@cj-tech-master/excelts 9.5.4 → 9.5.5

package/dist/cjs/modules/word/reader/paragraph-section-parsers.js

@@ -0,0 +1,505 @@
+"use strict";
+/**
+ * DOCX Reader - Paragraph and Section Properties Parsers
+ *
+ * Pure parsers for `w:pPr` (paragraph properties) and `w:sectPr` (section
+ * properties). Both are large and have inter-dependencies (paragraphs may
+ * embed section properties); they're co-located here.
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.parseParagraphProperties = parseParagraphProperties;
+exports.parseSectionProperties = parseSectionProperties;
+const parse_utils_1 = require("./parse-utils");
+const properties_parsers_1 = require("./properties-parsers");
+// =============================================================================
+// Paragraph Properties
+// =============================================================================
+function parseParagraphProperties(pPrEl) {
+    const pPr = {};
+    const pStyleEl = (0, parse_utils_1.findChildNs)(pPrEl, "pStyle");
+    if (pStyleEl) {
+        pPr.style = (0, parse_utils_1.attrVal)(pStyleEl, "val");
+    }
+    const jcEl = (0, parse_utils_1.findChildNs)(pPrEl, "jc");
+    if (jcEl) {
+        pPr.alignment = (0, parse_utils_1.attrVal)(jcEl, "val");
+    }
+    if ((0, parse_utils_1.findChildNs)(pPrEl, "keepNext")) {
+        pPr.keepNext = true;
+    }
+    if ((0, parse_utils_1.findChildNs)(pPrEl, "keepLines")) {
+        pPr.keepLines = true;
+    }
+    if ((0, parse_utils_1.findChildNs)(pPrEl, "pageBreakBefore")) {
+        pPr.pageBreakBefore = true;
+    }
+    if ((0, parse_utils_1.findChildNs)(pPrEl, "bidi")) {
+        pPr.bidi = true;
+    }
+    // New boolean toggles
+    const ctxSp = (0, parse_utils_1.boolToggle)(pPrEl, "contextualSpacing");
+    if (ctxSp !== undefined) {
+        pPr.contextualSpacing = ctxSp;
+    }
+    const suppLn = (0, parse_utils_1.boolToggle)(pPrEl, "suppressLineNumbers");
+    if (suppLn !== undefined) {
+        pPr.suppressLineNumbers = suppLn;
+    }
+    const suppHyph = (0, parse_utils_1.boolToggle)(pPrEl, "suppressAutoHyphens");
+    if (suppHyph !== undefined) {
+        pPr.suppressAutoHyphens = suppHyph;
+    }
+    const mirr = (0, parse_utils_1.boolToggle)(pPrEl, "mirrorIndents");
+    if (mirr !== undefined) {
+        pPr.mirrorIndents = mirr;
+    }
+    const wc = (0, parse_utils_1.boolToggle)(pPrEl, "widowControl");
+    if (wc !== undefined) {
+        pPr.widowControl = wc;
+    }
+    const ww = (0, parse_utils_1.boolToggle)(pPrEl, "wordWrap");
+    if (ww !== undefined) {
+        pPr.wordWrap = ww;
+    }
+    const stg = (0, parse_utils_1.boolToggle)(pPrEl, "snapToGrid");
+    if (stg !== undefined) {
+        pPr.snapToGrid = stg;
+    }
+    const ofp = (0, parse_utils_1.boolToggle)(pPrEl, "overflowPunct");
+    if (ofp !== undefined) {
+        pPr.overflowPunctuation = ofp;
+    }
+    const topLinePunct = (0, parse_utils_1.boolToggle)(pPrEl, "topLinePunct");
+    if (topLinePunct !== undefined) {
+        pPr.topLinePunctuation = topLinePunct;
+    }
