@docx4j/core-ts 0.6.0 → 0.7.0

package/dist/model/text/Extractor.mjs

@@ -1,8 +1,7 @@
 import { AlternativeFormatInputPart } from '../../parts/BinaryPart.mjs';
-import { Emulator } from '../listnumbering/Emulator.mjs';
 import { runItemsOf, selectedBranchOf, segmentsOf, hiddenRunsOf, SDT_TYPES } from '../content/tree.mjs';
 import { TextBlock, TextRun, TextRow, TextCell, TextField, TextNumbering, TextRevision, TextWarning, TextComment, TextImage, TextResult, } from './TextResult.mjs';
-import { calendarText } from './TextContext.mjs';
+import { revisionDateText } from '../content/tracking.mjs';
 import { normalize } from './normalize.mjs';
 const W_NS = 'http://schemas.openxmlformats.org/wordprocessingml/2006/main';
 const R_NS = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships';
@@ -146,8 +145,9 @@ function asciiFont(rPr) {
     var _a;
     return (_a = rPr === null || rPr === void 0 ? void 0 : rPr.rFonts) === null || _a === void 0 ? void 0 : _a.ascii;
 }
+/** A revision with its author and its stamp as REQ-702 writes it: the author's local time, the offset where known (CR-012 section 23.7). */
 function revisionOf(tc, type) {
-    return new TextRevision(type, tc.author, calendarText(tc.date));
+    return new TextRevision(type, tc.author, revisionDateText(tc.date, tc.dateUtc));
 }
 /**
  * Each item's place in `Paragraph.text` (CR-012 D2, section 4.4): the content API's own reader since phase 4
@@ -172,6 +172,8 @@ class FieldState {
         this.locked = false;
         /** A table of contents under `toc: false` (D12): nothing of it is read. */
         this.toc = false;
+        /** A SET field: its stored result is never shown (Word hides it; REQ-505, CR-012 section 26). */
+        this.hiddenResult = false;
         this.codeStart = 0;
         this.previousFloor = 0;
     }
@@ -185,6 +187,7 @@ class FieldState {
         f.computed = this.computed;
         f.link = this.link;
         f.toc = this.toc;
+        f.hiddenResult = this.hiddenResult;
         return f;
     }
 }
@@ -220,6 +223,8 @@ class Story {
         this.pending = [];
         this.tableDepth = 0;
         this.noteNumber = 0;
+        if (part)
+            x.storyParts.add(part);
     }
     /** The resolver the walk reads styles with, for `officeSpans`' hidden rule. */
     get resolver() {
@@ -268,6 +273,8 @@ export class Extractor {
         this.context = context;
         this.warnings = [];
         this.warned = new Set();
+        /** The parts this reading took a story from, for what the walk can't meet (`partWarningsUnmet`). */
+        this.storyParts = new Set();
         /** What every address this extraction makes starts with. */
         this.root = '';
         /** The run properties a paragraph style and character style give, by their ids. */
@@ -283,19 +290,39 @@ export class Extractor {
     }
     // --- the package ------------------------------------------------------------------------------
     extract() {
-        var _a;
+        var _a, _b, _c, _d, _e, _f, _g;
         const out = [];
         const comments = new Map();
+        // a name the container held twice, the first copy read (REQ-206, CR-024): the package's, so every reading of it says so
+        for (const d of (_b = (_a = this.trees.pkg) === null || _a === void 0 ? void 0 : _a.duplicateEntries) !== null && _b !== void 0 ? _b : []) {
+            this.warn(new TextWarning(TextWarning.PACKAGE_DUPLICATE_ENTRY, undefined, undefined, undefined, d.entry, d.count));
+        }
+        // more than one officeDocument relationship, the last's target read (REQ-206's statement, CR-024 section 9)
+        const mainParts = (_c = this.trees.pkg) === null || _c === void 0 ? void 0 : _c.multipleMainParts;
+        if (mainParts)
+            this.warn(new TextWarning(TextWarning.PACKAGE_MULTIPLE_MAIN_PARTS, undefined, undefined, undefined, mainParts.partName, mainParts.count));
+        // an External main relationship dropped, this package's own repair (REQ-206's NOTE, CR-031), and a relationships part read
+        // by local name (CR-031 section 2.3): the load's, so every reading says so
+        const external = (_d = this.trees.pkg) === null || _d === void 0 ? void 0 : _d.externalMainPart;
+        if (external)
+            this.warn(new TextWarning(TextWarning.PACKAGE_EXTERNAL_MAIN_PART, undefined, undefined, undefined, external.partName, external.urls[0]));
+        for (const r of (_f = (_e = this.trees.pkg) === null || _e === void 0 ? void 0 : _e.relationshipsNamespaces) !== null && _f !== void 0 ? _f : []) {
+            this.warn(new TextWarning(TextWarning.PACKAGE_RELATIONSHIPS_NAMESPACE, undefined, undefined, undefined, r.partName, r.count));
+        }
         const metadata = this.metadata();
         const main = this.trees.main;
         const body = main === null || main === void 0 ? void 0 : main.document.body;
-        if (!main || !body)
+        if (!main || !body) {
+            if (main)
+                this.storyParts.add(main.part);
+            this.partWarningsUnmet();
             return new TextResult(out, comments, this.warnings, metadata, this.options);
+        }
         // the main story, each text box after its anchor; sections recorded as they end
         const story = new Story(this, 'MAIN', `${this.root}body`, main.part);
         story.sections = [];
         const mainBlocks = [];
-        this.blocks((_a = body.content) !== null && _a !== void 0 ? _a : [], `${this.root}body`, story, mainBlocks);
