@cj-tech-master/excelts 9.5.4 → 9.5.5

package/dist/cjs/modules/word/reader/comments-parser.js

@@ -0,0 +1,96 @@
+"use strict";
+/**
+ * DOCX reader — comments parsers.
+ *
+ * Split out from `docx-reader.ts` to keep that file's part-orchestration
+ * scope manageable. We accept `parseParagraph` as an injected dependency
+ * because the body paragraph parser still lives in `docx-reader.ts` (it
+ * has many cross-references that would not be cheap to disentangle); a
+ * direct import would form a cycle.
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.parseCommentsXml = parseCommentsXml;
+exports.parseCommentsExtendedXml = parseCommentsExtendedXml;
+const dom_1 = require("../../xml/dom.js");
+const parse_utils_1 = require("./parse-utils");
+const reader_context_1 = require("./reader-context");
+/** Parse `word/comments.xml` into a list of CommentDef. */
+function parseCommentsXml(xmlStr, ctx, parseParagraph) {
+    // Comments are a separate part — reset the field state so a body-level
+    // unterminated field does not bleed into comment parsing.
+    const savedField = ctx.field;
+    ctx.field = (0, reader_context_1.createFieldState)();
+    try {
+        const doc = (0, dom_1.parseXml)(xmlStr);
+        const root = doc.root;
+        const comments = [];
+        for (const commentEl of (0, parse_utils_1.findChildrenNs)(root, "comment")) {
+            const id = (0, parse_utils_1.attrInt)(commentEl, "id");
+            const author = (0, parse_utils_1.attrVal)(commentEl, "author");
+            if (id === undefined || !author) {
+                continue;
+            }
+            const content = [];
+            for (const child of commentEl.children) {
+                if (child.type === "element" && child.name.replace(/^w:/, "") === "p") {
+                    content.push(parseParagraph(child, ctx));
+                }
+            }
+            const comment = { id, author, content };
+            const date = (0, parse_utils_1.attrVal)(commentEl, "date");
+            if (date) {
+                comment.date = date;
+            }
+            const initials = (0, parse_utils_1.attrVal)(commentEl, "initials");
+            if (initials) {
+                comment.initials = initials;
+            }
+            comments.push(comment);
+        }
+        return comments;
+    }
+    finally {
+        ctx.field = savedField;
+    }
+}
+/**
+ * Parse `word/commentsExtended.xml` into a paraId → metadata map.
+ *
+ * Modern Word stores comment "resolved" state and reply threading in a
+ * sidecar part keyed by the paragraph id of each comment's first
+ * paragraph. The reader merges this map into `CommentDef.done` /
+ * `CommentDef.parentId` once both parts have been parsed.
+ */
+function parseCommentsExtendedXml(xmlStr) {
+    const map = new Map();
+    const doc = (0, dom_1.parseXml)(xmlStr);
+    const root = doc.root;
+    for (const child of root.children) {
+        if (child.type !== "element") {
+            continue;
+        }
+        // w15:commentEx
+        const name = child.name;
+        if (!name.endsWith("commentEx")) {
+            continue;
+        }
+        const paraId = child.attributes["w15:paraId"] ?? child.attributes["paraId"];
+        if (!paraId) {
+            continue;
+        }
+        const entry = {};
+        const done = child.attributes["w15:done"] ?? child.attributes["done"];
+        if (done === "1" || done === "true") {
+            entry.done = true;
+        }
+        else if (done === "0" || done === "false") {
+            entry.done = false;
+        }
+        const pid = child.attributes["w15:paraIdParent"] ?? child.attributes["paraIdParent"];
+        if (pid) {
+            entry.parentId = pid;
+        }
+        map.set(paraId, entry);
+    }
+    return map;
+}