+    const kinsoku = (0, parse_utils_1.boolToggle)(pPrEl, "kinsoku");
+    if (kinsoku !== undefined) {
+        pPr.kinsoku = kinsoku;
+    }
+    const asd = (0, parse_utils_1.boolToggle)(pPrEl, "autoSpaceDE");
+    if (asd !== undefined) {
+        pPr.autoSpaceEastAsianText = asd;
+    }
+    const asdn = (0, parse_utils_1.boolToggle)(pPrEl, "autoSpaceDN");
+    if (asdn !== undefined) {
+        pPr.autoSpaceEastAsianDigit = asdn;
+    }
+    const textAlignEl = (0, parse_utils_1.findChildNs)(pPrEl, "textAlignment");
+    if (textAlignEl) {
+        pPr.textAlignment = (0, parse_utils_1.attrVal)(textAlignEl, "val");
+    }
+    const outlineLvlEl = (0, parse_utils_1.findChildNs)(pPrEl, "outlineLvl");
+    if (outlineLvlEl) {
+        pPr.outlineLevel = (0, parse_utils_1.attrInt)(outlineLvlEl, "val");
+    }
+    const textDirEl = (0, parse_utils_1.findChildNs)(pPrEl, "textDirection");
+    if (textDirEl) {
+        pPr.textDirection = (0, parse_utils_1.attrVal)(textDirEl, "val");
+    }
+    // Paragraph frame
+    const framePrEl = (0, parse_utils_1.findChildNs)(pPrEl, "framePr");
+    if (framePrEl) {
+        const f = {};
+        const dropCap = (0, parse_utils_1.attrVal)(framePrEl, "dropCap");
+        if (dropCap) {
+            f.dropCap = dropCap;
+        }
+        const lines = (0, parse_utils_1.attrInt)(framePrEl, "lines");
+        if (lines !== undefined) {
+            f.lines = lines;
+        }
+        const fw = (0, parse_utils_1.attrInt)(framePrEl, "w");
+        if (fw !== undefined) {
+            f.width = fw;
+        }
+        const fh = (0, parse_utils_1.attrInt)(framePrEl, "h");
+        if (fh !== undefined) {
+            f.height = fh;
+        }
+        const hSpace = (0, parse_utils_1.attrInt)(framePrEl, "hSpace");
+        if (hSpace !== undefined) {
+            f.hSpace = hSpace;
+        }
+        const vSpace = (0, parse_utils_1.attrInt)(framePrEl, "vSpace");
+        if (vSpace !== undefined) {
+            f.vSpace = vSpace;
+        }
+        const wrap = (0, parse_utils_1.attrVal)(framePrEl, "wrap");
+        if (wrap) {
+            f.wrap = wrap;
+        }
+        const hAnchor = (0, parse_utils_1.attrVal)(framePrEl, "hAnchor");
+        if (hAnchor) {
+            f.hAnchor = hAnchor;
+        }
+        const vAnchor = (0, parse_utils_1.attrVal)(framePrEl, "vAnchor");
+        if (vAnchor) {
+            f.vAnchor = vAnchor;
+        }
+        const x = (0, parse_utils_1.attrInt)(framePrEl, "x");
+        if (x !== undefined) {
+            f.x = x;
+        }
+        const xAlign = (0, parse_utils_1.attrVal)(framePrEl, "xAlign");
+        if (xAlign) {
+            f.xAlign = xAlign;
+        }
+        const y = (0, parse_utils_1.attrInt)(framePrEl, "y");
+        if (y !== undefined) {
+            f.y = y;
+        }
+        const yAlign = (0, parse_utils_1.attrVal)(framePrEl, "yAlign");
+        if (yAlign) {
+            f.yAlign = yAlign;
+        }
+        pPr.frame = f;
+    }
+    // Thematic break: check for bottom border with special pattern
+    const pBdrEl = (0, parse_utils_1.findChildNs)(pPrEl, "pBdr");