+        this.blocks((_g = body.content) !== null && _g !== void 0 ? _g : [], `${this.root}body`, story, mainBlocks);
         story.sections.push(body.sectPr);
         this.context.mainRead = true;
         const headers = this.options.includes.has('headers');
@@ -315,19 +342,35 @@ export class Extractor {
         // comments, rendered at their references (REQ-504)
         if (this.commentsIncluded())
             this.comments(comments);
+        this.partWarningsUnmet();
         return new TextResult(out, comments, this.warnings, metadata, this.options);
     }
-    /** The result's metadata: `fields_update_on_open` and the document properties (REQ-801, REQ-515). */
+    /** The result's metadata: `fields_update_on_open`, `repaired` and the document properties (REQ-801, REQ-207, REQ-515). */
     metadata() {
-        var _a;
-        const metadata = { [TextResult.FIELDS_UPDATE_ON_OPEN]: this.fieldsUpdateOnOpen() };
-        Object.assign(metadata, (_a = this.trees.properties) !== null && _a !== void 0 ? _a : {});
+        var _a, _b, _c;
+        const metadata = {
+            [TextResult.FIELDS_UPDATE_ON_OPEN]: this.fieldsUpdateOnOpen(),
+            [TextResult.REPAIRED]: (_b = (_a = this.trees.pkg) === null || _a === void 0 ? void 0 : _a.repaired) !== null && _b !== void 0 ? _b : 'none',
+        };
+        Object.assign(metadata, (_c = this.trees.properties) !== null && _c !== void 0 ? _c : {});
+        // a properties part read for the metadata written (REQ-515), where it could not be read (WD 0.76)
+        if (this.options.metadata)
+            for (const role of ['core', 'app', 'custom'])
+                this.consult(role);
         return metadata;
     }
+    /** `w:updateFields`, which every result's metadata carries: so a settings part that cannot be read is needed by every reading. */
     fieldsUpdateOnOpen() {
         var _a;
+        this.consult('settings');
         return on((_a = this.trees.settings) === null || _a === void 0 ? void 0 : _a.updateFields);
     }
+    /** A part the reading needs here: `part.not_read`, once, where it could not be read (REQ-801; CR-012 section 28). */
+    consult(role) {
+        const partName = this.trees.notRead.get(role);
+        if (partName !== undefined)
+            this.warn(new TextWarning(TextWarning.PART_NOT_READ, undefined, undefined, undefined, partName));
+    }
     commentsIncluded() {
         return this.view === 'markup' || this.options.comments;
     }
@@ -346,31 +389,35 @@ export class Extractor {
         const metadata = {};
         if (this.trees.pkg !== undefined)
             metadata[TextResult.FIELDS_UPDATE_ON_OPEN] = this.fieldsUpdateOnOpen();
+        this.partWarningsUnmet();
         return new TextResult(out, new Map(), this.warnings, metadata, this.options);
     }
     /** The notes of one kind, in the context's reference order (`Texts.of(notesPart)`). */
     extractNotes(type) {
         const out = [];
         this.notes(type, out);
+        this.partWarningsUnmet();
         return new TextResult(out, new Map(), this.warnings, { [TextResult.FIELDS_UPDATE_ON_OPEN]: this.fieldsUpdateOnOpen() }, this.options);
     }
     /** The comments (`Texts.of(commentsPart)`): what the view includes, in the comments part's order. */
     extractComments() {
         const comments = new Map();
         this.comments(comments);
+        this.partWarningsUnmet();
         return new TextResult([], comments, this.warnings, { [TextResult.FIELDS_UPDATE_ON_OPEN]: this.fieldsUpdateOnOpen() }, this.options);
     }
     // --- the glossary (REQ-516) -------------------------------------------------------------------
     glossary(out) {
-        var _a, _b, _c;
+        var _a, _b, _c, _d;
+        this.storyRead((_a = this.trees.main) === null || _a === void 0 ? void 0 : _a.part.glossaryDocumentPart);
         const g = this.trees.glossary;
-        const parts = (_b = (_a = g === null || g === void 0 ? void 0 : g.tree.docParts) === null || _a === void 0 ? void 0 : _a.docPart) !== null && _b !== void 0 ? _b : [];
+        const parts = (_c = (_b = g === null || g === void 0 ? void 0 : g.tree.docParts) === null || _b === void 0 ? void 0 : _b.docPart) !== null && _c !== void 0 ? _c : [];
         for (let i = 0; i < parts.length; i++) {
             const body = parts[i].docPartBody;
             if (!body)
                 continue;
             const prefix = `${this.root}glossary:${i}`;
-            this.blocks((_c = body.content) !== null && _c !== void 0 ? _c : [], prefix, new Story(this, 'GLOSSARY', prefix, g.part), out);
+            this.blocks((_d = body.content) !== null && _d !== void 0 ? _d : [], prefix, new Story(this, 'GLOSSARY', prefix, g.part), out);
         }
     }
     // --- block level ------------------------------------------------------------------------------
@@ -378,11 +425,15 @@ export class Extractor {
     blocks(items, prefix, story, sink, state = PLAIN) {
         if (!items)
             return;
+        const dropCap = {};
         for (let i = 0; i < items.length; i++)
-            i = this.block(items, i, prefix, story, sink, state);
+            i = this.block(items, i, prefix, story, sink, state, dropCap);
     }
-    /** The block at `items[i]` into `sink`, with the paragraphs a view joins to it; the index of the last item it took. */
-    block(items, i, prefix, story, sink, state) {
+    /**
+     * The block at `items[i]` into `sink`, with the paragraphs a view joins to it; the index of the last item it took.
