Crate rwml 

rwml — one native Rust reader for both Microsoft Word formats: legacy .doc (Word 97–2003 binary, [MS-DOC]) and modern .docx (OOXML WordprocessingML). No JVM, no Apache POI, no external .docx crate, no shelling out — Document::open format-detects from the magic bytes and both feed the same DocModel and Markdown/HTML exporters.

• .doc is an OLE2/CFB compound file. The text lives in the WordDocument stream; the piece table (CLX) in the 0Table/1Table stream maps character positions to byte offsets, and each piece is either UTF-16LE (Korean body text) or 8-bit text in the document’s ANSI codepage (fCompressed — cp1252 for Western, cp949 for Korean, from the FIB language id).
• .docx is a ZIP of XML parts (word/document.xml + styles, numbering, relationships, media), parsed with zip + quick-xml behind the default docx feature. Disable it (default-features = false) for a dependency-light .doc-only build.

// Works for either format — detection is automatic.
let bytes = std::fs::read("report.doc").unwrap();
let text = rwml::extract_text(&bytes).unwrap();
println!("{text}");

Two surfaces:

• Flat text — extract_text / Document::text, the same output as POI WordExtractor.getText() (fast, allocation-light).
• A full document model — Document::model (paragraphs, character runs with bold/italic/…, structured tables with colspan/rowspan, headings, lists, hyperlinks, and extracted images), plus Document::to_markdown and Document::to_html. Built lazily, so the flat path never pays for it.

Parsing untrusted input is panic-free / bounds-checked: a malformed or hostile .doc/.docx yields Error, never a crash. (The only expect is on the crate’s own compiled-in blank template behind the infallible Document::new/Default; use Document::try_new for a Result instead of that build-invariant panic.)

---

rwml (from WordprocessingML) is an independent project, not affiliated with or endorsed by Microsoft. Microsoft Word and the .doc/.docx formats are Microsoft trademarks, referenced only to indicate format compatibility; the crate is built solely from the public [MS-DOC]/[MS-CFB]/OOXML specifications.

Modules

wasm

Browser/WASM adapter functions over the same core crate::Document API.

Structs

AuthoredComment

A comment to author on a generated run.

AuthoredContentControl

Plain text content-control metadata to author on a generated run.

AuthoredNote

Footnote or endnote metadata to author after a generated run.

AuthoredRevision

Tracked revision metadata to author on a generated run.

Cell

A table cell — may hold block content and span rows/columns.

CellBuilder

Thin builder for a table cell with block content and layout metadata.

CellMargins

Table-cell margins in twips (1/20 point), ordered by physical side.

CharProps

Character-level formatting that affects rendering.

Chart

A block-level chart with literal category and numeric caches.

ChartBuilder

Thin builder for a block-level Chart.

ChartSeries

One named chart series with literal numeric values.

Color

An sRGB color.

Comment

A Word comment extracted from a .docx comments part or recovered from a legacy .doc annotation subdocument.

CommentBuilder

Thin builder for a generated Word comment body and metadata.

ContentControlBuilder

Thin builder for generated plain text content-control metadata.

CoreProperties

Core document properties extracted from docProps/core.xml or generated document setup metadata.

DocBuilder

Thin builder for creating a DocModel without filling every struct field by hand.

DocGrid

Document grid settings for a WordprocessingML section.

DocMeta

Document-level metadata.

DocModel

A whole .doc document as an ordered list of block-level nodes plus document-level metadata.

DocSetup

Document-level layout + metadata, for authoring and rendering. All fields are optional/default so existing read paths are unaffected.

Document

A parsed Word document — either legacy .doc (OLE2/[MS-DOC]) or modern .docx (OOXML). Document::open format-detects from the magic bytes and both backends feed the same DocModel and exporters, so text(), to_markdown(), to_html(), and images() behave identically regardless of which Word format the bytes are in.

DocumentReport

Summary of the opened document’s format, visible stats, observed feature markers, and warnings.

EditCapability

Whether package-preserving edit APIs are available for an opened document.

FeatureInventory

Counts of Word features observed while opening a document.

Field

A Word field observed in a .docx body.

FieldContext

Caller-supplied evaluation context for volatile or external-data fields.

FieldEvaluationReasonCount

Count of unsupported field evaluations for one reason.

FieldKindCount

Count of observed fields for one field kind.

FloatingShape

Floating shape geometry recovered from a Word document.

HeaderFooter

A running header/footer record recovered from a Word document.

Image

An embedded image. bytes/mime are present only when the picture was extracted; otherwise the node is a placeholder.

ImageBuilder

Thin builder for an embedded Image block.

Indent

Paragraph indentation in points; None = unset.

ListInfo

List membership of a paragraph.

