Use real WinAnsi encoding when writing appearance stream text.

3 files changed, 53 insertions, 6 deletions

diff --git a/source/pdf/pdf-appearance.c b/source/pdf/pdf-appearance.c
index b231b6b5..faea5b33 100644
--- a/source/pdf/pdf-appearance.c
+++ b/source/pdf/pdf-appearance.c
@@ -674,8 +674,8 @@ measure_simple_string(fz_context *ctx, fz_font *font, const char *text)
 	{
 		int c, g;
 		text += fz_chartorune(&c, text);
-		/* WinAnsi is close enough to Latin-1 to not matter. Use middle dot for unencodable characters. */
-		if (c >= 256) c = REPLACEMENT;
+		c = pdf_winansi_from_unicode(c);
+		if (c < 0) c = REPLACEMENT;
 		g = fz_encode_character(ctx, font, c);
 		w += fz_advance_glyph(ctx, font, g, 0);
 	}
@@ -690,8 +690,8 @@ write_simple_string(fz_context *ctx, fz_buffer *buf, const char *a, const char *
 	{
 		int c;
 		a += fz_chartorune(&c, a);
-		/* WinAnsi is close enough to Latin-1 to not matter. Use middle dot for unencodable characters. */
-		if (c >= 256) c = REPLACEMENT;
+		c = pdf_winansi_from_unicode(c);
+		if (c < 0) c = REPLACEMENT;
 		if (c == '(' || c == ')' || c == '\\')
 			fz_append_byte(ctx, buf, '\\');
 		fz_append_byte(ctx, buf, c);
@@ -872,8 +872,8 @@ write_comb_string(fz_context *ctx, fz_buffer *buf, const char *a, const char *b,
 		int c, g;
 
 		a += fz_chartorune(&c, a);
-		/* WinAnsi is close enough to Latin-1 to not matter. Use middle dot for unencodable characters. */
-		if (c >= 256) c = REPLACEMENT;
+		c = pdf_winansi_from_unicode(c);
+		if (c < 0) c = REPLACEMENT;
 
 		g = fz_encode_character(ctx, font, c);
 		gw = fz_advance_glyph(ctx, font, g, 0) * 1000;
diff --git a/source/pdf/pdf-encoding.c b/source/pdf/pdf-encoding.c
index 04874457..f4fe584c 100644
--- a/source/pdf/pdf-encoding.c
+++ b/source/pdf/pdf-encoding.c
@@ -123,3 +123,22 @@ int pdf_greek_from_unicode(int u)
 	}
 	return -1;
 }
+
+int pdf_winansi_from_unicode(int u)
+{
+	int l = 0;
+	int r = nelem(winansi_from_unicode) - 1;
+	if (u < 128)
+		return u;
+	while (l <= r)
+	{
+		int m = (l + r) >> 1;
+		if (u < winansi_from_unicode[m].u)
+			r = m - 1;
+		else if (u > winansi_from_unicode[m].u)
+			l = m + 1;
+		else
+			return winansi_from_unicode[m].c;
+	}
+	return -1;
+}
diff --git a/source/pdf/pdf-encodings.h b/source/pdf/pdf-encodings.h
index 3c5bedfd..f9e84c49 100644
--- a/source/pdf/pdf-encodings.h
+++ b/source/pdf/pdf-encodings.h
@@ -341,3 +341,31 @@ static const struct { unsigned short u, c; } iso8859_7_from_unicode[] = {
 	{0x03cc,252}, {0x03cd,253}, {0x03ce,254}, {0x2015,175}, {0x2018,161},
 	{0x2019,162}, {0x20ac,164},
 };
+
+static const struct { unsigned short u, c; } winansi_from_unicode[] = {
+	{0x00a0,160}, {0x00a1,161}, {0x00a2,162}, {0x00a3,163}, {0x00a4,164},
+	{0x00a5,165}, {0x00a6,166}, {0x00a7,167}, {0x00a8,168}, {0x00a9,169},
+	{0x00aa,170}, {0x00ab,171}, {0x00ac,172}, {0x00ad,173}, {0x00ae,174},
+	{0x00af,175}, {0x00b0,176}, {0x00b1,177}, {0x00b2,178}, {0x00b3,179},
+	{0x00b4,180}, {0x00b5,181}, {0x00b6,182}, {0x00b7,183}, {0x00b8,184},
+	{0x00b9,185}, {0x00ba,186}, {0x00bb,187}, {0x00bc,188}, {0x00bd,189},
+	{0x00be,190}, {0x00bf,191}, {0x00c0,192}, {0x00c1,193}, {0x00c2,194},
+	{0x00c3,195}, {0x00c4,196}, {0x00c5,197}, {0x00c6,198}, {0x00c7,199},
+	{0x00c8,200}, {0x00c9,201}, {0x00ca,202}, {0x00cb,203}, {0x00cc,204},
+	{0x00cd,205}, {0x00ce,206}, {0x00cf,207}, {0x00d0,208}, {0x00d1,209},
+	{0x00d2,210}, {0x00d3,211}, {0x00d4,212}, {0x00d5,213}, {0x00d6,214},
+	{0x00d7,215}, {0x00d8,216}, {0x00d9,217}, {0x00da,218}, {0x00db,219},
+	{0x00dc,220}, {0x00dd,221}, {0x00de,222}, {0x00df,223}, {0x00e0,224},
+	{0x00e1,225}, {0x00e2,226}, {0x00e3,227}, {0x00e4,228}, {0x00e5,229},
+	{0x00e6,230}, {0x00e7,231}, {0x00e8,232}, {0x00e9,233}, {0x00ea,234},
+	{0x00eb,235}, {0x00ec,236}, {0x00ed,237}, {0x00ee,238}, {0x00ef,239},
+	{0x00f0,240}, {0x00f1,241}, {0x00f2,242}, {0x00f3,243}, {0x00f4,244},
+	{0x00f5,245}, {0x00f6,246}, {0x00f7,247}, {0x00f8,248}, {0x00f9,249},
+	{0x00fa,250}, {0x00fb,251}, {0x00fc,252}, {0x00fd,253}, {0x00fe,254},
+	{0x00ff,255}, {0x0152,140}, {0x0153,156}, {0x0160,138}, {0x0161,154},
+	{0x0178,159}, {0x017d,142}, {0x017e,158}, {0x0192,131}, {0x02c6,136},
+	{0x02dc,152}, {0x2013,150}, {0x2014,151}, {0x2018,145}, {0x2019,146},
+	{0x201a,130}, {0x201c,147}, {0x201d,148}, {0x201e,132}, {0x2020,134},
+	{0x2021,135}, {0x2022,149}, {0x2026,133}, {0x2030,137}, {0x2039,139},
+	{0x203a,155}, {0x20ac,128}, {0x2122,153},
+};