1 files changed, 240 insertions, 64 deletions

diff --git a/pdf/pdf_page.c b/pdf/pdf_page.c
index f2389fea..b413b68d 100644
--- a/pdf/pdf_page.c
+++ b/pdf/pdf_page.c
@@ -56,7 +56,7 @@ static int pdfi_process_page_contents(pdf_context *ctx, pdf_dict *page_dict)
     if (code < 0)
         return code;
 
-    if (o->type == PDF_INDIRECT) {
+    if (pdfi_type_of(o) == PDF_INDIRECT) {
         if (((pdf_indirect_ref *)o)->ref_object_num == page_dict->object_num)
             return_error(gs_error_circular_reference);
 
@@ -77,7 +77,7 @@ static int pdfi_process_page_contents(pdf_context *ctx, pdf_dict *page_dict)
     }
 
     ctx->encryption.decrypt_strings = false;
-    if (o->type == PDF_ARRAY) {
+    if (pdfi_type_of(o) == PDF_ARRAY) {
         pdf_array *a = (pdf_array *)o;
 
         for (i=0;i < pdfi_array_size(a); i++) {
@@ -85,13 +85,13 @@ static int pdfi_process_page_contents(pdf_context *ctx, pdf_dict *page_dict)
             code = pdfi_array_get_no_deref(ctx, a, i, (pdf_obj **)&r);
             if (code < 0)
                 goto page_error;
-            if (r->type == PDF_STREAM) {
+            if (pdfi_type_of (r) == PDF_STREAM) {
                 code = pdfi_interpret_content_stream(ctx, NULL, (pdf_stream *)r, page_dict);
                 pdfi_countdown(r);
                 if (code < 0)
                     goto page_error;
             } else {
-                if (r->type != PDF_INDIRECT) {
+                if (pdfi_type_of(r) != PDF_INDIRECT) {
                         pdfi_countdown(r);
                         code = gs_note_error(gs_error_typecheck);
                         goto page_error;
@@ -108,7 +108,7 @@ static int pdfi_process_page_contents(pdf_context *ctx, pdf_dict *page_dict)
                             code = 0;
                         goto page_error;
                     }
-                    if (o1->type != PDF_STREAM) {
+                    if (pdfi_type_of(o1) != PDF_STREAM) {
                         pdfi_countdown(o1);
                         code = gs_note_error(gs_error_typecheck);
                         goto page_error;
@@ -123,7 +123,7 @@ static int pdfi_process_page_contents(pdf_context *ctx, pdf_dict *page_dict)
             }
         }
     } else {
-        if (o->type == PDF_STREAM) {
+        if (pdfi_type_of(o) == PDF_STREAM) {
             code = pdfi_interpret_content_stream(ctx, NULL, (pdf_stream *)o, page_dict);
         } else {
             pdfi_countdown(o);
@@ -295,14 +295,23 @@ static int pdfi_set_media_size(pdf_context *ctx, pdf_dict *page_dict)
     if (a == NULL) {
         code = pdfi_dict_get_type(ctx, page_dict, "CropBox", PDF_ARRAY, (pdf_obj **)&a);
         if (code >= 0 && pdfi_array_size(a) >= 4) {
+            pdf_obj *box_obj = NULL;
+
             for (i=0;i<4;i++) {
-                code = pdfi_array_get_number(ctx, a, i, &d_crop[i]);
-                d_crop[i] *= userunit;
+                code = pdfi_array_get_no_store_R(ctx, a, i, &box_obj);
+                if (code >= 0) {
+                    code = pdfi_obj_to_real(ctx, box_obj, &d_crop[i]);
+                    pdfi_countdown(box_obj);
+                }
+                if (code < 0)
+                    break;
             }
             pdfi_countdown(a);
-            normalize_rectangle(d_crop);
-            memcpy(ctx->page.Crop, d_crop, 4 * sizeof(double));
-            do_crop = true;
+            if (code >= 0) {
+                normalize_rectangle(d_crop);
+                memcpy(ctx->page.Crop, d_crop, 4 * sizeof(double));
+                do_crop = true;
+            }
         }
         a = default_media;
     }
@@ -313,7 +322,20 @@ static int pdfi_set_media_size(pdf_context *ctx, pdf_dict *page_dict)
     ctx->page.UserUnit = userunit;
 
