Add anchor and image validation checks

Author: kylebuss

.editorconfig

@@ -0,0 +1,13 @@
+# Editor configuration, see http://editorconfig.org
+root = true
+
+[*]
+charset = utf-8
+indent_style = space
+indent_size = 2
+insert_final_newline = true
+trim_trailing_whitespace = true
+
+[*.md]
+max_line_length = off
+trim_trailing_whitespace = false

locales.ts

@@ -0,0 +1,27 @@
+export interface HelpLocales {
+  docusaurus: string;
+  crowdin: string;
+}
+
+export const helpLocales = [
+  {
+    docusaurus: "de",
+    crowdin: "de",
+  },
+  {
+    docusaurus: "es",
+    crowdin: "es-ES",
+  },
+  {
+    docusaurus: "fr",
+    crowdin: "fr",
+  },
+  {
+    docusaurus: "pt-BR",
+    crowdin: "pt-BR",
+  },
+  {
+    docusaurus: "id",
+    crowdin: "id",
+  },
+] as HelpLocales[];

regex_validation.ts

@@ -0,0 +1,22 @@
+export interface RegexValidation {
+  expression: RegExp;
+  message: string;
+  filesToSkip?: string[];
+}
+
+export const regexValidations: RegexValidation[] = [
+  {
+    expression: /^---\n(.*?)\n---/s,
+    message: "front matter"
+  },
+  {
+    expression: /(?:(?=)({#[\d\w]+}))/gs,
+    message: "anchor",
+    filesToSkip: ["Community-Checking/enable-community-checking.md", "Draft-Generation/generating-a-draft.md"]
+  },
+  {
+    expression: /(?:(?=)(!\[\]\(\.\/[\d]+\.png\)))/gs,
+    message: "image",
+    filesToSkip: ["connect-paratext-project.md"] //["Draft-Generation/understanding-drafts.md"]
+  }
+]

update.mts

@@ -8,7 +8,8 @@
 // This doesn't fully solve the problem though, because metadata in the translated files may be different from the English files.
 
 import { copy, readerFromStreamReader } from "jsr:@std/io";
-import { walk } from "jsr:@std/fs/walk";
+import { helpLocales } from "./locales.ts";
+import { runChecks } from "./validate_docs.mts";
 
 const projectId = Deno.env.get("CROWDIN_PROJECT_ID");
 const apiKey = Deno.env.get("CROWDIN_API_KEY");
@@ -112,29 +113,6 @@ async function saveLatestBuild() {
   }
 }
 
-const helpLocales = [
-  {
-    docusaurus: "de",
-    crowdin: "de",
-  },
-  {
-    docusaurus: "es",
-    crowdin: "es-ES",
-  },
-  {
-    docusaurus: "fr",
-    crowdin: "fr",
-  },
-  {
-    docusaurus: "pt-BR",
-    crowdin: "pt-BR",
-  },
-  {
-    docusaurus: "id",
-    crowdin: "id",
-  },
-] as const;
-
 async function copyDirContentsRecursive(
   source: string,
   dest: string
@@ -216,129 +194,6 @@ async function fetchNotionDocs() {
   }
 }
 
-async function runChecks() {
-  // Check for malformed links in the i18n files
-
-  for await (const dirEntry of walk(`${projectRoot}/i18n`)) {
-    if (dirEntry.isFile && dirEntry.name.endsWith(".md")) {
-      const file = await Deno.readTextFile(dirEntry.path);
-      for (const [index, line] of file.split("\n").entries()) {
-        // Look for Markdown links that had a space added between the closing parenthesis and the opening bracket
-        if (/\]\s+\(/.test(line)) {
-          console.log(
-            `%cFound malformed link in ${dirEntry.path} at line ${index + 1}`,
-            "color: red"
-          );
-          console.log(line);
-          console.log(
-            " ".repeat(line.indexOf("]") + 1) + "%c^ Erroneous space",
-            "color: blue"
-          );
-        }
-      }
-    }
-  }
-
-  // Check that all original docs have a corresponding translation, and vice versa
-  const originalDocs = new Map<string, string>();
-  const docsDir = `${projectRoot}/docs`;
-  for await (const dirEntry of walk(docsDir)) {
-    if (dirEntry.isFile && dirEntry.name.endsWith(".md")) {
-      if (dirEntry.path.indexOf(docsDir) !== 0)
-        throw new Error("Unexpected path");
-      const relativePath = dirEntry.path.slice(docsDir.length + 1);
-      originalDocs.set(relativePath, await Deno.readTextFile(dirEntry.path));
-    }
-  }
-
-  for (const locale of helpLocales) {
-    const i18nDir = `${projectRoot}/i18n/${locale.docusaurus}/docusaurus-plugin-content-docs/current`;
-
-    const foundLocalizations = new Set<string>();
-
-    for await (const dirEntry of walk(i18nDir)) {
-      if (dirEntry.isFile && dirEntry.name.endsWith(".md")) {
-        if (dirEntry.path.indexOf(i18nDir) !== 0)
-          throw new Error("Unexpected path");
-
-        const relativePath = dirEntry.path.slice(i18nDir.length + 1);
-
-        foundLocalizations.add(relativePath);
-
-        if (!originalDocs.has(relativePath)) {
-          console.log(
-            `%cNo original document found for i18n file ${dirEntry.path}`,
-            "color: red"
-          );
-        }
-      }
-    }
-    for (const [relativePath, content] of originalDocs) {
-      if (!foundLocalizations.has(relativePath)) {
-        console.log(
-          `%cNo translation file found for ${relativePath} in ${i18nDir}`,
-          "color: red"
-        );
-      } else {
-        // Check that the front matter matches
-        const translation = await Deno.readTextFile(
-          `${i18nDir}/${relativePath}`
-        );
-        const frontMatterRegex = /^---\n(.*?)\n---/s;
-        const originalFrontMatterMatch = content.match(frontMatterRegex);
-        const translationFrontMatterMatch = translation.match(frontMatterRegex);
-        if (originalFrontMatterMatch == null) {
-          console.log(
-            `%cNo front matter found in original document ${relativePath}`,
-            "color: red"
-          );
-        }
-        if (translationFrontMatterMatch == null) {
-          console.log(
-            `%cNo front matter found in translation ${relativePath}`,
-            "color: red"
-          );
-        }
-
-        if (
-          originalFrontMatterMatch == null ||
-          translationFrontMatterMatch == null
-        )
-          continue;
-
-        const originalFrontMatter = originalFrontMatterMatch[1].split("\n");
-        const translationFrontMatter =
-          translationFrontMatterMatch[1].split("\n");
-
-        if (originalFrontMatter.length !== translationFrontMatter.length) {
-          console.log(
-            `%cFront matter length mismatch in ${relativePath} for locale ${locale.docusaurus}`,
-            "color: red"
-          );
-        }
-
-        for (const [index, line] of originalFrontMatter.entries()) {
-          // the title can change; everything else should match
-          if (line.indexOf("title: ") === 0) continue;
-          if (line !== translationFrontMatter[index]) {
-            console.log(
-              `%cFront matter mismatch in ${relativePath} for locale ${
-                locale.docusaurus
-              } at line ${index + 1}`,
-              "color: red"
-            );
-            console.log(`%cOriginal:    ${line}`, "color: blue");
-            console.log(
-              `%cTranslation: ${translationFrontMatter[index]}`,
-              "color: blue"
-            );
-          }
-        }
-      }
-    }
-  }
-}
-
 try {
   console.log("--- Deleting existing files ---");
   await deleteExistingFiles();