@cj-tech-master/excelts 9.5.4 → 9.5.5

package/dist/cjs/modules/word/query/data-binding.js

@@ -0,0 +1,395 @@
+"use strict";
+/**
+ * OpenDoPE Data Binding API
+ *
+ * Resolve data bindings in a document by evaluating XPath expressions
+ * against CustomXML parts and populating the SDT content with values.
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.resolveDataBindings = resolveDataBindings;
+const dom_1 = require("../../xml/dom.js");
+// =============================================================================
+// Public API
+// =============================================================================
+/**
+ * Resolve data bindings in a document by evaluating XPath expressions
+ * against CustomXML parts and populating the SDT content with values.
+ *
+ * Supports:
+ * - Simple value binding: SDT text is replaced with XPath evaluation result
+ * - Repeating sections: SDT with `repeatingSection` is duplicated per data row
+ * - Conditional display: SDT is removed if XPath returns empty/false
+ *
+ * @param doc - The document to process (returns a new modified copy).
+ * @param data - Optional override data map (storeItemId → XML string). If not
+ *              provided, uses the document's embedded CustomXML parts.
+ * @returns A new document with bindings resolved.
+ */
+function resolveDataBindings(doc, data) {
+    // Build a map of storeItemId → parsed XML content
+    const xmlStore = new Map();
+    if (data) {
+        for (const [key, value] of data) {
+            xmlStore.set(key.toLowerCase(), value);
+        }
+    }
+    else if (doc.customXmlParts) {
+        for (const part of doc.customXmlParts) {
+            xmlStore.set(part.itemId.toLowerCase(), part.xmlContent);
+        }
+    }
+    if (xmlStore.size === 0) {
+        return doc;
+    }
+    const newBody = doc.body.map(block => resolveBlockBinding(block, xmlStore));
+    return { ...doc, body: newBody };
+}
+// =============================================================================
+// Internal helpers
+// =============================================================================
+function resolveBlockBinding(block, xmlStore) {
+    if (block.type === "sdt") {
+        return resolveSdtBinding(block, xmlStore);
+    }
+    if (block.type === "table") {
+        return {
+            ...block,
+            rows: block.rows.map(r => ({
+                ...r,
+                cells: r.cells.map(c => ({
+                    ...c,
+                    content: c.content.map(inner => resolveBlockBinding(inner, xmlStore))
+                }))
+            }))
+        };
+    }
+    return block;
+}
+function resolveSdtBinding(sdt, xmlStore) {
+    const binding = sdt.properties?.dataBinding;
+    if (!binding) {
+        // No binding — recurse into content for nested SDTs/tables
+        return {
+            ...sdt,
+            content: sdt.content.map(c => {
+                if ("type" in c && c.type === "table") {
+                    return resolveBlockBinding(c, xmlStore);
+                }
+                return c;
+            })
+        };
+    }
+    // Resolve the value from CustomXML using simple XPath evaluation
+    const value = evaluateSimpleXPath(binding, xmlStore);
+    if (sdt.properties?.repeatingSection) {
+        // Repeating section: value should be a collection
+        // Each child in the data produces a copy of the repeatingSection item template
+        const values = evaluateRepeatingXPath(binding, xmlStore);
+        if (values.length === 0) {
+            // Remove the SDT entirely if no data
+            return { type: "paragraph", children: [] };
+        }
+        // Return the SDT with repeated content, with inner bindings resolved per item
+        const repeatedContent = [];
+        const template = sdt.content;
+        // Build parent xpath for resolving child element values
+        const parentXpath = binding.xpath;
+        const storeId = binding.storeItemId.replace(/[{}]/g, "").toLowerCase();
+        const xmlContent = xmlStore.get(storeId) ?? xmlStore.get(`{${storeId}}`);
+        for (let itemIdx = 0; itemIdx < values.length; itemIdx++) {
+            // Clone template content for each item
+            for (const item of template) {
+                const cloned = structuredClone(item);
+                // Resolve inner SDT bindings within this repeated item
+                if (xmlContent) {
+                    resolveInnerBindings(cloned, parentXpath, itemIdx + 1, xmlStore);
+                }
+                repeatedContent.push(cloned);
+            }
+        }
+        return { ...sdt, content: repeatedContent };
+    }
+    if (value === null) {
+        // Condition not met — remove SDT (return empty paragraph)
+        return { type: "paragraph", children: [] };
+    }
+    // Simple binding — replace text content of the SDT
+    const textRun = {
+        content: [{ type: "text", text: value }]
+    };
+    const textPara = { type: "paragraph", children: [textRun] };
+    return { ...sdt, content: [textPara] };
+}
+/**
+ * Evaluate an XPath expression against CustomXML stored as a string.
