Refactor to use paragraph formatters

.github/workflows/CODEOWNERS

@@ -0,0 +1 @@
+* @kkartch0

src/models/ParagraphFormatter.ts

@@ -0,0 +1,21 @@
+import { GospelStudyPluginSettings } from "./GospelStudyPluginSettings";
+
+/**
+ * Represents a paragraph formatter.
+ */
+export interface ParagraphFormatter {
+
+    /**
+     * Determines if the formatter is enabled based on the settings.
+     * @param settings - The plugin settings.
+     * @returns A boolean indicating if the formatter is enabled.
+     */
+    isEnabled(settings: GospelStudyPluginSettings): boolean;
+
+    /**
+     * Formats the paragraph.
+     * @param paragraph - The paragraph to be formatted.
+     * @returns The formatted paragraph.
+     */
+    format(paragraph: string): string;
+}

src/paragraphFormatting/fullyQualifyStudyLinksFormatter.ts

@@ -0,0 +1,24 @@
+import { ParagraphFormatter } from "src/models/ParagraphFormatter";
+
+/**
+ * Replaces all occurrences of "/study/" in the given text with "https://www.churchofjesuschrist.org/study/".
+ *
+ * @param text - The input text to be processed.
+ * @returns The modified text with fully qualified study links.
+ * @example
+ * const inputText = "This is a paragraph with <a href='/study/scriptures/bofm/1-ne/1.1'>study link</a> and <a href='/study/scriptures/bofm/1-ne/2.1'>another study link</a>.";
+ * const modifiedText = fullyQualifyStudyLinks(inputText);
+ * console.log(modifiedText);
+ * // Output: "This is a paragraph with <a href='https://www.churchofjesuschrist.org/study/scriptures/bofm/1-ne/1.1'>study link</a> and <a href='https://www.churchofjesuschrist.org/study/scriptures/bofm/1-ne/2.1'>another study link</a>."
+ */
+export const fullyQualifyStudyLinksFormatter: ParagraphFormatter = {
+	isEnabled(): boolean {
+		return true;
+	},
+
+	format(paragraph: string): string {
+		return paragraph.replace(/"\/study\//g, "\"https://www.churchofjesuschrist.org/study/");
+	}
+};
+
+export default fullyQualifyStudyLinksFormatter;

src/paragraphFormatting/index.ts

@@ -0,0 +1,44 @@
+import { ParagraphFormatter } from "src/models/ParagraphFormatter";
+import { GospelStudyPluginSettings } from "src/models/GospelStudyPluginSettings";
+import { registeredFormatters } from "./registeredFormatters";
+/**
+ * Retrieves the paragraphs with the specified ids from the document, formats, and then returns them.
+ * 
+ * @param document - The document object representing the HTML document.
+ * @param paragraphIds - An array of active paragraph IDs.
+ * @returns An array of formatted paragraphs.
+ */
+export function getFormattedParagraphs(document: Document, paragraphIds: string[], pluginSettings: GospelStudyPluginSettings): string[] {
+	const paragraphs: string[] = [];
+	const enabledFormatters = registeredFormatters.filter((formatter: ParagraphFormatter) => formatter.isEnabled(pluginSettings));
+
+	paragraphIds.forEach((id) => {
+		if (id === "-") { // Hyphen indicates that the next paragraph is not contiguous with the previous one, so we add an ellipsis.
+			paragraphs.push("…");
+			return;
+		}
+		const paragraphElement = document.getElementById(id);
+		if (!paragraphElement) return;
+
+		const formattedParagraph = formatParagraph(paragraphElement.innerHTML, enabledFormatters);
+
+		paragraphs.push(formattedParagraph);
+	});
+	return paragraphs;
+}
+
+
+/**
+ * Formats a paragraph by applying various transformations based on the plugin settings.
+ * 
+ * @param paragraph- The HTML content of the paragraph.
+ * @param pluginSettings - The settings of the Gospel Study plugin.
+ * @returns The formatted paragraph HTML.
+ */
+export function formatParagraph(paragraph: string, enabledFormatters: ParagraphFormatter[]): string {
+	enabledFormatters.forEach((formatter) => {
+		paragraph = formatter.format(paragraph);
+	});
+
+	return paragraph;
+}