MetafileInfo

Best-effort metadata for a preserved WMF/EMF package part.

Note

A footnote or endnote recovered from a Word document.

PageSetup

Page geometry, in points. Default is A4 portrait with 1-inch margins.

ParaProps

Paragraph-level properties.

Paragraph

A paragraph: inline runs plus paragraph-level properties.

ParagraphBuilder

Thin builder for a paragraph with runs and paragraph-level layout.

ParagraphStyle

A paragraph style definition for generated .docx output.

ParagraphStyleBuilder

Thin builder for a paragraph style definition.

RenderReport

Renderer metrics and warnings for a generated PDF.

RenderedPdf

PDF bytes plus the render report produced by the same pagination pass.

Revision

A tracked revision extracted from a .docx body.

RevisionBuilder

Thin builder for generated tracked insertion/deletion metadata.

Row

A table row.

Run

An inline run of text with uniform character properties.

RunBuilder

Thin builder for an inline Run with character formatting.

SectionSetup

Section-level layout recovered from or generated into .docx section properties.

ShapeDistance

Distance from a floating shape to surrounding text, in English Metric Units.

ShapeEffectExtent

Visual-effect extents for a recovered floating shape, in English Metric Units.

ShapeExtent

Drawing extent for a recovered floating shape, in English Metric Units.

ShapePoint

Absolute point for a recovered floating shape, in English Metric Units.

ShapePosition

Positioning metadata for a recovered floating shape.

ShapeWrapping

Text-wrapping policy declared by a floating shape anchor.

SourceRegion

A span of DocModel::blocks that came from one source subdocument.

Spacing

Paragraph spacing in points; None = unset (renderer/writer uses defaults).

Stats

Aggregate document statistics, mirroring the project-wide DocStats contract.

Table

A table: rows of cells.

TableBorderColors

Optional table border colors by physical side.

TableBorderSizes

Optional table border widths by physical side, in eighths of a point.

TableBorderStyles

Optional table border line styles by physical side.

TableBuilder

Thin builder for rich Table values.

TextAnchor

A text range in the main document body associated with an annotation.

TextBox

A text box recovered from a Word document.

WebExtensionTaskPane

A generated Office web-extension task pane package entry.

Enums

Align

Paragraph alignment.

Block

A block-level node.

ChartKind

Supported chart layouts for authored .docx output.

ChartShape

Supported shape styles for authored 3-D bar/column charts.

CoreProperty

Editable .docx core document properties supported by Document::set_core_property.

DocGridType

Document grid behavior for a WordprocessingML section.

DocumentFormat

Source document format detected by crate::Document::open.

DocumentWarning

Human- and machine-readable warnings derived from FeatureInventory.

EditReadOnlyReason

Reason package-preserving edits are unavailable for an opened document.

Error

Errors produced while opening or decoding a .doc file.

FieldEvaluationReason

Reason a field’s computed value is not available.

FieldKind

Known Word field instruction classes rwml distinguishes.

FieldRole

Whether a run is plain text or the cached result of a field.

FieldUnsupportedReason

Best-effort reason an imported field remains cached in the model.

HeaderFooterKind

Kind of running header/footer region recovered from a Word document.

MetafileFormat

Metafile image container observed in a package.

NoteKind

Kind of note recovered from a Word document.

PageNumberFormat

Display format for generated section page numbers.

RenderWarning

Human- and machine-readable warnings derived from features the current renderer cannot faithfully compute or draw yet.

RevisionKind

Kind of tracked revision marker observed in a .docx body.

RevisionView

Flat text view policy for tracked revisions.

SectionBreakKind

Section-break start behavior for a WordprocessingML section boundary.

SourceRegionKind

A coarse source subdocument region from the original Word file.

TableBorderSide

A physical side of a table border.

TableBorderStyle

Uniform table border line style.

TextDirection

Text flow direction for generated .docx section properties.

VCell

Vertical alignment of cell content.

VertAlign

Vertical alignment of a run (super/subscript).

Functions

extract_text

Convenience: decode .doc bytes into normalized plain text (all sub-documents — main body, then footnotes/endnotes/headers). Errors with Error::NoText if nothing indexable was found.

try_write_docx

Fallible variant of write_docx: returns the serialization error instead of an empty buffer if packaging ever fails. Prefer this when you need to surface write failures rather than silently emit nothing.

write_docx

Serialize a DocModel — one you built from data, built with DocBuilder, or read from a .doc/.docx — to a clean, Office-openable .docx byte buffer. This is the authoring entry point: construct a model (paragraphs/runs with fonts, sizes, colors; headings; styled/sized/shaded tables; images; page setup) and write a styled Word document. Available with the default docx feature.

Type Aliases

Result

Convenience alias.