Fix caching on systems with bad system time

meson.build

@@ -1,4 +1,4 @@
-project('layeredfs', 'c', 'cpp', version: '3.2',
+project('layeredfs', 'c', 'cpp', version: '3.3',
     default_options: [
         'cpp_std=c++17',
         'buildtype=release',

src/avs.h

@@ -205,7 +205,7 @@ X(int,        property_query_size, property_t prop) \
 X(void,       property_destroy, property_t prop) \
 /* md5sum *sha1 if needed) */ \
 X(mdigest_p,  mdigest_create, mdigest_algs_t algorithm) \
-X(void,       mdigest_update, mdigest_p digest, const char* data, int size) \
+X(void,       mdigest_update, mdigest_p digest, const void* data, int size) \
 X(void,       mdigest_finish, mdigest_p digest, uint8_t* hash, int size) \
 X(void,       mdigest_destroy, mdigest_p digest) \
 /* compression */ \

src/hook.cpp

@@ -220,34 +220,24 @@ void handle_texbin(HookFile &file) {
         return;
     }
 
+    auto starting = file.get_path_to_open();
     string out = CACHE_FOLDER "/" + file.norm_path;
     auto out_hashed = out + ".hashed";
+    auto cache_hasher = CacheHasher(out_hashed);
 
-    uint8_t hash[MD5_LEN];
-    hash_filenames(pngs_list, hash);
-
-    uint8_t cache_hash[MD5_LEN] = {0};
-    FILE* cache_hashfile;
-    cache_hashfile = fopen(out_hashed.c_str(), "rb");
-    if (cache_hashfile) {
-        fread(cache_hash, 1, sizeof(cache_hash), cache_hashfile);
-        fclose(cache_hashfile);
+    cache_hasher.add(starting);
+    for (auto &path : pngs_list) {
+        cache_hasher.add(path);
     }
+    cache_hasher.finish();
 
-    auto time_out = file_time(out.c_str());
-    auto newest = file_time(file.get_path_to_open().c_str());
-    for (auto &path : pngs_list)
-        newest = std::max(newest, file_time(path.c_str()));
     // no need to merge - timestamps all up to date, dll not newer, files haven't been deleted
-    if(time_out >= newest && time_out >= dll_time && memcmp(hash, cache_hash, sizeof(hash)) == 0) {
+    if(cache_hasher.matches()) {
         file.mod_path = out;
         log_misc("texbin cache up to date, skip");
         return;
     }
-    // log_verbose("Regenerating cache");
-    // log_verbose("  time_out >= newest == %d", time_out >= newest);
-    // log_verbose("  time_out >= dll_time == %d", time_out >= dll_time);
-    // log_verbose("  memcmp(hash, cache_hash, sizeof(hash)) == 0 == %d", memcmp(hash, cache_hash, sizeof(hash)) == 0);
+    log_verbose("Regenerating cache");
 
     Texbin texbin;
     auto _orig_data = file.load_to_vec();
@@ -284,11 +274,7 @@ void handle_texbin(HookFile &file) {
         return;
     }
 
-    cache_hashfile = fopen(out_hashed.c_str(), "wb");
-    if (cache_hashfile) {
-        fwrite(hash, 1, sizeof(hash), cache_hashfile);
-        fclose(cache_hashfile);
-    }
+    cache_hasher.commit();
     file.mod_path = out;
 
     log_misc("Texbin generation took %d ms", time() - start);

src/imagefs.cpp

@@ -458,24 +458,16 @@ void merge_xmls(HookFile &file) {
     auto starting = file.get_path_to_open();
     out = CACHE_FOLDER "/" + file.norm_path;
     auto out_hashed = out + ".hashed";
+    auto cache_hasher = CacheHasher(out_hashed);
 
-    uint8_t hash[MD5_LEN];
-    hash_filenames(to_merge, hash);
-
-    uint8_t cache_hash[MD5_LEN] = {0};
-    FILE* cache_hashfile = fopen(out_hashed.c_str(), "rb");
-    if (cache_hashfile) {
-        fread(cache_hash, 1, sizeof(cache_hash), cache_hashfile);
-        fclose(cache_hashfile);
+    cache_hasher.add(starting); // don't forget to take the input into account
+    for (auto &path : to_merge) {
+        cache_hasher.add(path);
     }
+    cache_hasher.finish();
 
