refactor(object, utilities): moved diff saving and reading to the object files

This commit is contained in:
lisk77 2025-09-19 19:22:40 +02:00
parent 13e3ba70b2
commit 4b2575ab96
4 changed files with 378 additions and 402 deletions

View file

@ -114,4 +114,377 @@ void* parse_object(char* hash, ObjectType type, size_t* size_out) {
default: free(content); return NULL;
}
}
static int mkdir_recursive(const char *path, mode_t mode) {
char *path_copy = strdup(path);
if (!path_copy) return -1;
char *p = path_copy;
if (*p == '/') p++;
while (*p) {
while (*p && *p != '/') p++;
char saved = *p;
*p = '\0';
if (mkdir(path_copy, mode) == -1 && errno != EEXIST) {
free(path_copy);
return -1;
}
*p = saved;
if (*p) p++;
}
free(path_copy);
return 0;
}
int save_diff(ActionList* diff, char* path, char* root, size_t diff_id, char* basefile_hash, File* modified_file, int tree) {
size_t buffer_size = 41;
buffer_size += 1 + snprintf(NULL, 0, "%d", diff->len);
for (size_t idx = 0; idx < diff->len; idx++) {
Action action = diff->actions[idx];
if (action.type == DELETE) {
// 4 = OPBIT SPACE ... SPACE ... SPACE
buffer_size +=
4 +
snprintf(NULL, 0, "%d", action.line_original) +
snprintf(NULL, 0, "%d", action.line_changed);
continue;
}
// 6 = OPBIT SPACE ... SPACE ... SPACE ... SPACE ... SPACE
buffer_size +=
6 +
snprintf(NULL, 0, "%d", action.line_original) +
snprintf(NULL, 0, "%d", action.line_changed) +
snprintf(NULL, 0, "%zu", strlen(modified_file->content[action.line_changed])) +
strlen(modified_file->content[action.line_changed]);
}
char tmp[buffer_size];
size_t offset = 0;
offset += snprintf(tmp, sizeof(tmp), "%s %zu ", basefile_hash, diff->len);
for (size_t idx = 0; idx < diff->len; idx++) {
Action action = diff->actions[idx];
size_t remaining = sizeof(tmp) - offset;
if (action.type == DELETE) {
offset += snprintf(tmp + offset,
remaining,
"0 %zu %zu ",
action.line_original,
action.line_changed
);
continue;
}
offset += snprintf(tmp + offset, remaining,
"1 %zu %zu %zu %s ",
action.line_original,
action.line_changed,
strlen(modified_file->content[action.line_changed]),
modified_file->content[action.line_changed]
);
}
char id[2+snprintf(NULL, 0, "%d", diff_id)+strlen(path)];
snprintf(id, sizeof(id), "%s %d", path, diff_id);
char hash[41];
if (tree) object_hash(TreeDiffObject, id, hash);
else object_hash(FileDiffObject, id, hash);
char dir_path[PATH_MAX];
char file_path[PATH_MAX];
snprintf(dir_path, sizeof(dir_path), "%s/.merk/objects/%.2s", root, hash);
mkdir(dir_path, 0755);
snprintf(file_path, sizeof(file_path), "%s/%s", dir_path, hash+2);
if (access(file_path, F_OK) == 0) {
return 1;
}
FILE* fp = fopen(file_path, "wb");
if (!fp) {
perror("ERROR: cannot open path in save_diff!\n");
return 0;
}
// Ensure the directory structure exists
if (mkdir_recursive(dir_path, 0755) < 0 && errno != EEXIST) {
perror("ERROR: failed to create directory structure in save_diff!");
return 0;
}
uLong originalLen = strlen(tmp) + 1;
uLong compressedLen = compressBound(originalLen);
Bytef* compressed = malloc(compressedLen);
if (!compressed) {
fclose(fp);
return 0;
}
if (compress(compressed, &compressedLen, (const Bytef*)tmp, originalLen) != Z_OK) {
perror("ERROR: compression failed in save_diff!");
free(compressed);
fclose(fp);
return 0;
}
fprintf(fp, "%lu ", (unsigned long)originalLen);
fwrite(compressed, 1, compressedLen, fp);
fclose(fp);
chmod(file_path, S_IRUSR | S_IRGRP | S_IROTH);
return 1;
}
int read_diff(char* content, char* basefile_hash, ActionList* diff_out) {
if (!content || !basefile_hash || !diff_out) {
return 0;
}
size_t idx = 0;
strncpy(basefile_hash, content, 40);
basefile_hash[40] = '\0';
idx += 41;
while (content[idx] && IS_DIGIT(content[idx])) {
idx++;
}
if (idx == 41) {
perror("ERROR: no length found at start of diff!");
return 0;
}
char* number_of_actions = calloc(idx - 41 + 1, sizeof(char));
if (!number_of_actions) {
return 0;
}
memcpy(number_of_actions, content + 41, idx - 41);
number_of_actions[idx - 41] = '\0';
char* end;
long actions = strtol(number_of_actions, &end, 10);
free(number_of_actions);
if (end == number_of_actions || *end != '\0' || actions <= 0) {
perror("ERROR: invalid number of actions in read_diff!");
return 0;
}
size_t action_idx = 0;
while (content[idx] && action_idx < (size_t)actions) {
idx++;
Action action = {0};
if (content[idx] == '0') {
idx++;
if (content[idx] != ' ') {
perror("ERROR: expected space after opbit in read_diff!");
return 0;
}
idx++;
char* endptr;
