entities 6.0.1 → 7.0.1

package/src/decode.spec.ts

@@ -1,320 +0,0 @@
-import { describe, it, expect, vitest } from "vitest";
-import * as entities from "./decode.js";
-
-describe("Decode test", () => {
-    const testcases = [
-        { input: "&", output: "&" },
-        { input: "&", output: "&" },
-        { input: "&", output: "&" },
-        { input: "&", output: "&" },
-        { input: "&", output: "&" },
-        { input: "&", output: "&" },
-        { input: "&", output: "&" },
-        { input: ":", output: ":" },
-        { input: ":", output: ":" },
-        { input: ":", output: ":" },
-        { input: ":", output: ":" },
-        { input: "&#", output: "&#" },
-        { input: "&>", output: "&>" },
-        { input: "id=770&#anchor", output: "id=770&#anchor" },
-    ];
-
-    for (const { input, output } of testcases) {
-        it(`should XML decode ${input}`, () =>
-            expect(entities.decodeXML(input)).toBe(output));
-        it(`should HTML decode ${input}`, () =>
-            expect(entities.decodeHTML(input)).toBe(output));
-    }
-
-    it("should HTML decode partial legacy entity", () => {
-        expect(entities.decodeHTMLStrict("&timesbar")).toBe("&timesbar");
-        expect(entities.decodeHTML("&timesbar")).toBe("×bar");
-    });
-
-    it("should HTML decode legacy entities according to spec", () =>
-        expect(entities.decodeHTML("?&image_uri=1&ℑ=2&image=3")).toBe(
-            "?&image_uri=1&ℑ=2&image=3",
-        ));
-
-    it("should back out of legacy entities", () =>
-        expect(entities.decodeHTML("&ampa")).toBe("&a"));
-
-    it("should not parse numeric entities in strict mode", () =>
-        expect(entities.decodeHTMLStrict("&#55")).toBe("&#55"));
-
-    it("should parse &nbsp followed by < (#852)", () =>
-        expect(entities.decodeHTML("&nbsp<")).toBe("\u00A0<"));
-
-    it("should decode trailing legacy entities", () => {
-        expect(entities.decodeHTML("&timesbar;&timesbar")).toBe("⨱×bar");
-    });
-
-    it("should decode multi-byte entities", () => {
-        expect(entities.decodeHTML("&NotGreaterFullEqual;")).toBe("≧̸");
-    });
-
-    it("should not decode legacy entities followed by text in attribute mode", () => {
-        expect(
-            entities.decodeHTML("&not", entities.DecodingMode.Attribute),
-        ).toBe("¬");
-
-        expect(
-            entities.decodeHTML("&noti", entities.DecodingMode.Attribute),
-        ).toBe("&noti");
-
-        expect(
-            entities.decodeHTML("&not=", entities.DecodingMode.Attribute),
-        ).toBe("&not=");
-
-        expect(entities.decodeHTMLAttribute("&notp")).toBe("&notp");
-        expect(entities.decodeHTMLAttribute("&notP")).toBe("&notP");
-        expect(entities.decodeHTMLAttribute("&not3")).toBe("&not3");
-    });
-});
-
-describe("EntityDecoder", () => {
-    it("should decode decimal entities", () => {
-        const callback = vitest.fn();
-        const decoder = new entities.EntityDecoder(
-            entities.htmlDecodeTree,
-            callback,
-        );
-
-        expect(decoder.write("&#5", 1)).toBe(-1);
-        expect(decoder.write("8;", 0)).toBe(5);
-
-        expect(callback).toHaveBeenCalledTimes(1);
-        expect(callback).toHaveBeenCalledWith(":".charCodeAt(0), 5);
-    });
-
-    it("should decode hex entities", () => {
-        const callback = vitest.fn();
-        const decoder = new entities.EntityDecoder(
-            entities.htmlDecodeTree,
-            callback,
-        );
-
-        expect(decoder.write("&#x3a;", 1)).toBe(6);
-
-        expect(callback).toHaveBeenCalledTimes(1);
-        expect(callback).toHaveBeenCalledWith(":".charCodeAt(0), 6);
-    });
-
-    it("should decode named entities", () => {
-        const callback = vitest.fn();
-        const decoder = new entities.EntityDecoder(
-            entities.htmlDecodeTree,
-            callback,
-        );
-
-        expect(decoder.write("&amp;", 1)).toBe(5);
-
-        expect(callback).toHaveBeenCalledTimes(1);