+     * `dropCap` is a drop cap's frame paragraph just read in the same container, for the paragraph next to take (REQ-521).
+     */
+    block(items, i, prefix, story, sink, state, dropCap = {}) {
         var _a, _b, _c;
         const item = items[i];
         if (this.removes(item))
@@ -390,6 +441,8 @@ export class Extractor {
         const v = unwrap(item);
         const address = `${prefix}/${i}`;
         const tn = typeOf(v);
+        if (tn !== P && !isMarker(item))
+            dropCap.block = undefined; // a table, a control, the end: left as it is
         if (tn === P) {
             const chain = [v];
             const addresses = [address];
@@ -411,9 +464,14 @@ export class Extractor {
             if ((_a = this.plan) === null || _a === void 0 ? void 0 : _a.joinsIntoTable(chain[chain.length - 1])) {
                 // a deleted mark with a table next and text left: into the table's first cell (docx4j CR-043 section 17)
                 this.pendingTableJoin = { chain, addresses };
+                dropCap.block = undefined;
                 return j;
             }
-            this.paragraph(chain, addresses, story, sink, state);
+            const taking = dropCap.block;
+            dropCap.block = undefined;
+            const block = this.paragraph(chain, addresses, story, sink, state, taking);
+            if (taking === undefined && block !== undefined && this.dropCapFrame(chain, block))
+                dropCap.block = block;
             return j;
         }
         if (tn === TBL) {
@@ -424,6 +482,7 @@ export class Extractor {
                 this.pendingTableJoin = undefined;
                 this.paragraph(chain, addresses, story, sink, state);
             }
+            dropCap.block = undefined;
         }
         else if (isSdt(v) && this.unwraps(item)) {
             // an inserted control rejected: its content in its place, its properties gone
@@ -454,7 +513,7 @@ export class Extractor {
         }
         else if (tn === AC) {
             const ac = v;
-            this.compareBranches(ac, address);
+            this.compareBranches(ac, address, story);
             this.blocks(selectedBranchOf(ac), address, story, sink, state);
         }
         else if (revisionHolder(item) !== undefined) {
@@ -532,15 +591,20 @@ export class Extractor {
         }
         return -1;
     }
-    paragraph(chain, addresses, story, sink, state) {
+    /**
+     * A paragraph chain's block into the sink, returned; undefined where it makes none. A drop cap's frame paragraph read
+     * just before (`dropCap`; REQ-521) is taken in at its start.
+     */
+    paragraph(chain, addresses, story, sink, state, dropCap) {
         var _a, _b, _c, _d;
         const first = chain[0];
         const last = chain[chain.length - 1];
         const block = new TextBlock('PARAGRAPH', story.kind, addresses[0]);
+        this.consult('styles'); // every block is resolved through the part's defaults
         if (this.fieldsOnly) {
             for (let c = 0; c < chain.length; c++)
                 this.runs(chain[c].content, new ParagraphState(block, chain[c], addresses[c], story), state);
-            return;
+            return undefined;
         }
         if (chain.length > 1)
             block.joined = addresses.slice(1);
@@ -577,7 +641,7 @@ export class Extractor {
             block.runs.length = 0;
             this.flushIfOutsideTable(story, sink);
             this.endSection(last, story);
-            return;
+            return undefined;
         }
         block.hidden = hiddenParagraph;
         if (!this.options.hidden) {
@@ -589,7 +653,7 @@ export class Extractor {
             // a table of contents' paragraph, under toc: false (D12): no line at all
             this.flushIfOutsideTable(story, sink);
             this.endSection(last, story);
-            return;
+            return undefined;
         }
         // one count per paragraph: the label (list_labels) and the numbering's value come from it
         if (this.trees.pkg !== undefined && (this.options.listLabels || block.numbering !== undefined)) {
@@ -600,9 +664,78 @@ export class Extractor {
                 block.numbering.value = counted.count;
         }
         block.section = (_d = (_c = story.sections) === null || _c === void 0 ? void 0 : _c.length) !== null && _d !== void 0 ? _d : 0;
-        sink.push(block);
+        if (dropCap === undefined || !this.joinDropCap(dropCap, block, story, sink))
+            sink.push(block);
         this.flushIfOutsideTable(story, sink);
         this.endSection(last, story);
+        return block;
+    }
+    /**
+     * Whether a paragraph just read is a drop cap's frame, for the paragraph next to take (REQ-521; CR-012 section 24.3):
+     * under `drop_caps: join`, one `w:p` whose `w:framePr/@w:dropCap` (its own, or its style's) is `drop` or `margin`, whose
+     * text is at most one word (no white space and no space character, U+00A0 included), and which ends no section.
