no-DB mode implementation #94

Manually merged
jabber.developer merged 34 commits from feat/no-db-mode into master 2026-05-05 19:32:42 +00:00
34 changed files with 8579 additions and 711 deletions
Showing only changes of commit 7ba7e53291 - Show all commits

View File

@@ -6729,6 +6729,54 @@ cmd_logging(ProfWin* window, const char* const command, gchar** args)
return TRUE;
}
static void
_show_integrity_issues(GSList* issues)
{
int errors = 0, warnings = 0, infos = 0;
for (GSList* l = issues; l; l = l->next) {
integrity_issue_t* issue = l->data;
const char* level_str;
switch (issue->level) {
case INTEGRITY_ERROR:
level_str = "ERROR";
jabber.developer2 marked this conversation as resolved Outdated

It future, this requirement should be removed as it doesn't make much sense. Offline capabilities are really lacking. But I see why it exists (local SQLite is getting disconnected on disconnect).

Feel free to click "resolve" on this comment.

It future, this requirement should be removed as it doesn't make much sense. Offline capabilities are really lacking. But I see why it exists (local SQLite is getting disconnected on disconnect). Feel free to click "resolve" on this comment.
errors++;
break;
case INTEGRITY_WARNING:
level_str = "WARN";
warnings++;
break;
case INTEGRITY_INFO:
level_str = "INFO";
infos++;
jabber.developer2 marked this conversation as resolved Outdated

Nice nesting: everything is flat. Easy to read, easy to maintain.

Resolveable.

Nice nesting: everything is flat. Easy to read, easy to maintain. Resolveable.
break;
default:
level_str = "???";
break;
}
if (issue->line > 0) {
if (issue->level == INTEGRITY_ERROR) {
cons_show_error("[%s] %s:%d — %s", level_str, issue->file, issue->line, issue->message);
} else {
cons_show("[%s] %s:%d — %s", level_str, issue->file, issue->line, issue->message);
}
} else {
if (issue->level == INTEGRITY_ERROR) {
cons_show_error("[%s] %s — %s", level_str, issue->file, issue->message);
} else {
cons_show("[%s] %s — %s", level_str, issue->file, issue->message);
}
}
}
if (!issues) {
cons_show("Verification complete: no issues found.");
} else {
cons_show("Verification complete: %d error(s), %d warning(s), %d info(s).",
errors, warnings, infos);
g_slist_free_full(issues, (GDestroyNotify)integrity_issue_free);
}
}
gboolean
cmd_history(ProfWin* window, const char* const command, gchar** args)
{
@@ -6773,50 +6821,7 @@ cmd_history(ProfWin* window, const char* const command, gchar** args)
cons_show("Verifying history integrity...");
GSList* issues = log_database_verify_integrity(contact_jid);
int errors = 0, warnings = 0, infos = 0;
for (GSList* l = issues; l; l = l->next) {
integrity_issue_t* issue = l->data;
const char* level_str;
switch (issue->level) {
case INTEGRITY_ERROR:
level_str = "ERROR";
errors++;
break;
case INTEGRITY_WARNING:
level_str = "WARN";
warnings++;
break;
case INTEGRITY_INFO:
level_str = "INFO";
infos++;
break;
default:
level_str = "???";
break;
}
if (issue->line > 0) {
if (issue->level == INTEGRITY_ERROR) {
cons_show_error("[%s] %s:%d — %s", level_str, issue->file, issue->line, issue->message);
} else {
cons_show("[%s] %s:%d — %s", level_str, issue->file, issue->line, issue->message);
}
} else {
if (issue->level == INTEGRITY_ERROR) {
cons_show_error("[%s] %s — %s", level_str, issue->file, issue->message);
} else {
cons_show("[%s] %s — %s", level_str, issue->file, issue->message);
}
}
}
if (!issues) {
cons_show("Verification complete: no issues found.");
} else {
cons_show("Verification complete: %d error(s), %d warning(s), %d info(s).",
errors, warnings, infos);
g_slist_free_full(issues, (GDestroyNotify)integrity_issue_free);
}
_show_integrity_issues(issues);
return TRUE;
}

