Generalise pdf_links to be fz_links.

Move to a non-pdf specific type for links. PDF specific parsing is
done in pdf_annots.c as before, but the essential type (and handling
functions for that type) are in a new file fitz/base_link.c.

The new type is more expressive than before; specifically all the
possible PDF modes are expressable in it. Hopefully this should
allow XPS links to be represented too.

10 files changed, 317 insertions, 99 deletions

diff --git a/android/jni/Core.mk b/android/jni/Core.mk
index 259f0398..0ed43751 100644
--- a/android/jni/Core.mk
+++ b/android/jni/Core.mk
@@ -25,6 +25,7 @@ LOCAL_SRC_FILES := \
 	$(MY_ROOT)/fitz/base_geometry.c \
 	$(MY_ROOT)/fitz/base_getopt.c \
 	$(MY_ROOT)/fitz/base_hash.c \
+	$(MY_ROOT)/fitz/base_link.c \
 	$(MY_ROOT)/fitz/base_memory.c \
 	$(MY_ROOT)/fitz/base_object.c \
 	$(MY_ROOT)/fitz/base_string.c \
diff --git a/apps/pdfapp.c b/apps/pdfapp.c
index 849b0d18..cf136261 100644
--- a/apps/pdfapp.c
+++ b/apps/pdfapp.c
@@ -238,7 +238,7 @@ void pdfapp_close(pdfapp_t *app)
 	app->page_text = NULL;
 
 	if (app->page_links)
-		pdf_free_link(app->ctx, app->page_links);
+		fz_free_link(app->ctx, app->page_links);
 	app->page_links = NULL;
 
 	if (app->doctitle)
@@ -394,7 +394,7 @@ static void pdfapp_showpage(pdfapp_t *app, int loadpage, int drawpage, int repai
 		if (app->page_text)
 			fz_free_text_span(app->ctx, app->page_text);
 		if (app->page_links)
-			pdf_free_link(app->ctx, app->page_links);
+			fz_free_link(app->ctx, app->page_links);
 
 		if (app->xref)
 			pdfapp_loadpage_pdf(app);
@@ -467,25 +467,13 @@ static void pdfapp_showpage(pdfapp_t *app, int loadpage, int drawpage, int repai
 	fz_flush_warnings(app->ctx);
 }
 
-static void pdfapp_gotouri(pdfapp_t *app, fz_obj *uri)
+static void pdfapp_gotouri(pdfapp_t *app, char *uri)
 {
-	char *buf;
-	int n = fz_to_str_len(uri);
-	buf = fz_malloc(app->ctx, n + 1);
-	memcpy(buf, fz_to_str_buf(uri), n);
-	buf[n] = 0;
-	winopenuri(app, buf);
-	fz_free(app->ctx, buf);
+	winopenuri(app, uri);
 }
 
-static void pdfapp_gotopage(pdfapp_t *app, fz_obj *obj)
+static void pdfapp_gotopage(pdfapp_t *app, int number)
 {
-	int number;
-
-	number = pdf_find_page_number(app->xref, obj);
-	if (number < 0)
-		return;
-
 	if (app->histlen + 1 == 256)
 	{
 		memmove(app->hist, app->hist + 1, sizeof(int) * 255);
@@ -1025,7 +1013,7 @@ void pdfapp_onkey(pdfapp_t *app, int c)
 
 void pdfapp_onmouse(pdfapp_t *app, int x, int y, int btn, int modifiers, int state)
 {
-	pdf_link *link;
+	fz_link *link;
 	fz_matrix ctm;
 	fz_point p;
 
