v0.1.1 | Richer IR type system, DOCX writer output, improved PPTX/XLSX IR renderers, and writer APIs in all language bindings

IR — extended type system

• TextSpan gains nine typography fields: font_name, font_size_half_pt,
  color, highlight, underline (UnderlineStyle enum), vertical_align
  (VerticalAlign: Superscript / Subscript / Baseline), all_caps,
  small_caps, char_spacing_half_pt.
• Paragraph gains twelve layout fields: alignment (ParagraphAlignment:
  Left / Center / Right / Justify / Distribute), indent_left_twips,
  indent_right_twips, first_line_indent_twips, space_before_twips,
  space_after_twips, line_spacing (LineSpacing: Auto / Multiple / Exact /
  AtLeast), background_color, border (ParagraphBorder), keep_with_next,
  keep_together, page_break_before, outline_level.
• Table gains width_twips, column_widths_twips, border
  (TableBorder), alignment (TableAlignment), indent_left_twips,
  cell_padding_twips, caption.
• TableRow gains height_twips, allow_break, repeat_as_header.
• TableCell gains background_color, border, vertical_align
  (CellVerticalAlign), text_align, width_twips, padding (CellPadding),
  text_direction (TextDirection).
• Image gains data, format (ImageFormat), pixel_width,
  pixel_height, display_width_emu, display_height_emu, decorative,
  positioning (ImagePositioning: Inline or Floating with FloatingImage).
• Section gains page_setup (PageSetup), columns (ColumnLayout),
  break_type (SectionBreakType), and six header/footer slots
  (header, footer, first_page_header, first_page_footer,
  even_page_header, even_page_footer).
• Metadata gains author, subject, keywords, created, modified,
  description (written to docProps/core.xml).
• New Element variants: TextBox, PageBreak, ColumnBreak,
  Footnote(Note), Endnote(Note), CodeBlock.
• List gains start_number, style (ListStyle), level.
  ListItem.content promoted from Vec<InlineContent> to Vec<Element>
  to allow block-level content (tables, images) inside list items.
• InlineContent gains FootnoteRef and EndnoteRef variants.
• New supporting types: BorderLine, TableBorder, ParagraphBorder,
  CellPadding, FloatingImage, HeaderFooter, TextBox, Note,
  FootnoteRef, CodeBlock, PageSetup, ColumnLayout.
• New enums: UnderlineStyle, ParagraphAlignment, LineSpacing,
  BorderStyle, CellVerticalAlign, TableAlignment, TextDirection,
  ImageFormat, ImagePositioning, SectionBreakType, VerticalAlign,
  FloatAnchor, TextWrap, ListStyle.
• All new fields are Option<_> or default to false/None — fully
  backwards compatible; existing callers require only ..Default::default()
  on struct literals.

DocxWriter — OOXML emission for all new fields

• Run properties: <w:rFonts>, <w:sz>/<w:szCs>, <w:color>,
  <w:shd> (highlight), <w:u>, <w:vertAlign>, <w:caps>,
  <w:smallCaps>, <w:spacing> (character spacing).
• Paragraph properties: <w:jc>, <w:ind>, <w:spacing> (before/after/
  line), <w:pBdr>, <w:shd>, <w:keepNext>, <w:keepLines>,
  <w:pageBreakBefore>, <w:outlineLvl>.
• Table: <w:tblW>, <w:tblInd>, <w:tblBorders>, <w:jc> (table),
  <w:tblCellMar>, <w:tblGrid>/<w:gridCol>.
• Table row: <w:trHeight>, <w:tblHeader>, <w:cantSplit>.
• Table cell: <w:tcW>, <w:gridSpan>, <w:vMerge>, <w:shd> (cell),
  <w:tcBorders>, <w:vAlign>, <w:textDirection>, <w:tcMar> (per-edge
  padding), cell-level text_align propagated to contained paragraphs.
• Table caption: emitted as a Caption-styled paragraph before <w:tbl>.
• Images: inline <wp:inline> and floating <wp:anchor> with
  <wp:wrapSquare> / <wp:wrapTight> / <wp:wrapThrough> /
  <wp:wrapTopAndBottom> / <wp:wrapNone>.
• Text boxes: <wp:anchor> + <wps:txbx> + <w:txbxContent>.
• Sections: <w:sectPr> with <w:pgSz>, <w:pgMar>, <w:cols> (uniform
  and per-column widths with separator rule), <w:type> (Continuous / NextPage
  / EvenPage / OddPage). Header/footer parts written to /word/header*.xml and
  /word/footer*.xml with correct relationship entries.
• Footnotes / endnotes: /word/footnotes.xml and /word/endnotes.xml
  parts; inline <w:footnoteReference> / <w:endnoteReference> runs with
  FootnoteReference / EndnoteReference character styles.
• {PAGE} / {NUMPAGES} sentinels in header/footer text spans emitted as
  <w:fldChar> / <w:instrText> field runs.
• Page break: <w:br w:type="page">. Column break: <w:br w:type="column">.
• Code blocks: Code-styled paragraph with preserved whitespace and
  line breaks.
• Lists: <w:numFmt> driven by ListStyle; <w:startOverride> for
  non-1 start_number; block-level item content (tables, images) written
  alongside the numbered paragraph.
• Metadata: author, subject, keywords, description, created,
  modified written to docProps/core.xml as Dublin Core properties.

