From: Lady <redacted>
Date: Sat, 21 May 2022 01:09:20 +0000 (-0700)
Subject: L·E·I·R·I support
X-Git-Tag: 0.1.0~11
X-Git-Url: https://git.ladys.computer/Pisces/commitdiff_plain/1c80eb9624a66d32800885321f37c4b05161d607?ds=inline;hp=edbef3a179c4404eb02a83860a76c1f11b9b63d2

L·E·I·R·I support
---

diff --git a/iri.js b/iri.js
index f1abe78..b9682d9 100644
--- a/iri.js
+++ b/iri.js
@@ -42,7 +42,6 @@ const path·noscheme = String.raw
 const path·absolute = String.raw
   `/(?:(?:${segment·nz})(?:/(?:${segment}))*)?`;
 const path·abempty = String.raw`(?:/(?:${segment}))*`;
-//deno-lint-ignore no-unused-vars
 const path = String.raw
   `${path·abempty}|${path·absolute}|${path·noscheme}|${path·rootless}|${path·empty}`;
 const reg·name = String.raw
@@ -85,7 +84,6 @@ const ipath·noscheme = String.raw
 const ipath·absolute = String.raw
   `/(?:(?:${isegment·nz})(?:/(?:${isegment}))*)?`;
 const ipath·abempty = String.raw`(?:/(?:${isegment}))*`;
-//deno-lint-ignore no-unused-vars
 const ipath = String.raw
   `${ipath·abempty}|${ipath·absolute}|${ipath·noscheme}|${ipath·rootless}|${ipath·empty}`;
 const ireg·name = String.raw
@@ -107,21 +105,88 @@ const IRI = String.raw
   `(?:${scheme}):(?:${ihier·part})(?:\?(?:${iquery}))?(?:#(?:${ifragment}))?`;
 const IRI·reference = String.raw`(?:${IRI})|(?:${irelative·ref})`;
 
