Merge pull request #34 from HackYourFuture-CPH/validate-links-take-2

Check that local image/link targets are valid

Author: adamblanchard

.prettierignore

@@ -0,0 +1 @@
+/dist/

.tool-versions

@@ -0,0 +1 @@
+nodejs 23.11.0

dist/parse.js

@@ -0,0 +1,35 @@
+import mit from "markdown-it";
+export const parse = (content) => {
+    const parser = mit();
+    const tokens = parser.parse(content, {});
+    const parsedLinks = [];
+    const parsedImages = [];
+    const scan = (tokens) => {
+        tokens.forEach((token, index) => {
+            if (token.type === "link_open") {
+                const indexOfNextClose = tokens.findIndex((t2, i2) => i2 > index && t2.type === "link_close");
+                if (indexOfNextClose > index) {
+                    parsedLinks.push({
+                        target: token.attrGet("href"),
+                        content: tokens
+                            .slice(index + 1, indexOfNextClose)
+                            .map((t) => t.content)
+                            .join(""),
+                    });
+                }
+            }
+            if (token.type === "image")
+                parsedImages.push({
+                    src: token.attrGet("src"),
+                    alt: token.content,
+                });
+            if (token.children)
+                scan(token.children);
+        });
+    };
+    scan(tokens);
+    return {
+        links: parsedLinks,
+        images: parsedImages,
+    };
+};

dist/validateLinks.js

@@ -0,0 +1,80 @@
+import { exec } from "node:child_process";
+import { readFile } from "node:fs/promises";
+import { parse } from "./parse.js";
+import path, { dirname, normalize } from "node:path/posix";
+import { isAbsolute } from "node:path";
+const findAllFilesInGit = async () => {
+    return await new Promise((resolve, reject) => {
+        exec("git ls-files -z", (error, stdout, stderr) => {
+            if (error)
+                reject(error);
+            if (stderr)
+                reject(new Error(`git ls-files outputted on stderr: ${stderr}`));
+            else
+                resolve(stdout.split("\0").filter(Boolean));
+        });
+    });
+};
+const findMarkdownFiles = (files) => {
+    const ignorePattern = /^(README|LICENSE|contributing\/)/;
+    return files.filter((f) => f.toLocaleLowerCase().endsWith(".md") && !ignorePattern.test(f));
+};
+const scanForLinks = async (filenames) => {
+    return Promise.all(filenames.map(async (filename) => {
+        const content = await readFile(filename, "utf-8");
+        return { filename, ...parse(content) };
+    }));
+};
+const externalLinkPattern = /^\w+:/;
+const isExternalLink = (t) => externalLinkPattern.test(t);
+const main = async () => {
+    const gitFiles = await findAllFilesInGit();
+    // For now, we assume that there are no case clashes
+    const lowercaseGitFiles = gitFiles.map((s) => s.toLocaleLowerCase());
+    const markdownFilenames = findMarkdownFiles(gitFiles);
+    const parsedFiles = await scanForLinks(markdownFilenames);
+    let errors = 0;
+    for (const parsedFile of parsedFiles) {
+        for (const img of parsedFile.images) {
+            if (!isExternalLink(img.src)) {
+                const resolved = path.join(dirname(parsedFile.filename), img.src);
+                const exists = lowercaseGitFiles.includes(resolved.toLocaleLowerCase());
+                if (!exists) {
+                    console.log(`error BROKEN-INTERNAL-IMAGE ${parsedFile.filename}:0 Broken internal image reference ${img.src}`);
+                    ++errors;
+                }
+            }
+        }
+        for (const link of parsedFile.links) {
+            if (link.target.startsWith("#")) {
+                // Already checked by the linter
+                continue;
+            }
+            if (!isExternalLink(link.target)) {
+                const target = link.target.split("#")[0];
+                let resolved;
+                if (isAbsolute(target)) {
+                    resolved = normalize(`./${target}`);
+                }
+                else {
+                    resolved = normalize(path.join(dirname(parsedFile.filename), target));
+                }
+                const isFile = lowercaseGitFiles.includes(resolved.toLocaleLowerCase());
+                const resolvedWithTrailingSlash = resolved.endsWith("/")
+                    ? resolved.toLocaleLowerCase()
+                    : `${resolved.toLocaleLowerCase()}/`;
+                const isDirectory = lowercaseGitFiles.some((s) => s.startsWith(resolvedWithTrailingSlash));
+                if (!isFile && !isDirectory) {
+                    console.log(`error BROKEN-INTERNAL-LINK ${parsedFile.filename}:0 Link target does not exist: ${target}`);
+                    ++errors;
+                }
+            }
+        }
+    }
+    if (errors > 0)
+        process.exit(1);
+};
+main().catch((error) => {
+    console.error(error);
+    process.exit(1);
+});

