@rangojs/router 0.0.0-experimental.11 → 0.0.0-experimental.111

package/src/router/pattern-matching.ts

@@ -6,6 +6,8 @@
 
 import type { RouteEntry, TrailingSlashMode } from "../types";
 import type { EntryData } from "../server/context";
+import { debugLog, isRouterDebugEnabled } from "./logging.js";
+import { safeDecodeURIComponent } from "./url-params.js";
 
 /**
  * Parsed segment info
@@ -15,6 +17,7 @@ export interface ParsedSegment {
   value: string; // static text, param name, or "*"
   optional: boolean;
   constraint?: string[]; // enum values like ["en", "gb"]
+  suffix?: string; // literal text after param in same segment (e.g., ".html")
 }
 
 /**
@@ -37,11 +40,22 @@ export function parsePattern(pattern: string): ParsedSegment[] {
   // - :param(a|b)
   // - :param(a|b)?
   // - *
-  const segmentRegex = /\/(:([a-zA-Z_][a-zA-Z0-9_]*)(\(([^)]+)\))?(\?)?|(\*)|([^/]+))/g;
+  const segmentRegex =
+    /\/(:([a-zA-Z_][a-zA-Z0-9_]*)(\(([^)]+)\))?(\?)?([^/]*)|(\*)|([^/]+))/g;
 
   let match;
   while ((match = segmentRegex.exec(pattern)) !== null) {
-    const [, , paramName, , constraint, optional, wildcard, staticText] = match;
+    const [
+      ,
+      ,
+      paramName,
+      ,
+      constraint,
+      optional,
+      suffix,
+      wildcard,
+      staticText,
+    ] = match;
 
     if (wildcard) {
       segments.push({ type: "wildcard", value: "*", optional: false });
@@ -51,6 +65,7 @@ export function parsePattern(pattern: string): ParsedSegment[] {
         value: paramName,
         optional: optional === "?",
         constraint: constraint ? constraint.split("|") : undefined,
+        suffix: suffix || undefined,
       });
     } else if (staticText) {
       segments.push({ type: "static", value: staticText, optional: false });
@@ -60,6 +75,55 @@ export function parsePattern(pattern: string): ParsedSegment[] {
   return segments;
 }
 
+/**
+ * Compiled pattern result containing regex, param metadata, and trailing slash info.
+ */
+export interface CompiledPattern {
+  regex: RegExp;
+  paramNames: string[];
+  optionalParams: Set<string>;
+  hasTrailingSlash: boolean;
+  /**
+   * Param-name → allowed values for constrained params (e.g. `:lang(en|gb)`).
+   * Validated against the **decoded** param value after regex extraction so
+   * a URL like `/en%20GB` still matches `:lang(en GB)` — matching the trie
+   * path's behavior (trie-matching.ts:validateAndBuild).
+   */
+  constraints?: Record<string, string[]>;
+}
+
+// Module-level cache for compiled patterns. Route patterns are a finite set
+// defined at build time, so this map is bounded by the number of routes.
+const compiledPatternCache = new Map<string, CompiledPattern>();
+
+/**
+ * Get a compiled pattern from cache or compile and cache it.
+ * Avoids O(routes) regex compilations per request in the fallback path.
+ */
+export function getCompiledPattern(pattern: string): CompiledPattern {
+  let compiled = compiledPatternCache.get(pattern);
+  if (compiled) return compiled;
+  compiled = compilePattern(pattern);
+  compiledPatternCache.set(pattern, compiled);
+  return compiled;
+}
+
+/**
+ * Return the current size of the compiled pattern cache.
+ * Exposed for testing.
+ */
+export function getPatternCacheSize(): number {
+  return compiledPatternCache.size;
+}
+
+/**
+ * Clear the compiled pattern cache.
+ * Exposed for testing.
+ */
+export function clearPatternCache(): void {
+  compiledPatternCache.clear();
+}
+
 /**
  * Compile a route pattern to regex
  *
@@ -77,12 +141,7 @@ export function parsePattern(pattern: string): ParsedSegment[] {
  * compilePattern("/:locale(en|gb)/blog")  // matches /en/blog or /gb/blog
  * compilePattern("/:locale(en|gb)?/blog") // matches /blog, /en/blog, or /gb/blog
  */
-export function compilePattern(pattern: string): {
-  regex: RegExp;
-  paramNames: string[];
-  optionalParams: Set<string>;
-  hasTrailingSlash: boolean;
-} {
+export function compilePattern(pattern: string): CompiledPattern {
   // Detect if pattern has trailing slash (but not just "/")
   const hasTrailingSlash = pattern.length > 1 && pattern.endsWith("/");
   // Remove trailing slash for parsing (we'll add it back to regex if needed)
@@ -91,6 +150,7 @@ export function compilePattern(pattern: string): {
   const segments = parsePattern(normalizedPattern);
   const paramNames: string[] = [];
   const optionalParams = new Set<string>();
+  let constraints: Record<string, string[]> | undefined;
 
