Support creating both serif and sans-serif CJK fonts in mutool create.

Also support Vertical WMode.

4 files changed, 65 insertions, 36 deletions

diff --git a/source/fitz/noto.c b/source/fitz/noto.c
index ee1dadcc..1d039794 100644
--- a/source/fitz/noto.c
+++ b/source/fitz/noto.c
@@ -188,7 +188,7 @@ fz_lookup_noto_font(fz_context *ctx, int script, int language, int serif, int *s
 	case UCDN_SCRIPT_KATAKANA:
 		return fz_lookup_cjk_font(ctx, FZ_ADOBE_JAPAN_1, serif, 0, size, NULL);
 	case UCDN_SCRIPT_BOPOMOFO:
-		return fz_lookup_cjk_font(ctx, FZ_ADOBE_GB_1, serif, 0, size, NULL);
+		return fz_lookup_cjk_font(ctx, FZ_ADOBE_CNS_1, serif, 0, size, NULL);
 	case UCDN_SCRIPT_HAN:
 		switch (language)
 		{
diff --git a/source/pdf/pdf-font.c b/source/pdf/pdf-font.c
index b00fe18f..3ed8cbb9 100644
--- a/source/pdf/pdf-font.c
+++ b/source/pdf/pdf-font.c
@@ -2158,27 +2158,28 @@ pdf_font_writing_supported(fz_font *font)
 
 /* Add a non-embedded UTF16-encoded CID-font for the CJK scripts: CNS1, GB1, Japan1, or Korea1 */
 pdf_obj *
-pdf_add_cjk_font(fz_context *ctx, pdf_document *doc, fz_font *fzfont, int script)
+pdf_add_cjk_font(fz_context *ctx, pdf_document *doc, fz_font *fzfont, int script, int wmode, int serif)
 {
 	pdf_obj *fref, *font, *subfont, *fontdesc;
 	pdf_obj *dfonts;
 	fz_rect bbox = { -200, -200, 1200, 1200 };
 	unsigned char digest[16];
+	int flags;
 
 	const char *basefont, *encoding, *ordering;
 	int supplement;
 
 	switch (script)
 	{
-	case FZ_ADOBE_CNS_1:
-		basefont = "Song";
-		encoding = "UniCNS-UTF16-H";
+	case FZ_ADOBE_CNS_1: /* traditional chinese */
+		basefont = serif ? "Ming" : "Fangti";
+		encoding = wmode ? "UniCNS-UTF16-V" : "UniCNS-UTF16-H";
 		ordering = "CNS1";
 		supplement = 7;
 		break;
-	case FZ_ADOBE_GB_1:
-		basefont = "Ming";
-		encoding = "UniGB-UTF16-H";
+	case FZ_ADOBE_GB_1: /* simplified chinese */
+		basefont = serif ? "Song" : "Heiti";
+		encoding = wmode ? "UniGB-UTF16-V" : "UniGB-UTF16-H";
 		ordering = "GB1";
 		supplement = 5;
 		break;
@@ -2186,19 +2187,23 @@ pdf_add_cjk_font(fz_context *ctx, pdf_document *doc, fz_font *fzfont, int script
 		script = FZ_ADOBE_JAPAN_1;
 		/* fall through */
 	case FZ_ADOBE_JAPAN_1:
-		basefont = "Mincho";
-		encoding = "UniJIS-UTF16-H";
+		basefont = serif ? "Mincho" : "Gothic";
+		encoding = wmode ? "UniJIS-UTF16-V" : "UniJIS-UTF16-H";
 		ordering = "Japan1";
 		supplement = 6;
 		break;
 	case FZ_ADOBE_KOREA_1:
-		basefont = "Batang";
-		encoding = "UniKS-UTF16-H";
+		basefont = serif ? "Batang" : "Dotum";
+		encoding = wmode ? "UniKS-UTF16-V" : "UniKS-UTF16-H";
 		ordering = "Korea1";
 		supplement = 2;
 		break;
 	}
 