+     */
+    dropCapFrame(chain, block) {
+        var _a, _b;
+        if (this.options.dropCaps !== 'join' || chain.length !== 1)
+            return false;
+        const p = chain[0];
+        if ((_a = p.pPr) === null || _a === void 0 ? void 0 : _a.sectPr)
+            return false;
+        // Java's Character.isWhitespace or isSpaceChar, docx4j's test
+        if (/[\t-\r\x1C-\x1F\p{Z}]/u.test(block.text))
+            return false;
+        let pPr = this.viewPPr(p);
+        if (this.resolver) {
+            try {
+                pPr = this.resolver.getEffectivePPr(pPr);
+            }
+            catch {
+                // the paragraph's own properties, as read
+            }
+        }
+        const dropCap = (_b = pPr === null || pPr === void 0 ? void 0 : pPr.framePr) === null || _b === void 0 ? void 0 : _b.dropCap;
+        return dropCap === 'drop' || dropCap === 'margin';
+    }
+    /**
+     * A drop cap's frame paragraph joined to the block after it (REQ-521): its runs and fields put at the block's start with
+     * no separator, the block (its address, its properties) in the frame's place in the sink, so that what the frame
+     * anchored still follows it; the warning `paragraph.drop_cap_joined` at the frame's address. The letter's run merges
+     * into the word's first run on docx4j's rule (the same kind, hidden and placeholder flags, revision and formatting; D6,
+     * Jason 2026-11-02), which replaces section 4.4's `officeOffset` contiguity rule at this one join, the two runs' text
+     * coming from two paragraphs' `Paragraph.text`: the merged run keeps the letter's source, and its `officeOffset` is
+     * undefined. False, and nothing done, where the frame is not in the sink.
+     */
+    joinDropCap(frame, block, story, sink) {
+        var _a;
+        const at = sink.lastIndexOf(frame);
+        if (at < 0)
+            return false;
+        const runs = [...frame.runs];
+        let first = 0;
+        if (runs.length > 0 && block.runs.length > 0) {
+            const left = runs[runs.length - 1];
+            const right = block.runs[0];
+            if (left.kind === 'TEXT' && right.kind === 'TEXT' && !left.math && !right.math && left.hidden === right.hidden
+                && left.placeholder === right.placeholder && sameRevision(left.revision, right.revision) && left.sameFormat(right)) {
+                const merged = new TextRun('TEXT', left.text + right.text, left.revision);
+                merged.hidden = left.hidden;
+                merged.placeholder = left.placeholder;
+                merged.copyFormat(left);
+                merged.source = left.source;
+                runs[runs.length - 1] = merged;
+                first = 1;
+            }
+        }
+        runs.push(...block.runs.slice(first));
+        block.runs.length = 0;
+        block.runs.push(...runs);
+        block.fields.unshift(...frame.fields);
+        block.revisions += frame.revisions;
+        block.hidden = block.hidden && frame.hidden;
+        sink[at] = block;
+        this.warn(new TextWarning(TextWarning.DROP_CAP_JOINED, frame.address, undefined, undefined, (_a = story.part) === null || _a === void 0 ? void 0 : _a.partName.name));
+        return true;
     }
     endSection(last, story) {
         var _a;
@@ -624,7 +757,10 @@ export class Extractor {
         }
     }
     // --- altChunk (REQ-513) -----------------------------------------------------------------------
-    /** An `altChunk`: not expanded until CR-012 phase 5e, which raises `altchunk.not_expanded` with its content type. */
+    /**
+     * An `altChunk`: not expanded until CR-012 phase 5e, which raises `altchunk.not_expanded` naming the chunk's part, its
+     * selector (WD 0.47, CR-012 section 22.2 (a)), with its content type.
+     */
     altChunk(chunk, address, story) {
         var _a;
         const part = related(story.part, chunk.id);
@@ -632,7 +768,7 @@ export class Extractor {
             return;
         const ct = (_a = part.contentType) !== null && _a !== void 0 ? _a : '';
         const semi = ct.indexOf(';');
-        this.warn(new TextWarning(TextWarning.ALTCHUNK_NOT_EXPANDED, address, undefined, (semi >= 0 ? ct.slice(0, semi) : ct).trim()));
+        this.warn(new TextWarning(TextWarning.ALTCHUNK_NOT_EXPANDED, address, undefined, (semi >= 0 ? ct.slice(0, semi) : ct).trim(), part.partName.name));
     }
     // --- bound values (REQ-509) -------------------------------------------------------------------
     /**
@@ -648,6 +784,10 @@ export class Extractor {
         const binding = (_a = sdtPrChild(pr, 'dataBinding')) === null || _a === void 0 ? void 0 : _a.value;
         if (!binding)
             return undefined;
+        // a store it names that could not be read: read as absent, the binding resolved by REQ-509's repair (CR-012 section 28, D3)
+        const unread = this.trees.unreadStoreOf(binding);
+        if (unread !== undefined)
+            this.warn(new TextWarning(TextWarning.PART_NOT_READ, undefined, undefined, undefined, unread));
         const value = this.trees.boundValue(binding);
         if (value === undefined)
             return undefined;
@@ -756,16 +896,16 @@ export class Extractor {
         const numId = String(numIdValue);
         if (Number(numIdValue) === 0)
             return undefined;
+        const emulator = this.numberingEmulator();
         const ilvlValue = (_b = numPr === null || numPr === void 0 ? void 0 : numPr.ilvl) === null || _b === void 0 ? void 0 : _b.val;
         const ilvl = ilvlValue === undefined || ilvlValue === null ? 0 : Number(ilvlValue);
         let format;
         let text;
         let abstractNumId;
         let start = 1;
-        const part = this.trees.numberingPart;
         const definitions = this.trees.numbering;
         try {
-            if (part && !directNumId && this.linkedElsewhere(part, numId, String(ilvl), pStyle))
+            if (emulator && !directNumId && this.linkedElsewhere(emulator, numId, String(ilvl), pStyle))
                 return undefined;
             const list = definitions === null || definitions === void 0 ? void 0 : definitions.list(numId);
             abstractNumId = (_c = list === null || list === void 0 ? void 0 : list.abstractDefinition) === null || _c === void 0 ? void 0 : _c.id;
@@ -784,10 +924,21 @@ export class Extractor {
         numbering.abstractNumId = abstractNumId;
         return numbering;
     }
+    /**
+     * The numbering a numbered paragraph is counted by (`PackageTrees.numberingFor`): where the numbering part is absent or
+     * cannot be read, Word's default definition, and `part.absent` or `part.not_read` naming the part, once (WD 0.76).