-        expect(callback).toHaveBeenCalledWith("&".charCodeAt(0), 5);
-    });
-
-    it("should decode legacy entities", () => {
-        const callback = vitest.fn();
-        const decoder = new entities.EntityDecoder(
-            entities.htmlDecodeTree,
-            callback,
-        );
-        decoder.startEntity(entities.DecodingMode.Legacy);
-
-        expect(decoder.write("&amp", 1)).toBe(-1);
-
-        expect(callback).toHaveBeenCalledTimes(0);
-
-        expect(decoder.end()).toBe(4);
-
-        expect(callback).toHaveBeenCalledTimes(1);
-        expect(callback).toHaveBeenCalledWith("&".charCodeAt(0), 4);
-    });
-
-    it("should decode named entity written character by character", () => {
-        const callback = vitest.fn();
-        const decoder = new entities.EntityDecoder(
-            entities.htmlDecodeTree,
-            callback,
-        );
-
-        for (const c of "amp") {
-            expect(decoder.write(c, 0)).toBe(-1);
-        }
-        expect(decoder.write(";", 0)).toBe(5);
-
-        expect(callback).toHaveBeenCalledTimes(1);
-        expect(callback).toHaveBeenCalledWith("&".charCodeAt(0), 5);
-    });
-
-    it("should decode numeric entity written character by character", () => {
-        const callback = vitest.fn();
-        const decoder = new entities.EntityDecoder(
-            entities.htmlDecodeTree,
-            callback,
-        );
-
-        for (const c of "#x3a") {
-            expect(decoder.write(c, 0)).toBe(-1);
-        }
-        expect(decoder.write(";", 0)).toBe(6);
-
-        expect(callback).toHaveBeenCalledTimes(1);
-        expect(callback).toHaveBeenCalledWith(":".charCodeAt(0), 6);
-    });
-
-    it("should decode hex entities across several chunks", () => {
-        const callback = vitest.fn();
-        const decoder = new entities.EntityDecoder(
-            entities.htmlDecodeTree,
-            callback,
-        );
-
-        for (const chunk of ["#x", "cf", "ff", "d"]) {
-            expect(decoder.write(chunk, 0)).toBe(-1);
-        }
-
-        expect(decoder.write(";", 0)).toBe(9);
-        expect(callback).toHaveBeenCalledTimes(1);
-        expect(callback).toHaveBeenCalledWith(0xc_ff_fd, 9);
-    });
-
-    it("should not fail if nothing is written", () => {
-        const callback = vitest.fn();
-        const decoder = new entities.EntityDecoder(
-            entities.htmlDecodeTree,
-            callback,
-        );
-
-        expect(decoder.end()).toBe(0);
-        expect(callback).toHaveBeenCalledTimes(0);
-    });
-
-    describe("errors", () => {
-        it("should produce an error for a named entity without a semicolon", () => {
-            const errorHandlers = {
-                missingSemicolonAfterCharacterReference: vitest.fn(),
-                absenceOfDigitsInNumericCharacterReference: vitest.fn(),
-                validateNumericCharacterReference: vitest.fn(),
-            };
-            const callback = vitest.fn();
-            const decoder = new entities.EntityDecoder(
-                entities.htmlDecodeTree,
-                callback,
-                errorHandlers,
-            );
-
-            decoder.startEntity(entities.DecodingMode.Legacy);
-            expect(decoder.write("&amp;", 1)).toBe(5);
-            expect(callback).toHaveBeenCalledTimes(1);
-            expect(callback).toHaveBeenCalledWith("&".charCodeAt(0), 5);
-            expect(
-                errorHandlers.missingSemicolonAfterCharacterReference,
-            ).toHaveBeenCalledTimes(0);
-
-            decoder.startEntity(entities.DecodingMode.Legacy);
-            expect(decoder.write("&amp", 1)).toBe(-1);
-            expect(decoder.end()).toBe(4);
-
-            expect(callback).toHaveBeenCalledTimes(2);
-            expect(callback).toHaveBeenLastCalledWith("&".charCodeAt(0), 4);
-            expect(
-                errorHandlers.missingSemicolonAfterCharacterReference,
-            ).toHaveBeenCalledTimes(1);
-        });
-
-        it("should produce an error for a numeric entity without a semicolon", () => {
-            const errorHandlers = {
