chore: office viewer 支持中英文空行；优化 tblpPr 的支持 (#6433)

packages/amis-core/src/utils/debug.tsx

@@ -385,6 +385,7 @@ export function enableDebug() {
 
 interface DebugWrapperProps {
   renderer: any;
+  children?: React.ReactNode;
 }
 
 export class DebugWrapper extends Component<DebugWrapperProps> {

packages/office-viewer/README.md

@@ -6,6 +6,10 @@ docx 渲染器，原理是将 docx 里的 xml 格式转成 html
 
 相对于 Canvas 渲染，这个实现方案比较简单，最终页面也可以很方便复制，但无法保证和原始 docx 文件展现一致，因为有部分功能难以在 HTML 中实现，比如图文环绕效果。
 
+## 还不支持的功能
+
+- wmf，需要使用 https://github.com/SheetJS/js-wmf
+
 ## 参考资料
 
 - [官方规范](https://www.ecma-international.org/publications-and-standards/standards/ecma-376/)

packages/office-viewer/__tests__/util/autoSpace.test.ts

@@ -0,0 +1,9 @@
+import {cjkspace} from '../../src/util/autoSpace';
+
+test('autoSpace', async () => {
+  expect(cjkspace('a中'.split(''))).toBe('a 中');
+});
+
+test('autoSpace 2', async () => {
+  expect(cjkspace('abc中def，测试'.split(''))).toBe('abc 中 def，测试');
+});

packages/office-viewer/src/openxml/word/Body.ts

@@ -52,6 +52,9 @@ export class Body {
           body.addChild(table);
           break;
 
+        case 'w:bookmarkEnd':
+          break;
+
         default:
           console.warn('Body.fromXML Unknown key', tagName, child);
       }

packages/office-viewer/src/openxml/word/Paragraph.ts

@@ -21,6 +21,11 @@ export interface ParagraphPr extends Properties {
   numPr?: NumberPr;
   runPr?: RunPr;
   tabs?: Tab[];
+
+  /**
+   * 其实是区分 autoSpaceDN 和 autoSpaceDE 的，但这里简化了
+   */
+  autoSpace?: boolean;
 }
 
 export type ParagraphChild =
@@ -45,6 +50,12 @@ export type ParagraphChild =
 // | CommentRangeEnd
 // | CommentReference;
 
+function parseAutoSpace(element: Element): boolean {
+  const autoSpaceDE = element.getElementsByTagName('w:autoSpaceDE').item(0);
+  const autoSpaceDN = element.getElementsByTagName('w:autoSpaceDN').item(0);
+  return !!autoSpaceDE || !!autoSpaceDN;
+}
+
 export class Paragraph {
   // 主要是为了方便调试用的
   paraId?: string;
@@ -78,7 +89,9 @@ export class Paragraph {
       tabs.push(Tab.fromXML(word, tabElement));
     }
 
-    return {cssStyle, pStyle, numPr, tabs};
+    const autoSpace = parseAutoSpace(element);
+
+    return {cssStyle, pStyle, numPr, tabs, autoSpace};
   }
 
   static fromXML(word: Word, element: Element): Paragraph {

packages/office-viewer/src/openxml/word/Table.ts

@@ -76,7 +76,8 @@ function parseTblJc(element: Element, cssStyle: CSSStyle) {
   switch (val) {
     case 'left':
     case 'start':
-      cssStyle['float'] = 'left';
+      // TODO: 会导致前面的文字掉下去，感觉还是不能支持这个功能
+      // cssStyle['float'] = 'left';
       break;
     case 'right':
     case 'end':
@@ -165,16 +166,26 @@ function parseTblLook(child: Element) {
  * http://webapp.docx4java.org/OnlineDemo/ecma376/WordML/tblpPr.html
  * 只支持部分
  */
-function parsetTlpPr(child: Element, style: CSSStyle) {
-  const topFromText = parseSize(child, 'w:topFromText');
-  const bottomFromText = parseSize(child, 'w:bottomFromText');
-  const rightFromText = parseSize(child, 'w:rightFromText');
-  const leftFromText = parseSize(child, 'w:leftFromText');
-  style['float'] = 'left';
-  style['margin-bottom'] = addSize(style['margin-bottom'], bottomFromText);
-  style['margin-left'] = addSize(style['margin-left'], leftFromText);
-  style['margin-right'] = addSize(style['margin-right'], rightFromText);
-  style['margin-top'] = addSize(style['margin-top'], topFromText);
+function parsetTlpPr(word: Word, child: Element, style: CSSStyle) {
+  // 如果设置 padding 会导致绝对定位不准确，所以一旦设置就不支持
+  if (typeof word.renderOptions.padding === 'undefined') {
+    const tplpX = parseSize(child, 'w:tblpX');
+    const tplpY = parseSize(child, 'w:tblpY');
+    style.position = 'absolute';
+    style.top = tplpY;
+    style.left = tplpX;
+  }
+
+  // 之前想用 float 来实现，但是会导致文字掉下去
+  // const topFromText = parseSize(child, 'w:topFromText');
+  // const bottomFromText = parseSize(child, 'w:bottomFromText');
+  // const rightFromText = parseSize(child, 'w:rightFromText');
+  // const leftFromText = parseSize(child, 'w:leftFromText');
+  // style['float'] = 'left';
+  // style['margin-bottom'] = addSize(style['margin-bottom'], bottomFromText);
+  // style['margin-left'] = addSize(style['margin-left'], leftFromText);
+  // style['margin-right'] = addSize(style['margin-right'], rightFromText);
+  // style['margin-top'] = addSize(style['margin-top'], topFromText);
 }
 