+const leiri·iprivate = String.raw
+  `[\u{E000}-\u{F8FF}\u{E0000}-\u{E0FFF}\u{F0000}-\u{FFFFD}\u{100000}-\u{10FFFD}]`;
+const leiri·ucschar = String.raw
+  `[ <>"{}|\\^${"`"}\u{0}-\u{1F}\u{7F}-\u{D7FF}\u{E000}-\u{FFFD}\u{10000}-\u{10FFFF}]`;
+const leiri·iunreserved = String.raw
+  `[A-Za-z0-9\-\._~]|${leiri·ucschar}`;
+const leiri·ipchar = String.raw
+  `${leiri·iunreserved}|${pct·encoded}|${sub·delims}|[:@]`;
+const leiri·ifragment = String.raw`(?:${leiri·ipchar}|[/?])*`;
+const leiri·iquery = String.raw
+  `(?:${leiri·ipchar}|${leiri·iprivate}|[/?])*`;
+const leiri·isegment·nz·nc = String.raw
+  `(?:${leiri·iunreserved}|${pct·encoded}|${sub·delims}|@)+`;
+const leiri·isegment·nz = String.raw`(?:${leiri·ipchar})+`;
+const leiri·isegment = String.raw`(?:${leiri·ipchar})*`;
+const leiri·ipath·empty = String.raw``;
+const leiri·ipath·rootless = String.raw
+  `(?:${leiri·isegment·nz})(?:/(?:${leiri·isegment}))*`;
+const leiri·ipath·noscheme = String.raw
+  `(?:${leiri·isegment·nz·nc})(?:/(?:${leiri·isegment}))*`;
+const leiri·ipath·absolute = String.raw
+  `/(?:(?:${leiri·isegment·nz})(?:/(?:${leiri·isegment}))*)?`;
+const leiri·ipath·abempty = String.raw`(?:/(?:${leiri·isegment}))*`;
+const leiri·ipath = String.raw
+  `${leiri·ipath·abempty}|${leiri·ipath·absolute}|${leiri·ipath·noscheme}|${leiri·ipath·rootless}|${leiri·ipath·empty}`;
+const leiri·ireg·name = String.raw
+  `(?:${leiri·iunreserved}|${pct·encoded}|${sub·delims})*`;
+const leiri·ihost = String.raw
+  `${IP·literal}|${IPv4address}|${leiri·ireg·name}`;
+const leiri·iuserinfo = String.raw
+  `(?:${leiri·iunreserved}|${pct·encoded}|${sub·delims}|:)*`;
+const leiri·iauthority = String.raw
+  `(?:(?:${leiri·iuserinfo})@)?(?:${leiri·ihost})(?::(?:${port}))?`;
+const leiri·irelative·part = String.raw
+  `//(?:${leiri·iauthority})(?:${leiri·ipath·abempty})|(?:${leiri·ipath·absolute})|(?:${leiri·ipath·noscheme})|(?:${leiri·ipath·empty})`;
+const leiri·irelative·ref = String.raw
+  `(?:${leiri·irelative·part})(?:\?(?:${leiri·iquery}))?(?:#(?:${leiri·ifragment}))?`;
+const leiri·ihier·part = String.raw
+  `//(?:${leiri·iauthority})(?:${leiri·ipath·abempty})|(?:${leiri·ipath·absolute})|(?:${leiri·ipath·rootless})|(?:${leiri·ipath·empty})`;
+const absolute·LEIRI = String.raw
+  `(?:${scheme}):(?:${leiri·ihier·part})(?:\?(?:${leiri·iquery}))?`;
+const LEIRI = String.raw
+  `(?:${scheme}):(?:${leiri·ihier·part})(?:\?(?:${leiri·iquery}))?(?:#(?:${leiri·ifragment}))?`;
+const LEIRI·reference = String.raw
+  `(?:${LEIRI})|(?:${leiri·irelative·ref})`;
+
 export const {
   isAbsoluteURI, // U·R·I with no fragment
   isURI,
+  isURIPath,
   isURIReference,
+  isURISuffix, // only authority, path, query, fragment
   isAbsoluteIRI, // I·R·I with no fragment
   isIRI,
+  isIRIPath,
   isIRIReference,
+  isIRISuffix, // only authority, path, query, fragment
+  isAbsoluteLEIRI, // L·E·I·R·I with no fragment
+  isLEIRI,
+  isLEIRIPath,
+  isLEIRIReference,
+  isLEIRISuffix, // only authority, path, query, fragment
 } = Object.fromEntries(
   Object.entries({
+    isAbsoluteLEIRI: absolute·LEIRI,
     isAbsoluteIRI: absolute·IRI,
     isAbsoluteURI: absolute·URI,
+    isLEIRI: LEIRI,
+    isLEIRIPath: leiri·ipath,
+    isLEIRIReference: LEIRI·reference,
+    isLEIRISuffix: String.raw
+      `(?:${leiri·iauthority})(?:${leiri·ipath·abempty})(?:\?(?:${leiri·iquery}))?(?:#(?:${leiri·ifragment}))?`,
     isIRI: IRI,
+    isIRIPath: ipath,
     isIRIReference: IRI·reference,
+    isIRISuffix: String.raw
+      `(?:${iauthority})(?:${ipath·abempty})(?:\?(?:${iquery}))?(?:#(?:${ifragment}))?`,
     isURI: URI,
+    isURIPath: path,
     isURIReference: URI·reference,
+    isURISuffix: String.raw
+      `(?:${authority})(?:${path·abempty})(?:\?(?:${query}))?(?:#(?:${fragment}))?`,
   }).map(([key, value]) => {
     const regExp = new RegExp(`^(?:${value})$`, "u");
     return [
@@ -143,7 +208,7 @@ export const {
 );
 