     for (i=0;i<4;i++) {
-        code = pdfi_array_get_number(ctx, a, i, &d[i]);
+        pdf_obj *box_obj = NULL;
+
+        code = pdfi_array_get_no_store_R(ctx, a, i, &box_obj);
+        if (code >= 0) {
+            code = pdfi_obj_to_real(ctx, box_obj, &d[i]);
+            pdfi_countdown(box_obj);
+        }
+
+        if (code < 0) {
+            pdfi_countdown(a);
+            pdfi_set_warning(ctx, code, NULL, W_PDF_BAD_MEDIABOX, "pdfi_get_media_size", NULL);
+            code = gs_erasepage(ctx->pgs);
+            return 0;
+        }
         d[i] *= userunit;
     }
     pdfi_countdown(a);
@@ -458,26 +480,21 @@ static void pdfi_setup_transfers(pdf_context *ctx)
     }
 }
 
-static int store_box(pdf_context *ctx, float *box, pdf_array *a)
-{
-    double f;
-    int code = 0, i;
-
-    for (i=0;i < 4;i++) {
-        code = pdfi_array_get_number(ctx, a, (uint64_t)i, &f);
-        if (code < 0)
-            return code;
-        box[i] = (float)f;
-    }
-    return 0;
-}
-
-int pdfi_page_info(pdf_context *ctx, uint64_t page_num, pdf_info_t *info)
+/* Return a dictionary containing information about the page. Basic information is that
+ * required to render the page; if extended is true then additionally contains an
+ * array of spot ink names and an array of dictionaries each of which contains
+ * information about a font used on the page. THis is normally only used for tools
+ * like pdf_info.ps
+ */
+int pdfi_page_info(pdf_context *ctx, uint64_t page_num, pdf_dict **info, bool extended)
 {
-    int code = 0;
-    pdf_dict *page_dict = NULL;
+    int code = 0, i=0;
+    pdf_dict *page_dict = NULL, *info_dict = NULL;
+    pdf_array *fonts_array = NULL, *spots_array = NULL;
     pdf_array *a = NULL;
-    double dbl = 0.0;
+    pdf_obj *o = NULL;
+    bool known = false;
+    double dummy;
 
     code = pdfi_page_get_dict(ctx, page_num, &page_dict);
     if (code < 0)
@@ -488,79 +505,202 @@ int pdfi_page_info(pdf_context *ctx, uint64_t page_num, pdf_info_t *info)
         goto done;
     }
 
-    code = pdfi_check_page(ctx, page_dict, false);
+    code = pdfi_dict_alloc(ctx, 6, &info_dict);
     if (code < 0)
         goto done;
 
-    info->boxes = BOX_NONE;
+    pdfi_countup(info_dict);
+
+    if (extended)
+        code = pdfi_check_page(ctx, page_dict, &fonts_array, &spots_array, false);
+    else
+        code = pdfi_check_page(ctx, page_dict, NULL, NULL, false);
+    if (code < 0)
+        goto done;
+
+    if (spots_array != NULL) {
+        code = pdfi_dict_put(ctx, info_dict, "Spots", (pdf_obj *)spots_array);
+        if (code < 0)
+            goto done;
+        pdfi_countdown(spots_array);
+    }
+
+    if (fonts_array != NULL) {
+        code = pdfi_dict_put(ctx, info_dict, "Fonts", (pdf_obj *)fonts_array);
+        if (code < 0)
+            goto done;
+        pdfi_countdown(fonts_array);
+    }
+
     code = pdfi_dict_get_type(ctx, page_dict, "MediaBox", PDF_ARRAY, (pdf_obj **)&a);
     if (code < 0)
         pdfi_set_warning(ctx, code, NULL, W_PDF_BAD_MEDIABOX, "pdfi_page_info", NULL);
 
