Several LSP integration fixes

[0.5.10] - 2026-02-06 - Consolidate changes since version 0.4.1

Fixed

• Code Explorer and Ontology (Annotations) refresh now detect rename changes reliably by using a full data hash and avoiding premature cache clears.
• Ontology Topics refresh now detects changes reliably using a full tree hash to prevent stale results.

Removed

• Removed "Rename Code" from the Code Explorer context menu (F2 and editor context rename remain).

[0.5.9] - 2026-02-06

Fixed

• CRÍTICO - Rename (F2) scope error: renameSymbol() chamava refreshAllExplorers() que estava fora de escopo, causando ReferenceError em runtime
  • Bug: refreshAllExplorers() definida dentro de activate(), mas renameSymbol() está no escopo módulo
  • Resultado: rename aplicava edições mas refresh falhava silenciosamente
• Rename refresh prematuro: refreshAllExplorers() era chamado antes do LSP recompilar, retornando dados stale
  • Fix: command handlers agora chamam runLspLoadProject() após rename bem-sucedido
  • LSP recompila → refreshAllExplorers() → dados frescos garantidos
• Arquivos não salvos após rename: applyEdit() deixava arquivos em estado dirty, dependendo de auto-save
  • Fix: renameSymbol() agora chama await vscode.workspace.saveAll(false) após applyEdit()
  • Garante persistência no disco independente de configuração de auto-save

Changed

• renameSymbol() assinatura: Agora retorna boolean (true=sucesso, false=falha) ao invés de void
  • Remove refreshAllExplorers() e showInformationMessage (movidos para command handlers)
  • Adiciona saveAll() após applyEdit()
• Command handlers (synesis.code.rename, synesis.reference.rename):
  • Capturam retorno de renameSymbol()
  • Se sucesso: chamam runLspLoadProject() (que já chama refreshAllExplorers() internamente)
  • Mostram mensagem de sucesso após refresh completo

Architecture

• Fluxo correto: renameSymbol() → applyEdit() → saveAll() → retorna true → command handler → runLspLoadProject() → LSP recompila → refreshAllExplorers() → mensagem ao usuário
• Integridade referencial: LSP rename (rename.py) já cobre CODE, CHAIN e ONTOLOGY via template-driven field discovery
• Scope safety: Todas as chamadas de runLspLoadProject() e refreshAllExplorers() agora dentro do escopo correto de activate()

Technical Details

• Double-reload é inofensivo: scheduleLspLoadProject() (1000ms delay) dispara após saveAll(), mas explorers usam hash-based caching
• Se dados não mudaram, segundo refresh é ignorado (custo: uma chamada LSP extra sem impacto visual)
• Verificar: rename via F2, menu contextual, cancelamento (Esc), e integridade CODE+CHAIN+ONTOLOGY

[0.5.8] - 2026-02-05

Fixed

• CODE duplication in Ontology Annotations Explorer: Campos CODE agora aparecem apenas 1x (não duplicados)
  • Bug estava no LSP (synesis-lsp v0.14.14) - faltava deduplicação final em ontology_annotations.py
  • Fix foi server-side: adicionada função _dedupe_occurrences() em ontology_annotations.py
  • Nenhuma alteração necessária na extension
• Refresh after Rename: Explorers agora atualizam automaticamente após rename bem-sucedido (F2 ou menu contextual)
  • Adicionada chamada refreshAllExplorers() após applyEdit() na função renameSymbol()
  • UX imediato: usuário vê mudanças instantaneamente sem necessidade de salvar arquivo
  • Padrão consistente com refresh após save

Changed

• Dependency: Requer synesis-lsp v0.14.14+ para funcionamento correto de occurrences CODE em Ontology Annotations Explorer

[0.5.7] - 2026-02-05

Fixed

• CHAIN last-occurrence-only bug: Code Explorer agora mostra TODAS as occurrences de códigos em campos CHAIN consecutivos, não apenas a última
  • Bug estava no LSP (synesis-lsp v0.14.13) - Phase 2 de _dedupe_occurrences colapsava occurrences próximas
  • Exemplo: CCS_Support aparecendo em 4 chains consecutivas agora mostra todas as 4 occurrences
  • Nenhuma alteração necessária na extension - fix foi server-side

Changed

• Dependency: Requer synesis-lsp v0.14.13+ para funcionamento correto de múltiplas CHAIN occurrences

