Regularize language and script names.

Drop the unused 'serif' argument to the CJK lookup functions.

Use the BCP 47 names for CJK scripts and languages:
    zh-Hant for traditional Chinese,
    zh-Hans for simplified Chinese,
    ja for Japanese,
    ko for Korean.

The lookup function also allows commonly used language+country codes:
    zh-TW and zh-HK for traditional Chinese,
    zh-CN for simplified Chinese.

5 files changed, 74 insertions, 63 deletions

diff --git a/source/fitz/font.c b/source/fitz/font.c
index f9609aec..733d91da 100644
--- a/source/fitz/font.c
+++ b/source/fitz/font.c
@@ -230,7 +230,7 @@ struct fz_font_context_s
 
 	/* Cached fallback fonts */
 	fz_font *base14[14];
-	fz_font *cjk[8];
+	fz_font *cjk[4];
 	struct { fz_font *serif, *sans; } fallback[256];
 	fz_font *symbol1, *symbol2;
 	fz_font *emoji;
@@ -416,7 +416,7 @@ fz_font *fz_load_fallback_font(fz_context *ctx, int script, int language, int se
 		*fontp = fz_load_system_fallback_font(ctx, script, language, serif, bold, italic);
 		if (!*fontp)
 		{
-			data = fz_lookup_noto_font(ctx, script, language, serif, &size, &subfont);
+			data = fz_lookup_noto_font(ctx, script, language, &size, &subfont);
 			if (data)
 				*fontp = fz_new_font_from_memory(ctx, NULL, data, size, subfont, 0);
 		}
@@ -697,20 +697,19 @@ fz_new_base14_font(fz_context *ctx, const char *name)
 }
 
 fz_font *
-fz_new_cjk_font(fz_context *ctx, int ordering, int serif)
+fz_new_cjk_font(fz_context *ctx, int ordering)
 {
 	const unsigned char *data;
 	int size, index;
-	int x = (ordering * 2) + !!serif;
-	if (x >= 0 && x < nelem(ctx->font->cjk))
+	if (ordering >= 0 && ordering < nelem(ctx->font->cjk))
 	{
-		if (ctx->font->cjk[x])
-			return fz_keep_font(ctx, ctx->font->cjk[x]);
-		data = fz_lookup_cjk_font(ctx, ordering, serif, &size, &index);
+		if (ctx->font->cjk[ordering])
+			return fz_keep_font(ctx, ctx->font->cjk[ordering]);
+		data = fz_lookup_cjk_font(ctx, ordering, &size, &index);
 		if (data)
 		{
-			ctx->font->cjk[x] = fz_new_font_from_memory(ctx, NULL, data, size, index, 0);
-			return fz_keep_font(ctx, ctx->font->cjk[x]);
+			ctx->font->cjk[ordering] = fz_new_font_from_memory(ctx, NULL, data, size, index, 0);
+			return fz_keep_font(ctx, ctx->font->cjk[ordering]);
 		}
 	}
 	fz_throw(ctx, FZ_ERROR_GENERIC, "cannot find builtin CJK font");
diff --git a/source/fitz/noto.c b/source/fitz/noto.c
index 8301929f..c3eea9d6 100644
--- a/source/fitz/noto.c
+++ b/source/fitz/noto.c
@@ -139,18 +139,18 @@ fz_lookup_builtin_font(fz_context *ctx, const char *name, int is_bold, int is_it
 }
 
 const unsigned char *
-fz_lookup_cjk_font(fz_context *ctx, int ordering, int serif, int *size, int *subfont)
+fz_lookup_cjk_font(fz_context *ctx, int ordering, int *size, int *subfont)
 {
 	*subfont = 0;
 #ifndef TOFU_CJK
 #ifndef TOFU_CJK_EXT
 #ifndef TOFU_CJK_LANG
 	switch (ordering) {
-	case FZ_ADOBE_JAPAN_1: *subfont=0; RETURN(han, SourceHanSerif_Regular_ttc);
-	case FZ_ADOBE_KOREA_1: *subfont=1; RETURN(han, SourceHanSerif_Regular_ttc);
-	case FZ_ADOBE_GB_1: *subfont=2; RETURN(han, SourceHanSerif_Regular_ttc);
+	case FZ_ADOBE_JAPAN: *subfont=0; RETURN(han, SourceHanSerif_Regular_ttc);
+	case FZ_ADOBE_KOREA: *subfont=1; RETURN(han, SourceHanSerif_Regular_ttc);
+	case FZ_ADOBE_GB: *subfont=2; RETURN(han, SourceHanSerif_Regular_ttc);
 	default:
-	case FZ_ADOBE_CNS_1: *subfont=3; RETURN(han, SourceHanSerif_Regular_ttc);
+	case FZ_ADOBE_CNS: *subfont=3; RETURN(han, SourceHanSerif_Regular_ttc);
 	}
 #else
 	RETURN(droid, DroidSansFallbackFull_ttf);
@@ -163,8 +163,21 @@ fz_lookup_cjk_font(fz_context *ctx, int ordering, int serif, int *size, int *sub
 #endif
 }
 