     if (code >= 0) {
-        code = store_box(ctx, (float *)&info->MediaBox, a);
+        pdf_obj *box_obj = NULL;
+
+        for (i = 0;i < pdfi_array_size(a); i++) {
+            code = pdfi_array_get_no_store_R(ctx, a, i, &box_obj);
+            if (code >= 0) {
+                code = pdfi_obj_to_real(ctx, box_obj, &dummy);
+                pdfi_countdown(box_obj);
+            }
+            if (code < 0) {
+                pdfi_set_warning(ctx, code, NULL, W_PDF_BAD_MEDIABOX, "pdfi_page_info", NULL);
+                goto done;
+            }
+        }
+
+        code = pdfi_dict_put(ctx, info_dict, "MediaBox", (pdf_obj *)a);
         if (code < 0)
             goto done;
-        info->boxes |= MEDIA_BOX;
         pdfi_countdown(a);
         a = NULL;
     }
 
     code = pdfi_dict_get_type(ctx, page_dict, "ArtBox", PDF_ARRAY, (pdf_obj **)&a);
     if (code >= 0) {
-        code = store_box(ctx, (float *)&info->ArtBox, a);
-        if (code < 0)
-            goto done;
-        info->boxes |= ART_BOX;
+        pdf_obj *box_obj = NULL;
+
+        for (i = 0;i < pdfi_array_size(a); i++) {
+            code = pdfi_array_get_no_store_R(ctx, a, i, &box_obj);
+            if (code >= 0) {
+                code = pdfi_obj_to_real(ctx, box_obj, &dummy);
+                pdfi_countdown(box_obj);
+            }
+            if (code < 0)
+                break;
+        }
+        if (code >= 0) {
+            code = pdfi_dict_put(ctx, info_dict, "ArtBox", (pdf_obj *)a);
+            if (code < 0)
+                goto done;
+        }
         pdfi_countdown(a);
         a = NULL;
     }
 
     code = pdfi_dict_get_type(ctx, page_dict, "CropBox", PDF_ARRAY, (pdf_obj **)&a);
     if (code >= 0) {
-        code = store_box(ctx, (float *)&info->CropBox, a);
-        if (code < 0)
-            goto done;
-        info->boxes |= CROP_BOX;
+        pdf_obj *box_obj = NULL;
+
+        for (i = 0;i < pdfi_array_size(a); i++) {
+            code = pdfi_array_get_no_store_R(ctx, a, i, &box_obj);
+            if (code >= 0) {
+                code = pdfi_obj_to_real(ctx, box_obj, &dummy);
+                pdfi_countdown(box_obj);
+            }
+            if (code < 0)
+                break;
+        }
+        if (code >= 0) {
+            code = pdfi_dict_put(ctx, info_dict, "CropBox", (pdf_obj *)a);
+            if (code < 0)
+                goto done;
+        }
         pdfi_countdown(a);
         a = NULL;
     }
 
     code = pdfi_dict_get_type(ctx, page_dict, "TrimBox", PDF_ARRAY, (pdf_obj **)&a);
     if (code >= 0) {
-        code = store_box(ctx, (float *)&info->TrimBox, a);
-        if (code < 0)
-            goto done;
-        info->boxes |= TRIM_BOX;
+        pdf_obj *box_obj = NULL;
+
+        for (i = 0;i < pdfi_array_size(a); i++) {
+            code = pdfi_array_get_no_store_R(ctx, a, i, &box_obj);
+            if (code >= 0) {
+                code = pdfi_obj_to_real(ctx, box_obj, &dummy);
+                pdfi_countdown(box_obj);
+            }
+            if (code < 0)
+                break;
+        }
+        if (code >= 0) {
+            code = pdfi_dict_put(ctx, info_dict, "TrimBox", (pdf_obj *)a);
+            if (code < 0)
+                goto done;
+        }
         pdfi_countdown(a);
         a = NULL;
     }
 
     code = pdfi_dict_get_type(ctx, page_dict, "BleedBox", PDF_ARRAY, (pdf_obj **)&a);
     if (code >= 0) {
-        code = store_box(ctx, (float *)&info->BleedBox, a);
-        if (code < 0)
-            goto done;
-        info->boxes |= BLEED_BOX;
+        pdf_obj *box_obj = NULL;
+
+        for (i = 0;i < pdfi_array_size(a); i++) {
+            code = pdfi_array_get_no_store_R(ctx, a, i, &box_obj);
+            if (code >= 0) {
+                code = pdfi_obj_to_real(ctx, box_obj, &dummy);
+                pdfi_countdown(box_obj);
+            }
+            if (code < 0)
+                break;
+        }
+        if (code >= 0) {
+            code = pdfi_dict_put(ctx, info_dict, "BleedBox", (pdf_obj *)a);
+            if (code < 0)
+                goto done;
+        }
         pdfi_countdown(a);
         a = NULL;
     }
     code = 0;
 
