From d6a9b41bacc3be065111e2244eeab96e78c16fbf Mon Sep 17 00:00:00 2001
From: Matthew Olsson <matthewcolsson@gmail.com>
Date: Mon, 10 May 2021 10:44:22 -0700
Subject: [PATCH] LibPDF: Parse page crop box and user units

---
 Userland/Libraries/LibPDF/Document.cpp | 20 ++++++++++++++++++--
 Userland/Libraries/LibPDF/Document.h   | 13 ++++++++++---
 2 files changed, 28 insertions(+), 5 deletions(-)

diff --git a/Userland/Libraries/LibPDF/Document.cpp b/Userland/Libraries/LibPDF/Document.cpp
index 7d7d01c79a6..c95966c8671 100644
--- a/Userland/Libraries/LibPDF/Document.cpp
+++ b/Userland/Libraries/LibPDF/Document.cpp
@@ -63,6 +63,8 @@ Page Document::get_page(u32 index)
     auto raw_page_object = resolve_to<DictObject>(get_or_load_value(page_object_index));
 
     auto resources = raw_page_object->get_dict(this, "Resources");
+    auto contents = raw_page_object->get_object(this, "Contents");
+
     auto media_box_array = raw_page_object->get_array(this, "MediaBox");
     auto media_box = Rectangle {
         media_box_array->at(0).to_float(),
@@ -70,9 +72,23 @@ Page Document::get_page(u32 index)
         media_box_array->at(2).to_float(),
         media_box_array->at(3).to_float(),
     };
-    auto contents = raw_page_object->get_object(this, "Contents");
 
-    Page page { resources, media_box, contents };
+    auto crop_box = media_box;
+    if (raw_page_object->contains("CropBox")) {
+        auto crop_box_array = raw_page_object->get_array(this, "CropBox");
+        crop_box = Rectangle {
+            crop_box_array->at(0).to_float(),
+            crop_box_array->at(1).to_float(),
+            crop_box_array->at(2).to_float(),
+            crop_box_array->at(3).to_float(),
+        };
+    }
+
+    float user_unit = 1.0f;
+    if (raw_page_object->contains("UserUnit"))
+        user_unit = raw_page_object->get_value("UserUnit").to_float();
+
+    Page page { move(resources), move(contents), media_box, crop_box, user_unit };
     m_pages.set(index, page);
     return page;
 }
diff --git a/Userland/Libraries/LibPDF/Document.h b/Userland/Libraries/LibPDF/Document.h
index f5210250879..5adc24a3ecb 100644
--- a/Userland/Libraries/LibPDF/Document.h
+++ b/Userland/Libraries/LibPDF/Document.h
@@ -24,8 +24,10 @@ struct Rectangle {
 
 struct Page {
     NonnullRefPtr<DictObject> resources;
-    Rectangle media_box;
     NonnullRefPtr<Object> contents;
+    Rectangle media_box;
+    Rectangle crop_box;
+    float user_unit;
 };
 
 class Document final : public RefCounted<Document> {
@@ -105,8 +107,13 @@ template<>
 struct Formatter<PDF::Page> : Formatter<StringView> {
     void format(FormatBuilder& builder, const PDF::Page& page)
     {
-        constexpr auto fmt_string = "Page {{\n  resources={}\n  contents={}\n  media_box={}\n}}";
-        auto str = String::formatted(fmt_string, page.resources->to_string(1), page.contents->to_string(1), page.media_box);
+        constexpr auto fmt_string = "Page {{\n  resources={}\n  contents={}\n  media_box={}\n  crop_box={}\n  user_unit={}\n}}";
+        auto str = String::formatted(fmt_string,
+            page.resources->to_string(1),
+            page.contents->to_string(1),
+            page.media_box,
+            page.crop_box,
+            page.user_unit);
         Formatter<StringView>::format(builder, str);
     }
 };