package/dist/cjs/modules/word/reader/doc-props-parsers.js

@@ -0,0 +1,194 @@
+"use strict";
+/**
+ * DOCX Reader - Document Properties Parsers
+ *
+ * Parses standard OPC document properties XML files:
+ * - docProps/core.xml (CoreProperties — title, author, dc:* fields)
+ * - docProps/app.xml (AppProperties — application info)
+ * - docProps/custom.xml (CustomProperty[] — user-defined name/value pairs)
+ * - word/fontTable.xml (FontDef[] — embedded font references)
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.parseCoreProps = parseCoreProps;
+exports.parseAppProps = parseAppProps;
+exports.parseCustomPropsXml = parseCustomPropsXml;
+exports.parseFontTableXml = parseFontTableXml;
+const dom_1 = require("../../xml/dom.js");
+const parse_utils_1 = require("./parse-utils");
+// =============================================================================
+// Core Properties Parser
+// =============================================================================
+function parseCoreProps(xmlStr) {
+    const doc = (0, dom_1.parseXml)(xmlStr);
+    const root = doc.root;
+    const props = {};
+    const fields = [
+        ["dc:title", "title"],
+        ["dc:subject", "subject"],
+        ["dc:creator", "creator"],
+        ["dc:description", "description"],
+        ["cp:keywords", "keywords"],
+        ["cp:lastModifiedBy", "lastModifiedBy"],
+        ["cp:revision", "revision"],
+        ["cp:category", "category"]
+    ];
+    for (const [tag, prop] of fields) {
+        const el = (0, dom_1.findChild)(root, tag);
+        if (el) {
+            const val = (0, dom_1.textContent)(el);
+            if (val) {
+                props[prop] = val;
+            }
+        }
+    }
+    const createdEl = (0, dom_1.findChild)(root, "dcterms:created");
+    if (createdEl) {
+        const val = (0, dom_1.textContent)(createdEl);
+        if (val) {
+            props.created = new Date(val);
+        }
+    }
+    const modifiedEl = (0, dom_1.findChild)(root, "dcterms:modified");
+    if (modifiedEl) {
+        const val = (0, dom_1.textContent)(modifiedEl);
+        if (val) {
+            props.modified = new Date(val);
+        }
+    }
+    return props;
+}
+// =============================================================================
+// App Properties Parser
+// =============================================================================
+function parseAppProps(xmlStr) {
+    const doc = (0, dom_1.parseXml)(xmlStr);
+    const root = doc.root;
+    const props = {};
+    const strFields = ["Application", "AppVersion", "Company", "Manager"];
+    const intFields = ["Pages", "Words", "Characters", "Lines", "Paragraphs"];
+    for (const field of strFields) {
+        const el = (0, dom_1.findChild)(root, field);
+        if (el) {
+            const val = (0, dom_1.textContent)(el);
+            if (val) {
+                props[field.charAt(0).toLowerCase() + field.slice(1)] = val;
+            }
+        }
+    }
+    for (const field of intFields) {
+        const el = (0, dom_1.findChild)(root, field);
+        if (el) {
+            const val = (0, dom_1.textContent)(el);
+            if (val) {
+                props[field.charAt(0).toLowerCase() + field.slice(1)] = parseInt(val, 10);
+            }
+        }
+    }
+    return props;
+}
+// =============================================================================
+// Custom Properties Parser
+// =============================================================================
+function parseCustomPropsXml(xmlStr) {
+    const doc = (0, dom_1.parseXml)(xmlStr);
+    const root = doc.root;
+    const props = [];
+    for (const propEl of root.children) {
+        if (propEl.type !== "element" || propEl.name !== "property") {
+            continue;
+        }
+        const name = propEl.attributes["name"];
+        if (!name) {
+            continue;
+        }
+        let value;
+        for (const child of propEl.children) {
+            if (child.type !== "element") {
+                continue;
+            }
+            const tn = child.name;
+            const tv = (0, dom_1.textContent)(child);
+            if (tn === "vt:lpwstr") {
+                value = { type: "string", value: tv };
+            }
+            else if (tn === "vt:i4") {
+                value = { type: "number", value: parseInt(tv, 10) };
+            }
+            else if (tn === "vt:r8") {
+                value = { type: "number", value: parseFloat(tv) };
+            }
+            else if (tn === "vt:bool") {
+                value = { type: "boolean", value: tv === "true" };
+            }
+            else if (tn === "vt:filetime") {
+                value = { type: "date", value: new Date(tv) };
+            }
+        }
+        if (value) {
+            props.push({ name, value });
+        }
+    }
+    return props;
+}
+// =============================================================================
+// Font Table Parser
+// =============================================================================
+function parseFontTableXml(xmlStr) {
+    const doc = (0, dom_1.parseXml)(xmlStr);
+    const root = doc.root;
+    const fonts = [];
+    for (const fontEl of (0, parse_utils_1.findChildrenNs)(root, "font")) {
+        const f = { name: (0, parse_utils_1.attrVal)(fontEl, "name") ?? "" };
+        const p1 = (0, parse_utils_1.findChildNs)(fontEl, "panose1");
+        if (p1) {
+            f.panose1 = (0, parse_utils_1.attrVal)(p1, "val");
+        }
+        const cs = (0, parse_utils_1.findChildNs)(fontEl, "charset");
+        if (cs) {
+            f.charset = (0, parse_utils_1.attrVal)(cs, "val");
+        }
+        const fam = (0, parse_utils_1.findChildNs)(fontEl, "family");
+        if (fam) {
+            f.family = (0, parse_utils_1.attrVal)(fam, "val");
+        }
+        const pitch = (0, parse_utils_1.findChildNs)(fontEl, "pitch");
+        if (pitch) {
+            f.pitch = (0, parse_utils_1.attrVal)(pitch, "val");
+        }
+        // Signature
+        const sigEl = (0, parse_utils_1.findChildNs)(fontEl, "sig");
+        if (sigEl) {
+            const sig = {};
+            for (const key of ["usb0", "usb1", "usb2", "usb3", "csb0", "csb1"]) {
+                const v = (0, parse_utils_1.attrVal)(sigEl, key);
+                if (v !== undefined) {
+                    sig[key] = v;
+                }
+            }
+            if (Object.keys(sig).length > 0) {
+                f.sig = sig;
+            }
+        }
+        // Embedded fonts
+        for (const [tag, rIdKey, keyKey] of [
+            ["embedRegular", "embedRegular", "embedRegularKey"],
+            ["embedBold", "embedBold", "embedBoldKey"],
+            ["embedItalic", "embedItalic", "embedItalicKey"],
+            ["embedBoldItalic", "embedBoldItalic", "embedBoldItalicKey"]
+        ]) {
+            const el = (0, parse_utils_1.findChildNs)(fontEl, tag);
+            if (el) {
+                const rId = el.attributes["r:id"] ?? el.attributes["id"];
+                if (rId) {
+                    f[rIdKey] = rId;
+                    const fontKey = (0, parse_utils_1.attrVal)(el, "fontKey");
+                    if (fontKey) {
+                        f[keyKey] = fontKey;
+                    }
+                }
+            }
+        }
+        fonts.push(f);
+    }
+    return fonts;
+}