epub: Add hyperlinks.

3 files changed, 188 insertions, 35 deletions

diff --git a/include/mupdf/html.h b/include/mupdf/html.h
index f9bd4d98..259333a4 100644
--- a/include/mupdf/html.h
+++ b/include/mupdf/html.h
@@ -207,6 +207,7 @@ struct fz_html_box_s
 	float em;
 	fz_html_box *up, *down, *last, *next;
 	fz_html_flow *flow_head, **flow_tail;
+	char *a_href;
 	fz_css_style style;
 };
 
@@ -276,6 +277,7 @@ void fz_add_css_font_faces(fz_context *ctx, fz_html_font_set *set, fz_archive *z
 fz_html *fz_parse_html(fz_context *ctx, fz_html_font_set *htx, fz_archive *zip, const char *base_uri, fz_buffer *buf, const char *user_css);
 void fz_layout_html(fz_context *ctx, fz_html *html, float w, float h, float em);
 void fz_draw_html(fz_context *ctx, fz_device *dev, const fz_matrix *ctm, fz_html *html, float page_top, float page_bot);
+fz_link *fz_load_html_links(fz_context *ctx, fz_html *html, int page, int page_h, const char *base_uri);
 void fz_drop_html(fz_context *ctx, fz_html *html);
 
 #endif
diff --git a/source/html/epub-doc.c b/source/html/epub-doc.c
index 246e343a..efc4ac18 100644
--- a/source/html/epub-doc.c
+++ b/source/html/epub-doc.c
@@ -50,13 +50,13 @@ find_anchor_flow(fz_html_flow *flow, const char *anchor, float page_h, int *page
 }
 
 static int
-find_anchor(fz_html_box *box, const char *anchor, float page_h, int *page)
+find_anchor_box(fz_html_box *box, const char *anchor, float page_h, int *page)
 {
 	while (box)
 	{
 		if (box->flow_head && find_anchor_flow(box->flow_head, anchor, page_h, page))
 			return 1;
-		if (box->down && find_anchor(box->down, anchor, page_h, page))
+		if (box->down && find_anchor_box(box->down, anchor, page_h, page))
 			return 1;
 		box = box->next;
 	}
@@ -64,35 +64,41 @@ find_anchor(fz_html_box *box, const char *anchor, float page_h, int *page)
 }
 
 static void
-epub_update_link_dests(fz_context *ctx, epub_document *doc, fz_outline *node)
+resolve_link_dest(fz_context *ctx, epub_document *doc, fz_link_dest *ld)
 {
 	epub_chapter *ch;
 
-	while (node)
+	if (ld->kind == FZ_LINK_GOTO)
 	{
-		if (node->dest.kind == FZ_LINK_GOTO)
-		{
-			const char *dest = node->dest.ld.gotor.dest;
-			const char *s = strchr(dest, '#');
-			int n = s ? s - dest : strlen(dest);
-			if (s && s[1] == 0)
-				s = NULL;
+		const char *dest = ld->ld.gotor.dest;
+		const char *s = strchr(dest, '#');
+		int n = s ? s - dest : strlen(dest);
+		if (s && s[1] == 0)
+			s = NULL;
 
-			for (ch = doc->spine; ch; ch = ch->next)
+		for (ch = doc->spine; ch; ch = ch->next)
+		{
+			if (strncmp(ch->path, dest, n) || ch->path[n] != 0)
+				continue;
+			ld->ld.gotor.page = ch->start;
+			if (s)
 			{
-				if (strncmp(ch->path, dest, n) || ch->path[n] != 0)
+				/* Search for a matching fragment */
+				if (find_anchor_box(ch->html->root, s+1, ch->page_h, &ld->ld.gotor.page))
 					continue;
-				node->dest.ld.gotor.page = ch->start;
-				if (s)
-				{
-					/* Search for a matching fragment */
-					if (find_anchor(ch->html->root, s+1, ch->page_h, &node->dest.ld.gotor.page))
-						continue;
-				}
-				break;
 			}
+			break;
 		}
-		epub_update_link_dests(ctx, doc, node->down);
+	}
+}
+
+static void
+epub_update_outline(fz_context *ctx, epub_document *doc, fz_outline *node)
+{
+	while (node)
+	{
+		resolve_link_dest(ctx, doc, &node->dest);
+		epub_update_outline(ctx, doc, node->down);
 		node = node->next;
 	}
 }
@@ -118,7 +124,7 @@ epub_layout(fz_context *ctx, fz_document *doc_, float w, float h, float em)
 		count += ceilf(ch->html->root->h / ch->page_h);
 	}
 
