add object names

This commit is contained in:
water 2020-09-03 20:11:31 -04:00
parent aa2dc99e75
commit bd15e47506
5 changed files with 3134 additions and 6 deletions

View file

@ -7,6 +7,7 @@
#include "ObjectFileDB.h" #include "ObjectFileDB.h"
#include <algorithm> #include <algorithm>
#include <set>
#include <cstring> #include <cstring>
#include <map> #include <map>
#include "LinkedObjectFileCreation.h" #include "LinkedObjectFileCreation.h"
@ -24,6 +25,39 @@ std::string ObjectFileRecord::to_unique_name() const {
return name + "-v" + std::to_string(version); return name + "-v" + std::to_string(version);
} }
std::string ObjectFileData::to_unique_name() const {
if (has_multiple_versions) {
std::string result = record.name + "-";
auto dgo_names_sorted = dgo_names;
std::sort(dgo_names_sorted.begin(), dgo_names_sorted.end());
for (auto x : dgo_names_sorted) {
auto ext = x.substr(x.length() - 4, 4);
if (ext == ".CGO" || ext == ".cgo" || ext == ".DGO" || ext == ".dgo") {
x = x.substr(0, x.length() - 4);
}
result += x + "-";
}
result.pop_back();
return result;
} else {
return record.name;
}
}
ObjectFileData& ObjectFileDB::lookup_record(ObjectFileRecord rec) {
ObjectFileData* result = nullptr;
for (auto& x : obj_files_by_name[rec.name]) {
if (x.record.version == rec.version) {
assert(x.record.hash == rec.hash);
assert(!result);
result = &x;
}
}
assert(result);
return *result;
}
/*! /*!
* Build an object file DB for the given list of DGOs. * Build an object file DB for the given list of DGOs.
*/ */
@ -69,6 +103,40 @@ void assert_string_empty_after(const char* str, int size) {
} }
} // namespace } // namespace
namespace {
std::string get_object_file_name(const std::string& original_name, uint8_t* data, int size) {
const char art_group_text[] =
"/src/next/data/art-group6/"; // todo, this may change in other games
const char suffix[] = "-ag.go";
int len = int(strlen(art_group_text));
for (int start = 0; start < size; start++) {
bool failed = false;
for (int i = 0; i < len; i++) {
if (start + i >= size || data[start + i] != art_group_text[i]) {
failed = true;
break;
}
}
if (!failed) {
for (int i = 0; i < int(original_name.length()); i++) {
if (start + len + i >= size || data[start + len + i] != original_name[i]) {
assert(false);
}
}
assert(int(strlen(suffix)) + start + len + int(original_name.length()) < size);
assert(!memcmp(data + start + len + original_name.length(), suffix, strlen(suffix) + 1));
return original_name + "-ag";
}
}
return original_name;
}
} // namespace
constexpr int MAX_CHUNK_SIZE = 0x8000; constexpr int MAX_CHUNK_SIZE = 0x8000;
/*! /*!
* Load the objects stored in the given DGO into the ObjectFileDB * Load the objects stored in the given DGO into the ObjectFileDB
@ -141,7 +209,9 @@ void ObjectFileDB::get_objs_from_dgo(const std::string& filename) {
assert(reader.bytes_left() >= obj_header.size); assert(reader.bytes_left() >= obj_header.size);
assert_string_empty_after(obj_header.name, 60); assert_string_empty_after(obj_header.name, 60);
add_obj_from_dgo(obj_header.name, reader.here(), obj_header.size, dgo_base_name); auto name = get_object_file_name(obj_header.name, reader.here(), obj_header.size);
add_obj_from_dgo(name, obj_header.name, reader.here(), obj_header.size, dgo_base_name);
reader.ffwd(obj_header.size); reader.ffwd(obj_header.size);
} }
@ -153,39 +223,63 @@ void ObjectFileDB::get_objs_from_dgo(const std::string& filename) {
* Add an object file to the ObjectFileDB * Add an object file to the ObjectFileDB
*/ */
void ObjectFileDB::add_obj_from_dgo(const std::string& obj_name, void ObjectFileDB::add_obj_from_dgo(const std::string& obj_name,
const std::string& name_in_dgo,
uint8_t* obj_data, uint8_t* obj_data,
uint32_t obj_size, uint32_t obj_size,
const std::string& dgo_name) { const std::string& dgo_name) {
stats.total_obj_files++; stats.total_obj_files++;
assert(obj_size > 128);
uint16_t version = *(uint16_t*)(obj_data + 8);
auto hash = crc32(obj_data, obj_size); auto hash = crc32(obj_data, obj_size);
bool duplicated = false;
// first, check to see if we already got it... // first, check to see if we already got it...
for (auto& e : obj_files_by_name[obj_name]) { for (auto& e : obj_files_by_name[obj_name]) {
if (e.data.size() == obj_size && e.record.hash == hash) { if (e.data.size() == obj_size && e.record.hash == hash) {
// just to make sure we don't have a hash collision.
assert(!memcmp(obj_data, e.data.data(), obj_size));
// already got it! // already got it!
e.reference_count++; e.reference_count++;
auto rec = e.record; auto& rec = e.record;
assert(name_in_dgo == e.name_in_dgo);
e.dgo_names.push_back(dgo_name);
obj_files_by_dgo[dgo_name].push_back(rec); obj_files_by_dgo[dgo_name].push_back(rec);
return; duplicated = true;
break;
} else {
e.has_multiple_versions = true;
} }
} }
if (duplicated) {
return;
}
// nope, have to add a new one. // nope, have to add a new one.
ObjectFileData data; ObjectFileData data;
data.data.resize(obj_size); data.data.resize(obj_size);
memcpy(data.data.data(), obj_data, obj_size); memcpy(data.data.data(), obj_data, obj_size);
data.record.hash = hash; data.record.hash = hash;
data.record.name = obj_name; data.record.name = obj_name;
data.dgo_names.push_back(dgo_name);
if (obj_files_by_name[obj_name].empty()) { if (obj_files_by_name[obj_name].empty()) {
// if this is the first time we've seen this object file name, add it in the order. // if this is the first time we've seen this object file name, add it in the order.
obj_file_order.push_back(obj_name); obj_file_order.push_back(obj_name);
} }
data.record.version = obj_files_by_name[obj_name].size(); data.record.version = obj_files_by_name[obj_name].size();
data.name_in_dgo = name_in_dgo;
data.obj_version = version;
obj_files_by_dgo[dgo_name].push_back(data.record); obj_files_by_dgo[dgo_name].push_back(data.record);
obj_files_by_name[obj_name].emplace_back(std::move(data)); obj_files_by_name[obj_name].emplace_back(std::move(data));
stats.unique_obj_files++; stats.unique_obj_files++;
stats.unique_obj_bytes += obj_size; stats.unique_obj_bytes += obj_size;
if (obj_files_by_name[obj_name].size() > 1) {
for (auto& e : obj_files_by_name[obj_name]) {
e.has_multiple_versions = true;
}
}
} }
/*! /*!
@ -202,8 +296,13 @@ std::string ObjectFileDB::generate_dgo_listing() {
for (const auto& name : dgo_names) { for (const auto& name : dgo_names) {
result += "(\"" + name + "\"\n"; result += "(\"" + name + "\"\n";
for (auto& obj : obj_files_by_dgo[name]) { for (auto& obj_rec : obj_files_by_dgo[name]) {
result += " " + obj.name + " :version " + std::to_string(obj.version) + "\n"; auto obj = lookup_record(obj_rec);
std::string extension = ".o";
if (obj.obj_version == 4 || obj.obj_version == 2) {
extension = ".go";
}
result += " (\"" + obj.to_unique_name() + extension + "\" \"" + obj.name_in_dgo + "\")\n";
} }
result += " )\n\n"; result += " )\n\n";
} }
@ -211,6 +310,25 @@ std::string ObjectFileDB::generate_dgo_listing() {
return result; return result;
} }
std::string ObjectFileDB::generate_obj_listing() {
std::string result;
std::set<std::string> all_unique_names;
int unique_count = 0;
for (auto& obj_file : obj_file_order) {
for (auto& x : obj_files_by_name.at(obj_file)) {
result +=
x.to_unique_name() + +", " + x.name_in_dgo + ", " + std::to_string(x.obj_version) + "\n";
unique_count++;
all_unique_names.insert(x.to_unique_name());
}
}
// this check is extremely important. It makes sure we don't have any repeat names. This could
// be caused by two files with the same name, in the same DGOs, but different data.
assert(int(all_unique_names.size()) == unique_count);
return result;
}
/*! /*!
* Process all of the linking data of all objects. * Process all of the linking data of all objects.
*/ */

