SuperSize: Caspian: Group dex methods by class

Bug: 1011921
Change-Id: Ieabbb0ba0b1727200d575f23ad9a5c6477b31557
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1884369Reviewed-by: Andrew Grieve <agrieve@chromium.org>
Commit-Queue: Jasper Chapman-Black <jaspercb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#710031}

tools/binary_size/libsupersize/caspian/caspian_web.cc

@@ -82,7 +82,7 @@ TreeNode* GetOrMakeParentNode(TreeNode* child_node) {
     parent->id_path = parent_path;
     parent->short_name_index = parent_path.find_last_of('/') + 1;
     // TODO: Container type might be a component instead of a directory.
-    parent->containerType = ContainerType::kDirectory;
+    parent->container_type = ContainerType::kDirectory;
   }
   if (child_node->parent != parent) {
     AttachToParent(child_node, parent);
@@ -90,11 +90,90 @@ TreeNode* GetOrMakeParentNode(TreeNode* child_node) {
   return parent;
 }
 
+/* Merges dex method symbols into containers based on the class of the dex
+ * method.
+ */
+void JoinDexMethodClasses(TreeNode* node) {
+  const bool is_file_node = node->container_type == ContainerType::kFile;
+  const bool has_dex =
+      node->node_stats.child_stats.count(SectionId::kDex) ||
+      node->node_stats.child_stats.count(SectionId::kDexMethod);
+  // Don't try to merge dex symbols for catch-all symbols under (No path).
+  const bool is_no_path = node->id_path == NO_NAME;
+
+  if (!is_file_node || !has_dex || is_no_path || node->children.empty()) {
+    return;
+  }
+
+  std::map<std::string_view, TreeNode*> java_class_containers;
+  std::vector<TreeNode*> other_symbols;
+
+  // Bucket dex symbols by their class.
+  for (TreeNode* child : node->children) {
+    // Unlike in .ndjson fields, Java classes loaded from .size files are just
+    // the classname, such as "android.support.v7.widget.toolbar".
+    // Method names contain the classname followed by the method definition,
+    // like "android.support.v7.widget.toolbar void onMeasure(int, int)".
+    const size_t split_index = child->id_path.find_first_of(' ');
+    // No return type / field type means it's a class node.
+    const bool is_class_node =
+        child->id_path.find_first_of(' ', child->short_name_index) ==
+        std::string_view::npos;
+    const bool has_class_prefix =
+        is_class_node || split_index != std::string_view::npos;
+
+    if (has_class_prefix) {
+      const std::string_view class_id_path =
+          split_index == std::string_view::npos
+              ? child->id_path
+              : child->id_path.substr(0, split_index);
+
+      // Strip package from the node name for classes in .java files since the
+      // directory tree already shows it.
+      int short_name_index = child->short_name_index;
+      size_t java_idx = node->id_path.find(".java");
+      if (java_idx != std::string_view::npos) {
+        size_t dot_idx = class_id_path.find_last_of('.');
+        short_name_index += dot_idx + 1;
+      }
+
+      TreeNode*& class_node = java_class_containers[class_id_path];
+      if (class_node == nullptr) {
+        class_node = new TreeNode();
+        class_node->id_path = class_id_path;
+        class_node->src_path = node->src_path;
+        class_node->component = node->component;
+        class_node->short_name_index = short_name_index;
+        class_node->container_type = ContainerType::kJavaClass;
+        parents[class_node->id_path] = class_node;
+      }
+
+      // Adjust the dex method's short name so it starts after the " "
+      if (split_index != std::string_view::npos) {
+        child->short_name_index = split_index + 1;
+      }
+      child->parent = nullptr;
+      AttachToParent(child, class_node);
+    } else {
+      other_symbols.push_back(child);
+    }
+  }
+
+  node->children = other_symbols;
+  for (auto& iter : java_class_containers) {
+    TreeNode* container_node = iter.second;
+    // Delay setting the parent until here so that `_attachToParent`
+    // doesn't add method stats twice
+    container_node->parent = node;
+    node->children.push_back(container_node);
+  }
+}
+
 void AddFileEntry(const std::string_view source_path,
                   const std::vector<const Symbol*>& symbols) {
   // Creates a single file node with a child for each symbol in that file.
   TreeNode* file_node = new TreeNode();
-  file_node->containerType = ContainerType::kFile;
+  file_node->container_type = ContainerType::kFile;
   if (source_path.empty()) {
     file_node->id_path = NO_NAME;
   } else {
@@ -107,7 +186,7 @@ void AddFileEntry(const std::string_view source_path,
   // Create symbol nodes.
   for (const auto sym : symbols) {
     TreeNode* symbol_node = new TreeNode();
-    symbol_node->containerType = ContainerType::kSymbol;
+    symbol_node->container_type = ContainerType::kSymbol;
     symbol_node->id_path = sym->full_name;
     symbol_node->size = sym->size;
     symbol_node->node_stats =
@@ -120,6 +199,8 @@ void AddFileEntry(const std::string_view source_path,
   while (orphan_node != &root) {
     orphan_node = GetOrMakeParentNode(orphan_node);
   }
+
+  JoinDexMethodClasses(file_node);
 }
 
