Nodejs provider: GetDocumentUris counts files twice when Location and workspaceFolders are both set #1035
Description
Nodejs Provider: GetDocumentUris counts files twice when Location and workspaceFolders are both set
Summary
The nodejs provider's GetDocumentUris() method in symbol_cache_helper.go counts files twice during provider preparation when both InitConfig.Location and ProviderSpecificConfig["workspaceFolders"] are set to the same directory.
Impact
- Provider preparation progress shows 2x the actual file count (e.g., 1604 instead of 802 files)
- Symbol cache operations process the same files multiple times
- Performance degradation during preparation phase
- Confusing user experience with incorrect progress reporting
Root Cause
File: external-providers/generic-external-provider/pkg/server_configurations/nodejs/symbol_cache_helper.go
Method: nodejsSymbolSearchHelper.GetDocumentUris() (lines 29-73)
Bug: The method adds workspace folders to additionalPaths without checking if they duplicate the primaryPath (which comes from Location).
func (h *nodejsSymbolSearchHelper) GetDocumentUris(conditionsByCap ...provider.ConditionsByCap) []uri.URI {
primaryPath := h.config.Location // Line 30: Set from InitConfig.Location
if after, ok := strings.CutPrefix(primaryPath, fmt.Sprintf("%s://", uri.FileScheme)); ok {
primaryPath = after
}
additionalPaths := []string{}
if val, ok := h.config.ProviderSpecificConfig["workspaceFolders"].([]string); ok {
for _, path := range val {
if after, prefixOk := strings.CutPrefix(path, fmt.Sprintf("%s://", uri.FileScheme)); prefixOk {
path = after
}
if primaryPath == "" {
primaryPath = path // Line 41: Only skips if primaryPath is empty
continue
}
additionalPaths = append(additionalPaths, path) // Line 44: BUG - adds duplicate!
}
}
// FileSearcher will scan both BasePath and AdditionalPaths
searcher := provider.FileSearcher{
BasePath: primaryPath, // Scans /path/to/source
AdditionalPaths: additionalPaths, // Also scans /path/to/source (duplicate!)
// ...
}Problem: When Location = "/path/to/source" and workspaceFolders = ["/path/to/source"], the code:
- Sets
primaryPath = "/path/to/source"(from Location) - Adds
/path/to/sourcetoadditionalPaths(from workspaceFolders) - FileSearcher scans the same directory twice
Reproduction
Test Case 1: tackle2-ui repository
config := provider.InitConfig{
Location: "/Users/tsanders/Workspace/tackle2-ui", // 802 JS/TS files
ProviderSpecificConfig: map[string]interface{}{
"workspaceFolders": []interface{}{"file:///Users/tsanders/Workspace/tackle2-ui"},
},
}Expected: 802 files
Actual: 1604 files (802 × 2)
Test Case 2: Empty Location (workaround)
config := provider.InitConfig{
Location: "", // Empty!
ProviderSpecificConfig: map[string]interface{}{
"workspaceFolders": []interface{}{"file:///Users/tsanders/Workspace/tackle2-ui"},
},
}Result: 802 files ✅ (correct because line 41-42 sets primaryPath from first workspace folder)
Proposed Fix
Option 1: Deduplicate paths (Recommended)
func (h *nodejsSymbolSearchHelper) GetDocumentUris(conditionsByCap ...provider.ConditionsByCap) []uri.URI {
primaryPath := h.config.Location
if after, ok := strings.CutPrefix(primaryPath, fmt.Sprintf("%s://", uri.FileScheme)); ok {
primaryPath = after
}
additionalPaths := []string{}
if val, ok := h.config.ProviderSpecificConfig["workspaceFolders"].([]string); ok {
for _, path := range val {
if after, prefixOk := strings.CutPrefix(path, fmt.Sprintf("%s://", uri.FileScheme)); prefixOk {
path = after
}
if primaryPath == "" {
primaryPath = path
continue
}
// FIX: Skip workspace folders that duplicate primaryPath
if path == primaryPath {
continue
}
additionalPaths = append(additionalPaths, path)
}
}
// ... rest of function
}Option 2: Document behavior and require callers to avoid duplicates
Add documentation to GetDocumentUris() clarifying that callers must not set both Location and workspaceFolders to the same value.
Recommendation: Option 1 is preferred because it's more defensive and prevents user errors.
Workaround (for downstream consumers)
Until this is fixed, consumers can work around the issue by:
// For nodejs provider only: Leave Location empty to prevent duplicate counting
location := util.SourceMountPath
if providerName == util.NodeJSProvider {
location = "" // workspaceFolders is already set in providerSpecificConfig
}
providerConfig := provider.Config{
InitConfig: []provider.InitConfig{
{
Location: location, // Empty for nodejs to avoid duplicate
ProviderSpecificConfig: map[string]interface{}{
"workspaceFolders": []interface{}{fmt.Sprintf("file://%s", util.SourceMountPath)},
},
},
},
}Related Issues
This issue was discovered while implementing provider preparation progress reporting in kantra (konveyor/kantra#XXX).
Questions for Maintainers
-
Is this the intended behavior? Should
workspaceFoldersbe additive toLocation, or is one meant to override the other? -
Does this affect other providers? The
genericprovider (Go, Python) doesn't seem to have this issue, but should we check them too? -
Breaking change concerns? Fixing this might change file counts for existing users. Should we add a deprecation period or config flag?
Environment
- analyzer-lsp version: v0.9.0-alpha.1.0.20251205151422-c5b85678c415
- Provider: nodejs (generic-external-provider)
- OS: macOS (but affects all platforms)
Additional Context
The nodejs provider is the only provider that processes workspaceFolders in this way. Other providers (Go, Python) don't set workspaceFolders in their config, so they don't exhibit this issue.