+int
+fz_lookup_cjk_ordering_by_language(const char *name)
+{
+	if (!strcmp(name, "zh-Hant")) return FZ_ADOBE_CNS;
+	if (!strcmp(name, "zh-TW")) return FZ_ADOBE_CNS;
+	if (!strcmp(name, "zh-HK")) return FZ_ADOBE_CNS;
+	if (!strcmp(name, "zh-Hans")) return FZ_ADOBE_GB;
+	if (!strcmp(name, "zh-CN")) return FZ_ADOBE_GB;
+	if (!strcmp(name, "ja")) return FZ_ADOBE_JAPAN;
+	if (!strcmp(name, "ko")) return FZ_ADOBE_KOREA;
+	return -1;
+}
+
 const unsigned char *
-fz_lookup_noto_font(fz_context *ctx, int script, int language, int serif, int *size, int *subfont)
+fz_lookup_noto_font(fz_context *ctx, int script, int language, int *size, int *subfont)
 {
 	/* TODO: Noto(SansSyriacEstrangela); */
 	/* TODO: Noto(SansSyriacWestern); */
@@ -180,20 +193,20 @@ fz_lookup_noto_font(fz_context *ctx, int script, int language, int serif, int *s
 		break;
 
 	case UCDN_SCRIPT_HANGUL:
-		return fz_lookup_cjk_font(ctx, FZ_ADOBE_KOREA_1, serif, size, subfont);
+		return fz_lookup_cjk_font(ctx, FZ_ADOBE_KOREA, size, subfont);
 	case UCDN_SCRIPT_HIRAGANA:
 	case UCDN_SCRIPT_KATAKANA:
-		return fz_lookup_cjk_font(ctx, FZ_ADOBE_JAPAN_1, serif, size, subfont);
+		return fz_lookup_cjk_font(ctx, FZ_ADOBE_JAPAN, size, subfont);
 	case UCDN_SCRIPT_BOPOMOFO:
-		return fz_lookup_cjk_font(ctx, FZ_ADOBE_CNS_1, serif, size, subfont);
+		return fz_lookup_cjk_font(ctx, FZ_ADOBE_CNS, size, subfont);
 	case UCDN_SCRIPT_HAN:
 		switch (language)
 		{
-		case FZ_LANG_ja: return fz_lookup_cjk_font(ctx, FZ_ADOBE_JAPAN_1, serif, size, subfont);
-		case FZ_LANG_ko: return fz_lookup_cjk_font(ctx, FZ_ADOBE_KOREA_1, serif, size, subfont);
-		case FZ_LANG_zh_Hans: return fz_lookup_cjk_font(ctx, FZ_ADOBE_GB_1, serif, size, subfont);
+		case FZ_LANG_ja: return fz_lookup_cjk_font(ctx, FZ_ADOBE_JAPAN, size, subfont);
+		case FZ_LANG_ko: return fz_lookup_cjk_font(ctx, FZ_ADOBE_KOREA, size, subfont);
+		case FZ_LANG_zh_Hans: return fz_lookup_cjk_font(ctx, FZ_ADOBE_GB, size, subfont);
 		default:
-		case FZ_LANG_zh_Hant: return fz_lookup_cjk_font(ctx, FZ_ADOBE_CNS_1, serif, size, subfont);
+		case FZ_LANG_zh_Hant: return fz_lookup_cjk_font(ctx, FZ_ADOBE_CNS, size, subfont);
 		}
 
 	case UCDN_SCRIPT_BRAILLE: break; /* no dedicated font; fallback to NotoSansSymbols will cover this */
diff --git a/source/pdf/pdf-font.c b/source/pdf/pdf-font.c
index 3dae9884..a0df70fd 100644
--- a/source/pdf/pdf-font.c
+++ b/source/pdf/pdf-font.c
@@ -409,15 +409,15 @@ pdf_load_substitute_cjk_font(fz_context *ctx, pdf_font_desc *fontdesc, const cha
 	if (!fontdesc->font)
 	{
 		const unsigned char *data;
-		int len;
-		int index;
+		int size;
+		int subfont;
 
