Feat/multi xlsx batch processing #1
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
BREAKING CHANGES: - Projects now enforce hard limits (200k tokens, 20 dataset summaries) - Context window overflow blocks execution for local LLMs (was warning-only) - Duplicate document detection prevents silent content duplication - Invalid --sheet-name now errors instead of silently falling back NEW FEATURES: - `docloom analyze --batch <dir>` processes multiple files efficiently - Batched embedding for --reindex (100 chunks/batch, prevents OOM) - Progress indicators for multi-file operations - Configurable timeout via --timeout-sec for large jobs CRITICAL FIXES (11 issues): - #1: Unbounded document accumulation now enforced at 200k tokens - #2: Duplicate file detection with absolute path comparison - #3: Memory freed immediately after outlier computation - #4: Context overflow blocks Ollama (prevents silent truncation) - #5: Cumulative --max-rows limits across project documents - #6: Chunked embedding prevents API batch size failures - #7: --sheet-name validation with available sheets listing - #8: 100k char limit on XLSX summaries with diagnostic errors - #12: Dataset summary basename collisions prevented with unique suffixes - #13: Prompt instructions deduplicated (40% token reduction) - #14: RAG chunker enforces hard maxTokens limit for large tables PERFORMANCE: - 10x100k-row XLSX files: 9.3GB → <2GB peak memory - Embedding reindex: processes 1000+ chunks without hanging - Prompt construction: 40% fewer tokens via deduplication DEVELOPER EXPERIENCE: - Context-aware error messages guide remediation - Validation fails fast with actionable suggestions - Memory profiling tests ensure sustained efficiency
… paths This resolves a critical bug in the XLSX parser where sheet relationships were incorrectly resolved if their target paths were absolute (e.g., '/xl/worksheets/sheet1.xml') instead of relative. ZIP archive entries do not include a leading slash, causing the parser to fail to locate and read the sheets. The fix introduces a `normalizeRelPath()` helper to strip leading slashes from relationship targets, ensuring paths are always relative to the ZIP root. This allows the parser to successfully read sheet data. **Key Changes & Improvements:** * **XLSX Fix:** Added `normalizeRelPath()` to resolve sheet and shared string relationship targets correctly, fixing the "0 columns" bug. * **Mixed-Input Batch:** Extended `analyze-batch` to gracefully handle non-tabular files (YAML, Markdown, Text, DocX) alongside structured data when a project path (`-p`) is provided. * **TSV Improvement:** Automatically sets the delimiter to tab for `.tsv` files if the `--delimiter` flag is not explicitly used. * **Testing:** Added a dedicated regression test for path normalization logic to prevent future regressions. This change unblocks proper analysis for projects relying on XLSX files and significantly improves the quality of input provided to the LLMs, which previously reported missing data due to the parser failure.
- Add CHANGELOG.md documenting all changes in v0.2.0 - Add PR template to standardize future contributions - Follows Keep a Changelog and Semantic Versioning standards
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
🎯 What This Solves
DocLoom could not reliably process multiple large XLSX files in a single project due to:
This PR makes multi-spreadsheet projects production-ready.
🚀 New Capabilities
docloom analyze-batch "data/*.xlsx"processes multiple files with progress.csv,.tsv,.xlsx+ non-tabular files (.yaml,.md, etc.)--sample-rows-project🐛 Critical Fixes (12 Issues)
XLSX Parser (NEW)
Memory & Context (Original 11)
📊 Performance Impact
✅ Testing
go test -race)🔄 Migration Guide
If you have existing projects with many XLSX files:
docloom list --docs -p <project>to audit document count--retrievalmode--max-rowslimits if neededBreaking changes:
--sheet-namenow errors with available sheet list