 export class Table {
@@ -257,7 +268,7 @@ export class Table {
           break;
 
         case 'w:tblpPr':
-          parsetTlpPr(child, tableStyle);
+          parsetTlpPr(word, child, tableStyle);
           break;
 
         default:

packages/office-viewer/src/parse/parsePr.ts

@@ -343,6 +343,29 @@ export function parsePr(word: Word, element: Element, type: 'r' | 'p' = 'p') {
         // 目前是自动计算的，所以不需要这个了
         break;
 
+      case 'w:bidi':
+        // http://webapp.docx4java.org/OnlineDemo/ecma376/WordML/bidi_1.html
+        // TODO: 还不清楚和 w:textDirection 是什么关系
+        if (getValBoolean(child, true)) {
+          console.warn('w:bidi is not supported.');
+        }
+        break;
+
+      case 'w:autoSpaceDE':
+      case 'w:autoSpaceDN':
+        // 这个在其它地方实现了
+        break;
+
+      case 'w:kinsoku':
+        // http://webapp.docx4java.org/OnlineDemo/ecma376/WordML/kinsoku.html
+        // 控制不了所以忽略了
+        break;
+
+      case 'w:overflowPunct':
+        // http://webapp.docx4java.org/OnlineDemo/ecma376/WordML/overflowPunct.html
+        // 支持不了
+        break;
+
       default:
         console.warn('parsePr Unknown tagName', tagName, child);
     }

packages/office-viewer/src/render/renderHyperLink.ts

@@ -3,11 +3,16 @@ import {appendChild, createElement} from '../util/dom';
 import Word from '../Word';
 import {Run} from '../openxml/word/Run';
 import renderRun from './renderRun';
+import type {Paragraph} from '../openxml/word/Paragraph';
 
 /**
  * 渲染链接
  */
-export function renderHyperLink(word: Word, hyperlink: Hyperlink): HTMLElement {
+export function renderHyperLink(
+  word: Word,
+  hyperlink: Hyperlink,
+  paragraph?: Paragraph
+): HTMLElement {
   const a = createElement('a') as HTMLAnchorElement;
 
   if (hyperlink.relation) {
@@ -24,7 +29,7 @@ export function renderHyperLink(word: Word, hyperlink: Hyperlink): HTMLElement {
 
   for (const child of hyperlink.children) {
     if (child instanceof Run) {
-      const span = renderRun(word, child);
+      const span = renderRun(word, child, paragraph);
       appendChild(a, span);
     }
   }

packages/office-viewer/src/render/renderNumbering.ts

@@ -78,6 +78,11 @@ export function renderNumbering(
     return null;
   }
 
+  if (!numbering) {
+    console.warn('renderNumbering: numbering is empty');
+    return null;
+  }
+
   const num = numbering.nums[numId];
 
   if (!num) {

packages/office-viewer/src/render/renderParagraph.ts

@@ -45,11 +45,11 @@ export default function renderParagraph(
 
   for (const child of paragraph.children) {
     if (child instanceof Run) {
-      appendChild(p, renderRun(word, child));
+      appendChild(p, renderRun(word, child, paragraph));
     } else if (child instanceof BookmarkStart) {
       appendChild(p, renderBookmarkStart(word, child));
     } else if (child instanceof Hyperlink) {
-      const hyperlink = renderHyperLink(word, child);
+      const hyperlink = renderHyperLink(word, child, paragraph);
       appendChild(p, hyperlink);
     } else if (child instanceof SmartTag) {
       renderInlineText(word, child, p);