-                missingSemicolonAfterCharacterReference: vitest.fn(),
-                absenceOfDigitsInNumericCharacterReference: vitest.fn(),
-                validateNumericCharacterReference: vitest.fn(),
-            };
-            const callback = vitest.fn();
-            const decoder = new entities.EntityDecoder(
-                entities.htmlDecodeTree,
-                callback,
-                errorHandlers,
-            );
-
-            decoder.startEntity(entities.DecodingMode.Legacy);
-            expect(decoder.write("&#x3a", 1)).toBe(-1);
-            expect(decoder.end()).toBe(5);
-
-            expect(callback).toHaveBeenCalledTimes(1);
-            expect(callback).toHaveBeenCalledWith(0x3a, 5);
-            expect(
-                errorHandlers.missingSemicolonAfterCharacterReference,
-            ).toHaveBeenCalledTimes(1);
-            expect(
-                errorHandlers.absenceOfDigitsInNumericCharacterReference,
-            ).toHaveBeenCalledTimes(0);
-            expect(
-                errorHandlers.validateNumericCharacterReference,
-            ).toHaveBeenCalledTimes(1);
-            expect(
-                errorHandlers.validateNumericCharacterReference,
-            ).toHaveBeenCalledWith(0x3a);
-        });
-
-        it("should produce an error for numeric entities without digits", () => {
-            const errorHandlers = {
-                missingSemicolonAfterCharacterReference: vitest.fn(),
-                absenceOfDigitsInNumericCharacterReference: vitest.fn(),
-                validateNumericCharacterReference: vitest.fn(),
-            };
-            const callback = vitest.fn();
-            const decoder = new entities.EntityDecoder(
-                entities.htmlDecodeTree,
-                callback,
-                errorHandlers,
-            );
-
-            decoder.startEntity(entities.DecodingMode.Legacy);
-            expect(decoder.write("&#", 1)).toBe(-1);
-            expect(decoder.end()).toBe(0);
-
-            expect(callback).toHaveBeenCalledTimes(0);
-            expect(
-                errorHandlers.missingSemicolonAfterCharacterReference,
-            ).toHaveBeenCalledTimes(0);
-            expect(
-                errorHandlers.absenceOfDigitsInNumericCharacterReference,
-            ).toHaveBeenCalledTimes(1);
-            expect(
-                errorHandlers.absenceOfDigitsInNumericCharacterReference,
-            ).toHaveBeenCalledWith(2);
-            expect(
-                errorHandlers.validateNumericCharacterReference,
-            ).toHaveBeenCalledTimes(0);
-        });
-
-        it("should produce an error for hex entities without digits", () => {
-            const errorHandlers = {
-                missingSemicolonAfterCharacterReference: vitest.fn(),
-                absenceOfDigitsInNumericCharacterReference: vitest.fn(),
-                validateNumericCharacterReference: vitest.fn(),
-            };
-            const callback = vitest.fn();
-            const decoder = new entities.EntityDecoder(
-                entities.htmlDecodeTree,
-                callback,
-                errorHandlers,
-            );
-
-            decoder.startEntity(entities.DecodingMode.Legacy);
-            expect(decoder.write("&#x", 1)).toBe(-1);
-            expect(decoder.end()).toBe(0);
-
-            expect(callback).toHaveBeenCalledTimes(0);
-            expect(
-                errorHandlers.missingSemicolonAfterCharacterReference,
-            ).toHaveBeenCalledTimes(0);
-            expect(
-                errorHandlers.absenceOfDigitsInNumericCharacterReference,
-            ).toHaveBeenCalledTimes(1);
-            expect(
-                errorHandlers.validateNumericCharacterReference,
-            ).toHaveBeenCalledTimes(0);
-        });
-    });
-});

package/src/encode.spec.ts

@@ -1,78 +0,0 @@
-import { describe, it, expect } from "vitest";
-import * as entities from "./index.js";
-
-describe("Encode->decode test", () => {
-    const testcases = [
-        {
