patent-xml-generator 1.0.0

package/dist/xml-utils.js

@@ -0,0 +1,95 @@
+"use strict";
+/**
+ * XML 工具函数
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.escapeXml = escapeXml;
+exports.escapeXmlPreservingInlineTags = escapeXmlPreservingInlineTags;
+exports.indent = indent;
+exports.newline = newline;
+exports.formatParagraphNum = formatParagraphNum;
+exports.checkGB18030 = checkGB18030;
+/** XML 特殊字符转义 */
+function escapeXml(text) {
+    return text
+        .replace(/&/g, '&')
+        .replace(/</g, '&lt;')
+        .replace(/>/g, '&gt;')
+        .replace(/"/g, '&quot;')
+        .replace(/'/g, '&apos;');
+}
+/** 检测文本中是否包含需要保留的行内标记 */
+const INLINE_TAGS = ['sup', 'sub', 'b', 'i', 'u', 'img', 'br', 'smallcaps', 'overscore'];
+const INLINE_TAG_PATTERN = new RegExp(`<(${INLINE_TAGS.join('|')})(\\s[^>]*)?>.*?</(${INLINE_TAGS.join('|')})>|<(br)\\s*/?>`, 'g');
+/**
+ * 智能转义：保留合法的行内XML标记，转义其余内容
+ * 支持 <sup>、<sub>、<b>、<i>、<u>、<br/> 等 DTD 允许的行内标记
+ */
+function escapeXmlPreservingInlineTags(text) {
+    // 找出所有合法的行内标记位置
+    const matches = [];
+    let match;
+    const pattern = new RegExp(INLINE_TAG_PATTERN.source, 'g');
+    while ((match = pattern.exec(text)) !== null) {
+        matches.push({
+            start: match.index,
+            end: match.index + match[0].length,
+            tag: match[0]
+        });
+    }
+    if (matches.length === 0) {
+        return escapeXml(text);
+    }
+    // 逐段处理：标记外的文本转义，标记内的保留
+    let result = '';
+    let cursor = 0;
+    for (const m of matches) {
+        if (cursor < m.start) {
+            result += escapeXml(text.slice(cursor, m.start));
+        }
+        result += m.tag;
+        cursor = m.end;
+    }
+    if (cursor < text.length) {
+        result += escapeXml(text.slice(cursor));
+    }
+    return result;
+}
+/**
+ * 生成缩进字符串
+ */
+function indent(level, useIndent) {
+    return useIndent ? '  '.repeat(level) : '';
+}
+/**
+ * 生成换行符
+ */
+function newline(useIndent) {
+    return useIndent ? '\n' : '';
+}
+/**
+ * 段落编号格式化：将数字转为4位补零字符串
+ * 1 -> "0001", 12 -> "0012"
+ */
+function formatParagraphNum(num) {
+    return num.toString().padStart(4, '0');
+}
+/**
+ * 检测文本是否为GB18030字符集范围内
+ * 超出范围的字符需要转为图片
+ */
+function checkGB18030(text) {
+    const invalidChars = [];
+    for (const char of text) {
+        const code = char.codePointAt(0);
+        // 基本判断：BMP范围内大部分CJK字符都在GB18030内
+        // 超出BMP的字符（如emoji、生僻字扩展区）可能不在GB18030内
+        if (code > 0xFFFF) {
+            invalidChars.push(char);
+        }
+    }
+    return {
+        valid: invalidChars.length === 0,
+        invalidChars
+    };
+}