+ *
+ * Supports the XPath 1.0 subset that real-world OpenDoPE-style bindings
+ * exercise:
+ *   - Absolute paths: `/root/child/grandchild`
+ *   - Namespace prefixes (matched by local name): `/ns0:root/ns0:child`
+ *   - Position predicates: `/root/item[1]`, `/root/item[last()]`
+ *   - Attribute access: `/root/element/@attr`
+ *   - Text content extraction from leaf nodes (CDATA-aware,
+ *     entity-decoded by the XML parser)
+ *
+ * Does NOT support: axes (ancestor::, following::), functions
+ * (count(), string()), arithmetic, union (|), or complex predicates.
+ *
+ * Implemented on top of the project's XML DOM parser. The previous
+ * regex-based scanner was unsafe — it returned entity-encoded text
+ * verbatim and could be confused by CDATA, comments, attributes
+ * containing `>`, etc.
+ *
+ * @returns The text content of the matched element/attribute, or null if
+ *          not found. Returns the empty string for an element that exists
+ *          but has no text — callers that need to distinguish empty from
+ *          missing must compare to `null` explicitly.
+ */
+function evaluateSimpleXPath(binding, xmlStore) {
+    const root = loadStoreRoot(binding, xmlStore);
+    if (!root) {
+        return null;
+    }
+    const xpath = binding.xpath;
+    // Attribute access: /path/to/element/@attr
+    const attrMatch = xpath.match(/^(.+)\/@([a-zA-Z0-9_:-]+)$/);
+    if (attrMatch) {
+        const elementPath = attrMatch[1];
+        const attrName = attrMatch[2];
+        const localAttr = stripPrefix(attrName);
+        const el = findElementByPath(root, elementPath);
+        if (!el) {
+            return null;
+        }
+        // Match attribute by local name, ignoring any prefix in the source.
+        for (const [name, value] of Object.entries(el.attributes)) {
+            if (stripPrefix(name) === localAttr) {
+                return value;
+            }
+        }
+        return null;
+    }
+    const el = findElementByPath(root, xpath);
+    if (!el) {
+        return null;
+    }
+    return collectTextContent(el);
+}
+function parsePathSegment(raw) {
+    let seg = stripPrefix(raw);
+    let position = 0;
+    const predMatch = seg.match(/\[(\d+|last\(\))\]$/);
+    if (predMatch) {
+        seg = seg.substring(0, seg.indexOf("["));
+        if (predMatch[1] === "last()") {
+            position = -1;
+        }
+        else {
+            position = parseInt(predMatch[1], 10);
+        }
+    }
+    return { localName: seg, position };
+}
+function stripPrefix(name) {
+    const i = name.indexOf(":");
+    return i >= 0 ? name.substring(i + 1) : name;
+}
+function elementLocal(el) {
+    return el.local ?? stripPrefix(el.name);
+}
+function loadStoreRoot(binding, xmlStore) {
+    const storeId = binding.storeItemId.replace(/[{}]/g, "").toLowerCase();
+    const xmlContent = xmlStore.get(storeId) ?? xmlStore.get(`{${storeId}}`);
+    if (!xmlContent) {
+        return null;
+    }
+    try {
+        return (0, dom_1.parseXml)(xmlContent).root;
+    }
+    catch {
+        // Malformed CustomXML store. Treat as empty rather than throwing —
+        // resolveDataBindings is best-effort by design.
+        return null;
+    }
+}
+/** Direct-child elements of `el` filtered by local name. */
+function directChildren(el, localName) {
+    const out = [];
+    for (const child of el.children) {
+        if (child.type === "element" && elementLocal(child) === localName) {
+            out.push(child);
+        }
+    }
+    return out;
+}
+/** Resolve a path of segments against `root`. The first segment is matched
+ *  against `root` itself (XPath absolute paths begin with the document root). */
+function findElementByPath(root, xpath) {
+    const segments = xpath
+        .split("/")
+        .filter(s => s.length > 0)
+        .map(parsePathSegment);
+    if (segments.length === 0) {
+        return root;
+    }
+    // First segment must match the document root by local name.
