdf_greek_from_unicode(int u)
+{
+	int l = 0;
+	int r = nelem(iso8859_7_from_unicode) - 1;
+	if (u < 128)
+		return u;
+	while (l <= r)
+	{
+		int m = (l + r) >> 1;
+		if (u < iso8859_7_from_unicode[m].u)
+			r = m - 1;
+		else if (u > iso8859_7_from_unicode[m].u)
+			l = m + 1;
+		else
+			return iso8859_7_from_unicode[m].c;
+	}
+	return -1;
+}
diff --git a/source/pdf/pdf-encodings.h b/source/pdf/pdf-encodings.h
index 025e9d03..a5d2e7ec 100644
--- a/source/pdf/pdf-encodings.h
+++ b/source/pdf/pdf-encodings.h
@@ -213,3 +213,107 @@ const char * const pdf_win_ansi[256] = { _notdef, _notdef, _notdef,
 	"divide", "oslash", "ugrave", "uacute", "ucircumflex", "udieresis",
 	"yacute", "thorn", "ydieresis"
 };
+
+const char * const pdf_glyph_name_from_koi8u[128] = {
+	_notdef, _notdef, _notdef, _notdef, _notdef, _notdef, _notdef, _notdef,
+	_notdef, _notdef, _notdef, _notdef, _notdef, _notdef, _notdef, _notdef,
+	_notdef, _notdef, _notdef, "integraltp", _notdef, "bulletoperator",
+	"radical", "approxequal", "lessequal", "greaterequal",
+	"nonbreakingspace", "integralbt", "degree", "twosuperior",
+	"periodcentered", "divide", _notdef, _notdef, _notdef, "iocyrillic",
+	"ecyrillic", _notdef, "icyrillic", "yicyrillic", _notdef, _notdef,
+	_notdef, _notdef, _notdef, "gheupturncyrillic", _notdef, _notdef,
+	_notdef, _notdef, _notdef, "afii10023", "afii10053", _notdef,
+	"afii10055", "afii10056", _notdef, _notdef, _notdef, _notdef, _notdef,
+	"afii10050", _notdef, "copyright", "iucyrillic", "afii10065",
+	"becyrillic", "tsecyrillic", "decyrillic", "iecyrillic", "efcyrillic",
+	"gecyrillic", "khacyrillic", "iicyrillic", "iishortcyrillic",
+	"kacyrillic", "elcyrillic", "emcyrillic", "encyrillic", "ocyrillic",
+	"pecyrillic", "iacyrillic", "ercyrillic", "escyrillic", "tecyrillic",
+	"ucyrillic", "zhecyrillic", "vecyrillic", "softsigncyrillic",
+	"yericyrillic", "zecyrillic", "shacyrillic", "ereversedcyrillic",
+	"shchacyrillic", "checyrillic", "hardsigncyrillic", "afii10048",
+	"afii10017", "afii10018", "afii10040", "afii10021", "afii10022",
+	"afii10038", "afii10020", "afii10039", "afii10026", "afii10027",
+	"afii10028", "afii10029", "afii10030", "afii10031", "afii10032",
+	"afii10033", "afii10049", "afii10034", "afii10035", "afii10036",
+	"afii10037", "afii10024", "afii10019", "afii10046", "afii10045",
+	"afii10025", "afii10042", "afii10047", "afii10043", "afii10041",
+	"afii10044",
+};
+
+static const struct { unsigned short u, c; } koi8u_from_unicode[] = {
+	{0x00a0,154}, {0x00a9,191}, {0x00b0,156}, {0x00b2,157}, {0x00b7,158},
+	{0x00f7,159}, {0x0401,179}, {0x0404,180}, {0x0406,182}, {0x0407,183},
+	{0x0410,225}, {0x0411,226}, {0x0412,247}, {0x0413,231}, {0x0414,228},
+	{0x0415,229}, {0x0416,246}, {0x0417,250}, {0x0418,233}, {0x0419,234},
+	{0x041a,235}, {0x041b,236}, {0x041c,237}, {0x041d,238}, {0x041e,239},
+	{0x041f,240}, {0x0420,242}, {0x0421,243}, {0x0422,244}, {0x0423,245},
+	{0x0424,230}, {0x0425,232}, {0x0426,227}, {0x0427,254}, {0x0428,251},
+	{0x0429,253}, {0x042a,255}, {0x042b,249}, {0x042c,248}, {0x042d,252},
+	{0x042e,224}, {0x042f,241}, {0x0430,193}, {0x0431,194}, {0x0432,215},
+	{0x0433,199}, {0x0434,196}, {0x0435,197}, {0x0436,214}, {0x0437,218},
+	{0x0438,201}, {0x0439,202}, {0x043a,203}, {0x043b,204}, {0x043c,205},
+	{0x043d,206}, {0x043e,207}, {0x043f,208}, {0x0440,210}, {0x0441,211},
+	{0x0442,212}, {0x0443,213}, {0x0444,198}, {0x0445,200}, {0x0446,195},
+	{0x0447,222}, {0x0448,219}, {0x0449,221}, {0x044a,223}, {0x044b,217},
+	{0x044c,216}, {0x044d,220}, {0x044e,192}, {0x044f,209}, {0x0451,163},
+	{0x0454,164}, {0x0456,166}, {0x0457,167}, {0x0490,189}, {0x0491,173},
+	{0x2219,149}, {0x221a,150}, {0x2248,151}, {0x2264,152}, {0x2265,153},
+	{0x2320,147}, {0x2321,155}, {0x2500,128}, {0x2502,129}, {0x250c,130},