+	flags = PDF_FD_SYMBOLIC;
+	if (serif)
+		flags |= PDF_FD_SERIF;
+
 	fref = pdf_find_font_resource(ctx, doc, PDF_CJK_FONT_RESOURCE, script, fzfont->buffer, digest);
 	if (fref)
 		return fref;
@@ -2222,8 +2227,8 @@ pdf_add_cjk_font(fz_context *ctx, pdf_document *doc, fz_font *fzfont, int script
 			{
 				pdf_dict_put(ctx, fontdesc, PDF_NAME(Type), PDF_NAME(FontDescriptor));
 				pdf_dict_put_text_string(ctx, fontdesc, PDF_NAME(FontName), basefont);
-				pdf_dict_put_int(ctx, fontdesc, PDF_NAME(Flags), 0);
 				pdf_dict_put_rect(ctx, fontdesc, PDF_NAME(FontBBox), &bbox);
+				pdf_dict_put_int(ctx, fontdesc, PDF_NAME(Flags), flags);
 				pdf_dict_put_int(ctx, fontdesc, PDF_NAME(ItalicAngle), 0);
 				pdf_dict_put_int(ctx, fontdesc, PDF_NAME(Ascent), 1000);
 				pdf_dict_put_int(ctx, fontdesc, PDF_NAME(Descent), -200);
diff --git a/source/tools/murun.c b/source/tools/murun.c
index e051be44..d9b5412e 100644
--- a/source/tools/murun.c
+++ b/source/tools/murun.c
@@ -3264,16 +3264,25 @@ static void ffi_PDFDocument_addCJKFont(js_State *J)
 	pdf_document *pdf = js_touserdata(J, 0, "pdf_document");
 	fz_font *font = js_touserdata(J, 1, "fz_font");
 	const char *on = js_tostring(J, 2);
+	const char *wm = js_tostring(J, 3);
+	const char *ss = js_tostring(J, 4);
 	int ord = FZ_ADOBE_JAPAN_1;
+	int wmode = 0;
+	int serif = 1;
 	pdf_obj *ind = NULL;
 
-	if (!strcmp(on, "CNS1") || !strcmp(on, "CN")) ord = FZ_ADOBE_CNS_1;
-	else if (!strcmp(on, "GB1") || !strcmp(on, "TW")) ord = FZ_ADOBE_GB_1;
+	if (!strcmp(on, "CNS1") || !strcmp(on, "TW") || !strcmp(on, "TC") || !strcmp(on, "Hant")) ord = FZ_ADOBE_CNS_1;
+	else if (!strcmp(on, "GB1") || !strcmp(on, "CN") || !strcmp(on, "SC") || !strcmp(on, "Hans")) ord = FZ_ADOBE_GB_1;
 	else if (!strcmp(on, "Korea1") || !strcmp(on, "KR") || !strcmp(on, "KO")) ord = FZ_ADOBE_KOREA_1;
 	else if (!strcmp(on, "Japan1") || !strcmp(on, "JP") || !strcmp(on, "JA")) ord = FZ_ADOBE_JAPAN_1;
 
+	if (!strcmp(wm, "V"))
+		wmode = 1;
+	if (!strcmp(ss, "sans") || !strcmp(ss, "sans-serif"))
+		serif = 0;
+
 	fz_try(ctx)
-		ind = pdf_add_cjk_font(ctx, pdf, font, ord);
+		ind = pdf_add_cjk_font(ctx, pdf, font, ord, wmode, serif);
 	fz_catch(ctx)
 		rethrow(J);
 
@@ -4676,7 +4685,7 @@ int murun_main(int argc, char **argv)
 		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, 2);
-		jsB_propfun(J, "PDFDocument.addCJKFont", ffi_PDFDocument_addCJKFont, 2);
+		jsB_propfun(J, "PDFDocument.addCJKFont", ffi_PDFDocument_addCJKFont, 4);
 		jsB_propfun(J, "PDFDocument.addFont", ffi_PDFDocument_addFont, 1);
 		jsB_propfun(J, "PDFDocument.addImage", ffi_PDFDocument_addImage, 1);
 		jsB_propfun(J, "PDFDocument.addPage", ffi_PDFDocument_addPage, 4);
diff --git a/source/tools/pdfcreate.c b/source/tools/pdfcreate.c
index 33892b45..f89bc2e4 100644
--- a/source/tools/pdfcreate.c
+++ b/source/tools/pdfcreate.c
@@ -22,7 +22,7 @@ static void usage(void)
 		"\t%%%%MediaBox LLX LLY URX URY\n"
 		"\t%%%%Rotate Angle\n"
 		"\t%%%%Font Name Filename (or base 14 font name)\n"
-		"\t%%%%CJKFont Name Ordering (CNS1, GB1, Japan1, or Korea1)\n"
+		"\t%%%%CJKFont Name Ordering WMode Style (Ordering=CNS1|GB1|Japan1|Korea1, WMode=H|V, Style=serif|sans)\n"
 		"\t%%%%Image Name Filename\n\n"
 		);
 	fputs(fz_pdf_write_options_usage, stderr);
@@ -67,20 +67,30 @@ static void add_font_res(pdf_obj *resources, char *name, char *path, char *encna
 	fz_drop_font(ctx, font);
 }
 
-static void add_cjkfont_res(pdf_obj *resources, char *name, char *on)
+static void add_cjkfont_res(pdf_obj *resources, char *name, char *on, char *wm, char *style)
 {
 	const unsigned char *data;
-	int size, index, ordering;
+	int size, index, ordering, wmode, serif;
 	fz_font *font;
 	pdf_obj *subres, *ref;
 