View file

@ -31,6 +31,11 @@ struct ObjectFileData {
std::vector<uint8_t> data; // raw bytes std::vector<uint8_t> data; // raw bytes
LinkedObjectFile linked_data; // data including linking annotations LinkedObjectFile linked_data; // data including linking annotations
ObjectFileRecord record; // name ObjectFileRecord record; // name
std::vector<std::string> dgo_names;
int obj_version = -1;
bool has_multiple_versions = false;
std::string name_in_dgo;
std::string to_unique_name() const;
uint32_t reference_count = 0; // number of times its used. uint32_t reference_count = 0; // number of times its used.
}; };
@ -38,6 +43,7 @@ class ObjectFileDB {
public: public:
ObjectFileDB(const std::vector<std::string>& _dgos); ObjectFileDB(const std::vector<std::string>& _dgos);
std::string generate_dgo_listing(); std::string generate_dgo_listing();
std::string generate_obj_listing();
void process_link_data(); void process_link_data();
void process_labels(); void process_labels();
void find_code(); void find_code();
@ -46,10 +52,12 @@ class ObjectFileDB {
void write_object_file_words(const std::string& output_dir, bool dump_v3_only); void write_object_file_words(const std::string& output_dir, bool dump_v3_only);
void write_disassembly(const std::string& output_dir, bool disassemble_objects_without_functions); void write_disassembly(const std::string& output_dir, bool disassemble_objects_without_functions);
void analyze_functions(); void analyze_functions();
ObjectFileData& lookup_record(ObjectFileRecord rec);
private: private:
void get_objs_from_dgo(const std::string& filename); void get_objs_from_dgo(const std::string& filename);
void add_obj_from_dgo(const std::string& obj_name, void add_obj_from_dgo(const std::string& obj_name,
const std::string& name_in_dgo,
uint8_t* obj_data, uint8_t* obj_data,
uint32_t obj_size, uint32_t obj_size,
const std::string& dgo_name); const std::string& dgo_name);

View file

@ -27,6 +27,7 @@ int main(int argc, char** argv) {
ObjectFileDB db(dgos); ObjectFileDB db(dgos);
write_text_file(combine_path(out_folder, "dgo.txt"), db.generate_dgo_listing()); write_text_file(combine_path(out_folder, "dgo.txt"), db.generate_dgo_listing());
write_text_file(combine_path(out_folder, "obj.txt"), db.generate_obj_listing());
db.process_link_data(); db.process_link_data();
db.find_code(); db.find_code();

1092
goal_src/build/all_objs.txt Normal file

File diff suppressed because it is too large Load diff

1909
goal_src/build/dgos.txt Normal file

File diff suppressed because it is too large Load diff