Normalize standalone CustomContent indentation in JA output

.github/workflows/test-ja-translation.yml

@@ -33,19 +33,19 @@ jobs:
       - name: Configure translation credentials
         shell: bash
         env:
-          GCP_KEY: ${{ secrets.GCP_KEY }}
+          GCP_CREDENTIALS_B64: ${{ secrets.GCP_CREDENTIALS_B64 }}
           GCP_PROJECT_ID: ${{ secrets.GCP_PROJECT_ID }}
           GCP_GLOSSARY_ID: ${{ secrets.GCP_GLOSSARY_ID }}
         run: |
           set -euo pipefail
 
-          if [ -z "${GCP_KEY}" ] || [ -z "${GCP_PROJECT_ID}" ] || [ -z "${GCP_GLOSSARY_ID}" ]; then
-            echo "Missing one or more required secrets: GCP_KEY, GCP_PROJECT_ID, GCP_GLOSSARY_ID" >&2
+          if [ -z "${GCP_CREDENTIALS_B64}" ] || [ -z "${GCP_PROJECT_ID}" ] || [ -z "${GCP_GLOSSARY_ID}" ]; then
+            echo "Missing one or more required secrets: GCP_CREDENTIALS_B64, GCP_PROJECT_ID, GCP_GLOSSARY_ID" >&2
             exit 1
           fi
 
           CREDENTIALS_FILE="$RUNNER_TEMP/gcp-key.json"
-          echo "${GCP_KEY}" | base64 --decode > "${CREDENTIALS_FILE}"
+          printf '%s' "${GCP_CREDENTIALS_B64}" | base64 --decode > "${CREDENTIALS_FILE}"
 
           echo "GOOGLE_APPLICATION_CREDENTIALS=${CREDENTIALS_FILE}" >> "$GITHUB_ENV"
           echo "PROJECT_ID=${GCP_PROJECT_ID}" >> "$GITHUB_ENV"

.gitignore

@@ -0,0 +1 @@
+.DS_Store

markdown-translator/src/customContentIndentation.js

@@ -0,0 +1,118 @@
+const CUSTOM_CONTENT_OPEN_RE = /^(\s*)<CustomContent\b[^>]*>\s*$/;
+const CUSTOM_CONTENT_CLOSE_RE = /^(\s*)<\/CustomContent>\s*$/;
+const CUSTOM_CONTENT_LINE_RE = /^(\s*)<\/?CustomContent\b[^>]*>\s*$/;
+const FENCE_RE = /^(\s*)(`{3,}|~{3,})/;
+
+const getLeadingSpaces = (line = "") => line.match(/^ */)?.[0].length ?? 0;
+
+const isStandaloneCustomContentLine = (line = "") =>
+  CUSTOM_CONTENT_LINE_RE.test(line);
+
+const getFenceMask = (lines) => {
+  const mask = Array(lines.length).fill(false);
+  let activeFence = null;
+
+  lines.forEach((line, index) => {
+    const fenceMatch = line.match(FENCE_RE);
+    if (!activeFence) {
+      if (fenceMatch) {
+        activeFence = {
+          marker: fenceMatch[2][0],
+          length: fenceMatch[2].length,
+        };
+        mask[index] = true;
+      }
+      return;
+    }
+
+    mask[index] = true;
+    if (
+      fenceMatch &&
+      fenceMatch[2][0] === activeFence.marker &&
+      fenceMatch[2].length >= activeFence.length
+    ) {
+      activeFence = null;
+    }
+  });
+
+  return mask;
+};
+
+const findContextIndent = (lines, fenceMask, openIndex, currentIndent) => {
+  for (let index = openIndex - 1; index >= 0; index--) {
+    if (fenceMask[index]) {
+      continue;
+    }
+
+    const line = lines[index];
+    if (!line.trim() || isStandaloneCustomContentLine(line)) {
+      continue;
+    }
+
+    if (getLeadingSpaces(line) > currentIndent) {
+      continue;
+    }
+
+    return getLeadingSpaces(line);
+  }
+
+  return null;
+};
+
+const findInnerContentIndent = (lines, fenceMask, openIndex, closeIndex) => {
+  for (let index = openIndex + 1; index < closeIndex; index++) {
+    if (fenceMask[index]) {
+      continue;
+    }
+
+    const line = lines[index];
+    if (!line.trim() || isStandaloneCustomContentLine(line)) {
+      continue;
+    }
+
+    return getLeadingSpaces(line);
+  }
+
+  return null;
+};
+
+export const normalizeStandaloneCustomContentIndentation = (markdown = "") => {
+  const lines = markdown.split("\n");
+  const fenceMask = getFenceMask(lines);
+  const stack = [];
+
+  lines.forEach((line, index) => {
+    if (fenceMask[index]) {
+      return;
+    }
+
+    if (CUSTOM_CONTENT_OPEN_RE.test(line)) {
+      stack.push({
+        openIndex: index,
+        targetIndent: findContextIndent(
+          lines,
+          fenceMask,
+          index,
+          getLeadingSpaces(line)
+        ),
+      });
+      return;
+    }
+
+    if (!CUSTOM_CONTENT_CLOSE_RE.test(line) || !stack.length) {
+      return;
+    }
+
+    const { openIndex, targetIndent } = stack.pop();
+    const normalizedIndent =
+      targetIndent ??
+      findInnerContentIndent(lines, fenceMask, openIndex, index) ??
+      getLeadingSpaces(lines[openIndex]);
+    lines[openIndex] = `${" ".repeat(normalizedIndent)}${lines[
+      openIndex
+    ].trimStart()}`;
+    lines[index] = `${" ".repeat(normalizedIndent)}${lines[index].trimStart()}`;
+  });
+
+  return lines.join("\n");
+};