PPTX IR renderer — ir_to_pptx

• Rich text runs: paragraphs and headings now emit styled <a:r> runs via
  add_rich_text() — bold, italic, font size, color, and font name from
  TextSpan are all preserved. Previously all formatting was stripped.
• Table elements: rendered as tab-separated cell text (rows joined with \n)
  instead of being silently dropped.
• Image elements: embedded as native PPTX media via the new
  SlideData::add_image() API — writes a <p:pic> shape with <p:blipFill>
  and an OPC media part. PNG, JPEG, and GIF are supported.
• CodeBlock elements: rendered with Courier New font run instead of being
  silently dropped.
• Slide dimensions: first Section.page_setup is now forwarded to
  PptxWriter::set_presentation_size() (1 twip = 914 400/1 440 EMU), fixing
  clipped output for landscape A4 and other non-16:9 documents.
• New PptxWriter::set_presentation_size(cx, cy) method; emits <p:sldSz>
  with the correct EMU values instead of always writing the 16:9 default.

XLSX IR renderer — ir_to_xlsx

• Header row styling: rows with TableRow.is_header = true are now written
  with bold weight and a grey (D3D3D3) background via set_cell_styled().
  TableCell.background_color overrides the default grey when set.
• Cell background color: non-header cells with TableCell.background_color
  set now get a solid fill style applied.
• Column widths: Table.column_widths_twips is now converted to Excel
  character-width units (twips × 96 / (1440 × 7), clamped 3–80) and written
  via set_column_width().
• Merged cells: TableCell.col_span and row_span > 1 now emit a
  <mergeCells>/<mergeCell ref="…"/> block in the worksheet XML instead of
  being ignored.
• New SheetData::merge_cells(row, col, row_span, col_span) method; inserts
  <mergeCells> between </sheetData> and </worksheet>.
• Row cursor tracks absolute position across all elements in a section, so
  paragraphs and headings interleaved with tables land in the correct rows.

Writer APIs — all language bindings

XlsxWriter and PptxWriter (previously Rust-only) are now callable from
every binding layer via a new index-based C FFI surface:

New C FFI symbols (include/office_oxide_c/office_oxide.h):

• office_xlsx_writer_new/free, office_xlsx_writer_add_sheet (returns sheet
  index), office_xlsx_sheet_set_cell, office_xlsx_sheet_set_cell_styled
  (bold + hex background), office_xlsx_sheet_merge_cells,
  office_xlsx_sheet_set_column_width, office_xlsx_writer_save,
  office_xlsx_writer_to_bytes
• office_pptx_writer_new/free, office_pptx_writer_set_presentation_size,
  office_pptx_writer_add_slide (returns slide index),
  office_pptx_slide_set_title, office_pptx_slide_add_text,
  office_pptx_slide_add_image (PNG/JPEG/GIF bytes + EMU geometry),
  office_pptx_writer_save, office_pptx_writer_to_bytes

Go — XlsxWriter and PptxWriter structs with CGo wrappers and
runtime.SetFinalizer for safe GC.

C# / .NET — new OfficeOxide.XlsxWriter and OfficeOxide.PptxWriter
classes (IDisposable); P/Invoke declarations added to NativeMethods.

Node.js — XlsxWriter and PptxWriter ESM + CJS classes; koffi
function prototypes; TypeScript ImageFormat type and class declarations.

Python — XlsxWriter and PyO3PptxWriter PyO3 classes calling Rust
directly (no C FFI round-trip); exported from office_oxide with full type
stubs in _native.pyi.

Bug fixes

• TableCell.padding (CellPadding) was defined in the IR but silently
  dropped by the writer; now emits <w:tcMar> with per-edge twip values.
• TableCell.text_align was defined in the IR but silently dropped; now
  propagated to contained paragraphs (respects pre-existing paragraph
  alignment, so explicit paragraph alignment is never overwritten).
• Table.caption was defined in the IR but silently dropped; now emitted
  as a Caption-styled paragraph immediately before the table.

