Merge branch 'cppdoc-cc:main' into main

Author: mumupika

migrate/migrate-bot.ts

@@ -5,6 +5,7 @@ import path, { join } from "path";
 import { fileURLToPath } from "url";
 import { execSync, spawnSync } from "child_process";
 import { visualizeTextDiff } from "./text-diff-visualizer";
+import { getTextFromDOM } from "./text-from-element";
 
 const __dirname = path.dirname(fileURLToPath(import.meta.url));
 
@@ -69,7 +70,17 @@ async function fetchPageContent(
   const html = await response.text();
   const dom = new JSDOM(html);
   const contentElement = dom.window.document.querySelector("#mw-content-text");
-  contentElement?.querySelector(".t-navbar")?.remove();
+
+  const selectorsToRemove = [
+    ".t-navbar",
+    ".t-example-live-link",
+    ".editsection",
+    "#toc",
+  ];
+  for (const selector of selectorsToRemove) {
+    const elements = contentElement?.querySelectorAll(selector);
+    elements?.forEach((el) => el.remove());
+  }
   const headingElement = dom.window.document.querySelector("#firstHeading");
   if (!contentElement) {
     throw new Error("Could not find #mw-content-text");
@@ -78,7 +89,7 @@ async function fetchPageContent(
     html: contentElement.innerHTML,
     title: headingElement?.textContent?.trim() || "",
     url,
-    innerText: contentElement.textContent?.trim() || "",
+    innerText: getTextFromDOM(contentElement),
   };
 }
 
@@ -91,7 +102,7 @@ async function convertToMDX(
     "{{LLM_DOCS}}",
     await readFile(
       __dirname +
-        "/../src/content/docs/development/guide/component-docs-for-llm.mdx",
+      "/../src/content/docs/development/guide/component-docs-for-llm.mdx",
       "utf8"
     )
   );
@@ -295,7 +306,14 @@ async function createPullRequest(
     .then((data) => {
       const dom = new JSDOM(data);
       const contentElement = dom.window.document.querySelector("main");
-      return contentElement?.textContent?.trim() || "";
+      const selectorsToRemove = [".sl-anchor-link"];
+      for (const selector of selectorsToRemove) {
+        const elements = contentElement?.querySelectorAll(selector);
+        elements?.forEach((el) => el.remove());
+      }
+
+      if (!contentElement) return "";
+      return getTextFromDOM(contentElement);
     })
     .catch(() => "");
 
@@ -423,19 +441,25 @@ async function main() {
       console.log(`  写入 ${filePath}`);
       await writeMDXFile(filePath, mdx, title);
 
-      console.log(`  尝试构建...`);
+      console.log(`  重新格式化...`);
+      spawnSync(`npm`, ["run", "format"], {
+        stdio: "inherit",
+        shell: true,
+      });
+
+      console.log(`  构建...`);
       const res = spawnSync(`npm`, ["run", "build"], {
         stdio: "inherit",
         shell: true,
       });
       if (res.status !== 0) {
         throw new Error(
           "构建失败，可能生成的MDX有问题：" +
-            res.stderr?.toString() +
-            res.stdout?.toString() +
-            res.error?.toString() +
-            " exit code " +
-            res.status
+          res.stderr?.toString() +
+          res.stdout?.toString() +
+          res.error?.toString() +
+          " exit code " +
+          res.status
         );
       }
 

