4 files changed, 42 insertions, 0 deletions
diff --git a/include/mupdf/page.h b/include/mupdf/page.h
index bf14dca2..4304fa22 100644
--- a/include/mupdf/page.h
+++ b/include/mupdf/page.h
@@ -32,6 +32,7 @@ struct pdf_textchar_s
 
 struct pdf_textline_s
 {
+	fz_point height;
 	int len, cap;
 	pdf_textchar *text;
 	pdf_textline *next;
diff --git a/include/mupdf/syntax.h b/include/mupdf/syntax.h
index 54eafa9e..87fc946a 100644
--- a/include/mupdf/syntax.h
+++ b/include/mupdf/syntax.h
@@ -28,6 +28,7 @@ fz_error *pdf_parseindobj(fz_obj **op, fz_file *f, char *buf, int cap, int *oid,
 fz_rect pdf_torect(fz_obj *array);
 fz_matrix pdf_tomatrix(fz_obj *array);
 fz_error *pdf_toutf8(char **dstp, fz_obj *src);
+fz_error *pdf_toucs2(unsigned short **dstp, fz_obj *src);
 
 /*
  * Encryption
diff --git a/mupdf/pdf_parse.c b/mupdf/pdf_parse.c
index 1bdd057e..e30bb3a2 100644
--- a/mupdf/pdf_parse.c
+++ b/mupdf/pdf_parse.c
@@ -77,6 +77,36 @@ pdf_toutf8(char **dstp, fz_obj *src)
 }
 
 fz_error *
+pdf_toucs2(unsigned short **dstp, fz_obj *src)
+{
+	unsigned char *srcptr = fz_tostrbuf(src);
+	unsigned short *dstptr;
+	int srclen = fz_tostrlen(src);
+	int i;
+
+	if (srclen > 2 && srcptr[0] == 254 && srcptr[1] == 255)
+	{
+		dstptr = *dstp = fz_malloc(((srclen - 2) / 2 + 1) * sizeof(short));
+		if (!dstptr)
+			return fz_outofmem;
+		for (i = 2; i < srclen; i += 2)
+			*dstptr++ = (srcptr[i] << 8) | srcptr[i+1];
+	}
+
+	else
+	{
+		dstptr = *dstp = fz_malloc((srclen + 1) * sizeof(short));
+		if (!dstptr)
+			return fz_outofmem;
+		for (i = 0; i < srclen; i++)
+			*dstptr++ = pdf_docencoding[srcptr[i]];
+	}
+
+	*dstptr = '\0';
+	return nil;
+}
+
+fz_error *
 pdf_parsearray(fz_obj **op, fz_file *file, char *buf, int cap)
 {
 	fz_error *error = nil;
diff --git a/mupdf/pdf_unicode.c b/mupdf/pdf_unicode.c
index d9103cd2..be57f7a2 100644
--- a/mupdf/pdf_unicode.c
+++ b/mupdf/pdf_unicode.c
@@ -99,6 +99,8 @@ pdf_newtextline(pdf_textline **linep)
 	line = *linep = fz_malloc(sizeof(pdf_textline));
 	if (!line)
 		return fz_outofmem;
+	line->height.x = 0;	/* bad default value... */
+	line->height.y = 10;
 	line->len = 0;
 	line->cap = 0;
 	line->text = nil;
@@ -161,6 +163,14 @@ extracttext(pdf_textline **line, fz_node *node, fz_matrix ctm)
 		int i, g, x, y;
 		int c;
 
+		/* get line height */
+		trm = fz_concat(tm, ctm);
+		trm.e = 0;
+		trm.f = 0;
+		p.x = 0;
+		p.y = 1;
+		(*line)->height = fz_transformpoint(trm, p);
+
 		for (i = 0; i < text->len; i++)
 		{
 			g = text->els[i].cid;