Follow embedded PDF links through Look

1 files changed, 121 insertions, 0 deletions

diff --git a/src/pdf_bridge.c b/src/pdf_bridge.c
index 8e433fde..bdc7b891 100644
--- a/src/pdf_bridge.c
+++ b/src/pdf_bridge.c
@@ -27,6 +27,8 @@ struct pardes_pdf_document {
     fz_stext_page *cached_text;
     fz_display_list *cached_display_list;
     fz_rect cached_bounds;
+    fz_link *cached_links;
+    int cached_links_loaded;
 
     /* Promote only a page which is rendered twice without changing pages. */
     int display_list_candidate_page_number;
@@ -126,12 +128,15 @@ pardes_pdf_free_document(pardes_pdf_document *document)
 static void
 pardes_pdf_drop_cached_page(pardes_pdf_document *document)
 {
+    fz_drop_link(document->ctx, document->cached_links);
     fz_drop_display_list(document->ctx, document->cached_display_list);
     fz_drop_stext_page(document->ctx, document->cached_text);
     fz_drop_page(document->ctx, document->cached_page);
     document->cached_display_list = NULL;
     document->cached_text = NULL;
     document->cached_page = NULL;
+    document->cached_links = NULL;
+    document->cached_links_loaded = 0;
     document->cached_page_number = -1;
     document->display_list_candidate_page_number = -1;
 }
@@ -994,6 +999,85 @@ pardes_pdf_point_inside_quad(pardes_pdf_point point, pardes_pdf_quad quad)
 }
 
 int
+pardes_pdf_link_at(
+    pardes_pdf_document *document,
+    int page_number,
+    pardes_pdf_point point,
+    pardes_pdf_link *out)
+{
+    fz_context *ctx;
+
+    if (out == NULL)
+        return PARDES_PDF_ERROR;
+    memset(out, 0, sizeof(*out));
+    out->page = -1;
+    if (document == NULL || page_number < 0 ||
+        page_number >= document->page_count)
+        return PARDES_PDF_ERROR;
+    if (!pardes_pdf_valid_point(point) || point.x < 0 || point.x > 1 ||
+        point.y < 0 || point.y > 1)
+        return PARDES_PDF_OK;
+
+    ctx = document->ctx;
+    fz_try(ctx)
+    {
+        fz_link *link;
+        fz_point page_point;
+        pardes_pdf_cache_page(document, page_number, 0);
+        if (!document->cached_links_loaded) {
+            document->cached_links = fz_load_links(ctx, document->cached_page);
+            document->cached_links_loaded = 1;
+        }
+        page_point = pardes_pdf_point_to_page(document, point);
+        for (link = document->cached_links; link != NULL; link = link->next) {
+            size_t len;
+            int destination_kind;
+            int destination_page = -1;
+            fz_link_dest destination = fz_make_link_dest_none();
+            if (!pardes_pdf_valid_bounds(link->rect) ||
+                !fz_is_point_inside_rect(page_point, link->rect) ||
+                link->uri == NULL || link->uri[0] == 0)
+                continue;
+            for (len = 0; len <= PARDES_PDF_MAX_LINK_URI_BYTES &&
+                link->uri[len] != 0; ++len)
+                ;
+            if (len > PARDES_PDF_MAX_LINK_URI_BYTES)
+                continue;
+            destination_kind = PARDES_PDF_OUTLINE_DESTINATION_EXTERNAL;
+            if (!fz_is_external_link(ctx, link->uri)) {
+                destination = fz_resolve_link_dest(ctx, document->doc, link->uri);
+                if (destination.loc.chapter >= 0 && destination.loc.page >= 0)
+                    destination_page = fz_page_number_from_location(
+                        ctx, document->doc, destination.loc);
+                if (destination_page >= 0 && destination_page < document->page_count)
+                    destination_kind = PARDES_PDF_OUTLINE_DESTINATION_INTERNAL;
+                else if (link->uri[0] == '#')
+                    continue;
+            }
+            out->quad = pardes_pdf_quad_from_page(
+                document, fz_quad_from_rect(link->rect));
+            out->destination_kind = destination_kind;
+            if (destination_kind == PARDES_PDF_OUTLINE_DESTINATION_INTERNAL) {
+                out->page = destination_page;
+                out->has_x = isfinite(destination.x);
+                out->has_y = isfinite(destination.y);
+                if (out->has_x) out->x = destination.x;
+                if (out->has_y) out->y = destination.y;
+            }
+            out->uri = (const unsigned char *)link->uri;
+            out->uri_len = len;
+            break;
+        }
+    }
+    fz_catch(ctx)
+    {
+        memset(out, 0, sizeof(*out));
+        return PARDES_PDF_ERROR;
+    }
+    return PARDES_PDF_OK;
+}
+
+int
 pardes_pdf_copy_selection(
     pardes_pdf_document *document,
     int page_number,
@@ -1041,6 +1125,43 @@ pardes_pdf_drop_owned_text(pardes_pdf_document *document, void *text)
         fz_free(document->ctx, text);
 }
 
+int
+pardes_pdf_copy_rectangle(
+    pardes_pdf_document *document,
+    int page_number,
+    pardes_pdf_quad quad,
+    pardes_pdf_owned_text *out)
+{
+    fz_context *ctx;
+    char *text = NULL;
+    fz_quad native;
+    if (document == NULL || out == NULL || page_number < 0 ||
+        page_number >= document->page_count || !pardes_pdf_valid_quad(quad))
+        return PARDES_PDF_ERROR;
+    memset(out, 0, sizeof(*out));
+    ctx = document->ctx;
+    fz_var(text);
+    fz_try(ctx)
+    {
+        pardes_pdf_cache_page(document, page_number, 1);
+        native.ul = pardes_pdf_point_to_page(document, quad.ul);
+        native.ur = pardes_pdf_point_to_page(document, quad.ur);
+        native.ll = pardes_pdf_point_to_page(document, quad.ll);
+        native.lr = pardes_pdf_point_to_page(document, quad.lr);
+        text = fz_copy_rectangle(ctx, document->cached_text,
+            fz_rect_from_quad(native), 0);
+    }
+    fz_catch(ctx)
+    {
+        fz_free(ctx, text);
+        return PARDES_PDF_ERROR;
+    }
+    out->handle = text;
+    out->data = (const unsigned char *)text;
+    out->len = text == NULL ? 0 : strlen(text);
+    return PARDES_PDF_OK;
+}
+
 typedef struct pardes_pdf_outline_measurement {
     size_t item_count;
     size_t bytes_len;