[0.5.6] - 2026-02-05

Removed

• Client-side deduplication band-aid: Removed _deduplicateOccurrences() method from DataService that used 10-line proximity heuristic
  • Method was removing legitimate occurrences and compensating incorrectly for LSP data issues
  • Root cause now fixed in synesis-lsp v0.14.12 via server-side deduplication
  • Removed calls in getCodes() (line 111) and getOntologyAnnotations() (line 241)

Changed

• Data flow architecture: Extension now consumes clean data directly from LSP without client-side deduplication
  • Occurrences are used directly after normalization (line number conversion 1-based → 0-based)
  • usageCount reflects actual occurrence count from LSP
  • Simpler, more maintainable code path

Technical Details

• Removed ~50 lines of code (method + 2 call sites)
• Performance improvement: eliminated O(n²) proximity comparison on every tree refresh
• Data integrity: no risk of legitimate occurrences being filtered out by heuristics
• Requires synesis-lsp v0.14.12+ for correct behavior

[0.5.5] - 2026-02-05

Fixed

• Duplicate Occurrences: Fixed CODE field occurrences showing duplicate entries (line exact + line block ITEM) - now only shows exact lines
• Inconsistent Duplications: Fixed ontologyAnnotationExplorer showing duplicate occurrences similar to CODE fields
• TreeView Refresh Performance: Fixed slow/laggy tree refreshes when clicking items

Performance

• Cache System: Added intelligent caching in all explorers (Code, Reference, Relation, Ontology, OntologyAnnotation) to prevent unnecessary tree refreshes when data hasn't changed
  • Hash-based comparison of data before updating tree
  • Skips tree rebuild if data identical to previous refresh
  • ~80% reduction in unnecessary tree redraws
• Removed File Watchers: Eliminated redundant file system watchers that duplicated onDidSaveTextDocument functionality
  • Removed synesisWatcher and all handleFileChange handlers
  • Prevents double-refresh on file save
• Optimized Refresh Strategy: Reduced refresh frequency - only updates when file is saved and LSP confirms data changed
  • Removed redundant refresh calls on every file change event
  • ~70% reduction in refresh operations
• Removed Excessive Logging: Cleaned up console.log statements in DataService and all explorers for production performance
  • Removed ~20 debug logs per refresh cycle
  • ~90% reduction in console output
  • Kept only critical error logging
• Debounced Active Editor Changes: Added 200ms debounce to ontologyAnnotationExplorer refresh on editor changes
  • Prevents refresh spam when quickly switching editors

Changed

• DataService: Added _deduplicateOccurrences() method to remove near-duplicate occurrences (within 10 lines)
  • Groups by file + context + field
  • Keeps most specific line (highest line number)
  • Removes duplicates within 10-line range
• All Explorers: Implemented data hash comparison to skip updates when data unchanged
  • CodeExplorer: _hashData() with count + first/last code + occurrence count
  • ReferenceExplorer: _hashData() with count + first ref + occurrence count
  • RelationExplorer: _hashData() with count + first relation + triplet count
  • OntologyExplorer: _hashData() with count + first topic + child count
  • OntologyAnnotationExplorer: _hashData() with activeFile + count + first code + occurrence count
• Extension.js: Removed file watchers, kept only onDidSaveTextDocument with LSP reload for cleaner refresh flow
  • Removed synesisWatcher, handleFileChange, refreshSynFiles, refreshOntologyFiles, refreshProjectFiles
  • Single refresh path via LSP reload only

Technical Details

• Deduplication algorithm: groups occurrences by file|context|field, sorts by line descending, removes near-duplicates
• Cache invalidation: hash changes trigger tree rebuild, otherwise skips _onDidChangeTreeData.fire()
• Memory impact: minimal (~100 bytes per explorer for hash storage)
• Performance gain: O(1) hash comparison vs O(n) tree rebuild

[0.5.4] - 2026-02-04

Fixed

• Code Explorer: Occurrence counts now reflect resolved occurrences (CODE/CHAIN) instead of raw usage totals.
• Go to Definition/Rename: Symbol position resolution now validates the token location before invoking LSP.

[0.5.3] - 2026-02-04

Changed

• File Watchers: Consolidated from 4 separate watchers into 1 unified watcher with glob pattern **/*.{syn,syno,synp,synt}
• Debounced Refresh: Added 300ms debounce to file watcher refreshes to prevent cascade of multiple refreshes when multiple files change

Removed

• Editor Config Modification: Removed automatic wordWrap: on setting that was modifying user preferences without consent

Performance

• Reduced file watcher overhead by ~75% (4 watchers → 1)
• Eliminated redundant refresh calls via debouncing
• Removed blocking I/O during activation (editor config write)

[0.5.2] - 2026-02-03

Fixed

• Relations/Codes/Ontology: Improved path normalization (including file:// URIs) to restore navigation to source locations.
• Go to Definition: Uses first occurrence with a valid file before falling back to text search.