-    dbl = info->Rotate = 0;
-    code = pdfi_dict_get_number(ctx, page_dict, "Rotate", &dbl);
-    code = 0;
-    info->Rotate = dbl;
+    code = pdfi_dict_get(ctx, page_dict, "Rotate", &o);
+    if (code >= 0) {
+        if (pdfi_type_of(o) == PDF_INT || pdfi_type_of(o) == PDF_REAL) {
+            code = pdfi_dict_put(ctx, info_dict, "Rotate", o);
+            if (code < 0)
+                goto done;
+        }
+        pdfi_countdown(o);
+    }
 
-    dbl = info->UserUnit = 1;
-    code = pdfi_dict_get_number(ctx, page_dict, "UserUnit", &dbl);
-    code = 0;
-    info->UserUnit = dbl;
+    code = pdfi_dict_get(ctx, page_dict, "UserUnit", &o);
+    if (code >= 0) {
+        if (pdfi_type_of(o) == PDF_INT || pdfi_type_of(o) == PDF_REAL) {
+            code = pdfi_dict_put(ctx, info_dict, "UserUnit", o);
+            if (code < 0)
+                goto done;
+        }
+        pdfi_countdown(o);
+    }
+
+    if (ctx->page.has_transparency)
+        code = pdfi_dict_put(ctx, info_dict, "UsesTransparency", PDF_TRUE_OBJ);
+    else
+        code = pdfi_dict_put(ctx, info_dict, "UsesTransparency", PDF_FALSE_OBJ);
+    if (code < 0)
+        goto done;
+
+    code = pdfi_dict_known(ctx, page_dict, "Annots", &known);
+    if (code >= 0 && known)
+        code = pdfi_dict_put(ctx, info_dict, "Annots", PDF_TRUE_OBJ);
+    else
+        code = pdfi_dict_put(ctx, info_dict, "Annots", PDF_FALSE_OBJ);
+    if (code < 0)
+        goto done;
 
-    info->HasTransparency = ctx->page.has_transparency;
-    info->NumSpots = ctx->page.num_spots;
+    code = pdfi_object_alloc(ctx, PDF_INT, 0, &o);
+    if (code >= 0) {
+        pdfi_countup(o);
+        ((pdf_num *)o)->value.i = ctx->page.num_spots;
+        code = pdfi_dict_put(ctx, info_dict, "NumSpots", o);
+        pdfi_countdown(o);
+        o = NULL;
+        if (code < 0)
+            goto done;
+    }
 
