[WIP] Start tinkering with IMAP

src/mail/replies.ts

@@ -0,0 +1,133 @@
+// EmailReplyParser is a library to parse plain text email content.
+// The goal is to identify quoted text, signatures, or original content.
+
+export class ReplyParser {
+    static VERSION = "0.5.11";
+
+    // Splits an email body into a list of Fragments.
+    static read(text: string): Email {
+        return new Email().read(text);
+    }
+
+    // Get the text of the visible portions of the given email body.
+    static parseReply(text: string): string {
+        return this.read(text).visibleText();
+    }
+}
+
+export class Email {
+    private fragments: Fragment[] = [];
+    private foundVisible: boolean = false;
+
+    // Gets the combined text of the visible fragments of the email body.
+    visibleText(): string {
+        return this.fragments
+            .filter((fragment) => !fragment.hidden)
+            .map((fragment) => fragment.toString())
+            .join("\n")
+            .trimEnd();
+    }
+
+    // Splits the given text into a list of Fragments.
+    read(text: string): this {
+        let modifiedText = text.slice();
+
+        // Normalize line endings.
+        modifiedText = modifiedText.replace(/\r\n/g, "\n");
+
+        // Handle multi-line reply headers.
+        const multiLineHeaderRegex = /^(?!On.*On\s.+?wrote:)(On\s(.+?)wrote:)$/m;
+        modifiedText = modifiedText.replace(multiLineHeaderRegex, (match) =>
+            match.replace(/\n/g, " ")
+        );
+
+        // Ensure proper splitting for lines of underscores.
+        modifiedText = modifiedText.replace(/([^\n])(?=\n_{7}_+)$/m, "$1\n");
+
+        // Reverse the text for parsing.
+        modifiedText = modifiedText.split("").reverse().join("");
+
+        this.foundVisible = false;
+        let fragment: Fragment | null = null;
+
+        const lines = modifiedText.split("\n");
+        for (const line of lines) {
+            const processedLine = line.trimEnd();
+            const isQuoted = processedLine.endsWith(">");
+
+            if (fragment && processedLine === "") {
+                if (Fragment.isSignature(fragment.lines[fragment.lines.length - 1])) {
+                    fragment.signature = true;
+                    this.finishFragment(fragment);
+                    fragment = null;
+                }
+            }
+
+            if (
+                fragment &&
+                (fragment.quoted === isQuoted ||
+                    (fragment.quoted &&
+                        (Fragment.isQuoteHeader(processedLine) || processedLine === "")))
+            ) {
+                fragment.lines.push(processedLine);
+            } else {
+                if (fragment) {
+                    this.finishFragment(fragment);
+                }
+                fragment = new Fragment(isQuoted, processedLine);
+            }
+        }
+
+        if (fragment) {
+            this.finishFragment(fragment);
+        }
+
+        this.fragments.reverse();
+        return this;
+    }
+
+    private finishFragment(fragment: Fragment): void {
+        fragment.finish();
+        if (!this.foundVisible) {
+            if (fragment.quoted || fragment.signature || fragment.toString().trim() === "") {
+                fragment.hidden = true;
+            } else {
+                this.foundVisible = true;
+            }
+        }
+        this.fragments.push(fragment);
+    }
+}
+
+class Fragment {
+    static SIGNATURE_REGEX = /(--\s*$|__\s*$|\w-$)|(^(\w+\s+){1,3}ym morf tneS$)/m;
+    static QUOTE_HEADER_REGEX = /^:etorw.*nO$|^.*:(morF|tneS|oT|tcejbuS)$/;
+
+    quoted: boolean;
+    signature: boolean = false;
+    hidden: boolean = false;
+    lines: string[];
+    private content: string | null = null;
+
+    constructor(quoted: boolean, firstLine: string) {
+        this.quoted = quoted;
+        this.lines = [firstLine];
+    }
+
+    static isSignature(line: string): boolean {
+        return this.SIGNATURE_REGEX.test(line);
+    }
+
+    static isQuoteHeader(line: string): boolean {
+        return this.QUOTE_HEADER_REGEX.test(line);
+    }
+
+    finish(): void {
+        this.content = this.lines.join("\n").split("").reverse().join("");
+        this.lines = [];
+    }
+
+    toString(): string {
+        return this.content || "";
+    }
+}