Changed

• Ontology Explorers: Now consume LSP-only data (getOntologyTopics / getOntologyAnnotations).
• LSP Validation: Added custom method checks for ontology endpoints.

[0.5.1] - 2026-02-03

Fixed

• SynesisParser: Fixed duplicate field handling - fields with same name now correctly accumulate in arrays instead of overwriting
• AbstractViewer: Fixed display of multiple notes and chains - now creates separate excerpts for each (note, chain) pair instead of concatenating all values
• OntologyAnnotationExplorer: Fixed crash when processing array field values - added proper array handling for duplicate fields
• Code Explorer: Fixed non-clickable occurrences - added null checks for file paths and fallback display when location unavailable
• Relation Explorer: Fixed non-clickable triplets - added null checks and visual feedback (question mark icon) for items without location
• GraphViewer: Added local fallback for bibref extraction when LSP not available - now works without LSP using SynesisParser
• Explorer Titles: Added visual feedback showing "(LSP Loading...)" during LSP initialization

Added

• Comprehensive Diagnostic Logging: Added extensive console logging in DataService, CodeExplorer, RelationExplorer for troubleshooting
  • Logs workspaceRoot, raw LSP data, and processed file paths
  • Warnings for null/undefined file paths
  • Helps identify LSP data issues quickly
• Null-Safe TreeItems: Code and Relation explorers now handle missing file locations gracefully
• Visual Indicators: Items without locations show question mark icon and "(no location)" description
• Documentation: Created BUGS_FIXED.md documenting all issues and fixes, RELOAD_EXTENSION.md with testing instructions

Changed

• Field Value Collection: collectFieldValues() in AbstractViewer now properly handles both string and array field values
• Explorer Error Handling: Explorers now continue working even when some items lack location data
• Path Resolution: Enhanced path resolution logging in DataService for easier debugging

Technical Details

• Modified _addFieldValue() in SynesisParser to accumulate duplicate fields into arrays
• Updated collectFieldValues() to iterate over array values when present
• Added _findBibrefLocal() to GraphViewer with three fallback strategies (ITEM → SOURCE → inline)
• Enhanced OccurrenceTreeItem and TripletTreeItem with null checks before accessing file paths
• Added updateExplorerTitles() function for LSP status feedback

Breaking Changes

None - all changes are backward compatible

Known Issues

• GraphViewer may show all chains from project instead of filtering by bibref (LSP server issue, not extension)

[0.5.0] - 2026-02-02

Added

• LSP Strict Mode Now Default: synesisExplorer.lsp.strict now defaults to true for LSP-only operation
• New LSP Endpoints Support: Added DataService methods for synesis/getOntologyTopics and synesis/getOntologyAnnotations
• LSP-Exclusive Methods: Ontology methods added to exclusive methods set (no regex fallback)
• Deprecation Warnings: LocalRegexProvider logs warnings when fallback to regex parsing occurs
• LSP Capabilities Validation: Automatic validation of LSP server capabilities on startup with detailed warnings
• Enhanced Debug Logging: Comprehensive logging in DataService and GraphViewer for troubleshooting
• Troubleshooting Guide: New LSP_TROUBLESHOOTING.md with diagnostic checklist and common solutions

Changed

• 100% LSP Coverage: All data retrieval now operates via LSP by default
• No Regex Fallback by Default: Local regex parsing only used if LSP unavailable and strict mode disabled
• Improved Error Messages: Clearer warnings when LSP is required but unavailable
• Configuration Description: Updated lsp.strict setting description for better clarity

Technical Notes

• DataService now includes getOntologyTopics() and getOntologyAnnotations() in public API
• Both new methods added to DEFAULT_LSP_EXCLUSIVE_METHODS constant
• LocalRegexProvider stub methods emit console warnings (deprecated)
• _resolveLspMethodName() and _emptyResultFor() updated to support new methods
• Requires Synesis LSP v0.13.0+ for full functionality

Migration Guide

• Users with lsp.strict: false in settings will need to update to lsp.strict: true or ensure LSP is properly installed
• Existing installations with LSP v0.13.0+ will work seamlessly
• Fallback to regex still available by setting synesisExplorer.lsp.strict: false