+    if (pBdrEl) {
+        const borders = {};
+        for (const side of ["top", "bottom", "left", "right", "between", "bar"]) {
+            const sideEl = (0, parse_utils_1.findChildNs)(pBdrEl, side);
+            if (sideEl) {
+                borders[side] = (0, properties_parsers_1.parseBorder)(sideEl);
+            }
+        }
+        pPr.borders = borders;
+    }
+    const spacingEl = (0, parse_utils_1.findChildNs)(pPrEl, "spacing");
+    if (spacingEl) {
+        const spacing = {};
+        const before = (0, parse_utils_1.attrInt)(spacingEl, "before");
+        const after = (0, parse_utils_1.attrInt)(spacingEl, "after");
+        const line = (0, parse_utils_1.attrInt)(spacingEl, "line");
+        const lineRule = (0, parse_utils_1.attrVal)(spacingEl, "lineRule");
+        if (before !== undefined) {
+            spacing.before = before;
+        }
+        if (after !== undefined) {
+            spacing.after = after;
+        }
+        if (line !== undefined) {
+            spacing.line = line;
+            // Per ECMA-376, w:lineRule defaults to "auto" when line is set
+            spacing.lineRule = (lineRule ?? "auto");
+        }
+        else if (lineRule) {
+            spacing.lineRule = lineRule;
+        }
+        const beforeAuto = (0, parse_utils_1.attrVal)(spacingEl, "beforeAutospacing");
+        if (beforeAuto === "1" || beforeAuto === "true") {
+            spacing.beforeAutoSpacing = true;
+        }
+        const afterAuto = (0, parse_utils_1.attrVal)(spacingEl, "afterAutospacing");
+        if (afterAuto === "1" || afterAuto === "true") {
+            spacing.afterAutoSpacing = true;
+        }
+        pPr.spacing = spacing;
+    }
+    const indEl = (0, parse_utils_1.findChildNs)(pPrEl, "ind");
+    if (indEl) {
+        const indent = {};
+        const left = (0, parse_utils_1.attrInt)(indEl, "left");
+        const right = (0, parse_utils_1.attrInt)(indEl, "right");
+        const hanging = (0, parse_utils_1.attrInt)(indEl, "hanging");
+        const firstLine = (0, parse_utils_1.attrInt)(indEl, "firstLine");
+        const start = (0, parse_utils_1.attrInt)(indEl, "start");
+        const end = (0, parse_utils_1.attrInt)(indEl, "end");
+        if (left !== undefined) {
+            indent.left = left;
+        }
+        if (right !== undefined) {
+            indent.right = right;
+        }
+        if (hanging !== undefined) {
+            indent.hanging = hanging;
+        }
+        if (firstLine !== undefined) {
+            indent.firstLine = firstLine;
+        }
+        if (start !== undefined) {
+            indent.start = start;
+        }
+        if (end !== undefined) {
+            indent.end = end;
+        }
+        pPr.indent = indent;
+    }
+    const numPrEl = (0, parse_utils_1.findChildNs)(pPrEl, "numPr");
+    if (numPrEl) {
+        const ilvlEl = (0, parse_utils_1.findChildNs)(numPrEl, "ilvl");
+        const numIdEl = (0, parse_utils_1.findChildNs)(numPrEl, "numId");
+        // Per OOXML schema, numId is required but ilvl is optional (defaults to 0).