-            input: "asdf & ÿ ü '",
-            xml: "asdf & ÿ ü '",
-            html: "asdf & ÿ ü '",
-        },
-        {
-            input: "&",
-            xml: "&",
-            html: "&",
-        },
-    ];
-
-    for (const { input, xml, html } of testcases) {
-        const encodedXML = entities.encodeXML(input);
-        it(`should XML encode ${input}`, () => expect(encodedXML).toBe(xml));
-        it(`should default to XML encode ${input}`, () =>
-            expect(entities.encode(input)).toBe(xml));
-        it(`should XML decode ${encodedXML}`, () =>
-            expect(entities.decodeXML(encodedXML)).toBe(input));
-        it(`should default to XML encode ${encodedXML}`, () =>
-            expect(entities.decode(encodedXML)).toBe(input));
-        it(`should default strict to XML encode ${encodedXML}`, () =>
-            expect(entities.decodeStrict(encodedXML)).toBe(input));
-
-        const encodedHTML5 = entities.encodeHTML5(input);
-        it(`should HTML5 encode ${input}`, () =>
-            expect(encodedHTML5).toBe(html));
-        it(`should HTML5 decode ${encodedHTML5}`, () =>
-            expect(entities.decodeHTML(encodedHTML5)).toBe(input));
-        it("should encode emojis", () =>
-            expect(entities.encodeHTML5("😄🍾🥳💥😇")).toBe(
-                "😄🍾🥳💥😇",
-            ));
-    }
-
-    it("should encode data URIs (issue #16)", () => {
-        const data =
-            "data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAALAAABAAEAAAIBRAA7";
-        expect(entities.decode(entities.encode(data))).toBe(data);
-    });
-
-    it("should HTML encode all ASCII characters", () => {
-        for (let index = 0; index < 128; index++) {
-            const char = String.fromCharCode(index);
-            const encoded = entities.encodeHTML(char);
-            const decoded = entities.decodeHTML(encoded);
-            expect(decoded).toBe(char);
-        }
-    });
-
-    it("should encode trailing parts of entities", () =>
-        expect(entities.encodeHTML("\uD835")).toBe("&#xd835;"));
-
-    it("should encode surrogate pair with first surrogate equivalent of entity, without corresponding entity", () =>
-        expect(entities.encodeHTML("\u{1D4A4}")).toBe("&#x1d4a4;"));
-});
-
-describe("encodeNonAsciiHTML", () => {
-    it("should encode all non-ASCII characters", () =>
-        expect(entities.encodeNonAsciiHTML("<test> #123! übermaßen")).toBe(
-            "&lt;test&gt; #123! &uuml;berma&szlig;en",
-        ));
-
-    it("should encode emojis", () =>
-        expect(entities.encodeNonAsciiHTML("😄🍾🥳💥😇")).toBe(
-            "&#x1f604;&#x1f37e;&#x1f973;&#x1f4a5;&#x1f607;",
-        ));
-
-    it("should encode chars above surrogates", () =>
-        expect(entities.encodeNonAsciiHTML("♒️♓️♈️♉️♊️♋️♌️♍️♎️♏️♐️♑️")).toBe(
-            "&#x2652;&#xfe0f;&#x2653;&#xfe0f;&#x2648;&#xfe0f;&#x2649;&#xfe0f;&#x264a;&#xfe0f;&#x264b;&#xfe0f;&#x264c;&#xfe0f;&#x264d;&#xfe0f;&#x264e;&#xfe0f;&#x264f;&#xfe0f;&#x2650;&#xfe0f;&#x2651;&#xfe0f;",
-        ));
-});

package/src/escape.spec.ts

@@ -1,14 +0,0 @@
-import { describe, it, expect } from "vitest";
-import * as entities from "./index.js";
-
-describe("escape HTML", () => {
-    it("should escape HTML attribute values", () =>
-        expect(entities.escapeAttribute('<a " attr > & value \u00A0!')).toBe(
-            "<a &quot; attr > &amp; value &nbsp;!",
-        ));
-
-    it("should escape HTML text", () =>
-        expect(entities.escapeText('<a " text > & value \u00A0!')).toBe(
-            '&lt;a " text &gt; &amp; value &nbsp;!',
-        ));
-});

package/src/index.spec.ts

@@ -1,125 +0,0 @@
-import { readFileSync } from "node:fs";
-import { describe, it, expect } from "vitest";
-import * as entities from "./index.js";
-import legacy from "../maps/legacy.json" with { type: "json" };
-
-const levels = ["xml", "entities"];
-
-describe("Documents", () => {
-    const levelDocuments = levels
-        .map((name) => new URL(`../maps/${name}.json`, import.meta.url))
-        .map((url) => JSON.parse(readFileSync(url, "utf8")))
