feat: add initial support for XML 1.1. edition 2

README.md

@@ -14,6 +14,7 @@ last updated, we had:
 
 * `xmlchars/xml/1.0/ed5` which covers XML 1.0 edition 5.
 * `xmlchars/xml/1.0/ed4` which covers XML 1.0 edition 4.
+* `xmlchars/xml/1.1/ed2` which covers XML 1.0 edition 2.
 * `xmlchars/xmlns/1.0/ed3` which covers XML Namespaces 1.0 edition 3.
 
 ## Features

src/xml/1.1/ed2.ts

@@ -0,0 +1,116 @@
+/**
+ * Character classes and associated utilities for the 2nd edition of XML 1.1.
+ *
+ * @author Louis-Dominique Dubeau
+ * @license MIT
+ * @copyright Louis-Dominique Dubeau
+ */
+
+//
+// Fragments.
+//
+export const CHAR = "\u0001-\uD7FF\uE000-\uFFFD\u{10000}-\u{10FFFF}";
+
+export const S = " \t\r\n";
+
+// tslint:disable-next-line:max-line-length
+export const NAME_START_CHAR = ":A-Z_a-z\u00C0-\u00D6\u00D8-\u00F6\u00F8-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD\u{10000}-\u{EFFFF}";
+
+export const NAME_CHAR =
+  `-${NAME_START_CHAR}.0-9\u00B7\u0300-\u036F\u203F-\u2040`;
+
+//
+// Regular expressions.
+//
+
+export const CHAR_RE = new RegExp(`^[${CHAR}]$`, "u");
+
+export const S_RE = new RegExp(`^[${S}]+$`, "u");
+
+export const NAME_START_CHAR_RE = new RegExp(`^[${NAME_START_CHAR}]$`, "u");
+
+export const NAME_CHAR_RE = new RegExp(`^[${NAME_CHAR}]$`, "u");
+
+export const NAME_RE = new RegExp(`^[${NAME_START_CHAR}][${NAME_CHAR}]*$`, "u");
+
+export const NMTOKEN_RE = new RegExp(`^[${NAME_CHAR}]+$`, "u");
+
+const TAB = 9;
+const NL = 0xA;
+const CR = 0xD;
+const SPACE = 0x20;
+
+//
+// Lists.
+//
+
+/** All characters in the ``S`` production. */
+export const S_LIST = [SPACE, NL, CR, TAB];
+
+/**
+ * Determines whether a codepoint matches the ``CHAR`` production.
+ *
+ * @param c The code point.
+ *
+ * @returns ``true`` if the codepoint matches ``CHAR``.
+ */
+export function isChar(c: number): boolean {
+  return (c >= 0x0001 && c <= 0xD7FF) ||
+    (c >= 0xE000 && c <= 0xFFFD) ||
+    (c >= 0x10000 && c <= 0x10FFFF);
+}
+
+/**
+ * Determines whether a codepoint matches the ``S`` (space) production.
+ *
+ * @param c The code point.
+ *
+ * @returns ``true`` if the codepoint matches ``S``.
+ */
+export function isS(c: number): boolean {
+  return S_LIST.includes(c);
+}
+
+const NAME_START_CHAR_LIST = [0x3A, 0x5F, 0x200C, 0x200D];
+
+/**
+ * Determines whether a codepoint matches the ``NAME_START_CHAR`` production.
+ *
+ * @param c The code point.
+ *
+ * @returns ``true`` if the codepoint matches ``NAME_START_CHAR``.
+ */
+// tslint:disable-next-line:cyclomatic-complexity
+export function isNameStartChar(c: number): boolean {
+  return ((c >= 0x41 && c <= 0x5A) ||
+          (c >= 0x61 && c <= 0x7A) ||
+          NAME_START_CHAR_LIST.includes(c) ||
+          (c >= 0xC0 && c <= 0xD6) ||
+          (c >= 0xD8 && c <= 0xF6) ||
+          (c >= 0x00F8 && c <= 0x02FF) ||
+          (c >= 0x0370 && c <= 0x037D) ||
+          (c >= 0x037F && c <= 0x1FFF) ||
+          (c >= 0x2070 && c <= 0x218F) ||
+          (c >= 0x2C00 && c <= 0x2FEF) ||
+          (c >= 0x3001 && c <= 0xD7FF) ||
+          (c >= 0xF900 && c <= 0xFDCF) ||
+          (c >= 0xFDF0 && c <= 0xFFFD) ||
+          (c >= 0x10000 && c <= 0xEFFFF));
+}
+
+const NAME_CHAR_LIST = [0x2D, 0x2E, 0xB7];
+
+/**
+ * Determines whether a codepoint matches the ``NAME_CHAR`` production.
+ *
+ * @param c The code point.
+ *
+ * @returns ``true`` if the codepoint matches ``NAME_CHAR``.
+ */
+export function isNameChar(c: number): boolean {
+  return isNameStartChar(c) ||
+    (c >= 0x30 && c <= 0x39) ||
+    NAME_CHAR_LIST.includes(c) ||
+    (c >= 0x0300 && c <= 0x036F) ||
+    (c >= 0x203F && c <= 0x2040);
+}