action.line_original = strtol(content + idx, &endptr, 10);
if (endptr == content + idx || *endptr != ' ') {
perror("ERROR: failed to parse line_original in read_diff!");
return 0;
}
idx = endptr - content + 1;
action.line_changed = strtol(content + idx, &endptr, 10);
if (endptr == content + idx || (*endptr != ' ' && *endptr != '\0')) {
perror("ERROR: failed to parse line_changed in read_diff!");
return 0;
}
idx = endptr - content;
} else if (content[idx] == '1') {
idx++;
if (content[idx] != ' ') {
perror("ERROR: expected space after opbit in read_diff!");
return 0;
}
idx++;
char* endptr;
action.line_original = strtol(content + idx, &endptr, 10);
if (endptr == content + idx || *endptr != ' ') {
perror("ERROR: failed to parse line_original in read_diff!");
return 0;
}
idx = endptr - content + 1;
action.line_changed = strtol(content + idx, &endptr, 10);
if (endptr == content + idx || *endptr != ' ') {
perror("ERROR: failed to parse line_changed in read_diff!");
return 0;
}
idx = endptr - content + 1;
size_t content_length = strtol(content + idx, &endptr, 10);
if (endptr == content + idx || *endptr != ' ') {
perror("ERROR: failed to parse content length in read_diff!");
return 0;
}
idx = endptr - content + 1;
action.content = strndup(content + idx, content_length);
if (!action.content) {
perror("ERROR: failed to allocate memory for action content in read_diff!");
return 0;
}
idx += content_length;
} else {
perror("ERROR: unknown action type in read_diff!");
return 0;
}
add_action(diff_out, action);
action_idx++;
}
return 1;
}
int save_file_diff(char* path, char* root, size_t diff_id, char* basefile_hash, Changes* changes) {
char basefile_location[2+strlen(root)+strlen("/.merk/objects/")+strlen(basefile_hash)];
snprintf(basefile_location, sizeof(basefile_location), "%s/.merk/objects/%.2s/%s", root, basefile_hash, basefile_hash+2);
size_t compressed_size;
unsigned char* compressed_data = (unsigned char*)get_file_content_with_size(basefile_location, &compressed_size);
if (!compressed_data) return 0;
size_t idx = 0;
while (idx < compressed_size && IS_DIGIT(compressed_data[idx])) {
idx++;
}
if (idx == 0) {
perror("ERROR: no length found at start of save_file_diff!");
free(compressed_data);
return 0;
}
char* size_str = calloc(idx + 1, sizeof(char));
if (!size_str) {
free(compressed_data);
return 0;
}
memcpy(size_str, compressed_data, idx);
size_str[idx] = '\0';
char* end;
long size = strtol(size_str, &end, 10);
if (end == size_str || *end != '\0') {
perror("ERROR: invalid length in base_file_list");
free(size_str);
free(compressed_data);
return 0;
}
size_t offset = strlen(size_str) + 1;
free(size_str);
char* basefile_content = calloc(size+1, sizeof(char));
if (!basefile_content) {
free(compressed_data);
return 0;
}
uLongf dest_len = (uLongf)size;
int result = uncompress((unsigned char*)basefile_content, &dest_len, compressed_data+offset, compressed_size);
free(compressed_data);
if (result != Z_OK) {
perror("ERROR: decompression of the base file list failed!");
free(basefile_content);
return 0;
}
basefile_content[dest_len] = '\0';
File* basefile = from_string(basefile_content);
char* modified_file_path = realpath(path, NULL);
File* modified_file = new_file(modified_file_path);
ActionList* diff = myers_diff(basefile, modified_file, 0, 0);
if (diff->len == 0) {
free(basefile_content);
return 1;
}
for (size_t idx = 0; idx < diff->len; idx++) {
Action action = diff->actions[idx];
if (action.type == INSERT) {
changes->insertions += 1;
} else if (action.type == DELETE) {
changes->deletions += 1;
}
}
save_diff(diff, path, root, diff_id, basefile_hash, modified_file, 0);
free(basefile_content);
return 1;
}
File* apply_diff(File* basefile, ActionList* diff) {
if (!basefile || !diff) return NULL;
File* copy = copy_file(basefile);
if (!copy) return NULL;
sort_action_list(diff);
for (size_t i = 0; i < diff->len; i++) {
Action* action = &diff->actions[i];
int success = 0;
switch (action->type) {
case INSERT:
success = insert_line(copy, action->content, action->line_changed);
if (!success) {
fprintf(stderr, "ERROR: Failed to insert line at index %zu\n", action->line_changed);
free_file(copy);
return NULL;
}
break;
case DELETE:
success = delete_line(copy, action->line_original);
if (!success) {
fprintf(stderr, "ERROR: Failed to delete line at index %zu\n", action->line_original);
free_file(copy);
return NULL;
}
break;
default:
fprintf(stderr, "ERROR: Unknown action type %d\n", action->type);
free_file(copy);
return NULL;
}
}
return copy;
}