command dump

dump

Serialize a document into a replayable batch script — the round-trip mechanism for editing a document by emit → modify → replay.

Synopsis

officecli dump <file> <path> [--format batch] [-o <out>] [--json]

Description

Walks the document and emits a JSON BatchItem[] array that, when replayed via officecli batch, reconstructs the source document. Currently .docx only.

The dump is portable: unstable IDs (paraId / rsidR / textId) and derived effective.* readbacks are filtered out. The OpenXML SDK regenerates IDs on save, so emit just stays out of the way.

Arguments

Name	Type	Required	Default	Description
file	path	Yes	-	Document path (.docx only in current release)
path	string	Yes	-	DOM path to dump. / emits the whole document; subtree paths emit just that subtree without bundling sibling resources. Supported: /, /body, /body/p[N], /body/tbl[N], and resource parts /theme, /settings, /numbering, /styles. Subtree emit uses last() xpath predicates so the script is safe to replay onto non-blank documents.

Options

Name	Type	Required	Default	Description
--format	string	No	batch	Output format. Currently only batch is supported.
-o / --out	path	No	-	Write output to file instead of stdout. Stdout output is the path on success.
--json	bool	No	false	Standard JSON envelope wrapper (the batch payload itself is always JSON).

What's emitted

v1.0.73 hardened the round-trip extensively: bookmarks (cross-paragraph spans), TOC fields with \t/\b switches, page-background color, hyperlink tooltip/tgtFrame/history, eastAsianLayout, paragraph-mark-only run formatting (markRPr.*), tables in headers/footers, columns + vAlign on inline section breaks, fldSimple/oMath inside hyperlinks/ins/del/footnotes, ruby/smartTag/customXml wrappers, cantSplit rows, tcW percent semantics, asymmetric tcMar padding, w:sym runs, noBreakHyphen/softHyphen, soft <w:br/> line breaks, ListItem SDT, MERGEFIELD whitespace quoting, complex-field HYPERLINKs, comment dates, PAGE field, header/footer types from sections, lineRule (atLeast/exact/auto), char-based indents, w14 ligatures/numForm/numSpacing, ins/del track-change attribution.

Layer	Mechanism
/styles	Emitted before body so paragraph styleId refs resolve on replay
/body paragraphs	Single-run paragraphs collapse into one add p row; multi-run paragraphs split into paragraph + run child rows
Tables and mixed body content	Typed add rows
Section page layout	set / on the root for page width/height/margins/columns/etc.
Inline section breaks	Section breaks inside the body emitted alongside their paragraph
docDefaults and document protection	Emitted alongside section layout
Headers and footers	Seed paragraph + appended content per-part
Comments / footnote refs / endnote refs	Anchored to the body paragraphs they reference
Numbering	Emitted wholesale via raw-set when document has list templates
Settings part	Emitted wholesale via raw-set
Theme part	Emitted wholesale via raw-set
Charts	Typed add (chartType + data string) — not raw-set
Pictures	Inlined as data URIs through the src= prop

Format keys are forwarded as-is; the OOXML schema reflection fallback in the Add side accepts arbitrary props, so emit doesn't need a per-key allowlist.

Examples

# Whole document to stdout
officecli dump report.docx /

# Write to a batch file
officecli dump report.docx / -o report.batch.json

# Subtree: just one paragraph
officecli dump report.docx /body/p[3]

# Subtree: a single table or a resource part
officecli dump report.docx /body/tbl[1]
officecli dump report.docx /numbering

# Round-trip: dump → batch
officecli dump report.docx / -o /tmp/r.json
officecli create rebuilt.docx --type docx
officecli batch rebuilt.docx --input /tmp/r.json

Notes

• --out - is treated as stdout (not a file literally named -).
• With --json, the envelope's data carries outputFile + itemCount metadata, not a bare path.
• TOC PAGEREF page numbers are preserved on round-trip but not recalculated — run refresh afterward to update them.

See Also

• batch — replay the emitted JSON (defaults to continue-on-error)
• refresh — recalculate TOC / PAGE fields after replay
• Word reference

---

Based on OfficeCLI v1.0.75