v0.1.0 — first release

• extract(input, options): a file path or bytes → numbered lines, section markers, sections, page starts for PDF, warnings, size and SHA-256 of the input.
• detect(input, options): the format and whether it can be extracted, without running an extractor.
• DoclinesError with codes unsupported, not_allowed, encrypted, no_text_layer, damaged, too_large, text_too_large, timeout, out_of_memory, binary, not_a_file.
• FORMATS and helpers sectionOf, pageOf, pageRanges, normalizeText, foldSpace, version.
• Formats: plain text, HTML, PDF (pdfjs-dist), DOCX/DOCM/DOTX/DOTM (mammoth), DOC (word-extractor), RTF, XLSX/XLSM/XLTX/XLTM, XLS (BIFF8), ODS, PPTX/PPTM/POTX/POTM/PPSX, PPT, ODT/OTT, ODP/OTP, EPUB 2 and 3, FB2 and FB2 in ZIP, EML (postal-mime), MSG (@kenjiuno/msgreader) with attachments.
• Recognised and refused with a label: XLSB, Excel 5/95, Word 6/95, PowerPoint 95, Apple iWork, XPS, DjVu, OpenDocument drawings, images, archives, SQLite, other OLE and ZIP files. Encrypted PDF, OOXML, DOC, XLS, PPT, ODF and EPUB are refused as encrypted.
• Extraction in a worker thread with a timeout and a heap cap; nothing written to stdout or stderr; --max-old-space-size on the host that overrides the cap is reported in warnings.
• Own ZIP reader on node:zlib with entry count, size and name checks; XML without entity expansion; detection on the calling thread reads only container directories and metadata entries.
• windows-1252 decoded by an own table on every Node version.
• CLI doclines <file> with --json, --detect, --formats, --max-text-bytes, --timeout.