lint

@@ -12,6 +12,31 @@ else
     fi
   ' 0
 
-  npm exec -- markdownlint --ignore 'node_modules/' '**/*.md'
-  npm exec -- prettier --check .
+  rc=0
+
+  echo "validateLinks: ..."
+  if npm exec -- node dist/validateLinks.js ; then
+    echo validateLinks: PASS
+  else
+    echo validateLinks: FAIL
+    rc=1
+  fi
+
+  echo "markdownlint: ..."
+  if npm exec -- markdownlint --ignore 'node_modules/' '**/*.md' ; then
+    echo "markdownlint: PASS"
+  else
+    echo "markdownlint: FAIL"
+    rc=1
+  fi
+
+  echo "prettier: ..."
+  if npm exec -- prettier --check . ; then
+    echo "prettier: PASS"
+  else
+    echo "prettier: FAIL"
+    rc=1
+  fi
+
+  exit $rc
 fi

package-lock.json

@@ -2,5 +2,11 @@
   "dependencies": {
     "markdownlint-cli": "^0.44.0",
     "prettier": "^3.5.3"
+  },
+  "type": "module",
+  "devDependencies": {
+    "@types/markdown-it": "^14.1.2",
+    "@types/node": "^22.14.1",
+    "typescript": "^5.8.3"
   }
 }

package.json

@@ -0,0 +1,60 @@
+import mit from "markdown-it";
+import type { Token } from "markdown-it/index.js";
+
+export type ParsedLink = {
+  readonly target: string;
+  readonly content: string;
+};
+
+export type ParsedImage = {
+  readonly src: string;
+  readonly alt: string;
+};
+
+export type ParseResult = {
+  readonly links: readonly ParsedLink[];
+  readonly images: readonly ParsedImage[];
+};
+
+export const parse = (content: string): ParseResult => {
+  const parser = mit();
+  const tokens = parser.parse(content, {});
+
+  const parsedLinks: ParsedLink[] = [];
+  const parsedImages: ParsedImage[] = [];
+
+  const scan = (tokens: Token[]) => {
+    tokens.forEach((token, index) => {
+      if (token.type === "link_open") {
+        const indexOfNextClose = tokens.findIndex(
+          (t2, i2) => i2 > index && t2.type === "link_close",
+        );
+
+        if (indexOfNextClose > index) {
+          parsedLinks.push({
+            target: token.attrGet("href") as string,
+            content: tokens
+              .slice(index + 1, indexOfNextClose)
+              .map((t) => t.content)
+              .join(""),
+          });
+        }
+      }
+
+      if (token.type === "image")
+        parsedImages.push({
+          src: token.attrGet("src") as string,
+          alt: token.content,
+        });
+
+      if (token.children) scan(token.children);
+    });
+  };
+
+  scan(tokens);
+
+  return {
+    links: parsedLinks,
+    images: parsedImages,
+  };
+};

parse.ts

@@ -0,0 +1,12 @@
+{
+  "compilerOptions": {
+    "target": "ESNext",
+    "module": "NodeNext",
+    "moduleResolution": "nodenext",
+    "outDir": "./dist/",
+    "esModuleInterop": true,
+    "forceConsistentCasingInFileNames": true,
+    "strict": true,
+    "skipLibCheck": true
+  }
+}