View File

@@ -195,6 +195,190 @@ typedef struct
int skipped; // counter
} export_ctx_t;
// Export a single contact's messages from SQLite to flatfile.
// Returns the number of newly exported messages, or -1 on error/skip.
static int
_export_one_contact(const char* contact)
{
// 1. Read existing flatfile lines for dedup
GHashTable* seen_keys = g_hash_table_new_full(g_str_hash, g_str_equal, g_free, NULL);
GSList* existing = _ff_read_all_lines(contact);
for (GSList* l = existing; l; l = l->next) {
ff_parsed_line_t* pl = l->data;
char* key = _make_dedup_key(pl->stanza_id, pl->timestamp_str, pl->from_jid, pl->message);
g_hash_table_add(seen_keys, key);
}
int existing_count = g_slist_length(existing);
// 2. Query SQLite for this contact — get ALL messages via direct SQL
GSList* sqlite_lines = db_sqlite_get_all_chat(contact);
if (!sqlite_lines) {
log_debug("export: no SQLite messages for %s", contact);
g_hash_table_destroy(seen_keys);
g_slist_free_full(existing, (GDestroyNotify)ff_parsed_line_free);
return -1;
}
// 3. Merge: write existing flatfile lines + new SQLite lines to temp file
auto_gchar gchar* log_path = ff_get_log_path(contact);
if (!log_path) {
g_hash_table_destroy(seen_keys);
g_slist_free_full(existing, (GDestroyNotify)ff_parsed_line_free);
g_slist_free_full(sqlite_lines, (GDestroyNotify)message_free);
return -1;
}
// Ensure directory
auto_gchar gchar* dir = g_path_get_dirname(log_path);
ff_ensure_dir(dir);
auto_gchar gchar* tmp_path = g_strdup_printf("%s.export.tmp", log_path);
// Use open() with O_NOFOLLOW|O_EXCL to prevent symlink attacks,
// and explicit 0600 permissions (chat history is private).
int tmp_fd = open(tmp_path, O_WRONLY | O_CREAT | O_EXCL | O_NOFOLLOW, S_IRUSR | S_IWUSR);
if (tmp_fd < 0) {
if (errno == EEXIST) {
// Stale temp file from previous failed export — remove and retry
unlink(tmp_path);
tmp_fd = open(tmp_path, O_WRONLY | O_CREAT | O_EXCL | O_NOFOLLOW, S_IRUSR | S_IWUSR);
}
}
if (tmp_fd < 0) {
log_error("export: cannot create %s: %s", tmp_path, strerror(errno));
g_hash_table_destroy(seen_keys);
g_slist_free_full(existing, (GDestroyNotify)ff_parsed_line_free);
g_slist_free_full(sqlite_lines, (GDestroyNotify)message_free);
return -1;
}
FILE* fp = fdopen(tmp_fd, "w");
if (!fp) {
log_error("export: fdopen failed for %s: %s", tmp_path, strerror(errno));
close(tmp_fd);
unlink(tmp_path);
g_hash_table_destroy(seen_keys);
g_slist_free_full(existing, (GDestroyNotify)ff_parsed_line_free);
g_slist_free_full(sqlite_lines, (GDestroyNotify)message_free);
return -1;
}
// Lock the temp file
int fd = fileno(fp);