@@ -1049,10 +1037,10 @@ void pdfapp_onmouse(pdfapp_t *app, int x, int y, int btn, int modifiers, int sta
 		wincursor(app, HAND);
 		if (btn == 1 && state == 1)
 		{
-			if (link->kind == PDF_LINK_URI)
-				pdfapp_gotouri(app, link->dest);
-			else if (link->kind == PDF_LINK_GOTO)
-				pdfapp_gotopage(app, fz_array_get(link->dest, 0)); /* [ pageobj ... ] */
+			if (link->kind == FZ_LINK_URI)
+				pdfapp_gotouri(app, link->dest.uri.uri);
+			else if (link->kind == FZ_LINK_GOTO)
+				pdfapp_gotopage(app, link->dest.gotor.page);
 			return;
 		}
 	}
diff --git a/apps/pdfapp.h b/apps/pdfapp.h
index c9fe1d4f..22b6ad0d 100644
--- a/apps/pdfapp.h
+++ b/apps/pdfapp.h
@@ -49,7 +49,7 @@ struct pdfapp_s
 	float page_rotate;
 	fz_display_list *page_list;
 	fz_text_span *page_text;
-	pdf_link *page_links;
+	fz_link *page_links;
 
 	/* snapback history */
 	int hist[256];
diff --git a/fitz/base_link.c b/fitz/base_link.c
new file mode 100644
index 00000000..d8d7e3ab
--- /dev/null
+++ b/fitz/base_link.c
@@ -0,0 +1,57 @@
+#include "fitz.h"
+
+static void
+free_link_dest(fz_context *ctx, fz_link_kind kind, fz_link_dest *dest)
+{
+	switch(kind)
+	{
+	case FZ_LINK_GOTO:
+		break;
+	case FZ_LINK_URI:
+		fz_free(ctx, dest->uri.uri);
+		break;
+	case FZ_LINK_LAUNCH:
+		fz_free(ctx, dest->launch.file_spec);
+		break;
+	case FZ_LINK_NAMED:
+		fz_free(ctx, dest->named.named);
+		break;
+	case FZ_LINK_GOTOR:
+		fz_free(ctx, dest->gotor.file_spec);
+		break;
+	}
+}
+
+fz_link *
+fz_new_link(fz_context *ctx, fz_link_kind kind, fz_rect bbox, fz_link_dest dest)
+{
+	fz_link *link;
+
+	fz_try(ctx)
+	{
+		link = fz_malloc_struct(ctx, fz_link);
+	}
+	fz_catch(ctx)
+	{
+		free_link_dest(ctx, kind, &dest);
+		fz_rethrow(ctx);
+	}
+	link->kind = kind;
+	link->rect = bbox;
+	link->next = NULL;
+	return link;
+}
+
+void
+fz_free_link(fz_context *ctx, fz_link *link)
+{
+	fz_link *next;
+
+	while (link)
+	{
+		next = link->next;
+		free_link_dest(ctx, link->kind, &link->dest);
+		fz_free(ctx, link);
+		link = next;
+	}
+}
diff --git a/fitz/fitz.h b/fitz/fitz.h
index 589f22ee..9ac43d48 100644
--- a/fitz/fitz.h
+++ b/fitz/fitz.h
@@ -1233,4 +1233,64 @@ enum
 	FZ_BLEND_KNOCKOUT = 32
 };
 
+/* Links */
+
+typedef struct fz_link_s fz_link;
+
+typedef union fz_link_dest_s fz_link_dest;
+
+typedef enum fz_link_kind_e
+{
+	FZ_LINK_GOTO = 0,
+	FZ_LINK_URI,
+	FZ_LINK_LAUNCH,
+	FZ_LINK_NAMED,
+	FZ_LINK_GOTOR
+} fz_link_kind;
+
+enum {
+	fz_link_flag_l_valid   = 1,  /* lt.x is valid */
+	fz_link_flag_t_valid   = 2,  /* lt.y is valid */
+	fz_link_flag_r_valid   = 4,  /* rb.x is valid */
+	fz_link_flag_b_valid   = 8,  /* rb.y is valid */
+	fz_link_flag_fit_h     = 16, /* Fit horizontally */
+	fz_link_flag_fit_v     = 32, /* Fit vertically */
+	fz_link_flag_r_is_zoom = 64  /* rb.x is actually a zoom figure */
+};
+
+union fz_link_dest_s
+{
+	struct {
+		int page;
+		int flags;
+		fz_point lt;
+		fz_point rb;
+		char *file_spec;
+		int new_window;
+	} gotor;
+	struct {
+		char *uri;
+		int is_map;
+	} uri;
+	struct {
+		char *file_spec;
+		int new_window;
+	} launch;
+	struct {
+		char *named;
+	} named;
+};
+
+
+struct fz_link_s
+{
+	fz_link_kind kind;
+	fz_rect rect;
+	fz_link_dest dest;
+	fz_link *next;
+};
+
+fz_link *fz_new_link(fz_context *ctx, fz_link_kind, fz_rect bbox, fz_link_dest dest);
+void fz_free_link(fz_context *ctx, fz_link *);
+
 #endif
diff --git a/pdf/mupdf.h b/pdf/mupdf.h
index 7db1ca36..ca173b12 100644
--- a/pdf/mupdf.h
+++ b/pdf/mupdf.h
@@ -409,26 +409,8 @@ void pdf_debug_font(pdf_font_desc *fontdesc);
  * Interactive features
  */
 
