reamkit 1.33.1 → 1.34.0

package/dist/esm/word/doc/doc-text.d.ts

@@ -6,6 +6,14 @@ export interface DocCharProps {
     readonly underlineKul?: number;
     /** Font size in half-points. */
     readonly sizeHalfPts?: number;
+    /** The font of ASCII characters, by name — `sprmCRgFtc0`. */
+    readonly fontAscii?: string;
+    /** The font of East Asian characters — `sprmCRgFtc1`. */
+    readonly fontEastAsia?: string;
+    /** The font of the characters neither ASCII nor East Asian — `sprmCRgFtc2`. */
+    readonly fontOther?: string;
+    /** The font of complex-script characters — `sprmCFtcBi`. */
+    readonly fontBi?: string;
     /** The character is special — `sprmCFSpec` (e.g. a picture placeholder). */
     readonly fSpec?: boolean;
     /** `sprmCPicLocation` — offset into the `Data` stream of the picture's PICF. */
@@ -80,6 +88,8 @@ export interface DocParaProps {
     readonly keepNext?: boolean;
     /** `sprmPFKeep` — every line on one page. */
     readonly keepLines?: boolean;
+    /** `sprmPFPageBreakBefore` — the paragraph starts a new page. */
+    readonly pageBreakBefore?: boolean;
 }
 /** One maximal same-formatting run of text, or an embedded picture. */
 export interface DocRun {
@@ -87,6 +97,15 @@ export interface DocRun {
     readonly props: DocCharProps;
     /** Set instead of `text` when the run is an embedded picture. */
     readonly picture?: DocPicture;
+    /** The run is a manual page break (0x0C), alone in a run with no text. */
+    readonly pageBreak?: boolean;
+    /** The run is a manual column break (0x0E), alone in a run whose text is `\n`. */
+    readonly columnBreak?: boolean;
+    /**
+     * The run is the result of a PAGE or NUMPAGES field, whose `text` is the
+     * number Word cached when it last saved.
+     */
+    readonly field?: 'PAGE' | 'NUMPAGES';
 }
 /** One paragraph: its runs plus the decoded paragraph properties. */
 export interface DocParagraph {
@@ -97,11 +116,59 @@ export interface DocParagraph {
      * closes a table cell. With `props.rowEnd` it also closes the row.
      */
     readonly cellMark?: boolean;
+    /**
+     * The paragraph was ended by the section mark of the (zero-based) section it
+     * is the last of — the index into {@link DocContent.sections}.
+     */
+    readonly sectionEnd?: number;
+    /** The paragraph mark's own character properties: what an empty line is set in. */
+    readonly mark?: DocCharProps;
+}
+/** One section of the main document (§2.8.26 PlcfSed): what its SEP says. */
+export interface DocSection {
+    /**
+     * `sprmSBkc` — where the section starts: 0 on the page in hand, 1 in the next
+     * column, 2 on a new page, 3 on an even page, 4 on an odd page. Absent when
+     * the SEP says nothing, which is a new page.
+     */
+    readonly bkc?: number;
+    /** The page size in points (xaPage/yaPage), when present and sane. */
+    readonly pageSize?: {
+        readonly widthPt: number;
+        readonly heightPt: number;
+    };
+    /** `sprmSFTitlePage` — the first page shows the first-page header and footer. */
+    readonly titlePage?: boolean;
+    /** The margins the SEP states, in twips; each absent one is Word's default. */
+    readonly margins?: DocMargins;
+    /**
+     * `sprmSFPgnRestart` with `sprmSPgnStart97` — the number the section's first
+     * page is printed with, when the section starts its count again; absent, the
+     * count runs on from the section before.
+     */
+    readonly pageNumberStart?: number;
+    /** `sprmSNfcPgn` — the MSONFC the section's page numbers print in; absent, arabic. */
+    readonly pageNumberNfc?: number;
+}
+/**
+ * A section's margins in twips (§2.6.4 sprmSDxaLeft … sprmSDzaGutter). A top or
+ * bottom margin is stored signed, negative for one the header or footer may not
+ * push; its size is the same either way.
+ */
+export interface DocMargins {
+    readonly leftTwips?: number;
+    readonly rightTwips?: number;
+    readonly topTwips?: number;
+    readonly bottomTwips?: number;
+    readonly headerTwips?: number;
+    readonly footerTwips?: number;
+    readonly gutterTwips?: number;
 }
 /**
- * The section's header/footer stories (best-effort: the binary story ordering
- * cannot be validated against a real file, so only well-formed, non-empty stories
- * are surfaced). `default` is the odd-page (primary) header/footer.
+ * One section's header/footer stories ([MS-DOC] Plcfhdd: six per section, after
+ * the six separator stories). `default` is the odd-page (primary) one. A story
+ * the section leaves empty is the previous section's, the very same array; one
+ * that holds no text is absent.
  */
 export interface DocHeaderFooters {
     readonly defaultHeader?: ReadonlyArray<DocParagraph>;
@@ -121,6 +188,8 @@ export interface DocLvl {
     readonly nfc: number;
     readonly iStartAt: number;
     readonly xst: ReadonlyArray<number>;
+    /** `ixchFollow` — what follows the number text: 0 a tab, 1 a space, 2 nothing. */
+    readonly follow: number;
 }
 /**
  * The document's list tables (DOC-10): an LFO → lsid map (a paragraph's 1-based
@@ -136,23 +205,23 @@ export interface DocListTables {
 /** Everything extracted from a `.doc`'s `WordDocument` stream. */
 export interface DocContent {
     readonly paragraphs: ReadonlyArray<DocParagraph>;
-    readonly headerFooters?: DocHeaderFooters;
+    /** Each section's header/footer stories, in section order. */
+    readonly headerFooters?: ReadonlyArray<DocHeaderFooters>;
+    /** `Dop.fFacingPages` — even pages show the even header and footer. */
+    readonly facingPages?: boolean;
     /** The list tables resolving a paragraph's (ilfo, ilvl) to its number format. */
     readonly listTables?: DocListTables;
     /**
-     * The first section's page size in points (xaPage/yaPage from its SEP), when
-     * present and sane; absent ⇒ the reader's default (US Letter) applies.
+     * Every section of the main document, in order, when the PlcfSed is usable. A
+     * section with no page size takes the reader's default (US Letter).
      */
-    readonly pageSize?: {
-        readonly widthPt: number;
-        readonly heightPt: number;
-    };
+    readonly sections?: ReadonlyArray<DocSection>;
     /** The file is encrypted/obfuscated — text cannot be read without the key. */
     readonly encrypted: boolean;
 }
 /**
  * Extract a `.doc`'s `WordDocument` stream into its main-document paragraphs plus
- * the section's header/footer stories, list tables and first-section page size.
+ * the section's header/footer stories, list tables and sections.
  * Joins the piece table (CLX → CPs/FCs), CHPX run formatting and PAPX paragraph
  * formatting, splitting at the CR / cell marks. Every offset is bounds-checked, so
  * a structurally odd file yields no text rather than throwing.