src/paragraphFormatting/registeredFormatters.ts

@@ -0,0 +1,24 @@
+import { ParagraphFormatter } from "src/models/ParagraphFormatter";
+import fs from 'fs';
+
+export let registeredFormatters: ParagraphFormatter[] = []; 
+
+/**
+ * Imports all formatters (i.e. files that end with "Formatter.ts") and assigns them to registeredFormatters.
+ * 
+ * @param pluginSettings - The settings of the Gospel Study plugin.
+ * @returns An array of enabled formatters.
+ */
+async function loadInAllFormatters() {
+    const files = fs.readdirSync('src/paragraphFormatting');
+    const formatterPromises = files.filter(file => file.endsWith('Formatter.ts')).map(async file => {
+        const module = await import(`./${file}`);
+        const formatter = module.default as ParagraphFormatter;
+        return formatter;
+    });
+
+    registeredFormatters = await Promise.all(formatterPromises);
+}
+
+
+loadInAllFormatters();

src/paragraphFormatting/remoteFootnotesFormatter.ts

@@ -0,0 +1,30 @@
+import { ParagraphFormatter } from "src/models/ParagraphFormatter";
+
+/**
+ * Removes footnotes from a paragraph of text.
+ *
+ * @param text - The input text containing footnotes.
+ * @returns The modified text with footnotes removed.
+ *
+ * @example
+ * const inputText = "This is a paragraph with a <a class='study-note-ref' href='#note1'><sup class='marker'>1</sup>footnote</a>."
+ * const modifiedText = removeFootnotesFromParagraph(inputText);
+ * console.log(modifiedText);
+ * // Output: "This is a paragraph with a footnote."
+ */
+export const footnotesFormatter: ParagraphFormatter = {
+	isEnabled(): boolean {
+		return true;
+	},
+
+	format(paragraph: string) {
+		paragraph = paragraph.replace(
+			/<a class="study-note-ref" href="[^>]*"><sup class="marker">[^<]*<\/sup>([^<]*)<\/a>/g,
+			"$1"
+		);
+
+		return paragraph;
+	}
+}
+
+export default footnotesFormatter;

src/paragraphFormatting/removeNonUsefulTagsFormatter.ts

@@ -0,0 +1,31 @@
+import { ParagraphFormatter } from "src/models/ParagraphFormatter";
+
+/**
+ * Removes tags that are not useful/functional in the Obsidian context such as associated content, page breaks, and note references.
+ *
+ * @param text - The input text containing HTML tags.
+ * @returns The modified text with non-useful tags removed.
+ */
+export const removeNonUsefulTagsFormatter: ParagraphFormatter = {
+	isEnabled(): boolean {
+		return true;
+	},
+
+	format(paragraph: string): string {
+		const parser = new DOMParser();
+		const doc = parser.parseFromString(paragraph, 'text/html');
+
+		const tagsToRemove = ['span.page-break', 'span[title="Associated Content"]', 'a.note-ref'];
+
+		tagsToRemove.forEach((tag) => {
+			const elements = doc.querySelectorAll(tag);
+			elements.forEach((element) => {
+				element.remove();
+			});
+		});
+
+		return doc.body.innerHTML;
+	}
+};
+
+export default removeNonUsefulTagsFormatter;

src/paragraphFormatting/removeScriptureReferenceLinksFormatter.ts

@@ -0,0 +1,35 @@
+import { GospelStudyPluginSettings } from "src/models/GospelStudyPluginSettings";
+import { ParagraphFormatter } from "src/models/ParagraphFormatter";
+
+/**
+ * Removes scripture reference links (e.g. <a class="scripture-ref">Some Reference</a>) from the given paragraph by replacing them with their text content.
+ *
+ * @param paragraphHTML - The HTML string representing the paragraph.
+ * @returns The modified paragraph HTML with scripture reference links removed.
+ * @example
+ * const inputText = "This is a paragraph with a <a class='scripture-ref'>scripture reference</a>."
+ * const modifiedText = removeScriptureReferenceLinks(inputText);
+ * console.log(modifiedText);
+ * // Output: "This is a paragraph with a scripture reference."
+ */
+export const removeScriptureReferenceLinksFormatter: ParagraphFormatter = {
+	isEnabled(settings: GospelStudyPluginSettings): boolean {
+		return !settings.retainScriptureReferenceLinks;
+	},
+
+	format(paragraph: string): string {
+
+		const parser = new DOMParser();
+		const doc = parser.parseFromString(paragraph, 'text/html');
+
+		const referenceLinks = doc.querySelectorAll('a.scripture-ref');
+		referenceLinks.forEach((element): void => {
+			// Replace the scripture reference link with the text content of the link.
+			element.replaceWith(element.textContent || '');
+		});
+		paragraph = doc.body.innerHTML;
+		return paragraph;
+	}
+}
+
+export default removeScriptureReferenceLinksFormatter;