-	epub_update_link_dests(ctx, doc, doc->outline);
+	epub_update_outline(ctx, doc, doc->outline);
 }
 
 static int
@@ -187,6 +193,41 @@ epub_run_page(fz_context *ctx, fz_page *page_, fz_device *dev, const fz_matrix *
 	}
 }
 
+static fz_link *
+epub_load_links(fz_context *ctx, fz_page *page_)
+{
+	epub_page *page = (epub_page*)page_;
+	epub_document *doc = page->doc;
+	epub_chapter *ch;
+	int n = page->number;
+	int count = 0;
+	fz_link *head, *link;
+
+	for (ch = doc->spine; ch; ch = ch->next)
+	{
+		int cn = ceilf(ch->html->root->h / ch->page_h);
+		if (n < count + cn)
+		{
+			head = fz_load_html_links(ctx, ch->html, n - count, ch->page_h, ch->path);
+			for (link = head; link; link = link->next)
+			{
+				/* Adjust for page margins */
+				link->rect.x0 += ch->page_margin[L];
+				link->rect.x1 += ch->page_margin[L];
+				link->rect.y0 += ch->page_margin[T];
+				link->rect.y1 += ch->page_margin[T];
+
+				/* Resolve local links */
+				resolve_link_dest(ctx, doc, &link->dest);
+			}
+			return head;
+		}
+		count += cn;
+	}
+
+	return NULL;
+}
+
 static fz_page *
 epub_load_page(fz_context *ctx, fz_document *doc_, int number)
 {
@@ -194,6 +235,7 @@ epub_load_page(fz_context *ctx, fz_document *doc_, int number)
 	epub_page *page = fz_new_page(ctx, sizeof *page);
 	page->super.bound_page = epub_bound_page;
 	page->super.run_page_contents = epub_run_page;
+	page->super.load_links = epub_load_links;
 	page->super.drop_page = epub_drop_page;
 	page->doc = doc;
 	page->number = number;
diff --git a/source/html/html-layout.c b/source/html/html-layout.c
index fb039061..323b0532 100644
--- a/source/html/html-layout.c
+++ b/source/html/html-layout.c
@@ -680,7 +680,7 @@ static void generate_boxes(fz_context *ctx, fz_xml *node, fz_html_box *top,
 
 			else if (display != DIS_NONE)
 			{
-				const char *dir, *lang, *id;
+				const char *dir, *lang, *id, *href;
 				int child_dir = markup_dir;
 				int child_lang = markup_lang;
 
@@ -716,6 +716,16 @@ static void generate_boxes(fz_context *ctx, fz_xml *node, fz_html_box *top,
 				else if (display == DIS_INLINE)
 				{
 					insert_inline_box(ctx, box, top, child_dir, g);
+					if (!strcmp(tag, "a"))
+					{
+						id = fz_xml_att(node, "id");
+						if (id)
+							generate_anchor(ctx, box, id, g);
+						href = fz_xml_att(node, "href");
+						if (href)
+							box->a_href = fz_pool_strdup(ctx, g->pool, href);
+					}
+
 				}
 				else
 				{
@@ -723,17 +733,6 @@ static void generate_boxes(fz_context *ctx, fz_xml *node, fz_html_box *top,
 					insert_box(ctx, box, BOX_BLOCK, top);
 				}
 
-				if (!strcmp(tag, "a"))
-				{
-					id = fz_xml_att(node, "id");
-					if (id)
-					{
-						/* We don't need to create a box here, because since <a> tags are inline
-						 * the DIS_INLINE case * above should already have done it for us. */
-						generate_anchor(ctx, box, id, g);
-					}
-				}
-
 				if (fz_xml_down(node))
 				{
 					int child_counter = list_counter;
@@ -1885,6 +1884,113 @@ fz_draw_html(fz_context *ctx, fz_device *dev, const fz_matrix *ctm, fz_html *htm
 	}
 }
 