packages/office-viewer/src/render/renderRun.ts

@@ -20,16 +20,27 @@ import {InstrText} from '../openxml/word/InstrText';
 import {renderInstrText} from './renderInstrText';
 import {Sym} from '../openxml/word/Sym';
 import {renderSym} from './renderSym';
+import {cjkspace} from '../util/autoSpace';
+import type {Paragraph} from './../openxml/word/Paragraph';
 
 const VARIABLE_CLASS_NAME = 'variable';
 
 /**
  * 对文本进行替换
  */
-function renderText(span: HTMLElement, word: Word, text: string) {
+function renderText(
+  span: HTMLElement,
+  word: Word,
+  text: string,
+  paragraph?: Paragraph
+) {
   // 简单过滤一下提升性能
   if (text.indexOf('{{') === -1) {
+    if (paragraph?.properties?.autoSpace) {
+      span.textContent = cjkspace(text.split(''));
+    } else {
       span.textContent = text;
+    }
   } else {
     span.dataset.originText = text;
     // 加个标识，后续可以通过它来查找哪些变量需要替换，这样就不用重新渲染整个文档了
@@ -53,7 +64,7 @@ export function updateVariableText(word: Word) {
 /**
  * 渲染 run 节点
  */
-export default function renderRun(word: Word, run: Run) {
+export default function renderRun(word: Word, run: Run, paragraph?: Paragraph) {
   const span = createElement('span');
 
   word.addClass(span, 'r');
@@ -62,12 +73,12 @@ export default function renderRun(word: Word, run: Run) {
 
   if (run.children.length === 1 && run.children[0] instanceof Text) {
     const text = run.children[0] as Text;
-    renderText(span, word, text.text);
+    renderText(span, word, text.text, paragraph);
   } else {
     for (const child of run.children) {
       if (child instanceof Text) {
         let newSpan = createElement('span');
-        renderText(span, word, child.text);
+        renderText(span, word, child.text, paragraph);
         appendChild(span, newSpan);
       } else if (child instanceof Break) {
         const br = renderBr(child);

packages/office-viewer/src/render/renderSection.ts

@@ -8,6 +8,8 @@ import Word from '../Word';
 export function renderSection(word: Word, section: Section) {
   const sectionEl = createElement('section') as HTMLElement;
 
+  // 用于后续绝对定位
+  sectionEl.style.position = 'relative';
   const props = section.properties;
   const pageSize = props.pageSize;
   if (pageSize) {

packages/office-viewer/src/util/autoSpace.ts

@@ -0,0 +1,41 @@
+/**
+ * 中英文间自动加空格，基于下面代码改的，去掉了 lodash 依赖
+ * https://gist.github.com/wyl8899/e0f31068681023480e20c34f6b19a275
+ */
+
+/* Partial implementation from https://zhuanlan.zhihu.com/p/33612593 */
+
+/* 标点 */
+const punctuationRegex = /\p{Punctuation}/u;
+/* 空格 */
+const spaceRegex = /\p{Separator}/u;
+/* CJK 字符，中日韩 */
+const cjkRegex =
+  /\p{Script=Han}|\p{Script=Katakana}|\p{Script=Hiragana}|\p{Script=Hangul}/u;
+
+const shouldSpace = (a: string, b: string): boolean => {
+  if (cjkRegex.test(a)) {
+    return !(
+      punctuationRegex.test(b) ||
+      spaceRegex.test(b) ||
+      cjkRegex.test(b)
+    );
+  } else {
+    return cjkRegex.test(b) && !punctuationRegex.test(a) && !spaceRegex.test(a);
+  }
+};
+
+const join = (
+  parts: string[],
+  sepFunc: (a: string, b: string) => string
+): string => {
+  return parts.reduce((r, p, i) => {
+    const sep = i !== 0 ? sepFunc(p, parts[i - 1]) : '';
+    return r + sep + p;
+  }, '');
+};
+
+export const cjkspace = (strings: string[]): string => {
+  const filtered = strings.filter(c => c !== undefined && c !== '') as string[];
+  return join(filtered, (a, b) => (shouldSpace(a, b) ? ' ' : ''));
+};