-typedef struct pdf_link_s pdf_link;
 typedef struct pdf_annot_s pdf_annot;
 
-typedef enum pdf_link_kind_e
-{
-	PDF_LINK_GOTO = 0,
-	PDF_LINK_URI,
-	PDF_LINK_LAUNCH,
-	PDF_LINK_NAMED,
-	PDF_LINK_ACTION,
-} pdf_link_kind;
-
-struct pdf_link_s
-{
-	pdf_link_kind kind;
-	fz_rect rect;
-	fz_obj *dest;
-	pdf_link *next;
-};
-
 struct pdf_annot_s
 {
 	fz_obj *obj;
@@ -438,15 +420,16 @@ struct pdf_annot_s
 	pdf_annot *next;
 };
 
+fz_link_dest pdf_parse_link_dest(pdf_xref *xref, fz_obj *dest);
+
 fz_obj *pdf_lookup_dest(pdf_xref *xref, fz_obj *needle);
 fz_obj *pdf_lookup_name(pdf_xref *xref, char *which, fz_obj *needle);
 fz_obj *pdf_load_name_tree(pdf_xref *xref, char *which);
 
 fz_outline *pdf_load_outline(pdf_xref *xref);
 
-pdf_link *pdf_load_link(pdf_xref *xref, fz_obj *dict);
-void pdf_load_links(pdf_link **, pdf_xref *, fz_obj *annots);
-void pdf_free_link(fz_context *ctx, pdf_link *link);
+fz_link *pdf_load_link(pdf_xref *xref, fz_obj *dict);
+void pdf_load_links(fz_link **, pdf_xref *, fz_obj *annots);
 
 void pdf_load_annots(pdf_annot **, pdf_xref *, fz_obj *annots);
 void pdf_free_annot(fz_context *ctx, pdf_annot *link);
@@ -464,7 +447,7 @@ struct pdf_page_s
 	int transparency;
 	fz_obj *resources;
 	fz_buffer *contents;
-	pdf_link *links;
+	fz_link *links;
 	pdf_annot *annots;
 };
 
diff --git a/pdf/pdf_annot.c b/pdf/pdf_annot.c
index 79bdb4ff..ba4d509d 100644
--- a/pdf/pdf_annot.c
+++ b/pdf/pdf_annot.c
@@ -1,21 +1,6 @@
 #include "fitz.h"
 #include "mupdf.h"
 
-void
-pdf_free_link(fz_context *ctx, pdf_link *link)
-{
-	pdf_link *next;
-
-	while (link)
-	{
-		next = link->next;
-		if (link->dest)
-			fz_drop_obj(link->dest);
-		fz_free(ctx, link);
-		link = next;
-	}
-}
-
 static fz_obj *
 resolve_dest(pdf_xref *xref, fz_obj *dest)
 {
@@ -42,14 +27,155 @@ resolve_dest(pdf_xref *xref, fz_obj *dest)
 	return NULL;
 }
 