-    auto time_out = file_time(out.c_str());
-    // don't forget to take the input into account
-    auto newest = file_time(starting.c_str());
-    for (auto &path : to_merge)
-        newest = std::max(newest, file_time(path.c_str()));
     // no need to merge - timestamps all up to date, dll not newer, files haven't been deleted
-    if(time_out >= newest && time_out >= dll_time && memcmp(hash, cache_hash, sizeof(hash)) == 0) {
+    if(cache_hasher.matches()) {
         file.mod_path = out;
         return;
     }
@@ -511,11 +503,7 @@ void merge_xmls(HookFile &file) {
     }
 
     rapidxml_dump_to_file(out, merged_xml);
-    cache_hashfile = fopen(out_hashed.c_str(), "wb");
-    if (cache_hashfile) {
-        fwrite(hash, 1, sizeof(hash), cache_hashfile);
-        fclose(cache_hashfile);
-    }
+    cache_hasher.commit();
     file.mod_path = out;
 
     log_misc("Merge took %d ms", time() - start);

src/utils.cpp

@@ -6,6 +6,7 @@
 #include "utils.hpp"
 #include "log.hpp"
 #include "avs.h"
+#include "hook.h"
 
 char* snprintf_auto(const char* fmt, ...) {
     va_list argList;
@@ -198,13 +199,42 @@ string basename_without_extension(string const & path) {
     return p > 0 && p != string::npos ? basename.substr(0, p) : basename;
 }
 
-void hash_filenames(std::vector<std::string> &filenames, uint8_t hash[MD5_LEN]) {
-    auto digest = mdigest_create(MD5);
+CacheHasher::CacheHasher(std::string hash_file): hash_file(hash_file) {
+    digest = mdigest_create(MD5);
 
-    for (auto &path : filenames) {
-        mdigest_update(digest, path.c_str(), (int)path.length());
+    // always hash the DLL time
+    mdigest_update(digest, &dll_time, sizeof(dll_time));
+
+    auto cache_hashfile = fopen(hash_file.c_str(), "rb");
+    if (cache_hashfile) {
+        fread(existing_hash, 1, MD5_LEN, cache_hashfile);
+        fclose(cache_hashfile);
     }
+}
 
-    mdigest_finish(digest, hash, MD5_LEN);
+CacheHasher::~CacheHasher() {
     mdigest_destroy(digest);
 }
+
+void CacheHasher::add(std::string &path) {
+    mdigest_update(digest, path.c_str(), (int)path.length());
+
+    auto ts = file_time(path.c_str());
+    mdigest_update(digest, &ts, sizeof(ts));
+}
+
+void CacheHasher::finish() {
+    mdigest_finish(digest, new_hash, MD5_LEN);
+}
+
+bool CacheHasher::matches() {
+    return memcmp(new_hash, existing_hash, sizeof(new_hash)) == 0;
+}
+
+void CacheHasher::commit() {
+    auto cache_hashfile = fopen(hash_file.c_str(), "wb");
+    if (cache_hashfile) {
+        fwrite(new_hash, 1, sizeof(new_hash), cache_hashfile);
+        fclose(cache_hashfile);
+    }
+}

src/utils.hpp

@@ -28,7 +28,28 @@ std::vector<std::string> folders_in_folder(const char* root);
 uint64_t file_time(const char* path);
 LONG time(void);
 std::string basename_without_extension(std::string const & path);
-void hash_filenames(std::vector<std::string> &filenames, uint8_t hash[MD5_LEN]);
+
+// Hashes the names and timestamps of input files into a rebuilt output.
+// Invalidates on DLL timestamp change, input timestamp change, or input change
+class CacheHasher {
+    public:
+    CacheHasher(std::string hash_file);
+    ~CacheHasher();
+    // add a path and its timestamp to the hash. Should not be called after `finish`
+    void add(std::string &path);
+    // complete the hashing op
+    void finish();
+    // check if the hashfile matches
+    bool matches();
+    // write out an updated hashfile. Should be called after `finish`
+    void commit();
+
+    private:
+    std::string hash_file;
+    mdigest_p digest;
+    uint8_t existing_hash[MD5_LEN] = {0};
+    uint8_t new_hash[MD5_LEN] = {0};
+};
 
 struct CaseInsensitiveCompare {
     bool operator() (const std::string& a, const std::string& b) const {