api utilities

wp-api-etl/cli/ucdlib-wp-find.js

@@ -0,0 +1,38 @@
+import { Command, Option } from 'commander';
+import {JSDOM} from "jsdom";
+
+import Config from "../lib/config.js";
+import WpApi from "../lib/wp-api.js";
+
+const program = new Command();
+
+program
+  .command('post-by-element')
+  .description('Find posts that contain a specified DOM element')
+  .argument('<query>', 'Args to pass to QuerySelector')
+  .option('-t, --type <type...>', 'Post type to search', ['page'])
+  .action(async (query, options) => {
+    const foundPosts = [];
+    const config = new Config();
+    const api = new WpApi(config);
+
+    for (const postType of options.type) {
+      const posts = await api.getPostsByTypeAll(postType);
+      for (const post of posts) {
+        const dom = new JSDOM(post.content.rendered);
+        const elements = dom.window.document.querySelectorAll(query);
+        if (elements.length > 0) {
+          foundPosts.push(post);
+        }
+      }
+    }
+
+    console.log(`Found ${foundPosts.length} posts.`);
+    for (const post of foundPosts) {
+      console.log(post.link);
+    }
+
+  }
+);
+
+program.parse(process.argv);

wp-api-etl/cli/ucdlib-wp.js

@@ -0,0 +1,10 @@
+#! /usr/bin/env node
+import { Command } from 'commander';
+
+const program = new Command();
+program
+  .name('ucdlib-wp-main')
+  .version('0.0.1')
+  .command('find', 'Query for posts')
+
+program.parse(process.argv);

wp-api-etl/lib/config.js

@@ -1,9 +1,23 @@
 import dotenv from "dotenv";
+import { dirname, join } from 'path';
+import { fileURLToPath } from 'url';
+import { existsSync } from 'fs';
 
+/**
+ * @class Config
+ * @description Configuration class for the WP API ETL. Sets defaults and reads from .env file if present.
+ * @param {String} envPath - Path to the .env file
+ */
 class Config {
   constructor(envPath) {
     if (envPath) {
       dotenv.config({ path: envPath });
+    } else {
+      const __dirname = dirname(fileURLToPath(import.meta.url));
+      const envFile = join(__dirname, '../.env');
+      if (existsSync(envFile)) {
+        dotenv.config({ path: envFile });
+      }
     }
     const env = process.env;
 

wp-api-etl/lib/http-cache.js

@@ -3,6 +3,10 @@ import { fileURLToPath } from 'url';
 import path from 'path';
 import crypto from 'crypto';
 
+/**
+ * @class HTTPCache
+ * @description Simple HTTP cache for the WP API ETL. Stores cached responses in .http-cache directory.
+ */
 class HTTPCache {
 
   constructor() {
@@ -15,6 +19,11 @@ class HTTPCache {
     this.cache = {};
   }
 
+  /**
+   * @description Get a cached response by URL
+   * @param {String} url - URL to retrieve
+   * @returns - Cached response or null if not found
+   */
   get(url) {
     url = this.hash(url);
     if( this.cache[url] ) {
@@ -30,20 +39,38 @@ class HTTPCache {
     return null;
   }
 
+  /**
+   * @description Set a cached response by URL
+   * @param {String} url - URL to set
+   * @param {*} response - Response to cache
+   */
   set(url, response) {
     let hash = this.hash(url);
     this.cache[hash] = {url, response};
     fs.writeFileSync(this.getCacheFile(hash), JSON.stringify(this.cache[hash]));
   }
 
+  /**
+   * @description Get the path to a cache file by hash
+   * @param {String} hash - Hash of the URL
+   * @returns {String}
+   */
   getCacheFile(hash) {
     return path.join(this.localCacheDir, hash+'.json');
   }
 
+  /**
+   * @description Hash a URL
+   * @param {String} url - URL to hash
+   * @returns - Hashed URL
+   */
   hash(url) {
     return crypto.createHash('md5').update(url).digest('hex');
   }
 
+  /**
+   * @description Clear the whole cache
+   */
   clearCache(){
     fs.rmdirSync(this.localCacheDir, { recursive: true });
     fs.mkdirSync(this.localCacheDir);

wp-api-etl/lib/wp-api.js

@@ -3,6 +3,11 @@ import fetch from "node-fetch";
 import Config from "./config.js";
 import cache from '../lib/http-cache.js';
 
+/**
+ * @class WpApi
+ * @description  Wrapper for the WP API
+ * @param {Config} config - Instantiated config object from wp-api-etl/lib/config.js
+ */
 class WpApi {
   constructor(config) {
     if ( config ) {
@@ -43,6 +48,18 @@ class WpApi {
     return type;
   }
 