-pdf_link *
+fz_link_dest
+pdf_parse_link_dest(pdf_xref *xref, fz_obj *dest)
+{
+	fz_link_dest ld;
+	fz_obj *obj;
+	int read = 0;
+
+	obj = fz_array_get(dest, 0);
+	if (fz_is_int(obj))
+		ld.gotor.page = fz_to_int(obj)-1;
+	else
+		ld.gotor.page = pdf_find_page_number(xref, obj);
+
+	ld.gotor.flags = 0;
+	ld.gotor.lt.x = 0;
+	ld.gotor.lt.y = 0;
+	ld.gotor.rb.x = 0;
+	ld.gotor.rb.y = 0;
+	ld.gotor.file_spec = NULL;
+	ld.gotor.new_window = 0;
+
+	obj = fz_array_get(dest, 1);
+	if (!fz_is_name(obj))
+		return ld;
+
+	if (!strcmp("XYZ", fz_to_name(obj)))
+	{
+		read = 1+16;
+		ld.gotor.flags |= fz_link_flag_r_is_zoom;
+	}
+	else if ((!strcmp("Fit", fz_to_name(obj))) || (!strcmp("FitB", fz_to_name(obj))))
+	{
+		read = 0;
+		ld.gotor.flags |= fz_link_flag_fit_h;
+		ld.gotor.flags |= fz_link_flag_fit_v;
+	}
+	else if ((!strcmp("FitH", fz_to_name(obj))) || (!strcmp("FitBH", fz_to_name(obj))))
+	{
+		read = 32;
+		ld.gotor.flags |= fz_link_flag_fit_h;
+	}
+	else if ((!strcmp("FitV", fz_to_name(obj))) || (!strcmp("FitBV", fz_to_name(obj))))
+	{
+		read = 1;
+		ld.gotor.flags |= fz_link_flag_fit_v;
+	}
+	else if (!strcmp("FitR", fz_to_name(obj)))
+	{
+		read = 1+2+4+8;
+		ld.gotor.flags |= fz_link_flag_fit_h;
+		ld.gotor.flags |= fz_link_flag_fit_v;
+	}
+
+	if (read & 1)
+	{
+		obj = fz_array_get(dest, 2);
+		if (fz_is_int(obj))
+		{
+			ld.gotor.flags |= fz_link_flag_l_valid;
+			ld.gotor.lt.x = fz_to_int(obj);
+		}
+		else if (fz_is_real(obj))
+		{
+			ld.gotor.flags |= fz_link_flag_l_valid;
+			ld.gotor.lt.x = fz_to_real(obj);
+		}
+	}
+	if (read & 2)
+	{
+		obj = fz_array_get(dest, 3);
+		if (fz_is_int(obj))
+		{
+			ld.gotor.flags |= fz_link_flag_b_valid;
+			ld.gotor.rb.y = fz_to_int(obj);
+		}
+		else if (fz_is_real(obj))
+		{
+			ld.gotor.flags |= fz_link_flag_b_valid;
+			ld.gotor.rb.y = fz_to_real(obj);
+		}
+	}
+	if (read & 4)
+	{
+		obj = fz_array_get(dest, 4);
+		if (fz_is_int(obj))
+		{
+			ld.gotor.flags |= fz_link_flag_r_valid;
+			ld.gotor.rb.x = fz_to_int(obj);
+		}
+		else if (fz_is_real(obj))
+		{
+			ld.gotor.flags |= fz_link_flag_r_valid;
+			ld.gotor.rb.x = fz_to_real(obj);
+		}
+	}
+	if (read & (8+16+32))
+	{
+		if (read & 8)
+			obj = fz_array_get(dest, 5);
+		else if (read & 16)
+			obj = fz_array_get(dest, 3);
+		else
+			obj = fz_array_get(dest, 2);
+		if (fz_is_int(obj))
+		{
+			ld.gotor.flags |= fz_link_flag_t_valid;
+			ld.gotor.lt.x = fz_to_int(obj);
+		}
+		else if (fz_is_real(obj))
+		{
+			ld.gotor.flags |= fz_link_flag_t_valid;
+			ld.gotor.lt.x = fz_to_real(obj);
+		}
+	}
+	if (read & 16)
+	{
+		obj = fz_array_get(dest, 4);
+		if (fz_is_int(obj))
+		{
+			ld.gotor.flags |= fz_link_flag_r_valid;
+			ld.gotor.rb.x = fz_to_int(obj);
+		}
+		else if (fz_is_real(obj))
+		{
+			ld.gotor.flags |= fz_link_flag_r_valid;
+			ld.gotor.rb.x = fz_to_real(obj);
+		}
+	}
+
+	/* Duplicate the values out for the sake of stupid clients */
+	if ((ld.gotor.flags & (fz_link_flag_l_valid | fz_link_flag_r_valid)) == fz_link_flag_l_valid)
+		ld.gotor.rb.x = ld.gotor.lt.x;
+	if ((ld.gotor.flags & (fz_link_flag_l_valid | fz_link_flag_r_valid | fz_link_flag_r_is_zoom)) == fz_link_flag_r_valid)
+		ld.gotor.lt.x = ld.gotor.rb.x;
+	if ((ld.gotor.flags & (fz_link_flag_t_valid | fz_link_flag_b_valid)) == fz_link_flag_t_valid)
+		ld.gotor.rb.y = ld.gotor.lt.y;
+	if ((ld.gotor.flags & (fz_link_flag_t_valid | fz_link_flag_b_valid)) == fz_link_flag_b_valid)
+		ld.gotor.lt.y = ld.gotor.rb.y;
+
+	return ld;
+}
+
+fz_link *
 pdf_load_link(pdf_xref *xref, fz_obj *dict)
 {
-	fz_obj *dest;
+	fz_obj *dest = NULL;
 	fz_obj *action;
 	fz_obj *obj;
 	fz_rect bbox;
-	pdf_link_kind kind;
 	fz_context *ctx = xref->ctx;
 
