feat: add codepoint tests

Testing by codepoint may make sense in some cases.
lddubeau · Jul 23, 2018 · 881e552 · 881e552
1 parent 3c538e9
commit 881e552
Show file tree

Hide file tree

Showing 2 changed files with 147 additions and 8 deletions.
diff --git a/src/xmlchars.ts b/src/xmlchars.ts
@@ -52,6 +52,84 @@ export namespace XML_1_0 {
       export const NMTOKEN = new RegExp(`^[${fragments.NAME_CHAR}]+$`, "u");
     }
 
+    const TAB = 9;
+    const NL = 0xA;
+    const CR = 0xD;
+    const SPACE = 0x20;
+    const MINUS = 0x2D;
+    const COLON = 0x3A;
+
+    /**
+     * 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 === TAB ||
+              c === NL ||
+              c === CR ||
+              (c >= SPACE && 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 c === SPACE || c === NL || c === CR || c === TAB;
+    }
+
+    /**
+     * 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 === COLON ||
+              (c >= 0x41 && c <= 0x5A) ||
+              c === 0x5F ||
+              (c >= 0x61 && c <= 0x7A) ||
+              (c >= 0xC0 && c <= 0xD6) ||
+              (c >= 0xD8 && c <= 0xF6) ||
+              (c >= 0x00F8 && c <= 0x02FF) ||
+              (c >= 0x0370 && c <= 0x037D) ||
+              (c >= 0x037F && c <= 0x1FFF) ||
+              (c >= 0x200C && c <= 0x200D) ||
+              (c >= 0x2070 && c <= 0x218F) ||
+              (c >= 0x2C00 && c <= 0x2FEF) ||
+              (c >= 0x3001 && c <= 0xD7FF) ||
+              (c >= 0xF900 && c <= 0xFDCF) ||
+              (c >= 0xFDF0 && c <= 0xFFFD) ||
+              (c >= 0x10000 && c <= 0xEFFFF));
+    }
+
+    /**
+     * 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 === MINUS ||
+         c === 0x2E ||
+         (c >= 0x30 && c <= 0x39) ||
+         c === 0x00B7 ||
+         (c >= 0x0300 && c <= 0x036F) ||
+         (c >= 0x203F && c <= 0x2040));
+    }
   }
 
   /**

diff --git a/test/xmlchars.spec.ts b/test/xmlchars.spec.ts
@@ -82,6 +82,11 @@ const leadingDigit: Fixture = {
   data: "1bar",
 };
 
+const bom: Fixture = {
+  name: "bom",
+  data: "\uFFFE",
+};
+
 // tslint:disable-next-line:mocha-no-side-effect-code
 const ALL_FIXTURES = new Set([
   x,
@@ -99,6 +104,7 @@ const ALL_FIXTURES = new Set([
   leadingDot,
   leadingDash,
   leadingDigit,
+  bom,
 ]);
 
 interface Case {
@@ -132,8 +138,39 @@ function makeTests(re: RegExp, testCase: Case): void {
   });
 }
 
+function makeCodePointTestTests(codePointTest: (c: number) => boolean,
+                                testCase: Case): void {
+  for (const fixture of testCase.matching) {
+    const { data, name } = fixture;
+    // tslint:disable-next-line:no-non-null-assertion
+    const code = data.codePointAt(0)!;
+    if (data.length > String.fromCodePoint(code).length) {
+      // We skip those fixtures that contain more than one character.
+      continue;
+    }
+    it(`matches ${name}`, () => {
+      expect(codePointTest(code)).to.be.true;
+    });
+  }
+
+  for (const fixture of ALL_FIXTURES) {
+    if (!testCase.matching.includes(fixture)) {
+      const { data, name } = fixture;
+      // tslint:disable-next-line:no-non-null-assertion
+      const code = data.codePointAt(0)!;
+      if (data.length > String.fromCodePoint(code).length) {
+        // We skip those fixtures that contain more than one character.
+        continue;
+      }
+      it(`does not match ${name}`, () => {
+        expect(codePointTest(code)).to.be.false;
+      });
+    }
+  }
+}
+
 describe("XML_1_0", () => {
-  describe("ED5.regexes", () => {
+  describe("ED5", () => {
     // tslint:disable-next-line:mocha-no-side-effect-code
     const cases: Record<ED5_REGEX_NAMES, Case> = {
       CHAR: {
@@ -156,15 +193,38 @@ describe("XML_1_0", () => {
         matching: [x, abc, ideographic, nameWithColon, leadingDot, leadingDash,
                    leadingDigit, combining, extender, digit, poo],
       },
+      // tslint:disable-next-line:no-any
     } as any;
+    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_0.ED5.regexes[name], cases[name]);
+        });
+      }
+    });
 
-    // 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_0.ED5.regexes[name], cases[name]);
-      });
-    }
+    describe(".isChar", () => {
+      // tslint:disable-next-line:mocha-no-side-effect-code
+      makeCodePointTestTests(XML_1_0.ED5.isChar, cases.CHAR);
+    });
+
+    describe(".isS", () => {
+      // tslint:disable-next-line:mocha-no-side-effect-code
+      makeCodePointTestTests(XML_1_0.ED5.isS, cases.S);
+    });
+
+    describe(".isNameStartChar", () => {
+      // tslint:disable-next-line:mocha-no-side-effect-code
+      makeCodePointTestTests(XML_1_0.ED5.isNameStartChar,
+                             cases.NAME_START_CHAR);
+    });
+
+    describe(".isNameChar", () => {
+      // tslint:disable-next-line:mocha-no-side-effect-code
+      makeCodePointTestTests(XML_1_0.ED5.isNameChar, cases.NAME_CHAR);
+    });
   });
 
   describe("ED4.regexes", () => {
@@ -205,6 +265,7 @@ describe("XML_1_0", () => {
         matching: [x, abc, ideographic, nameWithColon, leadingDot, leadingDash,
                    leadingDigit, combining, extender, digit],
       },
+      // tslint:disable-next-line:no-any
     } as any;
 
     // tslint:disable-next-line:mocha-no-side-effect-code