if (flock(fd, LOCK_EX) != 0) {
log_warning("export: flock(%s) failed: %s", tmp_path, strerror(errno));
}
fprintf(fp, "%s", FLATFILE_HEADER);
int contact_exported = 0;
int contact_skipped = 0;
// Collect ALL lines into a merged list for sorted output
GSList* merged = NULL;
// Add existing flatfile lines (mark as already seen in dedup)
for (GSList* l = existing; l; l = l->next) {
ff_parsed_line_t* pl = l->data;
// Create a shallow copy for the merged list (originals freed separately)
ff_parsed_line_t* copy = g_malloc0(sizeof(ff_parsed_line_t));
copy->timestamp_str = g_strdup(pl->timestamp_str);
copy->timestamp = pl->timestamp ? g_date_time_ref(pl->timestamp) : NULL;
copy->type = g_strdup(pl->type);
copy->enc = g_strdup(pl->enc);
copy->stanza_id = g_strdup(pl->stanza_id);
copy->archive_id = g_strdup(pl->archive_id);
copy->replace_id = g_strdup(pl->replace_id);
copy->from_jid = g_strdup(pl->from_jid);
copy->from_resource = g_strdup(pl->from_resource);
copy->to_jid = g_strdup(pl->to_jid);
copy->to_resource = g_strdup(pl->to_resource);
copy->marked_read = pl->marked_read;
copy->message = g_strdup(pl->message);
merged = g_slist_prepend(merged, copy);
}
// Add SQLite messages that aren't already in the flatfile
for (GSList* l = sqlite_lines; l; l = l->next) {
ProfMessage* msg = l->data;
if (!msg || !msg->timestamp)
continue;
auto_gchar gchar* ts = g_date_time_format_iso8601(msg->timestamp);
const char* from_jid = msg->from_jid ? msg->from_jid->barejid : "unknown";
const char* body = msg->plain ? msg->plain : "";
const char* sid = msg->id ? msg->id : "";
char* key = _make_dedup_key(sid, ts, from_jid, body);
if (g_hash_table_contains(seen_keys, key)) {
g_free(key);
contact_skipped++;
continue;
}
g_hash_table_add(seen_keys, key);
ff_parsed_line_t* pl = g_malloc0(sizeof(ff_parsed_line_t));
pl->timestamp_str = g_strdup(ts);
pl->timestamp = g_date_time_ref(msg->timestamp);
pl->type = g_strdup(ff_get_message_type_str(msg->type));
pl->enc = g_strdup(ff_get_message_enc_str(msg->enc));
pl->stanza_id = g_strdup(sid);
pl->archive_id = msg->stanzaid ? g_strdup(msg->stanzaid) : NULL;
pl->replace_id = NULL;
pl->from_jid = g_strdup(from_jid);
pl->from_resource = msg->from_jid ? g_strdup(msg->from_jid->resourcepart) : NULL;
pl->to_jid = msg->to_jid ? g_strdup(msg->to_jid->barejid) : NULL;
pl->to_resource = msg->to_jid ? g_strdup(msg->to_jid->resourcepart) : NULL;
pl->marked_read = msg->marked_read;
pl->message = g_strdup(body);
merged = g_slist_prepend(merged, pl);
contact_exported++;
}
jabber.developer marked this conversation as resolved Outdated