+	{0x2510,131}, {0x2514,132}, {0x2518,133}, {0x251c,134}, {0x2524,135},
+	{0x252c,136}, {0x2534,137}, {0x253c,138}, {0x2550,160}, {0x2551,161},
+	{0x2552,162}, {0x2554,165}, {0x2557,168}, {0x2558,169}, {0x2559,170},
+	{0x255a,171}, {0x255b,172}, {0x255d,174}, {0x255e,175}, {0x255f,176},
+	{0x2560,177}, {0x2561,178}, {0x2563,181}, {0x2566,184}, {0x2567,185},
+	{0x2568,186}, {0x2569,187}, {0x256a,188}, {0x256c,190}, {0x2580,139},
+	{0x2584,140}, {0x2588,141}, {0x258c,142}, {0x2590,143}, {0x2591,144},
+	{0x2592,145}, {0x2593,146}, {0x25a0,148}
+};
+
+const char * const pdf_glyph_name_from_iso8859_7[128] = {
+	/* the block drawing characters have been omitted */
+	_notdef, _notdef, _notdef, _notdef, _notdef, _notdef, _notdef, _notdef,
+	_notdef, _notdef, _notdef, _notdef, _notdef, _notdef, _notdef, _notdef,
+	_notdef, _notdef, _notdef, _notdef, _notdef, _notdef, _notdef, _notdef,
+	_notdef, _notdef, _notdef, _notdef, _notdef, _notdef, _notdef, _notdef,
+	"nonbreakingspace", "quoteleft", "quoteright", "sterling", "euro",
+	_notdef, "brokenbar", "section", "dieresis", "copyright",
+	"ypogegrammeni", "guillemotleft", "logicalnot", "softhyphen", _notdef,
+	"horizontalbar", "degree", "plusminus", "twosuperior", "threesuperior",
+	"tonos", "dieresistonos", "Alphatonos", "periodcentered",
+	"Epsilontonos", "Etatonos", "Iotatonos", "guillemotright",
+	"Omicrontonos", "onehalf", "Upsilontonos", "Omegatonos",
+	"iotadieresistonos", "Alpha", "Beta", "Gamma", "Deltagreek", "Epsilon",
+	"Zeta", "Eta", "Theta", "Iota", "Kappa", "Lambda", "Mu", "Nu", "Xi",
+	"Omicron", "Pi", "Rho", _notdef, "Sigma", "Tau", "Upsilon", "Phi",
+	"Chi", "Psi", "Omegagreek", "Iotadieresis", "Upsilondieresis",
+	"alphatonos", "epsilontonos", "etatonos", "iotatonos",
+	"upsilondieresistonos", "alpha", "beta", "gamma", "delta", "epsilon",
+	"zeta", "eta", "theta", "iota", "kappa", "lambda", "mugreek", "nu",
+	"xi", "omicron", "pi", "rho", "sigmafinal", "sigma", "tau", "upsilon",
+	"phi", "chi", "psi", "omega", "iotadieresis", "upsilondieresis",
+	"omicrontonos", "upsilontonos", "omegatonos", _notdef,
+};
+
+static const struct { unsigned short u, c; } iso8859_7_from_unicode[] = {
+	{0x00a0,160}, {0x00a3,163}, {0x00a6,166}, {0x00a7,167}, {0x00a8,168},
+	{0x00a9,169}, {0x00ab,171}, {0x00ac,172}, {0x00ad,173}, {0x00b0,176},
+	{0x00b1,177}, {0x00b2,178}, {0x00b3,179}, {0x00b7,183}, {0x00bb,187},
+	{0x00bd,189}, {0x037a,170}, {0x0384,180}, {0x0385,181}, {0x0386,182},
+	{0x0388,184}, {0x0389,185}, {0x038a,186}, {0x038c,188}, {0x038e,190},
+	{0x038f,191}, {0x0390,192}, {0x0391,193}, {0x0392,194}, {0x0393,195},
+	{0x0394,196}, {0x0395,197}, {0x0396,198}, {0x0397,199}, {0x0398,200},
+	{0x0399,201}, {0x039a,202}, {0x039b,203}, {0x039c,204}, {0x039d,205},
+	{0x039e,206}, {0x039f,207}, {0x03a0,208}, {0x03a1,209}, {0x03a3,211},
+	{0x03a4,212}, {0x03a5,213}, {0x03a6,214}, {0x03a7,215}, {0x03a8,216},
+	{0x03a9,217}, {0x03aa,218}, {0x03ab,219}, {0x03ac,220}, {0x03ad,221},
+	{0x03ae,222}, {0x03af,223}, {0x03b0,224}, {0x03b1,225}, {0x03b2,226},
+	{0x03b3,227}, {0x03b4,228}, {0x03b5,229}, {0x03b6,230}, {0x03b7,231},
+	{0x03b8,232}, {0x03b9,233}, {0x03ba,234}, {0x03bb,235}, {0x03bc,236},
+	{0x03bd,237}, {0x03be,238}, {0x03bf,239}, {0x03c0,240}, {0x03c1,241},
+	{0x03c2,242}, {0x03c3,243}, {0x03c4,244}, {0x03c5,245}, {0x03c6,246},
+	{0x03c7,247}, {0x03c8,248}, {0x03c9,249}, {0x03ca,250}, {0x03cb,251},
+	{0x03cc,252}, {0x03cd,253}, {0x03ce,254}, {0x2015,175}, {0x2018,161},
+	{0x2019,162}, {0x20ac,164},
+};
diff --git a/source/pdf/pdf-font.c b/source/pdf/pdf-font.c
index 6be72250..69044c53 100644
--- a/source/pdf/pdf-font.c
+++ b/source/pdf/pdf-font.c
@@ -2147,9 +2147,51 @@ pdf_add_cid_font(fz_context *ctx, pdf_document *doc, fz_font *font)
 	return fref;
 }
 