+     */
+    numberingEmulator() {
+        const missing = this.trees.numberingMissing;
+        if (missing !== undefined && this.trees.pkg !== undefined) {
+            this.warn(new TextWarning(missing.kind === 'part.absent' ? TextWarning.PART_ABSENT : TextWarning.PART_NOT_READ, undefined, undefined, undefined, missing.partName));
+        }
+        return this.trees.numberingFor();
+    }
     /** Whether the level belongs to a paragraph style that is neither the paragraph's nor one its style is based on. */
-    linkedElsewhere(part, numId, ilvl, pStyle) {
+    linkedElsewhere(emulator, numId, ilvl, pStyle) {
         var _a, _b, _c, _d;
-        const linked = Emulator.levelStyleId(part, numId, ilvl);
+        const linked = emulator.levelStyleId(numId, ilvl);
         if (linked === undefined || linked === '')
             return false;
         // a paragraph naming no style is in the default paragraph style, which can own a level
@@ -803,11 +954,11 @@ export class Extractor {
     }
     /** The paragraph counted in its story's numbering state; undefined where it is not numbered. */
     count(p, story) {
-        const pkg = this.trees.pkg;
-        if (!pkg || !this.trees.numberingPart)
+        const emulator = this.trees.emulator;
+        if (this.trees.pkg === undefined || !emulator)
             return undefined;
         try {
-            return Emulator.getNumber(pkg, p.pPr, story.numbering());
+            return emulator.getNumber(p.pPr, story.numbering());
         }
         catch {
             return undefined;
@@ -817,6 +968,7 @@ export class Extractor {
     table(tbl, address, story, sink, state) {
         var _a, _b;
         const block = new TextBlock('TABLE', story.kind, address);
+        this.consult('styles');
         block.section = (_b = (_a = story.sections) === null || _a === void 0 ? void 0 : _a.length) !== null && _b !== void 0 ? _b : 0;
         block.source = tbl;
         story.tableDepth++;
@@ -866,7 +1018,7 @@ export class Extractor {
             else if (typeOf(v) === AC) {
                 // a row-level mc:AlternateContent: its selected branch's rows, addressed from the element (CR-013)
                 const ac = v;
-                this.compareBranches(ac, address);
+                this.compareBranches(ac, address, story);
                 this.rows(selectedBranchOf(ac), address, story, table, state);
             }
             else {
@@ -917,7 +1069,7 @@ export class Extractor {
             else if (typeOf(v) === AC) {
                 // a cell-level mc:AlternateContent: its selected branch's cells, addressed from the element (CR-013)
                 const ac = v;
-                this.compareBranches(ac, address);
+                this.compareBranches(ac, address, story);
                 this.cells(selectedBranchOf(ac), address, story, row, state);
             }
             else {
@@ -945,7 +1097,7 @@ export class Extractor {
             }
             else if (tn === AC) {
                 const ac = v;
-                this.compareBranches(ac, para.block.address);
+                this.compareBranches(ac, para.block.address, para.story);
                 this.runs(selectedBranchOf(ac), para, state);
             }
             else if (isSdt(v) && this.unwraps(item)) {
@@ -1047,6 +1199,7 @@ export class Extractor {
                     const f = fields[0];
                     if (f) {
                         f.separated = true;
+                        f.hiddenResult = isSetField(f.instr);
                         f.link = fieldHyperlinkTarget(f.instr);
                         if (!this.options.toc && !suppressed(para.story) && isTocField(f.instr)) {
                             f.toc = true;
@@ -1188,7 +1341,7 @@ export class Extractor {
             }
             case 'AlternateContent': {
                 const ac = v;
-                this.compareBranches(ac, para.block.address);
+                this.compareBranches(ac, para.block.address, para.story);
                 for (const o of (_b = selectedBranchOf(ac)) !== null && _b !== void 0 ? _b : [])
                     this.runItem(o, r, hidden, para, state);
                 return;
@@ -1215,6 +1368,8 @@ export class Extractor {
         }
         if (suppressed(para.story))
             return;
+        if (para.story.fields.some((f) => f.hiddenResult))
+            return; // a SET's stored value: never shown
         if (!(hidden && !this.options.hidden)) {
             for (const f of para.story.fields)
                 f.result += s;
@@ -1271,7 +1426,7 @@ export class Extractor {
             this.warn(new TextWarning(TextWarning.FIELD_NO_RESULT, para.block.address));
     }
     simpleField(field, para, state) {
-        var _a;
+        var _a, _b, _c;
         if (suppressed(para.story))
             return; // inside a computed field's cached result
         const instr = (_a = field.instr) !== null && _a !== void 0 ? _a : '';
@@ -1281,6 +1436,15 @@ export class Extractor {
         }
         if (this.options.fieldCodes && !inCode(para.story))
             this.addText(para, instr, 'FIELD_CODE', false, state);
+        if (isSetField(instr)) {
+            // a SET's stored value is never shown (REQ-505), in this form as in the complex one (FieldState.hiddenResult)