why g_free if key is char? Normally, we do this:
g_free for gchar
free for char

why `g_free` if `key` is `char`? Normally, we do this: `g_free` for `gchar` `free` for `char`

Corrected

Corrected
// Sort merged list by timestamp (ISO8601 is lexicographically sortable)
merged = g_slist_sort(merged, _compare_parsed_lines_by_timestamp);
// Write all lines sorted
int written = 0;
for (GSList* l = merged; l; l = l->next) {
ff_parsed_line_t* pl = l->data;
ff_write_line(fp, pl->timestamp_str,
pl->type, pl->enc,
pl->stanza_id, pl->archive_id, pl->replace_id,
pl->from_jid, pl->from_resource,
pl->to_jid, pl->to_resource, pl->marked_read,
pl->message);
written++;
if (written % EXPORT_PROGRESS_INTERVAL == 0) {
cons_show(" ... %s: %d/%d written so far", contact, written, (int)g_slist_length(merged));
}
}
g_slist_free_full(merged, (GDestroyNotify)ff_parsed_line_free);
fflush(fp);
fsync(fd); // ensure data is on disk before atomic rename
fclose(fp); // also releases flock
// Atomic rename
int result = -1;
if (rename(tmp_path, log_path) != 0) {
log_error("export: rename %s -> %s failed: %s", tmp_path, log_path, strerror(errno));
cons_show_error("Export failed for %s: could not rename temp file (%s)", contact, strerror(errno));
unlink(tmp_path);
} else {
cons_show("Exported %s: %d new, %d skipped (already existed: %d)",
contact, contact_exported, contact_skipped, existing_count);
result = contact_exported;
}
g_hash_table_destroy(seen_keys);
g_slist_free_full(existing, (GDestroyNotify)ff_parsed_line_free);
g_slist_free_full(sqlite_lines, (GDestroyNotify)message_free);
return result;
}
int
log_database_export_to_flatfile(const gchar* const contact_jid)
{
@@ -234,182 +418,9 @@ log_database_export_to_flatfile(const gchar* const contact_jid)
for (GSList* c = contacts; c; c = c->next) {
const char* contact = c->data;
// 1. Read existing flatfile lines for dedup
GHashTable* seen_keys = g_hash_table_new_full(g_str_hash, g_str_equal, g_free, NULL);
GSList* existing = _ff_read_all_lines(contact);
for (GSList* l = existing; l; l = l->next) {
ff_parsed_line_t* pl = l->data;
char* key = _make_dedup_key(pl->stanza_id, pl->timestamp_str, pl->from_jid, pl->message);
g_hash_table_add(seen_keys, key);
}
int existing_count = g_slist_length(existing);
// 2. Query SQLite for this contact — get ALL messages via direct SQL
GSList* sqlite_lines = db_sqlite_get_all_chat(contact);
if (!sqlite_lines) {
log_debug("export: no SQLite messages for %s", contact);
g_hash_table_destroy(seen_keys);
g_slist_free_full(existing, (GDestroyNotify)ff_parsed_line_free);
continue;
}
// 3. Merge: write existing flatfile lines + new SQLite lines to temp file
auto_gchar gchar* log_path = ff_get_log_path(contact);
if (!log_path) {
g_hash_table_destroy(seen_keys);
g_slist_free_full(existing, (GDestroyNotify)ff_parsed_line_free);
g_slist_free_full(sqlite_lines, (GDestroyNotify)message_free);
continue;
}
// Ensure directory
auto_gchar gchar* dir = g_path_get_dirname(log_path);
ff_ensure_dir(dir);
auto_gchar gchar* tmp_path = g_strdup_printf("%s.export.tmp", log_path);
// Use open() with O_NOFOLLOW|O_EXCL to prevent symlink attacks,
// and explicit 0600 permissions (chat history is private).
int tmp_fd = open(tmp_path, O_WRONLY | O_CREAT | O_EXCL | O_NOFOLLOW, S_IRUSR | S_IWUSR);
if (tmp_fd < 0) {
if (errno == EEXIST) {
// Stale temp file from previous failed export — remove and retry
unlink(tmp_path);
tmp_fd = open(tmp_path, O_WRONLY | O_CREAT | O_EXCL | O_NOFOLLOW, S_IRUSR | S_IWUSR);
}
}
if (tmp_fd < 0) {
log_error("export: cannot create %s: %s", tmp_path, strerror(errno));
g_hash_table_destroy(seen_keys);
g_slist_free_full(existing, (GDestroyNotify)ff_parsed_line_free);
g_slist_free_full(sqlite_lines, (GDestroyNotify)message_free);
continue;
}
FILE* fp = fdopen(tmp_fd, "w");
if (!fp) {
log_error("export: fdopen failed for %s: %s", tmp_path, strerror(errno));
close(tmp_fd);
unlink(tmp_path);
g_hash_table_destroy(seen_keys);
g_slist_free_full(existing, (GDestroyNotify)ff_parsed_line_free);
g_slist_free_full(sqlite_lines, (GDestroyNotify)message_free);
