From 4a73375cfb49325b30efa647ed41e5e3e147f380 Mon Sep 17 00:00:00 2001
From: MillenniumEarl <MillenniumEarl@users.noreply.github.com>
Date: Mon, 15 Feb 2021 21:28:37 +0100
Subject: [PATCH] Add json-ld.ts

---
 src/scripts/json-ld.ts | 55 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 55 insertions(+)
 create mode 100644 src/scripts/json-ld.ts

diff --git a/src/scripts/json-ld.ts b/src/scripts/json-ld.ts
new file mode 100644
index 0000000..04699c8
--- /dev/null
+++ b/src/scripts/json-ld.ts
@@ -0,0 +1,55 @@
+/**
+ * Represents information contained in a JSON+LD tag.
+ */
+export type JSONLD = { [s: string]: string | JSONLD }
+
+/**
+ * Extracts and processes the JSON-LD values of the page.
+ * @param {cheerio.Cheerio} body Page `body` selector
+ * @returns {JSONLD[]} List of data obtained from the page
+ */
+export function getJSONLD(body: cheerio.Cheerio): JSONLD {
+    shared.logger.trace("Extracting JSON-LD data...");
+
+    // Fetch the JSON-LD data
+    const structuredDataElements = body.find(f95Selector.GT_JSONLD);
+
+    // Parse the data
+    const values = structuredDataElements.map((idx, el) => parseJSONLD(el)).get();
+
+    // Merge the data and return a single value
+    return mergeJSONLD(values);
+}
+
+//#region Private methods
+/**
+ * Merges multiple JSON+LD tags into one object.
+ * @param data List of JSON+LD tags
+ */
+function mergeJSONLD(data: JSONLD[]): JSONLD {
+    // Local variables
+    let merged: JSONLD = {};
+
+    for (const value of data) {
+        merged = Object.assign(merged, value);
+    }
+
+    return merged;
+}
+
+/**
+ * Parse a JSON-LD element source code.
+ */
+function parseJSONLD(element: cheerio.Element): JSONLD {
+    // Get the element HTML
+    const html = cheerio.load(element).html().trim();
+
+    // Obtain the JSON-LD
+    const data = html
+        .replace("<script type=\"application/ld+json\">", "")
+        .replace("</script>", "");
+
+    // Convert the string to an object
+    return JSON.parse(data);
+}
+//#endregion Private methods
\ No newline at end of file