 	dest = NULL;
@@ -63,8 +189,10 @@ pdf_load_link(pdf_xref *xref, fz_obj *dict)
 	obj = fz_dict_gets(dict, "Dest");
 	if (obj)
 	{
-		kind = PDF_LINK_GOTO;
+		fz_link_dest ld;
 		dest = resolve_dest(xref, obj);
+		ld = pdf_parse_link_dest(xref, dest);
+		return fz_new_link(ctx, FZ_LINK_GOTO, bbox, ld);
 	}
 
 	action = fz_dict_gets(dict, "A");
@@ -76,54 +204,54 @@ pdf_load_link(pdf_xref *xref, fz_obj *dict)
 	if (action)
 	{
 		obj = fz_dict_gets(action, "S");
-		if (fz_is_name(obj) && !strcmp(fz_to_name(obj), "GoTo"))
+		if (!fz_is_name(obj))
 		{
-			kind = PDF_LINK_GOTO;
-			dest = resolve_dest(xref, fz_dict_gets(action, "D"));
 		}
-		else if (fz_is_name(obj) && !strcmp(fz_to_name(obj), "URI"))
+		else if (!strcmp(fz_to_name(obj), "GoTo"))
 		{
-			kind = PDF_LINK_URI;
-			dest = fz_dict_gets(action, "URI");
+			fz_link_dest ld;
+			dest = resolve_dest(xref, fz_dict_gets(action, "D"));
+			ld = pdf_parse_link_dest(xref, dest);
+			return fz_new_link(ctx, FZ_LINK_GOTO, bbox, ld);
 		}
-		else if (fz_is_name(obj) && !strcmp(fz_to_name(obj), "Launch"))
+		else if (!strcmp(fz_to_name(obj), "URI"))
 		{
-			kind = PDF_LINK_LAUNCH;
-			dest = fz_dict_gets(action, "F");
+			fz_link_dest ld;
+			ld.uri.is_map = fz_to_int(fz_dict_gets(action, "IsMap"));
+			ld.uri.uri = pdf_to_utf8(ctx, fz_dict_gets(action, "URI"));
+			return fz_new_link(ctx, FZ_LINK_URI, bbox, ld);
 		}
-		else if (fz_is_name(obj) && !strcmp(fz_to_name(obj), "Named"))
+		else if (!strcmp(fz_to_name(obj), "Launch"))
 		{
-			kind = PDF_LINK_NAMED;
-			dest = fz_dict_gets(action, "N");
+			fz_link_dest ld;
+			ld.launch.file_spec = pdf_to_utf8(ctx, fz_dict_gets(action, "F"));
+			ld.launch.new_window = fz_to_int(fz_dict_gets(action, "NewWindow"));
+			return fz_new_link(ctx, FZ_LINK_LAUNCH, bbox, ld);
 		}
-		else if (fz_is_name(obj) && (!strcmp(fz_to_name(obj), "GoToR")))
+		else if (!strcmp(fz_to_name(obj), "Named"))
 		{
-			kind = PDF_LINK_ACTION;
-			dest = action;
+			fz_link_dest ld;
+			ld.named.named = pdf_to_utf8(ctx, fz_dict_gets(action, "N"));
+			return fz_new_link(ctx, FZ_LINK_NAMED, bbox, ld);
 		}
-		else
+		else if (!strcmp(fz_to_name(obj), "GoToR"))
 		{
-			dest = NULL;
+			fz_link_dest ld;
+			dest = resolve_dest(xref, fz_dict_gets(action, "D"));
+			ld = pdf_parse_link_dest(xref, dest);
+			ld.gotor.file_spec = pdf_to_utf8(ctx, fz_dict_gets(action, "F"));
+			ld.gotor.new_window = fz_to_int(fz_dict_gets(action, "NewWindow"));
+			return fz_new_link(ctx, FZ_LINK_GOTOR, bbox, ld);
 		}
 	}
 