-		data = fz_lookup_cjk_font(ctx, ros, serif, &len, &index);
+		data = fz_lookup_cjk_font(ctx, ros, &size, &subfont);
 		if (!data)
 			fz_throw(ctx, FZ_ERROR_GENERIC, "cannot find builtin CJK font");
 
 		/* A glyph bbox cache is too big for CJK fonts. */
-		fontdesc->font = fz_new_font_from_memory(ctx, fontname, data, len, index, 0);
+		fontdesc->font = fz_new_font_from_memory(ctx, fontname, data, size, subfont, 0);
 	}
 
 	fontdesc->font->flags.ft_substitute = 1;
@@ -451,13 +451,13 @@ pdf_load_system_font(fz_context *ctx, pdf_font_desc *fontdesc, const char *fontn
 	if (collection)
 	{
 		if (!strcmp(collection, "Adobe-CNS1"))
-			pdf_load_substitute_cjk_font(ctx, fontdesc, fontname, FZ_ADOBE_CNS_1, serif);
+			pdf_load_substitute_cjk_font(ctx, fontdesc, fontname, FZ_ADOBE_CNS, serif);
 		else if (!strcmp(collection, "Adobe-GB1"))
-			pdf_load_substitute_cjk_font(ctx, fontdesc, fontname, FZ_ADOBE_GB_1, serif);
+			pdf_load_substitute_cjk_font(ctx, fontdesc, fontname, FZ_ADOBE_GB, serif);
 		else if (!strcmp(collection, "Adobe-Japan1"))
-			pdf_load_substitute_cjk_font(ctx, fontdesc, fontname, FZ_ADOBE_JAPAN_1, serif);
+			pdf_load_substitute_cjk_font(ctx, fontdesc, fontname, FZ_ADOBE_JAPAN, serif);
 		else if (!strcmp(collection, "Adobe-Korea1"))
-			pdf_load_substitute_cjk_font(ctx, fontdesc, fontname, FZ_ADOBE_KOREA_1, serif);
+			pdf_load_substitute_cjk_font(ctx, fontdesc, fontname, FZ_ADOBE_KOREA, serif);
 		else
 		{
 			if (strcmp(collection, "Adobe-Identity") != 0)
@@ -2133,28 +2133,28 @@ pdf_add_cjk_font(fz_context *ctx, pdf_document *doc, fz_font *fzfont, int script
 
 	switch (script)
 	{
-	case FZ_ADOBE_CNS_1: /* traditional chinese */
+	default:
+		script = FZ_ADOBE_CNS;
+		/* fall through */
+	case FZ_ADOBE_CNS: /* traditional chinese */
 		basefont = serif ? "Ming" : "Fangti";
 		encoding = wmode ? "UniCNS-UTF16-V" : "UniCNS-UTF16-H";
 		ordering = "CNS1";
 		supplement = 7;
 		break;
-	case FZ_ADOBE_GB_1: /* simplified chinese */
+	case FZ_ADOBE_GB: /* simplified chinese */
 		basefont = serif ? "Song" : "Heiti";
 		encoding = wmode ? "UniGB-UTF16-V" : "UniGB-UTF16-H";
 		ordering = "GB1";
 		supplement = 5;
 		break;
-	default:
-		script = FZ_ADOBE_JAPAN_1;
-		/* fall through */
-	case FZ_ADOBE_JAPAN_1:
+	case FZ_ADOBE_JAPAN:
 		basefont = serif ? "Mincho" : "Gothic";
 		encoding = wmode ? "UniJIS-UTF16-V" : "UniJIS-UTF16-H";
 		ordering = "Japan1";
 		supplement = 6;
 		break;
-	case FZ_ADOBE_KOREA_1:
+	case FZ_ADOBE_KOREA:
 		basefont = serif ? "Batang" : "Dotum";
 		encoding = wmode ? "UniKS-UTF16-V" : "UniKS-UTF16-H";
 		ordering = "Korea1";
diff --git a/source/tools/murun.c b/source/tools/murun.c
index 810b32ce..961d6a7a 100644
--- a/source/tools/murun.c
+++ b/source/tools/murun.c
@@ -3285,9 +3285,12 @@ static void ffi_PDFDocument_addSimpleFont(js_State *J)
 	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;
+	if (!strcmp(encname, "Latin") || !strcmp(encname, "Latn"))
+		enc = PDF_SIMPLE_ENCODING_LATIN;
+	else if (!strcmp(encname, "Greek") || !strcmp(encname, "Grek"))
+		enc = PDF_SIMPLE_ENCODING_GREEK;
+	else if (!strcmp(encname, "Cyrillic") || !strcmp(encname, "Cyrl"))
+		enc = PDF_SIMPLE_ENCODING_CYRILLIC;
 
 	fz_try(ctx)
 		ind = pdf_add_simple_font(ctx, pdf, font, enc);
@@ -3302,18 +3305,15 @@ static void ffi_PDFDocument_addCJKFont(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 *on = js_tostring(J, 2);
+	const char *lang = 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 ordering;
 	int wmode = 0;
 	int serif = 1;
 	pdf_obj *ind = NULL;
 