+    const first = segments[0];
+    if (elementLocal(root) !== first.localName) {
+        return null;
+    }
+    let current = root;
+    for (let i = 1; i < segments.length; i++) {
+        if (!current) {
+            return null;
+        }
+        const seg = segments[i];
+        const matches = directChildren(current, seg.localName);
+        if (matches.length === 0) {
+            return null;
+        }
+        if (seg.position === -1) {
+            current = matches[matches.length - 1];
+        }
+        else if (seg.position > 0) {
+            current = matches[seg.position - 1] ?? null;
+        }
+        else {
+            current = matches[0];
+        }
+    }
+    return current;
+}
+/** Concatenate descendant text and CDATA content. */
+function collectTextContent(el) {
+    let out = "";
+    const walk = (node) => {
+        if (node.type === "text" || node.type === "cdata") {
+            out += node.value;
+            return;
+        }
+        if (node.type === "element") {
+            for (const child of node.children) {
+                walk(child);
+            }
+        }
+    };
+    for (const child of el.children) {
+        walk(child);
+    }
+    return out;
+}
+function evaluateRepeatingXPath(binding, xmlStore) {
+    const root = loadStoreRoot(binding, xmlStore);
+    if (!root) {
+        return [];
+    }
+    const segments = binding.xpath
+        .split("/")
+        .filter(s => s.length > 0)
+        .map(parsePathSegment);
+    if (segments.length === 0) {
+        return [];
+    }
+    // Walk to the parent of the last segment, then count occurrences of the
+    // last segment's local name as direct children.
+    const parentSegments = segments.slice(0, -1);
+    const last = segments[segments.length - 1];
+    // Resolve the parent path. If only the root segment was provided we
+    // treat root as parent; if the parent has its own segments we walk
+    // them through findElementByPath against the root using a synthesised
+    // path.
+    let parent;
+    if (parentSegments.length === 0) {
+        parent = elementLocal(root) === last.localName ? null : root;
+        // Special case: `/root` itself as the repeating xpath — treat as one item.
+        if (elementLocal(root) === last.localName) {
+            return ["0"];
+        }
+    }
+    else {
+        const parentPath = "/" +
+            parentSegments
+                .map(s => s.position > 0
+                ? `${s.localName}[${s.position}]`
+                : s.position === -1
+                    ? `${s.localName}[last()]`
+                    : s.localName)
+                .join("/");
+        parent = findElementByPath(root, parentPath);
+    }
+    if (!parent) {
+        return [];
+    }
+    const count = directChildren(parent, last.localName).length;
+    return Array.from({ length: count }, (_, i) => String(i));
+}
+/**
+ * Resolve inner SDT data bindings within a repeated section item.
+ * Rewrites XPath in inner bindings to reference the specific item index,
+ * then evaluates and replaces the SDT text content.
+ */
+function resolveInnerBindings(content, parentXpath, itemPosition, xmlStore) {
+    if (!content || typeof content !== "object") {
+        return;
+    }
+    const obj = content;
+    // If this is an SDT with a dataBinding, resolve it in context of the repeated item
+    if (obj.type === "sdt" &&
+        typeof obj.properties === "object" &&
+        obj.properties !== null &&
+        obj.properties.dataBinding) {
+        const innerBinding = obj.properties.dataBinding;
+        // Rewrite the xpath: if its leading path-segment sequence equals
+        // `parentXpath`, insert a position predicate at the parent boundary.
+        // The previous implementation used a plain string `startsWith` check,
+        // which produced false positives when `parentXpath` was a prefix of a
+        // sibling's name (e.g. `/root/items` is a string prefix of
+        // `/root/items_total/count` even though `/root/items` is not a parent
+        // path at the segment level).
+        let resolvedXpath = innerBinding.xpath;
+        if (xpathStartsWithSegments(resolvedXpath, parentXpath)) {
+            resolvedXpath = `${parentXpath}[${itemPosition}]${resolvedXpath.slice(parentXpath.length)}`;
+        }
+        const modifiedBinding = {
+            ...innerBinding,
+            xpath: resolvedXpath
+        };
+        const value = evaluateSimpleXPath(modifiedBinding, xmlStore);
+        if (value !== null) {
+            const textRun = { content: [{ type: "text", text: value }] };
+            const textPara = { type: "paragraph", children: [textRun] };
+            obj.content = [textPara];
+        }
+        return;
+    }
+    // Recurse into arrays and object properties
+    if (Array.isArray(content)) {
+        for (const item of content) {
+            resolveInnerBindings(item, parentXpath, itemPosition, xmlStore);
+        }
+    }
+    else {
+        for (const key of Object.keys(obj)) {
+            const val = obj[key];
+            if (val && typeof val === "object") {
+                resolveInnerBindings(val, parentXpath, itemPosition, xmlStore);
+            }
+        }
+    }
+}
+/**
+ * Test whether `xpath` starts with `prefix` *as a sequence of path
+ * segments* — i.e. the next character after `prefix` is `/`, `[`, or end
+ * of string. Plain string `startsWith` mismatches sibling names (e.g.