 done:
+    if (code < 0) {
+        pdfi_countdown(info_dict);
+        info_dict = NULL;
+        *info = NULL;
+    } else
+        *info = info_dict;
+
     pdfi_countdown(a);
     pdfi_countdown(page_dict);
     return code;
@@ -585,7 +725,7 @@ int pdfi_page_get_dict(pdf_context *ctx, uint64_t page_num, pdf_dict **dict)
         code = pdfi_dict_get(ctx, ctx->Root, "Pages", &o);
         if (code < 0)
             goto page_error;
-        if (o->type != PDF_DICT) {
+        if (pdfi_type_of(o) != PDF_DICT) {
             code = gs_note_error(gs_error_typecheck);
             goto page_error;
         }
@@ -663,14 +803,35 @@ int pdfi_page_get_number(pdf_context *ctx, pdf_dict *target_dict, uint64_t *page
 static void release_page_DefaultSpaces(pdf_context *ctx)
 {
     if (ctx->page.DefaultGray_cs != NULL) {
+        if (ctx->page.DefaultGray_cs->interpreter_data != NULL) {
+            pdf_obj *o = (pdf_obj *)(ctx->page.DefaultGray_cs->interpreter_data);
+            if (o != NULL && pdfi_type_of(o) == PDF_NAME) {
+                pdfi_countdown(o);
+                ctx->page.DefaultGray_cs->interpreter_data = NULL;
+            }
+        }
         rc_decrement(ctx->page.DefaultGray_cs, "pdfi_page_render");
         ctx->page.DefaultGray_cs = NULL;
     }
     if (ctx->page.DefaultRGB_cs != NULL) {
+        if (ctx->page.DefaultRGB_cs->interpreter_data != NULL) {
+            pdf_obj *o = (pdf_obj *)(ctx->page.DefaultRGB_cs->interpreter_data);
+            if (o != NULL && pdfi_type_of(o) == PDF_NAME) {
+                pdfi_countdown(o);
+                ctx->page.DefaultRGB_cs->interpreter_data = NULL;
+            }
+        }
         rc_decrement(ctx->page.DefaultRGB_cs, "pdfi_page_render");
         ctx->page.DefaultRGB_cs = NULL;
     }
     if (ctx->page.DefaultCMYK_cs != NULL) {
+        if (ctx->page.DefaultCMYK_cs->interpreter_data != NULL) {
+            pdf_obj *o = (pdf_obj *)(ctx->page.DefaultCMYK_cs->interpreter_data);
+            if (o != NULL && pdfi_type_of(o) == PDF_NAME) {
+                pdfi_countdown(o);
+                ctx->page.DefaultCMYK_cs->interpreter_data = NULL;
+            }
+        }
         rc_decrement(ctx->page.DefaultCMYK_cs, "pdfi_page_render");
         ctx->page.DefaultCMYK_cs = NULL;
     }
@@ -684,6 +845,12 @@ static int setup_page_DefaultSpaces(pdf_context *ctx, pdf_dict *page_dict)
     return(pdfi_setup_DefaultSpaces(ctx, page_dict));
 }
 
+static bool
+pdfi_pattern_purge_all_proc(gx_color_tile * ctile, void *proc_data)
+{
+    return true;
+}
+
 int pdfi_page_render(pdf_context *ctx, uint64_t page_num, bool init_graphics)
 {
     int code, code1=0;
@@ -714,7 +881,7 @@ int pdfi_page_render(pdf_context *ctx, uint64_t page_num, bool init_graphics)
 
     pdfi_device_set_flags(ctx);
 
-    code = pdfi_check_page(ctx, page_dict, init_graphics);
+    code = pdfi_check_page(ctx, page_dict, NULL, NULL, init_graphics);
     if (code < 0)
         goto exit3;
 
@@ -729,8 +896,12 @@ int pdfi_page_render(pdf_context *ctx, uint64_t page_num, bool init_graphics)
     }
 
     code = pdfi_dict_knownget_type(ctx, page_dict, "Group", PDF_DICT, (pdf_obj **)&group_dict);
+    /* Ignore errors retrieving the Group dictionary, we will just ignore it. This allows us
+     * to handle files such as Bug #705206 where the Group dictionary is a free object in a
+     * compressed object stream.
+     */
     if (code < 0)
-        goto exit3;
+        pdfi_set_error(ctx, 0, NULL, E_BAD_GROUP_DICT, "pdfi_page_render", NULL);
     if (group_dict != NULL)
         page_group_known = true;
 
@@ -864,7 +1035,12 @@ exit3:
 
     release_page_DefaultSpaces(ctx);
 
-    if (code == 0 || (!ctx->args.pdfstoponerror && code != gs_error_stackoverflow))
+    /* Flush any pattern tiles. We don't want to (potentially) return to PostScript
+     * with any pattern tiles referencing our objects, in case the garbager runs.
+     */
+    gx_pattern_cache_winnow(gstate_pattern_cache(ctx->pgs), pdfi_pattern_purge_all_proc, NULL);
+
+    if (code == 0 || (!ctx->args.pdfstoponerror && code != gs_error_pdf_stackoverflow))
         if (!page_dict_error && ctx->finish_page != NULL)
             code = ctx->finish_page(ctx);
     return code;