+        if (numIdEl) {
+            pPr.numbering = {
+                level: ilvlEl ? ((0, parse_utils_1.attrInt)(ilvlEl, "val") ?? 0) : 0,
+                numId: (0, parse_utils_1.attrInt)(numIdEl, "val") ?? 0
+            };
+        }
+    }
+    const tabsEl = (0, parse_utils_1.findChildNs)(pPrEl, "tabs");
+    if (tabsEl) {
+        const tabs = [];
+        for (const tabEl of (0, parse_utils_1.findChildrenNs)(tabsEl, "tab")) {
+            tabs.push({
+                type: ((0, parse_utils_1.attrVal)(tabEl, "val") ?? "left"),
+                position: (0, parse_utils_1.attrInt)(tabEl, "pos") ?? 0,
+                leader: (0, parse_utils_1.attrVal)(tabEl, "leader")
+            });
+        }
+        if (tabs.length > 0) {
+            pPr.tabs = tabs;
+        }
+    }
+    const shdEl = (0, parse_utils_1.findChildNs)(pPrEl, "shd");
+    if (shdEl) {
+        pPr.shading = (0, properties_parsers_1.parseShading)(shdEl);
+    }
+    const rPrEl = (0, parse_utils_1.findChildNs)(pPrEl, "rPr");
+    if (rPrEl) {
+        pPr.markRunProperties = (0, properties_parsers_1.parseRunProperties)(rPrEl);
+    }
+    const sectPrEl = (0, parse_utils_1.findChildNs)(pPrEl, "sectPr");
+    if (sectPrEl) {
+        pPr.sectionProperties = parseSectionProperties(sectPrEl);
+    }
+    // Conditional formatting style mask
+    const cnfStyleEl = (0, parse_utils_1.findChildNs)(pPrEl, "cnfStyle");
+    if (cnfStyleEl) {
+        pPr.cnfStyle = (0, parse_utils_1.attrVal)(cnfStyleEl, "val");
+    }
+    // Paragraph property change
+    const pPrChangeEl = (0, parse_utils_1.findChildNs)(pPrEl, "pPrChange");
+    if (pPrChangeEl) {
+        const rev = (0, properties_parsers_1.parseRevisionInfo)(pPrChangeEl);
+        if (rev) {
+            const prevPPrEl = (0, parse_utils_1.findChildNs)(pPrChangeEl, "pPr");
+            pPr.propertyChange = {
+                revision: rev,
+                previousProperties: prevPPrEl ? parseParagraphProperties(prevPPrEl) : undefined
+            };
+        }
+    }
+    // Paragraph mark insertion/deletion (w:pPr > w:rPr > w:ins/w:del)
+    const rPrInPPr = (0, parse_utils_1.findChildNs)(pPrEl, "rPr");
+    if (rPrInPPr) {
+        const insEl = (0, parse_utils_1.findChildNs)(rPrInPPr, "ins");
+        if (insEl) {
+            const rev = (0, properties_parsers_1.parseRevisionInfo)(insEl);
+            if (rev) {
+                pPr.paragraphInsertion = rev;
+            }
+        }
+        const delEl = (0, parse_utils_1.findChildNs)(rPrInPPr, "del");
+        if (delEl) {
+            const rev = (0, properties_parsers_1.parseRevisionInfo)(delEl);
+            if (rev) {
+                pPr.paragraphDeletion = rev;
+            }
+        }
+    }
+    return pPr;
+}
+// =============================================================================
+// Section Properties
+// =============================================================================
+function parseSectionProperties(sectPrEl) {
+    const sect = {};
+    const pgSzEl = (0, parse_utils_1.findChildNs)(sectPrEl, "pgSz");
+    if (pgSzEl) {
+        // Per ECMA-376, w:orient defaults to "portrait" when absent
+        const orient = (0, parse_utils_1.attrVal)(pgSzEl, "orient");
+        // Per ECMA-376 17.6.13, w:w/h are required attributes; default to US Letter
+        // (12240 × 15840 twips) when malformed input omits them.