-	if (dest)
-	{
-		pdf_link *link = fz_malloc_struct(ctx, pdf_link);
-		link->kind = kind;
-		link->rect = bbox;
-		link->dest = fz_keep_obj(dest);
-		link->next = NULL;
-		return link;
-	}
-
 	return NULL;
 }
 
 void
-pdf_load_links(pdf_link **linkp, pdf_xref *xref, fz_obj *annots)
+pdf_load_links(fz_link **linkp, pdf_xref *xref, fz_obj *annots)
 {
-	pdf_link *link, *head, *tail;
+	fz_link *link, *head, *tail;
 	fz_obj *obj;
 	int i, n;
 
diff --git a/pdf/pdf_outline.c b/pdf/pdf_outline.c
index 8a33d8e2..3f6b745b 100644
--- a/pdf/pdf_outline.c
+++ b/pdf/pdf_outline.c
@@ -5,7 +5,6 @@ static fz_outline *
 pdf_load_outline_imp(pdf_xref *xref, fz_obj *dict)
 {
 	fz_context *ctx = xref->ctx;
-	pdf_link *link;
 	fz_outline *node;
 	fz_obj *obj;
 
@@ -25,10 +24,8 @@ pdf_load_outline_imp(pdf_xref *xref, fz_obj *dict)
 
 	if (fz_dict_gets(dict, "Dest") || fz_dict_gets(dict, "A"))
 	{
-		link = pdf_load_link(xref, dict);
-		if (link && link->kind == PDF_LINK_GOTO)
-			node->page = pdf_find_page_number(xref, fz_array_get(link->dest, 0));
-		pdf_free_link(xref->ctx, link);
+		fz_link_dest ld = pdf_parse_link_dest(xref, dict);
+		node->page = ld.gotor.page;
 	}
 
 	obj = fz_dict_gets(dict, "First");
diff --git a/pdf/pdf_page.c b/pdf/pdf_page.c
index a9e99e52..05af3b60 100644
--- a/pdf/pdf_page.c
+++ b/pdf/pdf_page.c
@@ -363,7 +363,7 @@ pdf_free_page(fz_context *ctx, pdf_page *page)
 	if (page->contents)
 		fz_drop_buffer(ctx, page->contents);
 	if (page->links)
-		pdf_free_link(ctx, page->links);
+		fz_free_link(ctx, page->links);
 	if (page->annots)
 		pdf_free_annot(ctx, page->annots);
 	fz_free(ctx, page);
diff --git a/win32/libmupdf.vcproj b/win32/libmupdf.vcproj
index 663f0268..bd24cb54 100644
--- a/win32/libmupdf.vcproj
+++ b/win32/libmupdf.vcproj
@@ -355,6 +355,10 @@
 				>
 			</File>
 			<File
+				RelativePath="..\fitz\base_link.c"
+				>
+			</File>
+			<File
 				RelativePath="..\fitz\base_memory.c"
 				>
 			</File>