-	if (!strcmp(on, "CNS1") || !strcmp(on, "CN")) ordering = FZ_ADOBE_CNS_1;
-	else if (!strcmp(on, "GB1") || !strcmp(on, "TW")) ordering = FZ_ADOBE_GB_1;
+	if (!strcmp(on, "CNS1") || !strcmp(on, "TW") || !strcmp(on, "TC") || !strcmp(on, "Hant")) ordering = FZ_ADOBE_CNS_1;
+	else if (!strcmp(on, "GB1") || !strcmp(on, "CN") || !strcmp(on, "SC") || !strcmp(on, "Hans")) ordering = FZ_ADOBE_GB_1;
 	else if (!strcmp(on, "Japan1") || !strcmp(on, "JP") || !strcmp(on, "JA")) ordering = FZ_ADOBE_JAPAN_1;
 	else if (!strcmp(on, "Korea1") || !strcmp(on, "KR") || !strcmp(on, "KO")) ordering = FZ_ADOBE_KOREA_1;
 	else ordering = FZ_ADOBE_JAPAN_1;
 
-	data = fz_lookup_cjk_font(ctx, ordering, 0, 0, &size, &index);
+	if (wm && !strcmp(wm, "V"))
+		wmode = 1;
+	else
+		wmode = 0;
+
+	if (style && (!strcmp(style, "sans") || !strcmp(style, "sans-serif")))
+		serif = 0;
+	else
+		serif = 1;
+
+	data = fz_lookup_cjk_font(ctx, ordering, serif, wmode, &size, &index);
 	font = fz_new_font_from_memory(ctx, NULL, data, size, index, 0);
 
 	subres = pdf_dict_get(ctx, resources, PDF_NAME(Font));
@@ -90,7 +100,7 @@ static void add_cjkfont_res(pdf_obj *resources, char *name, char *on)
 		pdf_dict_put_drop(ctx, resources, PDF_NAME(Font), subres);
 	}
 
-	ref = pdf_add_cjk_font(ctx, doc, font, ordering);
+	ref = pdf_add_cjk_font(ctx, doc, font, ordering, wmode, serif);
 	pdf_dict_puts(ctx, subres, name, ref);
 	pdf_drop_obj(ctx, ref);
 
@@ -124,7 +134,7 @@ The input is a raw content stream, with commands embedded in comments:
 %%MediaBox LLX LLY URX URY
 %%Rotate Angle
 %%Font Name Filename (or base 14 font name) [Encoding (Latin, Greek or Cyrillic)]
-%%CJKFont Name Ordering (CNS1, GB1, Japan1, or Korea1)
+%%CJKFont Name Ordering WMode Style (Ordering=CNS1|GB1|Japan1|Korea1, WMode=H|V, Style=serif|sans)
 %%Image Name Filename
 */
 static void create_page(char *input)
@@ -133,7 +143,7 @@ static void create_page(char *input)
 	int rotate = 0;
 
 	char line[4096];
-	char *s, *t, *p;
+	char *s, *p;
 	fz_stream *stm;
 
 	fz_buffer *contents;
@@ -163,25 +173,30 @@ static void create_page(char *input)
 			}
 			else if (!strcmp(s, "%%Font"))
 			{
-				s = fz_strsep(&p, " ");
-				t = fz_strsep(&p, " ");
-				if (!s || !t)
+				char *name = fz_strsep(&p, " ");
+				char *path = fz_strsep(&p, " ");
+				char *enc = fz_strsep(&p, " ");
+				if (!name || !path)
 					fz_throw(ctx, FZ_ERROR_GENERIC, "Font directive missing arguments");
-				add_font_res(resources, s, t, p);
+				add_font_res(resources, name, path, enc);
 			}
 			else if (!strcmp(s, "%%CJKFont"))
 			{
-				s = fz_strsep(&p, " ");
-				if (!s || !p)
+				char *name = fz_strsep(&p, " ");
+				char *ordering = fz_strsep(&p, " ");
+				char *wmode = fz_strsep(&p, " ");
+				char *style = fz_strsep(&p, " ");
+				if (!name || !ordering)
 					fz_throw(ctx, FZ_ERROR_GENERIC, "CJKFont directive missing arguments");
-				add_cjkfont_res(resources, s, p);
+				add_cjkfont_res(resources, name, ordering, wmode, style);
 			}
 			else if (!strcmp(s, "%%Image"))
 			{
-				s = fz_strsep(&p, " ");
-				if (!s || !p)
+				char *name = fz_strsep(&p, " ");
+				char *path = fz_strsep(&p, " ");
+				if (!name || !path)
 					fz_throw(ctx, FZ_ERROR_GENERIC, "Image directive missing arguments");
-				add_image_res(resources, s, p);
+				add_image_res(resources, name, path);
 			}
 		}
 		else