+  /**
+   * @description Retrieve list of posts of a given type
+   * @param {String} postType - A registered post type - ie 'page', 'exhibit', 'post', etc.
+   * @param {Object} params - Params object with the following properties:
+   * - query: Object - URL Query params to pass to the API
+   * - host: String - Hostname to use for the API (defaults to class config value)
+   * - apiPath: String - Path of the API (defaults to class config value)
+   * - useCache: Boolean - Whether to use the cache (defaults to class config value)
+   * - writeToCache: Boolean - Whether to write to the cache (defaults to class config value)
+   *
+   * @returns {Array}
+   */
   async getPostsByType(postType, params = {}) {
     postType = this.getTypeName(postType);
     const queryParams = this.toSearchParamsObject(params.query);
@@ -69,7 +86,6 @@ class WpApi {
 
     const q = queryParams.toString();
     let url = `${host}${apiPath}/${postType}${q ? '?' + q : ''}`;
-    console.log(`Fetching ${url}`);
 
     if ( useCache ){
       const cached = cache.get(url);
@@ -92,6 +108,17 @@ class WpApi {
 
   }
 
+  /**
+   * @description Retrieve all posts of a given type
+   * @param {String} postType - A registered post type - ie 'page', 'exhibit', 'post', etc.
+   * @param {Object} params - Params object with the following properties:
+   * - query: Object - URL Query params to pass to the API
+   * - host: String - Hostname to use for the API (defaults to class config value)
+   * - apiPath: String - Path of the API (defaults to class config value)
+   * - useCache: Boolean - Whether to use the cache (defaults to class config value)
+   * - writeToCache: Boolean - Whether to write to the cache (defaults to class config value)
+   * @returns {Array}
+   */
   async getPostsByTypeAll(postType, params = {}) {
     const out = [];
     let page = 1;
@@ -107,8 +134,14 @@ class WpApi {
       }
       page++;
     }
+    return out;
   }
 
+  /**
+   * @description Convert an object to a URLSearchParams object and sort it
+   * @param {Object} params
+   * @returns
+   */
   toSearchParamsObject(params) {
     params = new URLSearchParams(params);
     params.sort();

wp-api-etl/lightbox/index.js

@@ -1,7 +1,9 @@
-import Config from "../lib/config.js";
+import {JSDOM} from "jsdom";
 import { dirname, join } from 'path';
-import { fileURLToPath, URLSearchParams } from 'url';
-import { existsSync, mkdirSync } from 'fs';
+import { fileURLToPath } from 'url';
+import { existsSync } from 'fs';
+
+import Config from "../lib/config.js";
 import WpApi from "../lib/wp-api.js";
 
 class LightBoxImageCheck {
@@ -21,15 +23,35 @@ class LightBoxImageCheck {
   }
 
   async run(){
-    this.findPostsWithMissingImageIds();
+    const posts = await this.findPostsWithMissingImageIds();
+    console.log(posts);
   }
 
   async findPostsWithMissingImageIds(postTypes = ['pages', 'exhibit']){
     if ( !Array.isArray(postTypes) ) {
       postTypes = [postTypes];
     }
     const out = {};
-    const p = await this.api.getPostsByTypeAll('exhibits');
+    for( let postType of postTypes ) {
+      const posts = await this.api.getPostsByTypeAll(postType);
+      for( let post of posts ) {
+        let imgCt = 0;
+        const dom = new JSDOM(post.content.rendered);
+        const galleries = dom.window.document.querySelectorAll('.wp-block-gallery');
+        for( let gallery of galleries ) {
+          const images = gallery.querySelectorAll('img');
+          for( let image of images ) {
+            if ( !image.hasAttribute('data-id') ) {
+              imgCt++;
+            }
+          }
+        }
+        if ( imgCt > 0 ) {
+          out[post.id] = {imgCt, post};
+        }
+      }
+    }
+    return out;
   }
 }
 

wp-api-etl/lightbox/readme.md

@@ -1,3 +1,5 @@
 # Image Correction for Lightbox Plugin.
 
 The WP Gallery block has functionality that allows a user to bulk set the "link to" property on all image blocks. However, in some cases, [this does not work on our site](https://github.com/UCDavisLibrary/main-wp-website/issues/85#issuecomment-1535287882). The issue seems to be that when the gallery was originally imported, the images were uploaded, but the `data-id` attribute was not set. 
+
+Luckily, it only affects 5 exhibits, which will just be manually fixed.