Module types 

Wire types for the Docs v1 REST API.

Field naming follows Docs’ camelCase JSON via per-field #[serde(rename = "...")], mirroring src/drive/types.rs and src/drive/sheets/types.rs. Only the subset the CLI actually renders is modelled; every unmodelled field is tolerated and dropped, so a Google response gaining a field never breaks a parse.

Four shapes here are load-bearing and easy to get wrong:

• startIndex is absent, not 0, for the first element. Docs serialises proto3 JSON, which omits zero-valued integers, so every real documents.get has body.content[0] (the leading sectionBreak) with no startIndex key at all. Every index field is therefore Option<i64>, read through the start_index()/end_index() accessors. This is the direct analogue of ValueRange::values being absent-not-empty for a blank sheet.
• Indices are UTF-16 code units, and endIndex is exclusive. Not chars, not bytes. Any code deriving an index from a &str must use s.encode_utf16().count(). The difference is invisible in ASCII and wrong the moment a document contains an emoji or a CJK astral character, so only the anchor resolver computes one — see crate::drive::docs::anchor.
• An empty message still serialises as {} when present. A sectionBreak with no fields set arrives as "sectionBreak": {}, so the union variants this module does not inspect are modelled as Option<serde_json::Value>: Some(Object({})) when present, None when absent, which is exactly the discriminator semantics needed.
• tabs and body are mutually exclusive. With includeTabsContent=true the content lives under tabs[] and the top-level body is absent; without it, only body is populated. That is normalised once, in Document::resolved_tabs, rather than at every call site.

Structs

Body

A document body — the main index segment.

Bullet

A list-item marker.

Document

A document’s full structural model, from documents.get.

DocumentTab

The document-flavoured content of a tab.

Footer

One page footer — same shape as Header; Google models these as distinct message types even though the field is identical, so this mirrors that instead of aliasing.

Footnote

One footnote body — same shape again. The footnote’s own id is the map key it is stored under, not a field on this struct.

Header

One page header — a segment addressed by the map key it is stored under.

InlineObjectElement

A reference to an inline object.

NamedRange

One named range.

NamedRanges

Every named range sharing one name.

Paragraph

A paragraph and its inline elements.

ParagraphElement

One inline element of a paragraph.

ParagraphStyle

A paragraph’s style.

Range

An index span within one segment of one tab.

ResolvedSegment

One header, footer or footnote segment, resolved and ready to flatten.

ResolvedTab

One tab’s content, with the legacy single-body response normalised into the same shape so callers never branch on which form arrived.

StructuralElement

One structural element: exactly one of the variant fields is present, and which one is the discriminator.

Tab

One tab of a document.

TabProperties

A tab’s identity and position within the tab tree.

Table

A table.

TableCell

One cell of a table row.

TableCellStyle

Read-only cell spans; other formatting remains outside the edit model.

TableOfContents

A table of contents.

TableRow

One row of a table.

TextRun

A run of text with uniform styling.