4 files changed, 33 insertions, 0 deletions
diff --git a/include/fitz/base.h b/include/fitz/base.h
index da0132d4..562bb1df 100644
--- a/include/fitz/base.h
+++ b/include/fitz/base.h
@@ -21,6 +21,12 @@
 
 #define STRIDE(n, bcp) (((bpc) * (n) + 7) / 8)
 
+int chartorune(int *rune, char *str);
+int runetochar(char *str, int *rune);
+int runelen(long c);
+int runenlen(int *r, int nrune);
+int fullrune(char *str, int n);
+
 typedef struct fz_error_s fz_error;
 
 struct fz_error_s
diff --git a/include/mupdf/content.h b/include/mupdf/content.h
index 6dae61da..f891c3c7 100644
--- a/include/mupdf/content.h
+++ b/include/mupdf/content.h
@@ -66,15 +66,18 @@ struct pdf_csi_s
 	fz_obj *stack[32];
 	int top;
 	int xbalance;
+	fz_obj *array;
 
 	/* path object state */
 	fz_pathnode *path;
 	fz_pathnode *clip;
 
 	/* text object state */
+	fz_node *textclip;
 	fz_textnode *text;
 	fz_matrix tlm;
 	fz_matrix tm;
+	int textmode;
 
 	fz_tree *tree;
 };
diff --git a/include/mupdf/page.h b/include/mupdf/page.h
index 0dd1001b..3bb27db6 100644
--- a/include/mupdf/page.h
+++ b/include/mupdf/page.h
@@ -7,6 +7,8 @@ typedef struct pdf_outline_s pdf_outline;
 typedef struct pdf_nametree_s pdf_nametree;
 typedef struct pdf_pagetree_s pdf_pagetree;
 typedef struct pdf_page_s pdf_page;
+typedef struct pdf_textline_s pdf_textline;
+typedef struct pdf_textchar_s pdf_textchar;
 
 struct pdf_outlinetree_s
 {
@@ -48,6 +50,19 @@ struct pdf_page_s
 	int rotate;
 	fz_obj *resources;
 	fz_tree *tree;
+	pdf_textline *text;
+};
+
+struct pdf_textchar_s
+{
+	int x, y, c;
+};
+
+struct pdf_textline_s
+{
+	int len, cap;
+	pdf_textchar *text;
+	pdf_textline *next;
 };
 
 /* outline.c */
@@ -73,3 +88,9 @@ void pdf_droppagetree(pdf_pagetree *pages);
 fz_error *pdf_loadpage(pdf_page **pagep, pdf_xref *xref, fz_obj *ref);
 void pdf_droppage(pdf_page *page);
 
+/* unicode.c */
+fz_error *pdf_loadtextfromtree(pdf_textline **linep, fz_tree *tree);
+void pdf_debugtextline(pdf_textline *line);
+fz_error *pdf_newtextline(pdf_textline **linep);
+void pdf_droptextline(pdf_textline *line);
+
diff --git a/include/mupdf/rsrc.h b/include/mupdf/rsrc.h
index bb29dc8a..eaf20fb2 100644
--- a/include/mupdf/rsrc.h
+++ b/include/mupdf/rsrc.h
@@ -161,6 +161,9 @@ fz_error *pdf_loadembeddedcmap(fz_cmap **cmapp, pdf_xref *xref, fz_obj *stmref);
 fz_error *pdf_loadsystemcmap(fz_cmap **cmapp, char *name);
 fz_error *pdf_makeidentitycmap(fz_cmap **cmapp, int wmode, int bytes);
 
+/* unicode.c */
+fz_error *pdf_loadtounicode(pdf_font *font, pdf_xref *xref, char **strings, char *collection, fz_obj *cmapstm);
+
 /* fontfile.c */
 fz_error *pdf_loadbuiltinfont(pdf_font *font, char *basefont);
 fz_error *pdf_loadembeddedfont(pdf_font *font, pdf_xref *xref, fz_obj *stmref);