+ * `/root/items` vs. `/root/items_total`).
+ */
+function xpathStartsWithSegments(xpath, prefix) {
+    if (!xpath.startsWith(prefix)) {
+        return false;
+    }
+    if (xpath.length === prefix.length) {
+        return true;
+    }
+    const next = xpath.charAt(prefix.length);
+    return next === "/" || next === "[";
+}

package/dist/cjs/modules/word/query/form-fields.js

@@ -0,0 +1,272 @@
+"use strict";
+/**
+ * Form Field Data API
+ *
+ * Extract and fill form field values in a document.
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.extractFormFields = extractFormFields;
+exports.fillFormFields = fillFormFields;
+const text_utils_1 = require("../core/text-utils");
+const walker_1 = require("../core/walker");
+// =============================================================================
+// Public API
+// =============================================================================
+/**
+ * Extract all form field values from a document.
+ *
+ * Traverses the document body (paragraphs, tables, headers, footers)
+ * and collects all legacy form field data.
+ *
+ * @param doc - The document to extract from.
+ * @returns Array of form field entries with their current values.
+ */
+function extractFormFields(doc) {
+    const results = [];
+    const visitor = {
+        enterRun(run) {
+            if (!run.content) {
+                return "continue";
+            }
+            for (const rc of run.content) {
+                if (rc.type === "field" && rc.formField) {
+                    const ff = rc.formField;
+                    let value;
+                    if (ff.type === "text") {
+                        value = ff.default ?? "";
+                    }
+                    else if (ff.type === "checkBox") {
+                        value = ff.checked ?? ff.default ?? false;
+                    }
+                    else {
+                        // dropdown
+                        value = ff.default ?? 0;
+                    }
+                    results.push({
+                        name: ff.name ?? "",
+                        type: ff.type,
+                        value,
+                        entries: ff.type === "dropDown" ? ff.entries : undefined
+                    });
+                }
+            }
+            return "continue";
+        }
+    };
+    (0, walker_1.walkDocument)(doc, visitor, {
+        includeHeaders: true,
+        includeFooters: true,
+        includeFootnotes: false,
+        includeEndnotes: false,
+        includeComments: false
+    });
+    return results;
+}
+/**
+ * Fill form field values throughout a document.
+ *
+ * The body, headers, footers, footnotes, endnotes and comments are all
+ * processed so a form field placed inside a header (a common case) still
+ * gets filled. Inside paragraphs the visitor descends into hyperlinks and
+ * track-change wrappers so a form field referenced from a tracked
+ * insertion is also reachable.
+ *
+ * @param doc - The document to fill (returns a new copy with filled values).
+ * @param values - Map of field name → new value.
+ * @returns A new document with form fields populated.
+ */
+function fillFormFields(doc, values) {
+    const newBody = doc.body.map(block => fillFieldsInBlock(block, values));
+    const newHeaders = doc.headers
+        ? new Map(Array.from(doc.headers, ([k, h]) => [
+            k,
+            {
+                ...h,
+                content: {
+                    ...h.content,
+                    children: h.content.children.map(c => fillFieldsInBlock(c, values))
+                }
+            }
+        ]))
+        : undefined;
+    const newFooters = doc.footers
+        ? new Map(Array.from(doc.footers, ([k, f]) => [
+            k,
+            {
+                ...f,
+                content: {
+                    ...f.content,
+                    children: f.content.children.map(c => fillFieldsInBlock(c, values))
+                }
+            }
+        ]))
+        : undefined;
+    const newFootnotes = doc.footnotes
+        ? doc.footnotes.map(fn => ({
+            ...fn,
+            content: fn.content.map(p => fillFieldsInParagraph(p, values))
+        }))
+        : undefined;
+    const newEndnotes = doc.endnotes
+        ? doc.endnotes.map(en => ({
+            ...en,
+            content: en.content.map(p => fillFieldsInParagraph(p, values))
+        }))
+        : undefined;
+    const newComments = doc.comments
+        ? doc.comments.map(cm => ({
+            ...cm,
+            content: cm.content.map(p => fillFieldsInParagraph(p, values))
+        }))
+        : undefined;
+    return {
+        ...doc,
+        body: newBody,
+        ...(newHeaders ? { headers: newHeaders } : {}),
+        ...(newFooters ? { footers: newFooters } : {}),
+        ...(newFootnotes ? { footnotes: newFootnotes } : {}),
+        ...(newEndnotes ? { endnotes: newEndnotes } : {}),