+        sect.pageSize = {
+            width: (0, parse_utils_1.attrInt)(pgSzEl, "w") ?? 12240,
+            height: (0, parse_utils_1.attrInt)(pgSzEl, "h") ?? 15840,
+            orientation: orient === "landscape" ? "landscape" : "portrait"
+        };
+    }
+    const pgMarEl = (0, parse_utils_1.findChildNs)(sectPrEl, "pgMar");
+    if (pgMarEl) {
+        sect.margins = {
+            top: (0, parse_utils_1.attrInt)(pgMarEl, "top") ?? 1440,
+            right: (0, parse_utils_1.attrInt)(pgMarEl, "right") ?? 1440,
+            bottom: (0, parse_utils_1.attrInt)(pgMarEl, "bottom") ?? 1440,
+            left: (0, parse_utils_1.attrInt)(pgMarEl, "left") ?? 1440,
+            header: (0, parse_utils_1.attrInt)(pgMarEl, "header"),
+            footer: (0, parse_utils_1.attrInt)(pgMarEl, "footer"),
+            gutter: (0, parse_utils_1.attrInt)(pgMarEl, "gutter")
+        };
+    }
+    const typeEl = (0, parse_utils_1.findChildNs)(sectPrEl, "type");
+    if (typeEl) {
+        sect.breakType = (0, parse_utils_1.attrVal)(typeEl, "val");
+    }
+    const colsEl = (0, parse_utils_1.findChildNs)(sectPrEl, "cols");
+    if (colsEl) {
+        const cols = {};
+        cols.space = (0, parse_utils_1.attrInt)(colsEl, "space");
+        cols.count = (0, parse_utils_1.attrInt)(colsEl, "num");
+        const eqw = (0, parse_utils_1.attrVal)(colsEl, "equalWidth");
+        if (eqw !== undefined) {
+            cols.equalWidth = eqw === "1";
+        }
+        const sep = (0, parse_utils_1.attrVal)(colsEl, "sep");
+        if (sep === "1" || sep === "true") {
+            cols.separator = true;
+        }
+        const colDefs = (0, parse_utils_1.findChildrenNs)(colsEl, "col");
+        if (colDefs.length > 0) {
+            cols.columns = colDefs.map(c => ({
+                width: (0, parse_utils_1.attrInt)(c, "w") ?? 0,
+                space: (0, parse_utils_1.attrInt)(c, "space")
+            }));
+        }
+        sect.columns = cols;
+    }
+    if ((0, parse_utils_1.findChildNs)(sectPrEl, "titlePg")) {
+        sect.titlePage = true;
+    }
+    const pgNumEl = (0, parse_utils_1.findChildNs)(sectPrEl, "pgNumType");
+    if (pgNumEl) {
+        sect.pageNumbering = {
+            start: (0, parse_utils_1.attrInt)(pgNumEl, "start"),
+            format: (0, parse_utils_1.attrVal)(pgNumEl, "fmt")
+        };
+    }
+    // Page borders
+    const pgBordersEl = (0, parse_utils_1.findChildNs)(sectPrEl, "pgBorders");
+    if (pgBordersEl) {
+        const pb = {};
+        for (const side of ["top", "left", "bottom", "right"]) {
+            const sideEl = (0, parse_utils_1.findChildNs)(pgBordersEl, side);
+            if (sideEl) {
+                pb[side] = (0, properties_parsers_1.parseBorder)(sideEl);
+            }
+        }
+        const display = (0, parse_utils_1.attrVal)(pgBordersEl, "display");
+        if (display) {
+            pb.display = display;
+        }
+        const offsetFrom = (0, parse_utils_1.attrVal)(pgBordersEl, "offsetFrom");
+        if (offsetFrom) {
+            pb.offsetFrom = offsetFrom;
+        }
+        const zOrder = (0, parse_utils_1.attrVal)(pgBordersEl, "zOrder");
+        if (zOrder) {
+            pb.zOrder = zOrder;
+        }
+        sect.pageBorders = pb;
+    }
+    // Vertical alignment
+    const vAlignEl = (0, parse_utils_1.findChildNs)(sectPrEl, "vAlign");
+    if (vAlignEl) {
+        sect.verticalAlign = (0, parse_utils_1.attrVal)(vAlignEl, "val");
+    }
+    // Text direction
+    const textDirEl = (0, parse_utils_1.findChildNs)(sectPrEl, "textDirection");
+    if (textDirEl) {
+        sect.textDirection = (0, parse_utils_1.attrVal)(textDirEl, "val");
+    }
+    // Bidi
+    const bidiToggle = (0, parse_utils_1.boolToggle)(sectPrEl, "bidi");