 /**
- * Recomposes an I·R·I reference from its component parts.
+ * Recomposes an (L·E·)I·R·I reference from its component parts.
  *
  * See §5.3 of R·F·C 3986.
  */
@@ -151,44 +216,115 @@ export const composeReference = ($) => {
   const result = [];
   const { scheme, authority, path, query, fragment } = $;
   if (scheme != null) {
+    // A scheme is present.
     result.push(scheme, ":");
   } else {
+    // No scheme is present.
     /* do nothing */
   }
   if (authority != null) {
+    // An authority is present.
     result.push("//", authority);
   } else {
+    // No authority is present.
     /* do nothing */
   }
   result.push(path ?? "");
   if (query != null) {
+    // A query is present.
     result.push("?", query);
   } else {
+    // No query is present.
     /* do nothing */
   }
   if (fragment != null) {
+    // A fragment is present.
     result.push("#", fragment);
   } else {
+    // No fragment is present.
     /* do nothing */
   }
   return result.join("");
 };
 
 /**
- * Converts an I·R·I to the corresponding U·R·I by percent‐encoding
+ * Converts an L·E·I·R·I to the corresponding I·R·I by percent‐encoding
  * unsupported characters.
  *
+ * This function is somewhat complex because the I·R·I syntax allows
+ * private·use characters *only* in the query.
+ */
+export const escapeForIRI = ($) => {
+  const components = parseReference($);
+  const encoder = new TextEncoder();
+
+  // The path will always be present (although perhaps empty) on a
+  // successful parse. If it isn’t (and parsing failed), treat the
+  // entire input as the path.
+  components.path ??= `${$}`;
+
+  // Escape disallowed codepoints in each component and compose an
+  // I·R·I from the result.
+  return composeReference(
+    Object.fromEntries(
+      Object.entries(components).map(
+        ([componentName, componentValue]) => [
+          componentName,
+          componentValue == null ? undefined : [...function* () {
+            for (const character of componentValue) {
+              if (
+                new RegExp(`${leiri·ucschar}|${leiri·iprivate}`, "u")
+                  .test(
+                    character,
+                  ) &&
+                !new RegExp(
+                  `${ucschar}${
+                    componentName == "query" ? `|${iprivate}` : ""
+                  }`,
+                  "u",
+                ).test(character)
+              ) {
+                // This codepoint needs to be escaped.
+                for (const byte of encoder.encode(character)) {
+                  yield `%${byte < 0x10 ? "0" : ""}${
+                    byte.toString(0x10).toUpperCase()
+                  }`;
+                }
+              } else {
+                // This codepoint does not need escaping.
+                yield character;
+              }
+            }
+          }()].join(""),
+        ],
+      ),
+    ),
+  );
+};
+
+/**
+ * Converts an (L·E·)I·R·I to the corresponding U·R·I by
+ * percent‐encoding unsupported characters.
+ *
  * This does not punycode the authority.
  */
-export const iri2uri = ($) =>
+export const escapeForURI = ($) =>
   [...function* () {
     const encoder = new TextEncoder();
-    for (const character of $) {
-      if (new RegExp(`${ucschar}|${iprivate}`, "u").test(character)) {
+    for (const character of `${$}`) {
+      if (
+        new RegExp(`${leiri·ucschar}|${leiri·iprivate}`, "u").test(
+          character,
+        )
+      ) {
+        // This codepoint needs to be escaped.
         for (const byte of encoder.encode(character)) {
-          yield `%${byte.toString(0x10).toUpperCase()}`;
+          yield `%${byte < 0x10 ? "0" : ""}${
+            byte.toString(0x10).toUpperCase()
+          }`;
         }
       } else {
+        // This codepoint doesn’t need escaping.
         yield character;
       }
     }
@@ -208,12 +344,15 @@ export const mergePaths = (base, reference) => {
 