-/* Creates simple font */
+/* Create simple (8-bit encoding) fonts */
+
+static void
+pdf_add_simple_font_encoding_imp(fz_context *ctx, pdf_document *doc, pdf_obj *font, const char * const glyph_names[])
+{
+	pdf_obj *enc, *diff;
+	int i, last;
+
+	pdf_dict_put_drop(ctx, font, PDF_NAME_Encoding, enc = pdf_new_dict(ctx, doc, 3));
+	pdf_dict_put(ctx, enc, PDF_NAME_BaseEncoding, PDF_NAME_WinAnsiEncoding);
+	pdf_dict_put_drop(ctx, enc, PDF_NAME_Differences, diff = pdf_new_array(ctx, doc, 129));
+	last = 0;
+	for (i = 128; i < 256; ++i)
+	{
+		const char *glyph = glyph_names[i-128];
+		if (glyph)
+		{
+			if (last != i-1)
+				pdf_array_push_int(ctx, diff, i);
+			last = i;
+			pdf_array_push_name(ctx, diff, glyph);
+		}
+	}
+}
+
+static void
+pdf_add_simple_font_encoding(fz_context *ctx, pdf_document *doc, pdf_obj *fobj, int encoding)
+{
+	switch (encoding)
+	{
+	default:
+	case PDF_SIMPLE_ENCODING_LATIN:
+		pdf_dict_put(ctx, fobj, PDF_NAME_Encoding, PDF_NAME_WinAnsiEncoding);
+		break;
+	case PDF_SIMPLE_ENCODING_GREEK:
+		pdf_add_simple_font_encoding_imp(ctx, doc, fobj, pdf_glyph_name_from_iso8859_7);
+		break;
+	case PDF_SIMPLE_ENCODING_CYRILLIC:
+		pdf_add_simple_font_encoding_imp(ctx, doc, fobj, pdf_glyph_name_from_koi8u);
+		break;
+	}
+}
+
 pdf_obj *
-pdf_add_simple_font(fz_context *ctx, pdf_document *doc, fz_font *font)
+pdf_add_simple_font(fz_context *ctx, pdf_document *doc, fz_font *font, int encoding)
 {
 	pdf_obj *fobj = NULL;
 	pdf_obj *fref = NULL;
@@ -2174,7 +2216,7 @@ pdf_add_simple_font(fz_context *ctx, pdf_document *doc, fz_font *font)
 		/* Before we add this font as a resource check if the same font
 		 * already exists in our resources for this doc. If yes, then
 		 * hand back that reference */
-		fref = pdf_find_font_resource(ctx, doc, PDF_SIMPLE_FONT_RESOURCE, 0, font->buffer, digest);
+		fref = pdf_find_font_resource(ctx, doc, PDF_SIMPLE_FONT_RESOURCE, encoding, font->buffer, digest);
 		if (fref == NULL)
 		{
 			fobj = pdf_new_dict(ctx, doc, 10);
@@ -2184,7 +2226,6 @@ pdf_add_simple_font(fz_context *ctx, pdf_document *doc, fz_font *font)
 			case TYPE1: pdf_dict_put(ctx, fobj, PDF_NAME_Subtype, PDF_NAME_Type1); break;
 			case TRUETYPE: pdf_dict_put(ctx, fobj, PDF_NAME_Subtype, PDF_NAME_TrueType); break;
 			}
-			pdf_dict_put(ctx, fobj, PDF_NAME_Encoding, PDF_NAME_WinAnsiEncoding);
 