continue;
}
// Lock the temp file
int fd = fileno(fp);
if (flock(fd, LOCK_EX) != 0) {
log_warning("export: flock(%s) failed: %s", tmp_path, strerror(errno));
}
fprintf(fp, "%s", FLATFILE_HEADER);
int contact_exported = 0;
int contact_skipped = 0;
// Collect ALL lines into a merged list for sorted output
GSList* merged = NULL;
// Add existing flatfile lines (mark as already seen in dedup)
for (GSList* l = existing; l; l = l->next) {
ff_parsed_line_t* pl = l->data;
// Create a shallow copy for the merged list (originals freed separately)
ff_parsed_line_t* copy = g_malloc0(sizeof(ff_parsed_line_t));
copy->timestamp_str = g_strdup(pl->timestamp_str);
copy->timestamp = pl->timestamp ? g_date_time_ref(pl->timestamp) : NULL;
copy->type = g_strdup(pl->type);
copy->enc = g_strdup(pl->enc);
copy->stanza_id = g_strdup(pl->stanza_id);
copy->archive_id = g_strdup(pl->archive_id);
copy->replace_id = g_strdup(pl->replace_id);
copy->from_jid = g_strdup(pl->from_jid);
copy->from_resource = g_strdup(pl->from_resource);
copy->to_jid = g_strdup(pl->to_jid);
copy->to_resource = g_strdup(pl->to_resource);
copy->marked_read = pl->marked_read;
copy->message = g_strdup(pl->message);
merged = g_slist_prepend(merged, copy);
}
// Add SQLite messages that aren't already in the flatfile
for (GSList* l = sqlite_lines; l; l = l->next) {
ProfMessage* msg = l->data;
if (!msg || !msg->timestamp)
continue;
auto_gchar gchar* ts = g_date_time_format_iso8601(msg->timestamp);
const char* from_jid = msg->from_jid ? msg->from_jid->barejid : "unknown";
const char* body = msg->plain ? msg->plain : "";
const char* sid = msg->id ? msg->id : "";
char* key = _make_dedup_key(sid, ts, from_jid, body);
if (g_hash_table_contains(seen_keys, key)) {
g_free(key);
contact_skipped++;
continue;
}
g_hash_table_add(seen_keys, key);
ff_parsed_line_t* pl = g_malloc0(sizeof(ff_parsed_line_t));
pl->timestamp_str = g_strdup(ts);
pl->timestamp = g_date_time_ref(msg->timestamp);
pl->type = g_strdup(ff_get_message_type_str(msg->type));
pl->enc = g_strdup(ff_get_message_enc_str(msg->enc));
pl->stanza_id = g_strdup(sid);
pl->archive_id = msg->stanzaid ? g_strdup(msg->stanzaid) : NULL;
pl->replace_id = NULL;
pl->from_jid = g_strdup(from_jid);
pl->from_resource = msg->from_jid ? g_strdup(msg->from_jid->resourcepart) : NULL;
pl->to_jid = msg->to_jid ? g_strdup(msg->to_jid->barejid) : NULL;
pl->to_resource = msg->to_jid ? g_strdup(msg->to_jid->resourcepart) : NULL;
pl->marked_read = msg->marked_read;
pl->message = g_strdup(body);
merged = g_slist_prepend(merged, pl);
contact_exported++;
}
// Sort merged list by timestamp (ISO8601 is lexicographically sortable)
merged = g_slist_sort(merged, _compare_parsed_lines_by_timestamp);
// Write all lines sorted
int written = 0;
for (GSList* l = merged; l; l = l->next) {
ff_parsed_line_t* pl = l->data;
ff_write_line(fp, pl->timestamp_str,
pl->type, pl->enc,
pl->stanza_id, pl->archive_id, pl->replace_id,
pl->from_jid, pl->from_resource,
pl->to_jid, pl->to_resource, pl->marked_read,
pl->message);
written++;
if (written % EXPORT_PROGRESS_INTERVAL == 0) {
cons_show(" ... %s: %d/%d written so far", contact, written, (int)g_slist_length(merged));
}
}
g_slist_free_full(merged, (GDestroyNotify)ff_parsed_line_free);
fflush(fp);
fsync(fd); // ensure data is on disk before atomic rename
fclose(fp); // also releases flock
// Atomic rename
if (rename(tmp_path, log_path) != 0) {
log_error("export: rename %s -> %s failed: %s", tmp_path, log_path, strerror(errno));
cons_show_error("Export failed for %s: could not rename temp file (%s)", contact, strerror(errno));
unlink(tmp_path);
} else {
cons_show("Exported %s: %d new, %d skipped (already existed: %d)",
contact, contact_exported, contact_skipped, existing_count);
total_exported += contact_exported;
}
g_hash_table_destroy(seen_keys);
g_slist_free_full(existing, (GDestroyNotify)ff_parsed_line_free);
g_slist_free_full(sqlite_lines, (GDestroyNotify)message_free);
int exported = _export_one_contact(contact);
if (exported > 0)
total_exported += exported;
}
g_slist_free_full(contacts, g_free);