 /**
  * Returns the `scheme`, `authority`, `path`, `query`, and `fragment`
- * of the provided I·R·I reference.
+ * of the provided (L·E·)I·R·I reference.
+ *
+ * `path` will always be defined for valid references, and will be
+ * undefined for values which are not valid L·E·I·R·Is.
  */
 export const parseReference = ($) => {
   const regExp = new RegExp(
     String.raw
-      `^(?:(?<absolute·scheme>${scheme}):(?://(?<absolute·authority>${iauthority})(?<absolute·patha>${ipath·abempty})|(?<absolute·pathb>(?:${ipath·absolute})|(?:${ipath·rootless})|(?:${ipath·empty})))(?:\?(?<absolute·query>${iquery}))?(?:#(?<absolute·fragment>${ifragment}))?|(?://(?<relative·authority>${iauthority})(?<relative·patha>${ipath·abempty})|(?<relative·pathb>(?:${ipath·absolute})|(?:${ipath·noscheme})|(?:${ipath·empty})))(?:\?(?<relative·query>${iquery}))?(?:#(?<relative·fragment>${ifragment}))?)$`,
+      `^(?:(?<absolute·scheme>${scheme}):(?://(?<absolute·authority>${leiri·iauthority})(?<absolute·patha>${leiri·ipath·abempty})|(?<absolute·pathb>(?:${leiri·ipath·absolute})|(?:${leiri·ipath·rootless})|(?:${leiri·ipath·empty})))(?:\?(?<absolute·query>${leiri·iquery}))?(?:#(?<absolute·fragment>${leiri·ifragment}))?|(?://(?<relative·authority>${leiri·iauthority})(?<relative·patha>${leiri·ipath·abempty})|(?<relative·pathb>(?:${leiri·ipath·absolute})|(?:${leiri·ipath·noscheme})|(?:${leiri·ipath·empty})))(?:\?(?<relative·query>${leiri·iquery}))?(?:#(?<relative·fragment>${leiri·ifragment}))?)$`,
     "u",
   );
   const {
@@ -240,7 +379,7 @@ export const parseReference = ($) => {
 };
 
 /**
- * Removes all dot segments ("." or "..") from the provided I·R·I.
+ * Removes all dot segments ("." or "..") from the provided (L·E·)I·R·I.
  *
  * See §5.2.4 of R·F·C 3986.
  */
@@ -310,7 +449,8 @@ export const removeDotSegments = ($) => {
 };
 
 /**
- * Resolves the provided reference relative to the provided base I·R·I.
+ * Resolves the provided reference relative to the provided base
+ * (L·E·)I·R·I.
  *
  * See §5.2 of R·F·C 3986.
  */
@@ -322,10 +462,13 @@ export const resolveReference = (R, Base = location ?? "") => {
     query: Base·query,
   } = parseReference(Base);
   if (Base·scheme == null) {
+    // Base I·R·I’s must be valid I·R·I’s, meaning they must have a
+    // scheme.
     throw new TypeError(
-      `Piscēs: Base I·R·I did not have a scheme: ${Base}.`,
+      `Piscēs: Base did not have a scheme: ${Base}.`,
     );
   } else {
+    // The provided Base I·R·I is valid.
     const {
       scheme: R·scheme,
       authority: R·authority,
diff --git a/iri.test.js b/iri.test.js
index 51749cd..a586629 100644
--- a/iri.test.js
+++ b/iri.test.js
@@ -9,24 +9,24 @@
 
 import {
   composeReference,
-  iri2uri,
+  escapeForIRI,
+  escapeForURI,
   isAbsoluteIRI,
+  isAbsoluteLEIRI,
   isAbsoluteURI,
   isIRI,
   isIRIReference,
+  isLEIRI,
+  isLEIRIReference,
   isURI,
   isURIReference,
   parseReference,
   removeDotSegments,
   resolveReference,
 } from "./iri.js";
-import {
-  assert,
-  assertEquals,
-  assertStrictEquals,
-} from "./dev-deps.js";
+import { assertEquals, assertStrictEquals } from "./dev-deps.js";
 