+            this.resultOf(field.content, para, state, undefined, true);
+            const stored = ((_c = (_b = field.content) === null || _b === void 0 ? void 0 : _b.length) !== null && _c !== void 0 ? _c : 0) > 0;
+            para.block.fields.push(new TextField(instr, stored ? '' : undefined));
+            if (!stored)
+                this.warn(new TextWarning(TextWarning.FIELD_NO_RESULT, para.block.address));
+            return;
+        }
         const before = para.block.runs.length;
         const text = this.resultOf(field.content, para, state, fieldHyperlinkTarget(instr));
         const hasResult = text !== '' || para.block.runs.length > before;
@@ -1288,11 +1452,12 @@ export class Extractor {
         if (!hasResult)
             this.warn(new TextWarning(TextWarning.FIELD_NO_RESULT, para.block.address));
     }
-    /** The runs of a simple field's result, added to the paragraph; their text. */
-    resultOf(content, para, state, link) {
+    /** The runs of a simple field's result, added to the paragraph; their text. `hidden`: a result never shown (a SET's). */
+    resultOf(content, para, state, link, hidden = false) {
         const capture = new FieldState();
         capture.separated = true;
         capture.link = link;
+        capture.hiddenResult = hidden;
         para.story.fields.push(capture);
         try {
             this.runs(content, para, state);
@@ -1377,7 +1542,7 @@ export class Extractor {
                         return false;
                     case AC: {
                         const ac = value;
-                        this.compareBranches(ac, para.block.address);
+                        this.compareBranches(ac, para.block.address, para.story);
                         for (const o of (_a = selectedBranchOf(ac)) !== null && _a !== void 0 ? _a : [])
                             walk(unwrap(o), 1, visitor);
                         return false;
@@ -1502,13 +1667,13 @@ export class Extractor {
      * differ after REQ-1201's normalisation. A branch's text is every `w:t`, `m:t` and `a:t` in it, whatever
      * the view and hidden or not, as docx4j reads it: a comparison made at load knows no view, and a branch
      * that differs only in hidden or deleted text is one an agent should hear of (Jason, 2026-10-21, for WD
-     * 0.29; CR-012 section 15.6).
+     * 0.29; CR-012 section 15.6). `mce.nothing_selected` (REQ-305): where the element has Choices and nothing is selected.
      */
-    compareBranches(ac, address) {
+    compareBranches(ac, address, story) {
         var _a, _b, _c, _d;
         if (!this.options.warnings || this.fieldsOnly)
             return;
-        const selected = selectedBranchOf(ac);
+        const selected = selectedBranchOf(ac, { onNothingSelected: ({ namespaces }) => this.nothingSelected(address, story.part, namespaces) });
         if (selected === undefined)
             return;
         const mine = branchText(selected);
@@ -1529,6 +1694,44 @@ export class Extractor {
             }
         }
     }
+    /** `mce.nothing_selected`, its selector the part and its `detail` the namespaces (REQ-801). */
+    nothingSelected(address, part, namespaces) {
+        this.warn(new TextWarning(TextWarning.MCE_NOTHING_SELECTED, address, undefined, undefined, part === null || part === void 0 ? void 0 : part.partName.name, namespaces.join(' ')));
+    }
+    /**
+     * What the walk can't meet, from the warnings the load's passes recorded on each part this reading took, none with an address:
+     * `mce.nothing_selected` for an element a resolving load took out or one at a part's root (CR-029 section 3.5);
+     * `mce.undeclared_prefix` (REQ-323, CR-030), on a story part or the styles, numbering or settings part the reading consults;
+     * and `part.unexpected_root` (REQ-320, CR-028 D11).
+     */
+    partWarningsUnmet() {
+        var _a, _b;
+        if (!this.options.warnings || this.fieldsOnly)
+            return;
+        const main = (_a = this.trees.main) === null || _a === void 0 ? void 0 : _a.part;
+        const consulted = [main === null || main === void 0 ? void 0 : main.styleDefinitionsPart, main === null || main === void 0 ? void 0 : main.numberingDefinitionsPart, main === null || main === void 0 ? void 0 : main.documentSettingsPart];
+        for (const part of [...this.storyParts, ...consulted]) {
+            if (!part)
+                continue;
+            for (const w of (_b = part.warnings) !== null && _b !== void 0 ? _b : []) {
+                if (w.code === 'mce.nothing_selected' && (w.removed || w.root))
+                    this.nothingSelected(undefined, part, w.namespaces);
+                else if (w.code === 'mce.undeclared_prefix')
+                    this.warn(new TextWarning(TextWarning.MCE_UNDECLARED_PREFIX, undefined, undefined, undefined, part.partName.name, w.prefixes.join(' ')));
+                else if (w.code === 'part.unexpected_root')
+                    this.warn(new TextWarning(TextWarning.PART_UNEXPECTED_ROOT, undefined, undefined, undefined, w.partName, w.root));
+            }
+        }
+    }
+    /**
+     * A story part the reading takes that could not be read: none of its text, and `content.not_read` naming it (REQ-801),
+     * at the story's address prefix for a header or footer; docx4j's `storyContent` (`4fdd80063`).