migrate/text-from-element.ts

@@ -0,0 +1,143 @@
+import { JSDOM } from "jsdom"
+
+interface GetTextOptions {
+    treatBlockAsNewline?: boolean;
+    collapseSpaces?: boolean;
+    trimResult?: boolean;
+}
+
+const BLOCK_ELEMENTS = [
+    'div', 'p', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6',
+    'ul', 'ol', 'li', 'table', 'tr', 'td', 'th',
+    'section', 'article', 'header', 'footer', 'nav',
+    'aside', 'main', 'figure', 'figcaption', 'blockquote',
+    'pre', 'form', 'fieldset', 'legend', 'dl', 'dt', 'dd',
+    'hr', 'br'
+];
+
+const Node = new JSDOM('').window.Node;
+function isBlockElement(node: Node): boolean {
+    if (node.nodeType !== Node.ELEMENT_NODE) return false;
+    const element = node as HTMLElement;
+
+    if (BLOCK_ELEMENTS.includes(element.tagName.toLowerCase())) {
+        return true;
+    }
+
+    const style = window.getComputedStyle(element);
+    return style.display === 'block' ||
+        style.display === 'flex' ||
+        style.display === 'grid' ||
+        style.display.startsWith('table');
+}
+
+function isElementVisible(element: HTMLElement): boolean {
+    const style = window.getComputedStyle(element);
+    return style.display !== 'none' &&
+        style.visibility !== 'hidden' &&
+        style.opacity !== '0';
+}
+
+export function getTextFromDOM(
+    node: Node,
+    options: GetTextOptions = {}
+): string {
+    const {
+        treatBlockAsNewline = true,
+        collapseSpaces = true,
+        trimResult = true
+    } = options;
+
+    let result = '';
+    let lastChar = '';
+
+    function processNode(currentNode: Node, isBlockContext: boolean) {
+        if (!currentNode) return;
+        if (currentNode.nodeType === Node.ELEMENT_NODE) {
+            const element = currentNode as HTMLElement;
+            if (!isElementVisible(element)) return;
+
+            const isBlock = isBlockElement(currentNode);
+            const tagName = element.tagName.toLowerCase();
+
+            if (tagName === 'br') {
+                result += '\n';
+                lastChar = '\n';
+                return;
+            }
+
+            if (tagName === 'hr') {
+                result += '\n---\n';
+                lastChar = '\n';
+                return;
+            }
+
+
+            if (tagName === 'pre') {
+                const text = element.textContent || '';
+                if (text) {
+                    result += text;
+                    lastChar = text[text.length - 1] || '';
+                }
+                return;
+            }
+
+            const shouldAddNewline = treatBlockAsNewline && isBlock;
+            const separator = shouldAddNewline ? '\n' : ' ';
+
+            if (isBlock && result.length > 0 && lastChar !== '\n') {
+                result += separator;
+                lastChar = separator;
+            }
+
+            const currentIsBlockContext = isBlock || isBlockContext;
+            for (const childNode of Array.from(element.childNodes)) {
+                processNode(childNode, currentIsBlockContext);
+            }
+
+            if (isBlock && result.length > 0 && lastChar !== '\n') {
+                result += separator;
+                lastChar = separator;
+            }
+        } else if (currentNode.nodeType === Node.TEXT_NODE) {
+            let text = currentNode.textContent || '';
+            if (text.trim() === '') return;
+            text = text.replace(/\s+/g, ' ');
+            if (text.startsWith(' ')) {
+                if (result.length > 0 && lastChar !== ' ' && lastChar !== '\n') {
+                    result += ' ';
+                    lastChar = ' ';
+                }
+                text = text.substring(1);
+            }
+
+            if (text) {
+                const endsWithSpace = text.endsWith(' ');
+                const cleanText = endsWithSpace ? text.slice(0, -1) : text;
+                result += cleanText;
+                lastChar = cleanText[cleanText.length - 1] || '';
+                if (endsWithSpace && lastChar !== ' ' && lastChar !== '\n') {
+                    result += ' ';
+                    lastChar = ' ';
+                }
+            }
+        } else {
+            return;
+        }
+    }
+
+
+    const initialIsBlock = isBlockElement(node);
+    processNode(node, initialIsBlock);
+
+    if (collapseSpaces) {
+        result = result.replace(/[ \t]+/g, ' ');
+        result = result.replace(/\n{3,}/g, '\n\n');
+    }
+
+    if (trimResult) {
+        result = result.trim();
+    }
+
+    return result;
+}