-const exampleURIs = {
+const exampleURIReferences = {
   "ftp://ftp.is.co.za/rfc/rfc1808.txt": {
     scheme: "ftp",
     authority: "ftp.is.co.za",
@@ -71,17 +71,13 @@ const exampleURIs = {
     query: "name=ferret",
     fragment: "nose",
   },
-};
-
-const exampleURIReferences = {
-  ...exampleURIs,
   "./this:that": {
     path: "./this:that",
   },
 };
 
-const exampleIRIs = {
-  ...exampleURIs,
+const exampleIRIReferences = {
+  ...exampleURIReferences,
   "http://ヒキワリ.ナットウ.ニホン": {
     scheme: "http",
     authority: "ヒキワリ.ナットウ.ニホン",
@@ -92,18 +88,40 @@ const exampleIRIs = {
     authority: "JP納豆.例.jp",
     path: "/dir1/引き割り.html",
   },
+  "/dir1/引き割り.html": {
+    path: "/dir1/引き割り.html",
+  },
 };
 
-const exampleIRIReferences = {
-  ...exampleURIReferences,
-  ...exampleIRIs,
+const exampleLEIRIReferences = {
+  ...exampleIRIReferences,
+  "http://example.com/ foo /": {
+    scheme: "http",
+    authority: "example.com",
+    path: "/ foo /",
+  },
+  "\0": {
+    path: "\0",
+  },
+};
+
+const exampleReferences = {
+  ...exampleLEIRIReferences,
+  "\uD800": {},
+  "\uFFFE": {},
+  "\uFFFF": {},
 };
 
