Support for Display String

changelog.md

@@ -5,6 +5,7 @@ ChangeLog
 ------------------
 
 * Support for a new 'Date' type, from draft [draft-ietf-httpbis-sfbis-02][7].
+* Support for the "Display String" type.
 * Now requires Node 18.
 
 

readme.md

@@ -59,12 +59,17 @@ The following are examples of `item` headers:
 Parsed as string
 
 ```
-# Parsed as string
+# Parsed an ASCII string
 Header: "foo"
 
 # A simple string, called a 'Token' in the spec
 Header: foo
 
+# A Unicode string, called a 'Display String' in the spec. They use
+# percent encoding, but encode a different set of characters than
+# URLs.
+Header %"Frysl%C3%A2n"
+
 # Parsed as number
 Header: 5
 Header: -10
@@ -84,7 +89,6 @@ Header: "Hello world"; a="5"
 Header: @1686634251
 ```
 
-
 To parse these header values, use the `parseItem`:
 
 ```typescript
@@ -217,6 +221,9 @@ serializeItem([5.5, new Map()]);
 // Returns "hello world"
 serializeItem(["hello world", new Map()]);
 
+// Returns %"Frysl%C3%A2n"
+serializeItem(["Fryslân", new Map()]);
+
 // Returns ?1
 serializeItem([true, new Map()]);
 

src/displaystring.ts

@@ -0,0 +1,16 @@
+export class DisplayString {
+
+  private value: string;
+  constructor(value: string) {
+
+    this.value = value;
+
+  }
+
+  toString(): string {
+
+    return this.value;
+
+  }
+
+}

src/index.ts

@@ -3,3 +3,4 @@ export * from './parser';
 export * from './types';
 export * from './util';
 export { Token } from './token';
+export { DisplayString } from './displaystring';

src/parser.ts

@@ -11,6 +11,7 @@ import {
 import { Token } from './token';
 
 import { isAscii } from './util';
+import { DisplayString } from './displaystring';
 
 export function parseDictionary(input: string): Dictionary {
 
@@ -185,6 +186,9 @@ export default class Parser {
     if (char === '@') {
       return this.parseDate();
     }
+    if (char === '%') {
+      return this.parseDisplayString();
+    }
     throw new ParseError(this.pos, 'Unexpected input');
 
   }
@@ -299,6 +303,49 @@ export default class Parser {
 
   }
 
+  private parseDisplayString(): DisplayString {
+
+    const chars = this.getChars(2);
+    if (chars !== '%"') {
+      throw new ParseError(this.pos, 'Unexpected character. Display strings should start with %=');
+    }
+
+    const result:number[] = [];
+
+    while (!this.eof()) {
+
+      const char = this.getChar();
+      if (char.charCodeAt(0) <= 0x1F || (char.charCodeAt(0) >= 0x7F && char.charCodeAt(0) <= 0xFF)) {
+        throw new ParseError(this.pos, 'Invalid char found in DisplayString. Did you forget to escape?');
+      }
+
+      if (char==='%') {
+        const hexChars = this.getChars(2);
+        if (/^[0-9a-f]{2}$/.test(hexChars)) {
+          result.push(parseInt(hexChars, 16));
+        } else {
+          throw new ParseError(this.pos, `Unexpected sequence after % in DispalyString: "${hexChars}". Note that hexidecimals must be lowercase`);
+        }
+        continue;
+      }
+      if (char==='"') {
+        const textDecoder = new TextDecoder('utf-8', {
+          fatal: true
+        });
+        try {
+          return new DisplayString(
+            textDecoder.decode(new Uint8Array(result))
+          );
+        } catch (err) {
+          throw new ParseError(this.pos, 'Fatal error decoding UTF-8 sequence in Display String');
+        }
+      }
+      result.push(char.charCodeAt(0));
+    }
+    throw new ParseError(this.pos, 'Unexpected end of input');
+
+  }
+
   private parseToken(): Token {
 
     // The specification wants this check, but it's an unreachable code block.
@@ -429,6 +476,16 @@ export default class Parser {
 
     return this.input[this.pos++];
 
+  }
+  private getChars(count: number): string {
+
+    const result = this.input.substr(
+      this.pos,
+      count
+    );
+    this.pos += count;
+    return result;
+
   }
   private eof():boolean {
 

src/serializer.ts

@@ -11,6 +11,7 @@ import {
 import { Token } from './token';
 
 import { isAscii, isInnerList, isValidKeyStr } from './util';
+import { DisplayString } from './displaystring';
 
 export class SerializeError extends Error {}
 
@@ -80,6 +81,9 @@ export function serializeBareItem(input: BareItem): string {
   if (input instanceof ByteSequence) {
     return serializeByteSequence(input);
   }
+  if (input instanceof DisplayString) {
+    return serializeDisplayString(input);
+  }
   if (input instanceof Date) {
     return serializeDate(input);
   }
@@ -114,6 +118,24 @@ export function serializeString(input: string): string {
   return `"${input.replace(/("|\\)/g, (v) => '\\' + v)}"`;
 }
 
+export function serializeDisplayString(input: DisplayString): string {
+  let out = '%"';
+  const textEncoder = new TextEncoder();
+  for (const char of textEncoder.encode(input.toString())) {
+    if (
+      char === 0x25 // %
+      || char === 0x22 // "
+      || char <= 0x1f
+      || char >= 0x7f
+    ) {
+      out += '%' + char.toString(16);
+    } else {
+      out += String.fromCharCode(char);
+    }
+  }
+  return out + '"';
+}
+
 export function serializeBoolean(input: boolean): string {
   return input ? '?1' : '?0';
 }

src/types.ts

@@ -1,4 +1,5 @@
 import { Token } from './token';
+import { DisplayString } from './displaystring';
 
 /**
  * Lists are arrays of zero or more members, each of which can be an Item
@@ -47,6 +48,6 @@ export class ByteSequence {
 
 }
 
-export type BareItem = number | string | Token | ByteSequence | Date | boolean;
+export type BareItem = number | string | Token | ByteSequence | Date | boolean | DisplayString;
 
 export type Item = [BareItem, Parameters];

test/httpwg-tests.js

@@ -10,7 +10,7 @@ const {
 
   ParseError,
 } = require('../dist');
-const { Token, ByteSequence } = require('../dist');
+const { Token, ByteSequence, DisplayString } = require('../dist');
 const base32Encode = require('base32-encode');
 const base32Decode = require('base32-decode');
 const fs = require('fs');
@@ -24,6 +24,7 @@ describe('HTTP-WG tests', () => {
     'string',
     'token',
     'date',
+    'display-string',
 
     'item',
 
@@ -284,6 +285,12 @@ function packTestValue(input) {
       value: input.toString()
     }
   }
+  if(input instanceof DisplayString) {
+    return {
+      __type: 'displaystring',
+      value: input.toString()
+    }
+  }
   if (input instanceof ByteSequence) {
     return {
       __type: 'binary',
@@ -340,6 +347,8 @@ function unpackTestValue(input) {
         return new ByteSequence(Buffer.from(base32Decode(input.value, 'RFC4648')).toString('base64'));
       case 'date' :
         return new Date(input.value * 1000);
+      case 'displaystring' :
+        return new DisplayString(input.value);
       default:
         throw new Error('Unknown input __type: ' + input.__type);
     }