 			if (!is_builtin_font(ctx, font))
 			{
@@ -2213,6 +2254,8 @@ pdf_add_simple_font(fz_context *ctx, pdf_document *doc, fz_font *font)
 				pdf_dict_put_name(ctx, fobj, PDF_NAME_BaseFont, clean_font_name(font->name));
 			}
 
+			pdf_add_simple_font_encoding(ctx, doc, fobj, encoding);
+
 			fref = pdf_add_object(ctx, doc, fobj);
 
 			/* Add ref to our font resource hash table. */
diff --git a/source/tools/murun.c b/source/tools/murun.c
index 911fcf35..8b9cd29d 100644
--- a/source/tools/murun.c
+++ b/source/tools/murun.c
@@ -3251,10 +3251,16 @@ static void ffi_PDFDocument_addSimpleFont(js_State *J)
 	fz_context *ctx = js_getcontext(J);
 	pdf_document *pdf = js_touserdata(J, 0, "pdf_document");
 	fz_font *font = js_touserdata(J, 1, "fz_font");
+	const char *encname = js_tostring(J, 2);
 	pdf_obj *ind = NULL;
+	int enc = PDF_SIMPLE_ENCODING_LATIN;
+
+	if (!strcmp(encname, "Latin")) enc = PDF_SIMPLE_ENCODING_LATIN;
+	else if (!strcmp(encname, "Greek")) enc = PDF_SIMPLE_ENCODING_GREEK;
+	else if (!strcmp(encname, "Cyrillic")) enc = PDF_SIMPLE_ENCODING_CYRILLIC;
 
 	fz_try(ctx)
-		ind = pdf_add_simple_font(ctx, pdf, font);
+		ind = pdf_add_simple_font(ctx, pdf, font, enc);
 	fz_catch(ctx)
 		rethrow(J);
 
