Script to Check the Table of Contents in the Wiki and CI workflow

.github/workflows/wiki.yml

@@ -10,6 +10,24 @@ on:
   gollum:
 
 jobs:
+  table_of_contents_check:
+    name: Wiki Check Table of Contents match Headers
+    runs-on: ubuntu-20.04
+    env:
+      CACHE_DIRECTORY: ~/.bazel_cache
+    steps:
+      - uses: actions/checkout@v2
+
+      - name: Set up Bazel
+        uses: abhinavsingh/setup-bazel@v3
+        with:
+          version: 4.0.0
+
+      - name: Wiki Check Table of Contents
+        id: wikichecktoc
+        run: |
+          bazel run //scripts:wiki_table_of_contents_check -- ${GITHUB_WORKSPACE}
+
   wiki-deploy:
     runs-on: ${{ matrix.os }}
     strategy:

scripts/BUILD.bazel

@@ -227,6 +227,13 @@ kt_jvm_binary(
     ],
 )
 
+kt_jvm_binary(
+    name = "wiki_table_of_contents_check",
+    testonly = True,
+    main_class = "org.oppia.android.scripts.wiki.WikiTableOfContentsCheckKt",
+    runtime_deps = ["//scripts/src/java/org/oppia/android/scripts/wiki:wiki_table_of_contents_check_lib"],
+)
+
 # Note that this is intentionally not test-only since it's used by the app build pipeline. Also,
 # this apparently needs to be a java_binary to set up runfiles correctly when executed within a
 # Starlark rule as a tool.

scripts/src/java/org/oppia/android/scripts/wiki/BUILD.bazel

@@ -0,0 +1,18 @@
+"""
+Libraries corresponding to scripting tools that help with continuous integration workflows.
+"""
+
+load("@io_bazel_rules_kotlin//kotlin:kotlin.bzl", "kt_jvm_library")
+
+kt_jvm_library(
+    name = "wiki_table_of_contents_check_lib",
+    testonly = True,
+    srcs = [
+        "WikiTableOfContentsCheck.kt",
+    ],
+    visibility = ["//scripts:oppia_script_binary_visibility"],
+    deps = [
+        "//scripts/src/java/org/oppia/android/scripts/common:bazel_client",
+        "//scripts/src/java/org/oppia/android/scripts/common:git_client",
+    ],
+)

scripts/src/java/org/oppia/android/scripts/wiki/WikiTableOfContentsCheck.kt

@@ -0,0 +1,93 @@
+package org.oppia.android.scripts.wiki
+
+import java.io.File
+
+/**
+ * Script for ensuring that the table of contents in each wiki page matches with its respective headers.
+ *
+ * Usage:
+ *   bazel run //scripts:wiki_sample -- <path_to_default_working_directory>
+ *
+ * Arguments:
+ * - path_to_default_working_directory: The default working directory on the runner for steps, and the default location of repository.
+ *
+ * Example:
+ *   bazel run //scripts:wiki_sample -- ${GITHUB_WORKSPACE}
+ */
+fun main(vararg args: String) {
+  // Path to the repo's wiki.
+  val githubWorkspace = "${args[0]}/wiki/"
+  val wikiDirectory = File(githubWorkspace)
+
+  // Check if the wiki directory exists
+  if (wikiDirectory.exists() && wikiDirectory.isDirectory) {
+    processWikiDirectory(wikiDirectory)
+  } else {
+    println("No contents found in the Wiki directory.")
+  }
+}
+
+/**
+ * Checks every file in the wiki repo
+ *
+ * @param wikiDirectory the default working directory
+ */
+fun processWikiDirectory(wikiDirectory: File) {
+  wikiDirectory.listFiles()?.forEach { file ->
+    processWikiFile(file)
+  }
+}
+
+/**
+ * Processes the contents of a single wiki file to ensure the accuracy of the Table of Contents.
+ *
+ * @param file The wiki file to process.
+ */
+fun processWikiFile(file: File) {
+  var inTableOfContents = false
+  var skipBlankLine = false
+
+  file.forEachLine { line ->
+    when {
+      // Checking for Table of Contents section
+      line.trim() == "## Table of Contents" -> {
+        inTableOfContents = true
+        skipBlankLine = true
+      }
+      // Checking to skip the blank line immediately after the ## Table of Contents
+      skipBlankLine && line.isBlank() -> skipBlankLine = false
+      // Validating the contents in the Table of Content
+      inTableOfContents && line.trimStart().startsWith("- [") && !line.contains("https://") -> {
+        validateTableOfContents(file, line)
+      }
+      // Checking for end of Table of Contents section
+      inTableOfContents && line.isBlank() -> inTableOfContents = false
+    }
+  }
+}
+
+/**
+ * Validates the accuracy of a Table of Contents entry in a wiki file.
+ *
+ * @param file The wiki file being validated.
+ * @param line The line containing the Table of Contents entry.
+ */
+fun validateTableOfContents(file: File, line: String) {
+  val titleRegex = "\\[(.*?)\\]".toRegex()
+  val title = titleRegex.find(line)?.groupValues?.get(1)?.replace('-', ' ')
+    ?.replace(Regex("[?&./:’'*!,(){}\\[\\]+]"), "")
+    ?.trim()
+
+  val linkRegex = "\\(#(.*?)\\)".toRegex()
+  val link = linkRegex.find(line)?.groupValues?.get(1)?.removePrefix("#")?.replace('-', ' ')
+    ?.replace(Regex("[?&./:’'*!,(){}\\[\\]+]"), "")
+    ?.replace("confetti_ball", "")?.trim()
+
+  // Checks if the table of content title matches with the header link text
+  val matches = title.equals(link, ignoreCase = true)
+  if (!matches) {
+    throw Exception("\nMismatch of Table of Content with headers in the File: ${file.name}. " +
+      "\nThe Title: '${titleRegex.find(line)?.groupValues?.get(1)}' " +
+      "doesn't match with its corresponding Link: '${linkRegex.find(line)?.groupValues?.get(1)}'.")
+  }
+}