feat(gatsby-source-wordpress): Add searchAndReplace

packages/gatsby-source-wordpress/__tests__/process-node.test.js

@@ -4,6 +4,7 @@ import {
   getImgSrcRemoteFileMatchesFromNodeString,
   getImgTagMatchesWithUrl,
   getWpLinkRegex,
+  searchAndReplaceNodeStrings,
 } from "../dist/steps/source-nodes/create-nodes/process-node"
 
 const wpUrl = `wp.fakesite.com`
@@ -43,3 +44,28 @@ test(`HTML link transformation regex matches links`, async () => {
 
   expect(matches.length).toBe(2)
 })
+
+test(`Search and replace node strings using regex matches`, async () => {
+  const nodeString = `Some stuff in a random string
+
+  A new line with some stuff!
+
+  We need to test some <a href=\\"https://old-site.com/hi\\" />link</a> as well!`
+
+  const result = searchAndReplaceNodeStrings({
+    nodeString,
+    node: { __typename: "FakeTypeName", id: "cG9zdDo0OQ==" },
+    pluginOptions: {
+      searchAndReplace: [
+        { search: "(S|s)ome stuff", replace: "some other thing" },
+        { search: "https://old-site\.com", replace: "https://new-site.com" },
+      ]
+    }
+  })
+
+  expect(result).toBe(`some other thing in a random string
+
+  A new line with some other thing!
+
+  We need to test some <a href=\\"https://new-site.com/hi\\" />link</a> as well!`)
+})

packages/gatsby-source-wordpress/docs/plugin-options.md

@@ -38,6 +38,9 @@
   - [schema.requestConcurrency](#schemarequestconcurrency)
   - [schema.previewRequestConcurrency](#schemapreviewrequestconcurrency)
 - [excludeFieldNames](#excludefieldnames)
+- [searchAndReplace](#searchandreplace)
+  - [searchAndReplace[].search](#searchandreplacesearch)
+  - [searchAndReplace[].replace](#searchandreplacereplace)
 - [html](#html)
   - [html.useGatsbyImage](#htmlusegatsbyimage)
   - [html.imageMaxWidth](#htmlimagemaxwidth)
@@ -778,6 +781,67 @@ A list of field names to globally exclude from the ingested schema.
 
 ```
 
+## searchAndReplace
+
+An array of options to search and replace strings in nodes. See below for options.
+
+**Field type**: `Array`
+
+```js
+{
+  resolve: `gatsby-source-wordpress`,
+  options: {
+    searchAndReplace: [
+      {
+        search: "https://some-url.com",
+        replace: "https://some-new-url.com",
+      },
+    ],
+  },
+}
+
+```
+
+### searchAndReplace[].search
+
+The regex rule used to search a terme.
+
+**Field type**: `String`
+
+```js
+{
+  resolve: `gatsby-source-wordpress`,
+  options: {
+    searchAndReplace: [
+      {
+        search: "https://some-url.com",
+      },
+    ],
+  },
+}
+
+```
+
+### searchAndReplace[].replace
+
+The replacement string for each regex match.
+
+**Field type**: `String`
+
+```js
+{
+  resolve: `gatsby-source-wordpress`,
+  options: {
+    searchAndReplace: [
+      {
+        replace: "https://some-new-url.com",
+      },
+    ],
+  },
+}
+
+```
+
 ## html
 
 Options related to html field processing.

packages/gatsby-source-wordpress/src/steps/declare-plugin-options-schema.ts

@@ -530,6 +530,45 @@ When using this option, be sure to gitignore the wordpress-cache directory in th
             excludeFieldNames: [\`viewer\`],
           `),
       }),
+    searchAndReplace: Joi.array()
+      .description(`An array of options to search and replace strings in nodes. See below for options.`)
+      .allow(null)
+      .items(
+        Joi.object({
+          search: Joi.string()
+            .description(`The regex rule used to search a terme.`)
+            .meta({
+              example: wrapOptions(`
+                searchAndReplace: [
+                  {
+                    search: "https://some-url\.com"
+                  },
+                ],
+              `),
+            }),
+          replace: Joi.string()
+            .description(`The replacement string for each regex match.`)
+            .meta({
+              example: wrapOptions(`
+                searchAndReplace: [
+                  {
+                    replace: "https://some-new-url.com",
+                  },
+                ],
+              `),
+            })
+        })
+      )
+      .meta({
+        example: wrapOptions(`
+          searchAndReplace: [
+            {
+              search:  "https://some-url\.com",
+              replace: "https://some-new-url.com",
+            },
+          ],
+        `),
+      }),
     html: Joi.object({
       useGatsbyImage: Joi.boolean()
         .default(true)

packages/gatsby-source-wordpress/src/steps/source-nodes/create-nodes/process-node.js

@@ -850,6 +850,36 @@ const replaceNodeHtmlLinks = ({ wpUrl, nodeString, node }) => {
   return nodeString
 }
 
+// replaces specific string or regex with a given string from the plugin options config
+const searchAndReplaceNodeStrings = ({ nodeString, node, pluginOptions }) => {
+  if (Array.isArray(pluginOptions?.searchAndReplace)) {
+    pluginOptions.searchAndReplace.forEach(({ search, replace }) => {
+      const searchRegex = new RegExp(search, `g`)
+
+      const stringMatches = execall(searchRegex, nodeString)
+
+      if (stringMatches.length) {
+        stringMatches.forEach(({ match }) => {
+          if (match) {
+            try {
+              nodeString = nodeString.replace(search, replace)
+            } catch (e) {
+              console.error(e)
+              console.warn(
+                formatLogMessage(
+                  `Failed to process search and replace string in ${node.__typename} ${node.id}`
+                )
+              )
+            }
+          }
+        })
+      }
+    })
+  }
+
+  return nodeString
+}
+
 const processNodeString = async ({
   nodeString,
   node,
@@ -861,6 +891,7 @@ const processNodeString = async ({
     replaceNodeHtmlImages,
     replaceFileLinks,
     replaceNodeHtmlLinks,
+    searchAndReplaceNodeStrings,
   ]
 
   for (const nodeStringFilter of nodeStringFilters) {