+static int is_internal_uri(const char *uri)
+{
+	while (*uri >= 'a' && *uri <= 'z')
+		++uri;
+	if (uri[0] == ':' && uri[1] == '/' && uri[2] == '/')
+		return 0;
+	return 1;
+}
+
+static fz_link *load_link_flow(fz_context *ctx, fz_html_flow *flow, fz_link *head, int page, int page_h, const char *dir, const char *file)
+{
+	fz_link *link;
+	fz_html_flow *next;
+	char path[2048];
+	fz_rect bbox;
+	fz_link_dest dest;
+	char *href;
+	float w;
+
+	while (flow)
+	{
+		href = flow->box->a_href;
+		next = flow->next;
+		if (href && (int)(flow->y / page_h) == page)
+		{
+			/* Coalesce contiguous flow boxes into one link node */
+			w = flow->w;
+			while (next &&
+					next->y == flow->y &&
+					next->h == flow->h &&
+					next->box->a_href &&
+					!strcmp(href, next->box->a_href))
+			{
+				w += next->w;
+				next = next->next;
+			}
+
+			bbox.x0 = flow->x;
+			bbox.y0 = flow->y - page * page_h;
+			bbox.x1 = bbox.x0 + w;
+			bbox.y1 = bbox.y0 + flow->h;
+			if (flow->type != FLOW_IMAGE)
+			{
+				/* flow->y is the baseline, adjust bbox appropriately */
+				bbox.y0 -= 0.8 * flow->h;
+				bbox.y1 -= 0.8 * flow->h;
+			}
+
+			if (is_internal_uri(href))
+			{
+				if (href[0] == '#')
+				{
+					fz_strlcpy(path, file, sizeof path);
+					fz_strlcat(path, href, sizeof path);
+				}
+				else
+				{
+					fz_strlcpy(path, dir, sizeof path);
+					fz_strlcat(path, "/", sizeof path);
+					fz_strlcat(path, href, sizeof path);
+				}
+				fz_urldecode(path);
+				fz_cleanname(path);
+
+				memset(&dest, 0, sizeof dest);
+				dest.kind = FZ_LINK_GOTO;
+				dest.ld.gotor.dest = fz_strdup(ctx, path);
+				dest.ld.gotor.page = 0; /* computed in epub_load_links */
+			}
+			else
+			{
+				memset(&dest, 0, sizeof dest);
+				dest.kind = FZ_LINK_URI;
+				dest.ld.uri.uri = fz_strdup(ctx, href);
+				dest.ld.uri.is_map = 0;
+			}
+
+			link = fz_new_link(ctx, &bbox, dest);
+			link->next = head;
+			head = link;
+		}
+		flow = next;
+	}
+	return head;
+}
+
+static fz_link *load_link_box(fz_context *ctx, fz_html_box *box, fz_link *head, int page, int page_h, const char *dir, const char *file)
+{
+	while (box)
+	{
+		if (box->flow_head)
+			head = load_link_flow(ctx, box->flow_head, head, page, page_h, dir, file);
+		if (box->down)
+			head = load_link_box(ctx, box->down, head, page, page_h, dir, file);
+		box = box->next;
+	}
+	return head;
+}
+
+fz_link *
+fz_load_html_links(fz_context *ctx, fz_html *html, int page, int page_h, const char *file)
+{
+	char dir[2048];
+	fz_dirname(dir, file, sizeof dir);
+	return load_link_box(ctx, html->root, NULL, page, page_h, dir, file);
+}
+
 static char *concat_text(fz_context *ctx, fz_xml *root)
 {
 	fz_xml *node;
@@ -2085,6 +2191,9 @@ fz_print_html_box(fz_context *ctx, fz_html_box *box, int pstyle, int level)
 		if (box->list_item)
 			printf(" list=%d", box->list_item);
 
+		if (box->a_href)
+			printf(" href='%s'", box->a_href);
+
 		if (box->down || box->flow_head)
 			printf(" {\n");
 		else