-	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;
+	ordering = fz_lookup_cjk_ordering_by_language(lang);
 
 	if (!strcmp(wm, "V"))
 		wmode = 1;
@@ -3321,7 +3321,7 @@ static void ffi_PDFDocument_addCJKFont(js_State *J)
 		serif = 0;
 
 	fz_try(ctx)
-		ind = pdf_add_cjk_font(ctx, pdf, font, ord, wmode, serif);
+		ind = pdf_add_cjk_font(ctx, pdf, font, ordering, wmode, serif);
 	fz_catch(ctx)
 		rethrow(J);
 
diff --git a/source/tools/pdfcreate.c b/source/tools/pdfcreate.c
index 52d16dbe..043936e4 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 WMode Style (Ordering=CNS1|GB1|Japan1|Korea1, WMode=H|V, Style=serif|sans)\n"
+		"\t%%%%CJKFont Name Language WMode Style (Language=zh-Hant|zh-Hans|ja|ko, WMode=H|V, Style=serif|sans)\n"
 		"\t%%%%Image Name Filename\n\n"
 		);
 	fputs(fz_pdf_write_options_usage, stderr);
@@ -55,9 +55,12 @@ static void add_font_res(pdf_obj *resources, char *name, char *path, char *encna
 	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;
+		if (!strcmp(encname, "Latin") || !strcmp(encname, "Latn"))
+			enc = PDF_SIMPLE_ENCODING_LATIN;
+		else if (!strcmp(encname, "Greek") || !strcmp(encname, "Grek"))
+			enc = PDF_SIMPLE_ENCODING_GREEK;
+		else if (!strcmp(encname, "Cyrillic") || !strcmp(encname, "Cyrl"))
+			enc = PDF_SIMPLE_ENCODING_CYRILLIC;
 	}
 
 	ref = pdf_add_simple_font(ctx, doc, font, enc);
@@ -67,18 +70,14 @@ 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, char *wm, char *style)
+static void add_cjkfont_res(pdf_obj *resources, char *name, char *lang, char *wm, char *style)
 {
 	const unsigned char *data;
 	int size, index, ordering, wmode, serif;
 	fz_font *font;
 	pdf_obj *subres, *ref;
 
-	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;
+	ordering = fz_lookup_cjk_ordering_by_language(lang);
 
 	if (wm && !strcmp(wm, "V"))
 		wmode = 1;
@@ -90,7 +89,7 @@ static void add_cjkfont_res(pdf_obj *resources, char *name, char *on, char *wm,
 	else
 		serif = 1;
 
-	data = fz_lookup_cjk_font(ctx, ordering, serif, &size, &index);
+	data = fz_lookup_cjk_font(ctx, ordering, &size, &index);
 	font = fz_new_font_from_memory(ctx, NULL, data, size, index, 0);
 
 	subres = pdf_dict_get(ctx, resources, PDF_NAME(Font));
@@ -134,7 +133,7 @@ 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 WMode Style (Ordering=CNS1|GB1|Japan1|Korea1, WMode=H|V, Style=serif|sans)
+%%CJKFont Name Language WMode Style (Language=zh-Hant|zh-Hans|ja|ko, WMode=H|V, Style=serif|sans)
 %%Image Name Filename
 */
 static void create_page(char *input)
@@ -183,12 +182,12 @@ static void create_page(char *input)
 			else if (!strcmp(s, "%%CJKFont"))
 			{
 				char *name = fz_strsep(&p, " ");
-				char *ordering = fz_strsep(&p, " ");
+				char *lang = fz_strsep(&p, " ");
 				char *wmode = fz_strsep(&p, " ");
 				char *style = fz_strsep(&p, " ");
-				if (!name || !ordering)
+				if (!name || !lang)
 					fz_throw(ctx, FZ_ERROR_GENERIC, "CJKFont directive missing arguments");
-				add_cjkfont_res(resources, name, ordering, wmode, style);
+				add_cjkfont_res(resources, name, lang, wmode, style);
 			}
 			else if (!strcmp(s, "%%Image"))
 			{