v2.0.0

dir-tree v2.0.0

18 commits · 99 files changed · +8574 / -4961

Full Changelog: v1.0.5...v2.0.0

Overview

• fix(reader): isolate state for each read
• fix(reader): distinguish cleared excludes from defaults
• refactor(api): separate structural nodes from browser sources
• fix(metadata): share lookup indexes across package entries
• feat(reader): distinguish source failures and metadata warnings
• fix(parser): reject conflicting imported tree paths
• perf(reader): index ZIPs and bound filesystem work per read
• refactor(core): split implementation by module responsibility
• docs(api): document the 2.0 reader and migration contracts
• test(package): verify installed entries and the minimum Node runtime
• fix(parser): accept displayed dot roots in tree text
• fix(tree): share name invariants between edits and imports
• fix(browser): read only reachable legacy ignore rules
• fix(tree): preserve special path keys across tree operations
• refactor(zip): isolate record parsing and verify producer fixtures
• docs(v2): clarify contracts, migration status, and commit workflow
• refactor(playground): migrate examples to the v2 reader contract
• feat(reader): report per-read preparation and tree timings

Changelog

• Add optional per-read onTiming reports for source preparation, tree processing, and measured total duration across built-in adapters, isolating synchronous timing callback failures from successful reads

• Accept displayed dot roots in tree text and share structural name validation between editing and importing

• Skip legacy directory ignore files outside the current read scope, loading parent rules first

• Preserve special path keys across metadata, annotations, visibility, selection, and JSON transfer

• Reuse the metadata index builder and verify ZIP record parsing with Python and .NET archive fixtures

• Isolate read overrides and ignore rules per call, and let an explicit empty exclude list clear constructor defaults

• Move browser directory adapters and dropped-entry types to the /browser entry

• Keep file nodes structural and remove FileNode.handle and stripFileNodeHandles; adapters retain their sources for repeated reads

• Share tree lookup metadata between independently loaded ESM and CommonJS entries

• Report unreadable ignore files with stable error codes and expose optional metadata warnings through onWarning

• Reject duplicate normalized paths and invalid node structures before publishing imported trees

• Index ZIP trees without inflating ordinary entries, bound filesystem I/O per read, and apply batch annotation patches with one map copy

• Resolve separate TypeScript declarations for ESM and CommonJS and verify the package on Node.js 18.18.1