+    if (bidiToggle !== undefined) {
+        sect.bidi = bidiToggle;
+    }
+    // RTL gutter
+    const rtlGutterEl = (0, parse_utils_1.findChildNs)(sectPrEl, "rtlGutter");
+    if (rtlGutterEl) {
+        sect.rtlGutter = true;
+    }
+    // Form protection
+    const formProtEl = (0, parse_utils_1.findChildNs)(sectPrEl, "formProt");
+    if (formProtEl) {
+        const v = (0, parse_utils_1.attrVal)(formProtEl, "val");
+        sect.formProtection = v === "1" || v === "true";
+    }
+    // Document grid
+    const docGridEl = (0, parse_utils_1.findChildNs)(sectPrEl, "docGrid");
+    if (docGridEl) {
+        const dg = {};
+        const linePitch = (0, parse_utils_1.attrInt)(docGridEl, "linePitch");
+        if (linePitch !== undefined) {
+            dg.linePitch = linePitch;
+        }
+        const charSpace = (0, parse_utils_1.attrInt)(docGridEl, "charSpace");
+        if (charSpace !== undefined) {
+            dg.charSpace = charSpace;
+        }
+        const gridType = (0, parse_utils_1.attrVal)(docGridEl, "type");
+        if (gridType) {
+            dg.type = gridType;
+        }
+        sect.docGrid = dg;
+    }
+    // Line numbers
+    const lnNumEl = (0, parse_utils_1.findChildNs)(sectPrEl, "lnNumType");
+    if (lnNumEl) {
+        const ln = {};
+        const countBy = (0, parse_utils_1.attrInt)(lnNumEl, "countBy");
+        if (countBy !== undefined) {
+            ln.countBy = countBy;
+        }
+        const start = (0, parse_utils_1.attrInt)(lnNumEl, "start");
+        if (start !== undefined) {
+            ln.start = start;
+        }
+        const restart = (0, parse_utils_1.attrVal)(lnNumEl, "restart");
+        if (restart) {
+            ln.restart = restart;
+        }
+        const distance = (0, parse_utils_1.attrInt)(lnNumEl, "distance");
+        if (distance !== undefined) {
+            ln.distance = distance;
+        }
+        sect.lineNumbers = ln;
+    }
+    // Footnote properties
+    const fnPrEl = (0, parse_utils_1.findChildNs)(sectPrEl, "footnotePr");
+    if (fnPrEl) {
+        sect.footnoteProperties = (0, parse_utils_1.parseNoteProperties)(fnPrEl);
+    }
+    // Endnote properties
+    const enPrEl = (0, parse_utils_1.findChildNs)(sectPrEl, "endnotePr");
+    if (enPrEl) {
+        sect.endnoteProperties = (0, parse_utils_1.parseNoteProperties)(enPrEl);
+    }
+    // Headers/Footers refs
+    const headerRefs = [];
+    for (const hRef of (0, parse_utils_1.findChildrenNs)(sectPrEl, "headerReference")) {
+        headerRefs.push({
+            type: ((0, parse_utils_1.attrVal)(hRef, "type") ?? "default"),
+            rId: hRef.attributes["r:id"] ?? ""
+        });
+    }
+    if (headerRefs.length > 0) {
+        sect.headers = headerRefs;
+    }
+    const footerRefs = [];
+    for (const fRef of (0, parse_utils_1.findChildrenNs)(sectPrEl, "footerReference")) {
+        footerRefs.push({
+            type: ((0, parse_utils_1.attrVal)(fRef, "type") ?? "default"),
+            rId: fRef.attributes["r:id"] ?? ""
+        });
+    }
+    if (footerRefs.length > 0) {
+        sect.footers = footerRefs;
+    }
+    // sectPrChange (track changes for section properties)
+    const sectPrChangeEl = (0, parse_utils_1.findChildNs)(sectPrEl, "sectPrChange");
+    if (sectPrChangeEl) {
+        const rev = (0, properties_parsers_1.parseRevisionInfo)(sectPrChangeEl);
+        if (rev) {
+            const prevSectPrEl = (0, parse_utils_1.findChildNs)(sectPrChangeEl, "sectPr");
+            const change = {
+                revision: rev,
+                previousProperties: prevSectPrEl ? parseSectionProperties(prevSectPrEl) : undefined
+            };
+            sect.propertyChange = change;
+        }
+    }
+    return sect;
+}