+     */
+    storyRead(part, address) {
+        if (part && this.trees.unreadable.has(part)) {
+            this.warn(new TextWarning(TextWarning.CONTENT_NOT_READ, address, undefined, undefined, part.partName.name));
+        }
+    }
     warn(w) {
         if (this.warned.has(w.key))
             return;
@@ -1538,6 +1741,7 @@ export class Extractor {
     // --- headers and footers by section (REQ-502) ---------------------------------------------------
     sectionsWithHeaders(sections, mainBlocks, headers, footers, out) {
         var _a, _b;
+        this.consult('settings');
         const evenAndOdd = on((_a = this.trees.settings) === null || _a === void 0 ? void 0 : _a.evenAndOddHeaders);
         const emitted = new Set();
         const lastHeader = new Map();
@@ -1581,15 +1785,19 @@ export class Extractor {
         emitted.add(rId);
         const main = this.trees.main;
         const part = main ? related(main.part, rId) : undefined;
+        const prefix = `${this.root}${kind === 'HEADER' ? 'header:' : 'footer:'}${rId}`;
+        this.storyRead(part, prefix);
         const content = part ? this.trees.headersAndFooters.get(part) : undefined;
         if (!content)
             return;
-        const prefix = `${this.root}${kind === 'HEADER' ? 'header:' : 'footer:'}${rId}`;
         const story = new Story(this, kind, prefix, part);
         this.blocks(blockItems(content), prefix, story, out);
     }
     /** A header or footer part's content as its own story (`Texts.of(headerPart)`). */
     headerOrFooterPart(part, kind, rId) {
+        // read alone, a part that could not be read throws why, as the main part does (CR-012 section 25, D1)
+        if (this.trees.unreadable.has(part))
+            throw this.trees.unreadable.get(part);
         const out = [];
         const content = this.trees.headersAndFooters.get(part);
         const prefix = `${this.root}${kind === 'HEADER' ? 'header:' : 'footer:'}${rId}`;
@@ -1599,12 +1807,13 @@ export class Extractor {
     }
     // --- notes (REQ-503) ------------------------------------------------------------------------
     notes(type, out) {
-        var _a;
+        var _a, _b, _c;
+        this.storyRead(type === 'footnote' ? (_a = this.trees.main) === null || _a === void 0 ? void 0 : _a.part.footnotesPart : (_b = this.trees.main) === null || _b === void 0 ? void 0 : _b.part.endnotesPart);
         const holder = type === 'footnote' ? this.trees.footnotes : this.trees.endnotes;
         const numbers = type === 'footnote' ? this.context.footnotes : this.context.endnotes;
         if (!holder || numbers.size === 0)
             return;
-        const notes = ((_a = (type === 'footnote' ? holder.tree.footnote : holder.tree.endnote)) !== null && _a !== void 0 ? _a : []);
+        const notes = ((_c = (type === 'footnote' ? holder.tree.footnote : holder.tree.endnote)) !== null && _c !== void 0 ? _c : []);
         const byId = new Map();
         for (const note of notes)
             if (note.id !== undefined && note.id !== null)
@@ -1649,19 +1858,20 @@ export class Extractor {
     }
     // --- comments (REQ-504) ---------------------------------------------------------------------
     comments(out) {
-        var _a, _b, _c, _d, _e;
+        var _a, _b, _c, _d, _e, _f;
+        this.storyRead((_a = this.trees.main) === null || _a === void 0 ? void 0 : _a.part.commentsPart);
         const part = this.trees.comments;
         if (!part)
             return;
         const extended = new Map();
-        for (const ex of ((_b = (_a = this.trees.commentsExtended) === null || _a === void 0 ? void 0 : _a.commentEx) !== null && _b !== void 0 ? _b : [])) {
+        for (const ex of ((_c = (_b = this.trees.commentsExtended) === null || _b === void 0 ? void 0 : _b.commentEx) !== null && _c !== void 0 ? _c : [])) {
             if (ex.paraId)
                 extended.set(ex.paraId.toUpperCase(), ex);
         }
         const idByParaId = new Map();
         const anchored = this.trees.anchoredComments();
-        for (const c of ((_c = part.tree.comment) !== null && _c !== void 0 ? _c : [])) {
-            if ((_d = this.plan) === null || _d === void 0 ? void 0 : _d.removesComment(c.id))
+        for (const c of ((_d = part.tree.comment) !== null && _d !== void 0 ? _d : [])) {
+            if ((_e = this.plan) === null || _e === void 0 ? void 0 : _e.removesComment(c.id))
                 continue;
             const id = c.id === undefined || c.id === null ? '' : String(c.id);
             // a comment with no anchor in the document is none of its text: Word drops it on save (REQ-504)
@@ -1670,10 +1880,11 @@ export class Extractor {
             const comment = new TextComment(id);
             comment.author = c.author;
             comment.initials = c.initials;
-            comment.date = calendarText(c.date);
+            const paraId = lastParaId(c.content);
+            // REQ-702's form: w:date's wall clock, the offset where the comments extensible part's w16cex:dateUtc gives it
+            comment.date = revisionDateText(c.date, paraId === undefined ? undefined : this.trees.commentDatesUtc.get(paraId.toUpperCase()));
             const prefix = `${this.root}comment:${id}`;
             this.blocks(c.content, prefix, new Story(this, 'COMMENT', prefix, part.part), comment.blocks);
-            const paraId = lastParaId(c.content);
             if (paraId !== undefined) {
                 idByParaId.set(paraId.toUpperCase(), id);
                 const ex = extended.get(paraId.toUpperCase());
@@ -1690,7 +1901,7 @@ export class Extractor {
             const parent = idByParaId.get(comment.parentId.toUpperCase());
             comment.parentId = parent;
             if (parent !== undefined)
-                (_e = out.get(parent)) === null || _e === void 0 ? void 0 : _e.replies.push(comment);
+                (_f = out.get(parent)) === null || _f === void 0 ? void 0 : _f.replies.push(comment);
         }
     }
     // --- the view -------------------------------------------------------------------------------
@@ -1727,20 +1938,42 @@ export class Extractor {
     /**
      * A run's formatting beyond its paragraph style's (CR-046 phase 3): bold, italic, struck through,
      * monospaced, each as the run's own property gives it, else its styles', where the paragraph style
-     * alone doesn't give it. A run carrying only that.
