fix: detect new PDFs in already-indexed KB folders
All checks were successful
BotServer CI/CD / build (push) Successful in 3m11s
All checks were successful
BotServer CI/CD / build (push) Successful in 3m11s
- Don't skip entire GBKB scan when all KBs are indexed - Instead, skip individual files that are already tracked (not new) - This allows new PDFs added to existing KB folders to be detected and indexed Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
This commit is contained in:
parent
6d987c0eea
commit
32fbdb4b17
1 changed files with 12 additions and 12 deletions
|
|
@ -1543,7 +1543,7 @@ let file_state = FileState {
|
||||||
debug!("[GBKB] Found {} files total, acquiring file_states lock...", current_files.len());
|
debug!("[GBKB] Found {} files total, acquiring file_states lock...", current_files.len());
|
||||||
|
|
||||||
// Check if ALL KBs for this bot are already indexed in Qdrant
|
// Check if ALL KBs for this bot are already indexed in Qdrant
|
||||||
// If so, skip the entire scan to avoid deadlock and unnecessary downloads
|
// If so, only scan for NEW files - skip re-indexing existing ones
|
||||||
let mut kb_folders: HashSet<String> = HashSet::new();
|
let mut kb_folders: HashSet<String> = HashSet::new();
|
||||||
for (path, _) in current_files.iter() {
|
for (path, _) in current_files.iter() {
|
||||||
let parts: Vec<&str> = path.split('/').collect();
|
let parts: Vec<&str> = path.split('/').collect();
|
||||||
|
|
@ -1565,19 +1565,19 @@ let file_state = FileState {
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
if all_indexed && !kb_folders.is_empty() {
|
|
||||||
trace!("[GBKB] All {} KB folders already indexed, skipping scan for bot {}",
|
|
||||||
kb_folders.len(), self.bot_id);
|
|
||||||
return Ok(());
|
|
||||||
}
|
|
||||||
|
|
||||||
let mut file_states = self.file_states.write().await;
|
let mut file_states = self.file_states.write().await;
|
||||||
debug!("[GBKB] file_states lock acquired, processing {} files", current_files.len());
|
debug!("[GBKB] file_states lock acquired, processing {} files (all_indexed={})", current_files.len(), all_indexed);
|
||||||
|
|
||||||
for (path, current_state) in current_files.iter() {
|
for (path, current_state) in current_files.iter() {
|
||||||
let is_new = !file_states.contains_key(path);
|
let is_new = !file_states.contains_key(path);
|
||||||
debug!("[GBKB] DEBUG: path={} in_file_states={}", path, !is_new);
|
debug!("[GBKB] DEBUG: path={} in_file_states={}", path, !is_new);
|
||||||
|
|
||||||
|
// When all KBs are indexed, skip files that are already tracked (not new)
|
||||||
|
if all_indexed && !is_new {
|
||||||
|
trace!("[GBKB] Skipping already indexed file: {}", path);
|
||||||
|
continue;
|
||||||
|
}
|
||||||
|
|
||||||
// Use last_modified as primary change detector (more stable than ETag)
|
// Use last_modified as primary change detector (more stable than ETag)
|
||||||
// ETags can change due to metadata updates even when content is identical
|
// ETags can change due to metadata updates even when content is identical
|
||||||
let is_modified = if let Some(prev) = file_states.get(path) {
|
let is_modified = if let Some(prev) = file_states.get(path) {
|
||||||
|
|
|
||||||
Loading…
Add table
Reference in a new issue