test/xmlchars.spec.ts

@@ -2,6 +2,7 @@
 import { expect } from "chai";
 import * as xml_1_0_ed4 from "../src/xml/1.0/ed4";
 import * as xml_1_0_ed5 from "../src/xml/1.0/ed5";
+import * as xml_1_1_ed2 from "../src/xml/1.1/ed2";
 import * as xmlns_1_0_ed3 from "../src/xmlns/1.0/ed3";
 
 interface Fixture {
@@ -94,6 +95,11 @@ const bom: Fixture = {
   data: "\uFFFE",
 };
 
+const one: Fixture = {
+  name: "\\u0001",
+  data: "\u0001",
+};
+
 // tslint:disable-next-line:mocha-no-side-effect-code
 const ALL_FIXTURES = new Set([
   x,
@@ -113,6 +119,7 @@ const ALL_FIXTURES = new Set([
   leadingDash,
   leadingDigit,
   bom,
+  one,
 ]);
 
 interface Case {
@@ -287,6 +294,66 @@ describe("xml/1.0", () => {
   });
 });
 
+describe("xml/1.1", () => {
+  describe("ed2", () => {
+    // tslint:disable-next-line:mocha-no-side-effect-code
+    const cases: Record<FilterPropertyName<typeof xml_1_1_ed2, RegExp>,
+    Case> = {
+      CHAR_RE: {
+        matching: [one, x, poo, colon, space, tab, newline, cr, ideographic,
+                   combining, digit, extender],
+      },
+      S_RE: {
+        matching: [space, tab, newline, cr],
+      },
+      NAME_START_CHAR_RE: {
+        matching: [x, ideographic, poo, colon],
+      },
+      NAME_CHAR_RE: {
+        matching: [x, ideographic, poo, colon, combining, extender, digit],
+      },
+      NAME_RE: {
+        matching: [x, abc, ideographic, colon, nameWithColon, poo],
+      },
+      NMTOKEN_RE: {
+        matching: [x, abc, ideographic, colon, nameWithColon, leadingDot,
+                   leadingDash, leadingDigit, combining, extender, digit, poo],
+      },
+    };
+
+    describe("regexes", () => {
+      // tslint:disable-next-line:mocha-no-side-effect-code
+      for (const name of (Object.keys(cases) as (keyof typeof cases)[])) {
+        describe(name, () => {
+          // tslint:disable-next-line:mocha-no-side-effect-code
+          makeTests(xml_1_1_ed2[name], cases[name]);
+        });
+      }
+    });
+
+    describe(".isChar", () => {
+      // tslint:disable-next-line:mocha-no-side-effect-code
+      makeCodePointTestTests(xml_1_1_ed2.isChar, cases.CHAR_RE);
+    });
+
+    describe(".isS", () => {
+      // tslint:disable-next-line:mocha-no-side-effect-code
+      makeCodePointTestTests(xml_1_1_ed2.isS, cases.S_RE);
+    });
+
+    describe(".isNameStartChar", () => {
+      // tslint:disable-next-line:mocha-no-side-effect-code
+      makeCodePointTestTests(xml_1_1_ed2.isNameStartChar,
+                             cases.NAME_START_CHAR_RE);
+    });
+
+    describe(".isNameChar", () => {
+      // tslint:disable-next-line:mocha-no-side-effect-code
+      makeCodePointTestTests(xml_1_1_ed2.isNameChar, cases.NAME_CHAR_RE);
+    });
+  });
+});
+
 describe("xmlns/1.0", () => {
   describe("ed3", () => {
     // tslint:disable-next-line:mocha-no-side-effect-code