 void AddSymbolsAndFileNodes(SizeInfo* size_info) {
@@ -143,7 +224,7 @@ bool LoadSizeFile(const char* compressed, size_t size) {
   writer.reset(Json::StreamWriterBuilder().newStreamWriter());
   ParseSizeInfo(compressed, size, &info);
   // Build tree
-  root.containerType = ContainerType::kDirectory;
+  root.container_type = ContainerType::kDirectory;
   root.id_path = "/";
   parents[""] = &root;
   AddSymbolsAndFileNodes(&info);

tools/binary_size/libsupersize/caspian/model.cc

@@ -30,7 +30,7 @@ SectionId SizeInfo::ShortSectionName(const char* section_name) {
     } else if (!strcmp(section_name, ".dex")) {
       ret = SectionId::kDex;
     } else if (!strcmp(section_name, ".dex.method")) {
-      ret = SectionId::kDex;
+      ret = SectionId::kDexMethod;
     } else if (!strcmp(section_name, ".other")) {
       ret = SectionId::kOther;
     } else if (!strcmp(section_name, ".rodata")) {
@@ -61,8 +61,8 @@ void TreeNode::WriteIntoJson(Json::Value* out, int depth) {
   (*out)["shortNameIndex"] = this->short_name_index;
   // TODO: Put correct information.
   std::string type;
-  if (containerType != ContainerType::kSymbol) {
-    type += static_cast<char>(containerType);
+  if (container_type != ContainerType::kSymbol) {
+    type += static_cast<char>(container_type);
   }
   SectionId biggest_section = this->node_stats.ComputeBiggestSection();
   type += static_cast<char>(biggest_section);
@@ -95,12 +95,12 @@ NodeStats::NodeStats(SectionId sectionId,
                      int32_t count,
                      int32_t highlight,
                      int32_t size) {
-  childStats[sectionId] = {count, highlight, size};
+  child_stats[sectionId] = {count, highlight, size};
 }
 
 void NodeStats::WriteIntoJson(Json::Value* out) const {
   (*out) = Json::Value(Json::objectValue);
-  for (const auto kv : this->childStats) {
+  for (const auto kv : this->child_stats) {
     const std::string sectionId = std::string(1, static_cast<char>(kv.first));
     const Stat stats = kv.second;
     (*out)[sectionId] = Json::Value(Json::objectValue);
@@ -111,8 +111,8 @@ void NodeStats::WriteIntoJson(Json::Value* out) const {
 }
 
 NodeStats& NodeStats::operator+=(const NodeStats& other) {
-  for (const auto& it : other.childStats) {
-    childStats[it.first] += it.second;
+  for (const auto& it : other.child_stats) {
+    child_stats[it.first] += it.second;
   }
   return *this;
 }
@@ -120,7 +120,7 @@ NodeStats& NodeStats::operator+=(const NodeStats& other) {
 SectionId NodeStats::ComputeBiggestSection() const {
   SectionId ret = SectionId::kNone;
   int32_t max = 0;
-  for (auto& pair : childStats) {
+  for (auto& pair : child_stats) {
     if (pair.second.size > max) {
       ret = pair.first;
       max = pair.second.size;

tools/binary_size/libsupersize/caspian/model.h

@@ -98,7 +98,7 @@ struct NodeStats {
   NodeStats& operator+=(const NodeStats& other);
   SectionId ComputeBiggestSection() const;
 
-  std::map<SectionId, Stat> childStats;
+  std::map<SectionId, Stat> child_stats;
 };
 
 struct TreeNode {
@@ -120,7 +120,7 @@ struct TreeNode {
     childStats,
   */
 
-  ContainerType containerType = ContainerType::kSymbol;
+  ContainerType container_type = ContainerType::kSymbol;
 
   std::vector<TreeNode*> children;
   TreeNode* parent = nullptr;