-        .map((document, index) => [index, document]);
-
-    for (const [level, document] of levelDocuments) {
-        describe("Decode", () => {
-            it(levels[level], () => {
-                for (const entity of Object.keys(document)) {
-                    for (let l = level; l < levels.length; l++) {
-                        expect(entities.decode(`&${entity};`, l)).toBe(
-                            document[entity],
-                        );
-                        expect(
-                            entities.decode(`&${entity};`, { level: l }),
-                        ).toBe(document[entity]);
-                    }
-                }
-            });
-        });
-
-        describe("Decode strict", () => {
-            it(levels[level], () => {
-                for (const entity of Object.keys(document)) {
-                    for (let l = level; l < levels.length; l++) {
-                        expect(entities.decodeStrict(`&${entity};`, l)).toBe(
-                            document[entity],
-                        );
-                        expect(
-                            entities.decode(`&${entity};`, {
-                                level: l,
-                                mode: entities.DecodingMode.Strict,
-                            }),
-                        ).toBe(document[entity]);
-                    }
-                }
-            });
-        });
-
-        describe("Encode", () => {
-            it(levels[level], () => {
-                for (const entity of Object.keys(document)) {
-                    for (let l = level; l < levels.length; l++) {
-                        const encoded = entities.encode(document[entity], l);
-                        const decoded = entities.decode(encoded, l);
-                        expect(decoded).toBe(document[entity]);
-                    }
-                }
-            });
-
-            it("should only encode non-ASCII values if asked", () =>
-                expect(
-                    entities.encode("Great #'s of 🎁", {
-                        level,
-                        mode: entities.EncodingMode.ASCII,
-                    }),
-                ).toBe("Great #&apos;s of &#x1f381;"));
-        });
-    }
-
-    describe("Legacy", () => {
-        const legacyMap: Record<string, string> = legacy;
-        it("should decode", () => {
-            for (const entity of Object.keys(legacyMap)) {
-                expect(entities.decodeHTML(`&${entity}`)).toBe(
-                    legacyMap[entity],
-                );
-                expect(
-                    entities.decodeStrict(`&${entity}`, {
-                        level: entities.EntityLevel.HTML,
-                        mode: entities.DecodingMode.Legacy,
-                    }),
-                ).toBe(legacyMap[entity]);
-            }
-        });
-    });
-});
-
-const astral = [
-    ["1d306", "\uD834\uDF06"],
-    ["1d11e", "\uD834\uDD1E"],
-];
-
-const astralSpecial = [
-    ["80", "\u20AC"],
-    ["110000", "\uFFFD"],
-];
-
-describe("Astral entities", () => {
-    for (const [c, value] of astral) {
-        it(`should decode ${value}`, () =>
-            expect(entities.decode(`&#x${c};`)).toBe(value));
-
-        it(`should encode ${value}`, () =>
-            expect(entities.encode(value)).toBe(`&#x${c};`));
-
-        it(`should escape ${value}`, () =>
-            expect(entities.escape(value)).toBe(`&#x${c};`));
-    }
-
-    for (const [c, value] of astralSpecial) {
-        it(`should decode special \\u${c}`, () =>
-            expect(entities.decode(`&#x${c};`)).toBe(value));
-    }
-});
-
-describe("Escape", () => {
-    it("should always decode ASCII chars", () => {
-        for (let index = 0; index < 0x7f; index++) {
-            const c = String.fromCharCode(index);
-            expect(entities.decodeXML(entities.escape(c))).toBe(c);
-        }
-    });
-
-    it("should keep UTF8 characters", () =>
-        expect(entities.escapeUTF8('ß < "ü"')).toBe(`ß &lt; &quot;ü&quot;`));
-});