 Deno.test({
   name: "Identifies U·R·Is.",
   fn: () => {
-    for (const uri of Object.keys(exampleURIs)) {
-      assert(isURI(uri));
+    for (
+      const [uri, { scheme }] of Object.entries(exampleReferences)
+    ) {
+      assertStrictEquals(
+        isURI(uri),
+        uri in exampleURIReferences && scheme != null,
+      );
     }
   },
 });
@@ -111,8 +129,16 @@ Deno.test({
 Deno.test({
   name: "Identifies absolute U·R·Is.",
   fn: () => {
-    for (const [uri, { fragment }] of Object.entries(exampleURIs)) {
-      assertStrictEquals(isAbsoluteURI(uri), fragment == null);
+    for (
+      const [uri, { scheme, fragment }] of Object.entries(
+        exampleReferences,
+      )
+    ) {
+      assertStrictEquals(
+        isAbsoluteURI(uri),
+        uri in exampleURIReferences && scheme != null &&
+          fragment == null,
+      );
     }
   },
 });
@@ -120,8 +146,11 @@ Deno.test({
 Deno.test({
   name: "Identifies U·R·I references.",
   fn: () => {
-    for (const uri of Object.keys(exampleURIReferences)) {
-      assert(isURIReference(uri));
+    for (const uri of Object.keys(exampleReferences)) {
+      assertStrictEquals(
+        isURIReference(uri),
+        uri in exampleURIReferences,
+      );
     }
   },
 });
@@ -129,8 +158,13 @@ Deno.test({
 Deno.test({
   name: "Identifies I·R·Is.",
   fn: () => {
-    for (const iri of Object.keys(exampleIRIs)) {
-      assert(isIRI(iri));
+    for (
+      const [iri, { scheme }] of Object.entries(exampleReferences)
+    ) {
+      assertStrictEquals(
+        isIRI(iri),
+        iri in exampleIRIReferences && scheme != null,
+      );
     }
   },
 });
@@ -138,8 +172,16 @@ Deno.test({
 Deno.test({
   name: "Identifies absolute I·R·Is.",
   fn: () => {
-    for (const [iri, { fragment }] of Object.entries(exampleIRIs)) {
-      assertStrictEquals(isAbsoluteIRI(iri), fragment == null);
+    for (
+      const [iri, { scheme, fragment }] of Object.entries(
+        exampleReferences,
+      )
+    ) {
+      assertStrictEquals(
+        isAbsoluteIRI(iri),
+        iri in exampleIRIReferences && scheme != null &&
+          fragment == null,
+      );
     }
   },
 });
@@ -147,8 +189,54 @@ Deno.test({
 Deno.test({
   name: "Identifies I·R·I references.",
   fn: () => {
-    for (const iri of Object.keys(exampleIRIReferences)) {
-      assert(isIRIReference(iri));
+    for (const iri of Object.keys(exampleReferences)) {
+      assertStrictEquals(
+        isIRIReference(iri),
+        iri in exampleIRIReferences,
+      );
+    }
+  },
+});
+
+Deno.test({
+  name: "Identifies L·E·I·R·Is.",
+  fn: () => {
+    for (
+      const [leiri, { scheme }] of Object.entries(exampleReferences)
+    ) {
+      assertStrictEquals(
+        isLEIRI(leiri),
+        leiri in exampleLEIRIReferences && scheme != null,
+      );
+    }
+  },
+});
+
+Deno.test({
+  name: "Identifies absolute L·E·I·R·Is.",
+  fn: () => {
+    for (
+      const [leiri, { scheme, fragment }] of Object.entries(
+        exampleReferences,
+      )
+    ) {
+      assertStrictEquals(
+        isAbsoluteLEIRI(leiri),
+        leiri in exampleLEIRIReferences && scheme != null &&
+          fragment == null,
+      );
+    }
+  },
+});
+
+Deno.test({
+  name: "Identifies L·E·I·R·I references.",
+  fn: () => {
+    for (const leiri of Object.keys(exampleReferences)) {
+      assertStrictEquals(
+        isLEIRIReference(leiri),
+        leiri in exampleLEIRIReferences,
+      );
     }
   },
 });
@@ -156,7 +244,7 @@ Deno.test({
 Deno.test({
   name: "Correctly parses references.",
   fn: () => {
-    for (const [iri, value] of Object.entries(exampleIRIReferences)) {
+    for (const [iri, value] of Object.entries(exampleReferences)) {
       assertEquals(parseReference(iri), {
         scheme: undefined,
         authority: undefined,
@@ -172,19 +260,43 @@ Deno.test({
 Deno.test({
   name: "Correctly composes references.",
   fn: () => {
-    for (const [iri, value] of Object.entries(exampleIRIReferences)) {
+    for (
+      const [iri, value] of Object.entries(exampleLEIRIReferences)
+    ) {
       assertStrictEquals(composeReference(value), iri);
     }
   },
 });
 
 Deno.test({
-  name: "Converts IRIs to URIs.",
+  name: "Converts (L·E·)I·R·Is to U·R·Is.",
   fn: () => {
     assertStrictEquals(
-      iri2uri("/dir1/引き割り.html"),
+      escapeForURI("/dir1/引き割り.html"),
       "/dir1/%E5%BC%95%E3%81%8D%E5%89%B2%E3%82%8A.html",
     );
+    assertStrictEquals(
+      escapeForURI(" æ\0"),
+      "%20%C3%A6%00",
+    );
+    assertStrictEquals(
+      escapeForURI("\u{F0000}?\u{F0000}#\u{F0000}"),
+      "%F3%B0%80%80?%F3%B0%80%80#%F3%B0%80%80",
+    );
+  },
+});
+
+Deno.test({
+  name: "Converts L·E·I·R·Is to I·R·Is.",
+  fn: () => {
+    assertStrictEquals(
+      escapeForIRI(" æ\0"),
+      "%20æ%00",
+    );
+    assertStrictEquals(
+      escapeForIRI("\u{F0000}?\u{F0000}#\u{F0000}"),
+      "%F3%B0%80%80?\u{F0000}#%F3%B0%80%80",
+    );
   },
 });