---

Install

Rust   cargo add office_oxide

Python   pip install office-oxide

JavaScript (WASM, universal)   npm install office-oxide-wasm

Node.js (native)   npm install office-oxide

Go  

go get github.com/yfedoseev/office_oxide/go
# fetch the native library matching your platform:
go run github.com/yfedoseev/office_oxide/go/cmd/install@latest

C# / .NET   dotnet add package OfficeOxide

CLI

cargo binstall office_oxide_cli      # pre-built binary
brew install yfedoseev/tap/office-oxide
scoop bucket add yfedoseev https://github.com/yfedoseev/scoop-bucket && scoop install office-oxide

Raw C FFI — download the native-<platform>-<arch> asset below and include include/office_oxide_c/office_oxide.h.

Changelog

Full history: [CHANGELOG...

v0.1.0 | Initial public release

Cross-language bindings

• Rust core (office_oxide on crates.io): unified Document handle for
  all six formats, EditableDocument for DOCX/XLSX/PPTX editing, format-
  agnostic DocumentIR.
• Python (office-oxide on PyPI): context-manager Document /
  EditableDocument, os.PathLike support, complete type stubs in
  _native.pyi (Literal format names, _Path alias).
• Go (github.com/yfedoseev/office_oxide/go): CGo wrapper over the C FFI
  with idiomatic Open / Close / error-return API, go/cmd/install helper
  that fetches the matching native archive and prints the
  CGO_CFLAGS / CGO_LDFLAGS to export.
• C# / .NET (OfficeOxide on NuGet): LibraryImport P/Invoke,
  IDisposable, async/await, IsAotCompatible=true, IsTrimmable=true.
  Four SetCell overloads + SetCellEmpty. Net 8 and net 10 target
  frameworks.
• Node.js native (office-oxide on npm): koffi-based,
  no node-gyp, ESM + CJS entry points with an exports map, TypeScript
  definitions, Symbol.dispose support, platform prebuilds staged into
  prebuilds/<platform>-<arch>/.
• WASM (office-oxide-wasm on npm): three sub-path exports — default
  ESM for bundlers, office-oxide-wasm/node for CJS, office-oxide-wasm/web
  for native-ESM browser imports. TypeScript definitions shipped.
• C FFI (include/office_oxide_c/office_oxide.h): stable
  office_document_* / office_editable_* surface with out-param error
  codes and explicit memory ownership. Exported from the cdylib + staticlib;
  the substrate that Go, C#, and Node-native link against.

Tooling

• CLI (office-oxide binary): text, markdown, html, info, ir
  subcommands.
• MCP server (office-oxide-mcp binary): extract and info tools
  over JSON-RPC 2.0 / stdio.

Performance

• Up to 100× faster than python-docx, openpyxl, python-pptx, xlrd.
• Beats calamine on XLSX and all Rust / Python alternatives on .xls.
• 100% pass rate on valid Office files (6,062-file corpus: LibreOffice,
  Apache POI, python-pptx, python-docx, Pandoc, etc.). All 97 non-passing
  files are invalid inputs — corrupted ZIPs, missing required parts, malformed
  XML, or non-Office files with Office extensions.

Documentation & examples

• Per-language getting-started guides in docs/: Rust, Python,
  Go, C#, JavaScript (native), WASM, and C / raw FFI.
• Per-binding READMEs in python/, go/,
  csharp/OfficeOxide/,
  js/, wasm-pkg/.
• Identical extract / replace / read_xlsx demos per language under
  examples/.

Release CI

• Version parity across Cargo.toml, pyproject.toml,
  wasm-pkg/package.json, js/package.json, and
  csharp/OfficeOxide/OfficeOxide.csproj.
• 6-target native-lib build matrix producing .tar.gz / .zip archives
  with the shared library, static archive, and public header.
• 3-target WASM build (bundler / nodejs / web) with per-target module-type
  hints so Node + bundlers + browsers each load the right code.
• NuGet packaging with runtimes/<rid>/native/ prebuilts, Node
  prebuilds/<platform>-<arch>/ staging, and a go/v* module tag for the
  Go module proxy.

---

Install

Rust   cargo add office_oxide

Python   pip install office-oxide

JavaScript (WASM, universal)   npm install office-oxide-wasm

Node.js (native)   npm install office-oxide

Go  

go get github.com/yfedoseev/office_oxide/go
# fetch the native library matching your platform:
go run github.com/yfedoseev/office_oxide/go/cmd/install@latest

C# / .NET   dotnet add package OfficeOxide

CLI

cargo binstall office_oxide_cli      # pre-built binary
brew install yfedoseev/tap/office-oxide
scoop bucket add yfedoseev https://github.com/yfedoseev/scoop-bucket && scoop install office-oxide

Raw C FFI — download the native-<platform>-<arch> asset below and include include/office_oxide_c/office_oxide.h.

Changelog

Full history: CHANGELOG.md