markdown-translator/src/gcpTranslator.js

@@ -21,6 +21,7 @@ import { visit } from "unist-util-visit";
 // } from "remark-comment-o";
 
 import { getMdFileList, writeFileSync, handleAstNode } from "./lib.js";
+import { normalizeStandaloneCustomContentIndentation } from "./customContentIndentation.js";
 
 const pSum = {
   sum: 0,
@@ -74,7 +75,9 @@ export const gcpTranslator = async (filePath, outputFilePath) => {
       gfmToMarkdown(),
     ],
   });
-  const result = newFile.replaceAll(/(#+.+)(\\{)(#.+})/g, `$1{$3`);
+  const result = normalizeStandaloneCustomContentIndentation(
+    newFile.replaceAll(/(#+.+)(\\{)(#.+})/g, `$1{$3`)
+  );
   writeFileSync(outputFilePath, result);
 };
 

markdown-translator/src/index_ja.js

@@ -24,6 +24,7 @@ import { getMdFileList, writeFileSync, handleAstNode } from "./lib.js";
 import { loadVariables, variablesReplace } from "./variables.js";
 import { postProcessFileFrontmatterAliases } from "./frontmatterAliases.js";
 import { postProcessFileGithubMentions } from "./GithubMentions.js";
+import { normalizeStandaloneCustomContentIndentation } from "./customContentIndentation.js";
 
 const pSum = {
   sum: 0,
@@ -77,7 +78,9 @@ const translateSingleMdToJa = async (filePath, outputFilePath) => {
       gfmToMarkdown(),
     ],
   });
-  const result = newFile.replaceAll(/(#+.+)(\\{)(#.+})/g, `$1{$3`);
+  const result = normalizeStandaloneCustomContentIndentation(
+    newFile.replaceAll(/(#+.+)(\\{)(#.+})/g, `$1{$3`)
+  );
   writeFileSync(outputFilePath, result);
 };
 

markdown-translator/test/customContentIndentation.test.js

@@ -0,0 +1,118 @@
+import test from "node:test";
+import assert from "node:assert/strict";
+
+import { normalizeStandaloneCustomContentIndentation } from "../src/customContentIndentation.js";
+
+test("normalizes standalone CustomContent indentation in ordered-list export sections", () => {
+  const input = [
+    "    -   export intro",
+    "",
+    "        3.  before",
+    "",
+    "     <CustomContent plan=\"starter,essential\">",
+    "",
+    "    -   export bullet",
+    "",
+    "        1.  nested step",
+    "",
+    "     </CustomContent>",
+    "",
+    "     <CustomContent plan=\"premium\">",
+    "",
+    "    -   another export bullet",
+    "",
+    "     </CustomContent>",
+  ].join("\n");
+
+  const expected = [
+    "    -   export intro",
+    "",
+    "        3.  before",
+    "",
+    "    <CustomContent plan=\"starter,essential\">",
+    "",
+    "    -   export bullet",
+    "",
+    "        1.  nested step",
+    "",
+    "    </CustomContent>",
+    "",
+    "    <CustomContent plan=\"premium\">",
+    "",
+    "    -   another export bullet",
+    "",
+    "    </CustomContent>",
+  ].join("\n");
+
+  assert.equal(normalizeStandaloneCustomContentIndentation(input), expected);
+});
+
+test("normalizes standalone CustomContent indentation in unordered-list descriptions", () => {
+  const input = [
+    "-   description",
+    "",
+    "    extra detail",
+    "",
+    "      <CustomContent platform=\"tidb\">",
+    "",
+    "    tidb-only note",
+    "",
+    "      </CustomContent>",
+    "",
+    "      <CustomContent platform=\"tidb-cloud\">",
+    "",
+    "    cloud-only note",
+    "",
+    "      </CustomContent>",
+  ].join("\n");
+
+  const expected = [
+    "-   description",
+    "",
+    "    extra detail",
+    "",
+    "    <CustomContent platform=\"tidb\">",
+    "",
+    "    tidb-only note",
+    "",
+    "    </CustomContent>",
+    "",
+    "    <CustomContent platform=\"tidb-cloud\">",
+    "",
+    "    cloud-only note",
+    "",
+    "    </CustomContent>",
+  ].join("\n");
+
+  assert.equal(normalizeStandaloneCustomContentIndentation(input), expected);
+});
+
+test("does not rewrite inline CustomContent or fenced code examples", () => {
+  const input = [
+    "Inline <CustomContent plan=\"ja\">value</CustomContent> stays inline.",
+    "",
+    "```md",
+    "      <CustomContent plan=\"starter,essential\">",
+    "    literal example",
+    "      </CustomContent>",
+    "```",
+  ].join("\n");
+
+  assert.equal(normalizeStandaloneCustomContentIndentation(input), input);
+});
+
+test("keeps already-correct nested CustomContent indentation unchanged", () => {
+  const input = [
+    "    -   parent bullet",
+    "",
+    "        1.  nested step",
+    "",
+    "        <CustomContent plan=\"starter,essential\">",
+    "",
+    "        already aligned",
+    "",
+    "        </CustomContent>",
+  ].join("\n");
+
+  assert.equal(normalizeStandaloneCustomContentIndentation(input), input);
+});