From 50643df49e46d0817fc1f57c83cdf5aa2f98dfea Mon Sep 17 00:00:00 2001
From: Martin Michelsen <fuzziqersoftware+g@gmail.com>
Date: Mon, 8 Aug 2022 23:20:54 -0700
Subject: [PATCH] add PatchFileIndex

---
 src/PatchFileIndex.cc | 66 +++++++++++++++++++++++++++++++++++++++++++
 src/PatchFileIndex.hh | 48 +++++++++++++++++++++++++++++++
 2 files changed, 114 insertions(+)
 create mode 100644 src/PatchFileIndex.cc
 create mode 100644 src/PatchFileIndex.hh

diff --git a/src/PatchFileIndex.cc b/src/PatchFileIndex.cc
new file mode 100644
index 00000000..eb43370b
--- /dev/null
+++ b/src/PatchFileIndex.cc
@@ -0,0 +1,66 @@
+#include "PatchFileIndex.hh"
+
+#include <stdio.h>
+#include <string.h>
+
+#include <functional>
+#include <stdexcept>
+#include <phosg/Filesystem.hh>
+#include <phosg/Hash.hh>
+#include <phosg/Strings.hh>
+
+#include "Loggers.hh"
+
+using namespace std;
+
+
+
+void PatchFileIndex::File::load_data(const string& root_dir) {
+  string full_path = root_dir + '/' + join(this->path_directories, "/") + '/' + this->name;
+  auto f = fopen_unique(full_path, "rb");
+
+  this->size = 0;
+  this->crc32 = 0;
+  while (!feof(f.get())) {
+    auto& chunk = this->chunks.emplace_back();
+    chunk.data = fread(f.get(), 0x4000);
+    this->size += chunk.data.size();
+    this->crc32 = ::crc32(chunk.data.data(), chunk.data.size(), this->crc32);
+  }
+}
+
+PatchFileIndex::PatchFileIndex(const string& root_dir)
+  : root_dir(root_dir) {
+
+  vector<string> path_directories;
+  function<void(const string&)> collect_dir = [&](const string& dir) -> void {
+    path_directories.emplace_back(dir);
+
+    string full_dir_path = root_dir + '/' + join(path_directories, "/");
+    patch_index_log.info("Listing directory %s", full_dir_path.c_str());
+
+    for (const auto& item : list_directory(full_dir_path)) {
+      // Skip invisible files (e.g. .DS_Store on macOS)
+      if (starts_with(item, ".")) {
+        continue;
+      }
+
+      string full_item_path = full_dir_path + '/' + item;
+      if (isdir(full_item_path)) {
+        collect_dir(item);
+      } else if (isfile(full_item_path)) {
+        shared_ptr<File> f(new File());
+        f->path_directories = path_directories;
+        f->name = item;
+        f->load_data(root_dir);
+        this->files.emplace_back(f);
+        patch_index_log.info("Added file %s (%zu bytes; %zu chunks; %08" PRIX32 ")",
+            full_item_path.c_str(), f->size, f->chunks.size(), f->crc32);
+      }
+    }
+
+    path_directories.pop_back();
+  };
+
+  collect_dir(".");
+}
diff --git a/src/PatchFileIndex.hh b/src/PatchFileIndex.hh
new file mode 100644
index 00000000..ef5ae912
--- /dev/null
+++ b/src/PatchFileIndex.hh
@@ -0,0 +1,48 @@
+#pragma once
+
+#include <inttypes.h>
+
+#include <string>
+#include <unordered_map>
+#include <map>
+#include <vector>
+#include <memory>
+
+
+
+struct PatchFileIndex {
+  explicit PatchFileIndex(const std::string& root_dir);
+
+  struct File {
+    struct Chunk {
+      std::string data;
+      uint32_t crc32;
+    };
+    std::vector<std::string> path_directories;
+    std::string name;
+    std::vector<Chunk> chunks;
+    size_t size;
+    uint32_t crc32;
+
+    File() : size(0), crc32(0) { }
+    void load_data(const std::string& root_dir);
+  };
+
+  std::vector<std::shared_ptr<File>> files;
+  std::string root_dir;
+};
+
+struct PatchFileChecksumRequest {
+  std::shared_ptr<const PatchFileIndex::File> file;
+  uint32_t crc32;
+  uint32_t size;
+  bool response_received;
+
+  explicit PatchFileChecksumRequest(std::shared_ptr<const PatchFileIndex::File> file)
+    : file(file), crc32(0), size(0), response_received(false) { }
+  inline bool needs_update() const {
+    return !this->response_received ||
+           (this->crc32 != this->file->crc32) ||
+           (this->size != this->file->size);
+  }
+};