@@ -4697,7 +4703,7 @@ int murun_main(int argc, char **argv)
 		jsB_propfun(J, "PDFDocument.addObject", ffi_PDFDocument_addObject, 1);
 		jsB_propfun(J, "PDFDocument.addStream", ffi_PDFDocument_addStream, 2);
 		jsB_propfun(J, "PDFDocument.addRawStream", ffi_PDFDocument_addRawStream, 2);
-		jsB_propfun(J, "PDFDocument.addSimpleFont", ffi_PDFDocument_addSimpleFont, 1);
+		jsB_propfun(J, "PDFDocument.addSimpleFont", ffi_PDFDocument_addSimpleFont, 2);
 		jsB_propfun(J, "PDFDocument.addCJKFont", ffi_PDFDocument_addCJKFont, 2);
 		jsB_propfun(J, "PDFDocument.addFont", ffi_PDFDocument_addFont, 1);
 		jsB_propfun(J, "PDFDocument.addImage", ffi_PDFDocument_addImage, 1);
diff --git a/source/tools/pdfcreate.c b/source/tools/pdfcreate.c
index 93dff5fb..032f32d5 100644
--- a/source/tools/pdfcreate.c
+++ b/source/tools/pdfcreate.c
@@ -32,10 +32,10 @@ static void usage(void)
 static fz_context *ctx = NULL;
 static pdf_document *doc = NULL;
 
-static void add_font_res(pdf_obj *resources, char *name, char *path)
+static void add_font_res(pdf_obj *resources, char *name, char *path, char *encname)
 {
 	const unsigned char *data;
-	int size;
+	int size, enc;
 	fz_font *font;
 	pdf_obj *subres, *ref;
 
@@ -52,7 +52,15 @@ static void add_font_res(pdf_obj *resources, char *name, char *path)
 		pdf_dict_put_drop(ctx, resources, PDF_NAME_Font, subres);
 	}
 
-	ref = pdf_add_simple_font(ctx, doc, font);
+	enc = PDF_SIMPLE_ENCODING_LATIN;
+	if (encname)
+	{
+		if (!strcmp(encname, "Latin")) enc = PDF_SIMPLE_ENCODING_LATIN;
+		else if (!strcmp(encname, "Greek")) enc = PDF_SIMPLE_ENCODING_GREEK;
+		else if (!strcmp(encname, "Cyrillic")) enc = PDF_SIMPLE_ENCODING_CYRILLIC;
+	}
+
+	ref = pdf_add_simple_font(ctx, doc, font, enc);
 	pdf_dict_puts(ctx, subres, name, ref);
 	pdf_drop_obj(ctx, ref);
 
@@ -125,7 +133,7 @@ static void create_page(char *input)
 	int rotate = 0;
 
 	char line[4096];
-	char *s, *p;
+	char *s, *t, *p;
 	fz_stream *stm;
 
 	fz_buffer *contents;
@@ -156,7 +164,8 @@ static void create_page(char *input)
 			else if (!strcmp(s, "%%Font"))
 			{
 				s = fz_strsep(&p, " ");
-				add_font_res(resources, s, p);
+				t = fz_strsep(&p, " ");
+				add_font_res(resources, s, t, p);
 			}
 			else if (!strcmp(s, "%%CJKFont"))
 			{
diff --git a/source/tools/pdfportfolio.c b/source/tools/pdfportfolio.c
index 9a0efc01..8e252d43 100644
--- a/source/tools/pdfportfolio.c
+++ b/source/tools/pdfportfolio.c
@@ -255,7 +255,7 @@ int pdfportfolio_main(int argc, char **argv)
 
 			data = fz_lookup_base14_font(ctx, "Times-Roman", &size);
 			font = fz_new_font_from_memory(ctx, "Times-Roman", data, size, 0, 0);
-			font_obj = pdf_add_simple_font(ctx, doc, font);
+			font_obj = pdf_add_simple_font(ctx, doc, font, PDF_SIMPLE_ENCODING_LATIN);
 			fz_drop_font(ctx, font);
 
 			resources = pdf_add_object_drop(ctx, doc, pdf_new_dict(ctx, doc, 1));