+     * alone doesn't give it. And, whatever the paragraph style says, what the text shows (docx4j CR-069;
+     * CR-012 section 24): its case, folded the same way, a run under both caps and small caps being caps
+     * (WD REQ-607); its vertical alignment (REQ-710); and a cased run's language. A run carrying only that.
      */
     formatOf(r, p) {
-        var _a;
+        var _a, _b;
         const own = r.rPr;
         const styles = this.styleRPr((_a = own === null || own === void 0 ? void 0 : own.rStyle) === null || _a === void 0 ? void 0 : _a.val, p);
         const paragraph = this.styleRPr(undefined, p);
         const f = new TextRun('TEXT', '', undefined);
         f.bold = onOf(own === null || own === void 0 ? void 0 : own.b, styles === null || styles === void 0 ? void 0 : styles.b) && !on(paragraph === null || paragraph === void 0 ? void 0 : paragraph.b);
         f.italic = onOf(own === null || own === void 0 ? void 0 : own.i, styles === null || styles === void 0 ? void 0 : styles.i) && !on(paragraph === null || paragraph === void 0 ? void 0 : paragraph.i);
-        f.strike = onOf(own === null || own === void 0 ? void 0 : own.strike, styles === null || styles === void 0 ? void 0 : styles.strike) && !on(paragraph === null || paragraph === void 0 ? void 0 : paragraph.strike);
+        // struck once or twice (w:dstrike: docx4j CR-069 phase 4, CR-012 section 24.5), the paragraph style's either cancelling it
+        f.strike = (onOf(own === null || own === void 0 ? void 0 : own.strike, styles === null || styles === void 0 ? void 0 : styles.strike) || onOf(own === null || own === void 0 ? void 0 : own.dstrike, styles === null || styles === void 0 ? void 0 : styles.dstrike))
+            && !(on(paragraph === null || paragraph === void 0 ? void 0 : paragraph.strike) || on(paragraph === null || paragraph === void 0 ? void 0 : paragraph.dstrike));
         f.monospace = isMonospace(this.runFont(r, p)) && !isMonospace(asciiFont(paragraph));
+        f.caps = onOf(own === null || own === void 0 ? void 0 : own.caps, styles === null || styles === void 0 ? void 0 : styles.caps);
+        f.smallCaps = !f.caps && onOf(own === null || own === void 0 ? void 0 : own.smallCaps, styles === null || styles === void 0 ? void 0 : styles.smallCaps);
+        const vertAlign = (_b = own === null || own === void 0 ? void 0 : own.vertAlign) !== null && _b !== void 0 ? _b : styles === null || styles === void 0 ? void 0 : styles.vertAlign;
+        if ((vertAlign === null || vertAlign === void 0 ? void 0 : vertAlign.val) === 'superscript')
+            f.verticalAlignment = 'SUPERSCRIPT';
+        if ((vertAlign === null || vertAlign === void 0 ? void 0 : vertAlign.val) === 'subscript')
+            f.verticalAlignment = 'SUBSCRIPT';
+        if (f.caps || f.smallCaps)
+            f.language = this.language(own, styles);
         return f;
     }
+    /**
+     * The language a cased run's capitals are mapped by (WD REQ-607): its own `w:lang/@w:val`, else its styles' (with the
+     * document defaults'), else the `default_language` option. `w:themeFontLang` is no link (WD 0.80, CR-012 section 24.8):
+     * it chooses among the theme's fonts, and Word cases such a run by its editing language, which the option stands in for.
+     */
+    language(own, styles) {
+        var _a, _b;
+        return (_b = (_a = langOf(own === null || own === void 0 ? void 0 : own.lang)) !== null && _a !== void 0 ? _a : langOf(styles === null || styles === void 0 ? void 0 : styles.lang)) !== null && _b !== void 0 ? _b : this.options.defaultLanguage;
+    }
     /** The run's font for `monospace`: its own `w:rFonts/@w:ascii`, else its styles'. */
     runFont(r, p) {
         var _a, _b, _c;
@@ -1830,6 +2063,12 @@ function textLength(block) {
 function sameRevision(a, b) {
     return a === undefined ? b === undefined : a.equals(b);
 }
+/** A `w:lang`'s `w:val`, trimmed; undefined for none or an empty one. */
+function langOf(l) {
+    var _a;
+    const v = (_a = l === null || l === void 0 ? void 0 : l.val) === null || _a === void 0 ? void 0 : _a.trim();
+    return v === undefined || v === '' ? undefined : v;
+}
 /** An on/off property: the run's own where it has one, else its styles'. */
 function onOf(own, styles) {
     if (own !== undefined)
@@ -1878,6 +2117,11 @@ export function isTocField(instr) {
     const tokens = fieldTokens(instr);
     return tokens.length > 0 && tokens[0].toUpperCase() === 'TOC';
 }
+/** A SET field's code: Word stores its value as its result, under a bookmark of its name, and shows nothing (REQ-505). */
+function isSetField(instr) {
+    const tokens = fieldTokens(instr);
+    return tokens.length > 0 && tokens[0].toUpperCase() === 'SET';
+}
 /** A table of contents' code, shown under `field_codes` until it was known to be one, taken out again. */
 function removeFieldCode(f, para) {
     // what follows the field may merge with what preceded it, as it would with no field codes