docs:update (#19)

Author: erikbrinkman

README.md

@@ -5,10 +5,10 @@ MHTML Stream
 [![npm](https://img.shields.io/npm/v/mhtml-stream)](https://www.npmjs.com/package/mhtml-stream)
 [![license](https://img.shields.io/github/license/erikbrinkman/mhtml-stream)](LICENSE)
 
-Zero-dependency library for parsing MHTML data as streams using modern WHATWG
-streams and async iterators. Because it relies on modern cross javascript
-standards it works out-of-the-box in all javascript environments, with only a
-little tweaking necessary for module definitions.
+Library for parsing MHTML data as streams using modern WHATWG streams and async
+iterators. Because it relies on modern cross javascript standards it works
+out-of-the-box in all javascript environments, with only a little tweaking
+necessary for module definitions.
 
 Usage
 -----
@@ -39,3 +39,7 @@ Notes
   comes to whether whitespace should be added when unfolding. This currently
   uses the first whitespace character to indicate folding, and preservers any
   others.
+- Decoded part content preserves the original MIME line endings. The parser
+  splits the stream on `\r\n`, and the quoted-printable and 7bit/8bit decoders
+  re-insert that `\r\n` between lines, so extraction is lossless. Both decoders
+  accept a `newLine` argument if you'd rather normalize to `\n`.

package.json

@@ -1,7 +1,7 @@
 {
   "name": "mhtml-stream",
   "version": "2.0.2",
-  "description": "Zero dependency stream MHTML parser",
+  "description": "Streaming MHTML parser",
   "keywords": [
     "mhtml",
     "stream"

src/index.ts

@@ -240,7 +240,6 @@ export async function* parseMhtml(
     const headers = await parseHeaders(lines);
     const [boundary, terminus] = bound ?? (bound = getBoundary(headers));
 
-    // Content-Transfer-Encoding token values are case-insensitive (RFC 2045)
     const encoding = (
       headers.get("Content-Transfer-Encoding") ?? "7bit"
     ).toLowerCase();

src/utils.ts

@@ -127,7 +127,6 @@ export async function collect(
   return concat(chunks);
 }
 
-// CRLF, the canonical MIME line separator the stream is split on
 const crlf = new Uint8Array([13, 10]);
 
 /** whether a character code is a hex digit (0-9, A-F, a-f) */
@@ -145,9 +144,8 @@ export function isHexDigit(code: number): boolean {
  * If quoted printable "lines" aren't escaped with an "=" then a new line needs
  * to be inserted. We use `newLine`, which defaults to CRLF to match the
  * canonical MIME form; pass a custom separator (e.g. a single "\n") to
- * normalize instead. The separator is emitted between lines, never after the
- * last one, since the CRLF preceding the MIME boundary belongs to the
- * delimiter, not the body.
+ * normalize instead. The separator goes between lines; the CRLF before the MIME
+ * boundary belongs to the delimiter, not the body.
  */
 export async function* decodeQuotedPrintable(
   lines: AsyncIterable<Uint8Array>,
@@ -234,8 +232,8 @@ export async function* decodeBase64(
  * payload as-is. parseMhtml splits the stream on CRLF to find part boundaries,
  * so we re-insert `newLine` (defaulting to CRLF) between lines to restore the
  * original bytes exactly. Pass `newLine` (e.g. a single "\n") to normalize line
- * endings instead. The separator is emitted between lines, never after the last
- * one, since the CRLF preceding the boundary belongs to the delimiter.
+ * endings instead. The separator goes between lines; the CRLF before the
+ * boundary belongs to the delimiter, not the body.
  */
 export async function* decodeIdentity(
   lines: AsyncIterable<Uint8Array>,