View File

@@ -163,23 +163,12 @@ _ff_maybe_index_line(ff_contact_state_t* state, const char* buf, off_t pos)
g_date_time_unref(dt);
jabber.developer marked this conversation as resolved Outdated

Please, extract bom-check to a different method. it is violates DRY and SRP. database_flatfile_validate.c:123

Please, extract bom-check to a different method. it is violates DRY and SRP. `database_flatfile_validate.c:123`

Corrected

Corrected
}
static gboolean
_ff_state_build(ff_contact_state_t* state)
// Scan lines from an already-positioned file pointer, caching IDs and
// building index entries. Returns the number of messages scanned.
static size_t
_ff_scan_lines(ff_contact_state_t* state, FILE* fp)
{
FILE* fp = fopen(state->filepath, "r");
if (!fp)
return FALSE;
state->bom_len = ff_skip_bom(fp);
state->n_entries = 0;
state->total_lines = 0;
size_t msg_count = 0;
// Clear ID caches for full rebuild
g_hash_table_remove_all(state->archive_ids);
g_hash_table_remove_all(state->stanza_senders);
size_t count = 0;
while (1) {
off_t pos = ftell(fp);
gboolean truncated = FALSE;
@@ -197,16 +186,34 @@ _ff_state_build(ff_contact_state_t* state)
}
state->total_lines++;
// Extract IDs for O(1) dedup/LMC cache
_ff_cache_line_ids(state, buf);
if (msg_count % FF_INDEX_STEP == 0) {
if (count % FF_INDEX_STEP == 0) {
_ff_maybe_index_line(state, buf, pos);
}
msg_count++;
count++;
free(buf);
}
return count;
}
jabber.developer marked this conversation as resolved Outdated

please add comment explaining purpose

please add comment explaining purpose

Corrected

Corrected
static gboolean
_ff_state_build(ff_contact_state_t* state)
{
FILE* fp = fopen(state->filepath, "r");
if (!fp)
jabber.developer marked this conversation as resolved Outdated

Are !dt and !space cases ok to be ignored?

Are `!dt` and `!space` cases ok to be ignored?

Postponed — discussion. Current behaviour is "skip silently" (lines outside index step are not indexed); this is by design. No specific change requested.

Postponed — discussion. Current behaviour is "skip silently" (lines outside index step are not indexed); this is by design. No specific change requested.
return FALSE;
jabber.developer marked this conversation as resolved Outdated

please, consider reduction of nesting

please, consider reduction of nesting

Corrected

Corrected
state->bom_len = ff_skip_bom(fp);
state->n_entries = 0;
state->total_lines = 0;
// Clear ID caches for full rebuild
g_hash_table_remove_all(state->archive_ids);
g_hash_table_remove_all(state->stanza_senders);
_ff_scan_lines(state, fp);
fclose(fp);
@@ -247,37 +254,8 @@ _ff_state_extend(ff_contact_state_t* state)
}
}
// Use a local counter for new messages so index step alignment
// doesn't depend on the total from the initial build.
size_t new_count = 0;
while (1) {
off_t pos = ftell(fp);
gboolean truncated = FALSE;
char* buf = ff_readline(fp, &truncated);
if (!buf)
break;
if (truncated) {
free(buf);
break;
}
if (buf[0] == '\0' || buf[0] == '#') {
free(buf);
continue;
}
state->total_lines++;
// Extract IDs for O(1) dedup/LMC cache
_ff_cache_line_ids(state, buf);
if (new_count % FF_INDEX_STEP == 0) {
_ff_maybe_index_line(state, buf, pos);
}
new_count++;
free(buf);
}
// Use _ff_scan_lines for the shared read/index/cache loop.
_ff_scan_lines(state, fp);
fclose(fp);