+        ...(newComments ? { comments: newComments } : {})
+    };
+}
+// =============================================================================
+// Internal helpers
+// =============================================================================
+function fillFieldsInBlock(block, values) {
+    if (block.type === "paragraph") {
+        return fillFieldsInParagraph(block, values);
+    }
+    if (block.type === "table") {
+        return {
+            ...block,
+            rows: block.rows.map(r => ({
+                ...r,
+                cells: r.cells.map(c => ({
+                    ...c,
+                    content: c.content.map(inner => fillFieldsInBlock(inner, values))
+                }))
+            }))
+        };
+    }
+    if (block.type === "sdt") {
+        return {
+            ...block,
+            content: block.content.map(c => {
+                if ("type" in c && (c.type === "paragraph" || c.type === "table")) {
+                    return fillFieldsInBlock(c, values);
+                }
+                if (c &&
+                    typeof c === "object" &&
+                    !("type" in c) &&
+                    "content" in c &&
+                    Array.isArray(c.content)) {
+                    // Inline (run-only) SDT child — fill any field run-content nodes
+                    // it carries directly.
+                    return fillFieldsInRun(c, values);
+                }
+                return c;
+            })
+        };
+    }
+    if (block.type === "textBox") {
+        return {
+            ...block,
+            content: block.content.map(p => fillFieldsInParagraph(p, values))
+        };
+    }
+    if (block.type === "drawingShape") {
+        if (!block.textContent || block.textContent.length === 0) {
+            return block;
+        }
+        return {
+            ...block,
+            textContent: block.textContent.map(p => fillFieldsInParagraph(p, values))
+        };
+    }
+    if (block.type === "tableOfContents") {
+        if (!block.cachedParagraphs || block.cachedParagraphs.length === 0) {
+            return block;
+        }
+        return {
+            ...block,
+            cachedParagraphs: block.cachedParagraphs.map(p => fillFieldsInParagraph(p, values))
+        };
+    }
+    return block;
+}
+function fillFieldsInParagraph(para, values) {
+    let modified = false;
+    const newChildren = para.children.map(child => {
+        // Hyperlink: descend into its run children.
+        if ((0, text_utils_1.isHyperlink)(child)) {
+            let hlModified = false;
+            const newRuns = child.children.map(r => {
+                const filled = fillFieldsInRun(r, values);
+                if (filled !== r) {
+                    hlModified = true;
+                }
+                return filled;
+            });
+            if (hlModified) {
+                modified = true;
+                return { ...child, children: newRuns };
+            }
+            return child;
+        }
+        // Track-change wrappers around a run.
+        if ("type" in child &&
+            (child.type === "insertedRun" ||
+                child.type === "movedToRun")) {
+            const wrapper = child;
+            const filled = fillFieldsInRun(wrapper.run, values);
+            if (filled !== wrapper.run) {
+                modified = true;
+                return { ...wrapper, run: filled };
+            }
+            return child;
+        }
+        if (!(0, text_utils_1.isRun)(child)) {
+            return child;
+        }
+        const filled = fillFieldsInRun(child, values);
+        if (filled !== child) {
+            modified = true;
+        }
+        return filled;
+    });
+    return modified ? { ...para, children: newChildren } : para;
+}
+function fillFieldsInRun(run, values) {
+    if (!run.content) {
+        return run;
+    }
+    let runModified = false;
+    const newContent = run.content.map(rc => {
+        if (rc.type !== "field" || !rc.formField) {
+            return rc;
+        }
+        const ff = rc.formField;
+        const name = ff.name ?? "";
+        if (!values.has(name)) {
+            return rc;
+        }
+        const newVal = values.get(name);
+        runModified = true;
+        let newFF;
+        if (ff.type === "text") {
+            newFF = { ...ff, default: String(newVal) };
+        }
+        else if (ff.type === "checkBox") {
+            newFF = { ...ff, checked: Boolean(newVal) };
+        }
+        else {
+            // dropdown: value is the selected index
+            newFF = {
+                ...ff,
+                default: typeof newVal === "number" ? newVal : parseInt(String(newVal), 10) || 0
+            };
+        }
+        return { ...rc, formField: newFF };
+    });
+    return runModified ? { ...run, content: newContent } : run;
+}