   let regexPattern = "";
 
@@ -100,16 +160,22 @@ export function compilePattern(pattern: string): {
       regexPattern += "/(.*)";
     } else if (segment.type === "param") {
       paramNames.push(segment.value);
-      const valuePattern = segment.constraint
-        ? `(${segment.constraint.join("|")})`
-        : "([^/]+)";
+      const suffixPattern = segment.suffix ? escapeRegex(segment.suffix) : "";
+      // Constrained params capture anything here; the allowed values are
+      // checked post-decode in findMatch so URL-encoded constraint values
+      // (e.g. `:lang(en GB)` via `/en%20GB`) still match.
+      const valuePattern = segment.suffix ? "([^/]+?)" : "([^/]+)";
+
+      if (segment.constraint) {
+        (constraints ??= {})[segment.value] = segment.constraint;
+      }
 
       if (segment.optional) {
         optionalParams.add(segment.value);
         // Optional: make the whole /segment optional
-        regexPattern += `(?:/${valuePattern})?`;
+        regexPattern += `(?:/${valuePattern}${suffixPattern})?`;
       } else {
-        regexPattern += `/${valuePattern}`;
+        regexPattern += `/${valuePattern}${suffixPattern}`;
       }
     } else {
       // Static segment
@@ -122,6 +188,20 @@ export function compilePattern(pattern: string): {
     regexPattern = "/";
   }
 
+  // Patterns of only optional segments (e.g. `/:locale?`, `/:a?/:b?`) need
+  // an explicit `/` alternative so a bare `/` matches the absent form. The
+  // optional template `(?:/X)?` matches `/X` or empty string, but pathnames
+  // are never empty. Arises from `include("/:locale?", routes)` + inner
+  // `path("/")`. Skip when an explicit trailing slash already anchors the
+  // match.
+  const hasOnlyOptionalSegments =
+    !hasTrailingSlash &&
+    segments.length > 0 &&
+    segments.every((segment) => segment.type === "param" && segment.optional);
+  if (hasOnlyOptionalSegments) {
+    regexPattern = `(?:/|${regexPattern})`;
+  }
+
   // Add trailing slash to regex if pattern has one
   if (hasTrailingSlash) {
     regexPattern += "/";
@@ -132,9 +212,35 @@ export function compilePattern(pattern: string): {
     paramNames,
     optionalParams,
     hasTrailingSlash,
+    ...(constraints ? { constraints } : {}),
   };
 }
 
+/**
+ * Validate decoded params against a compiled pattern's constraints.
+ * Returns false if any constrained param has a non-empty value not in the
+ * allowed list. Absent optionals (key missing or `undefined`) are allowed;
+ * `""` is also tolerated as "absent" so user-provided params or fixtures
+ * that pass empty strings explicitly behave the same way.
+ */
+function satisfiesConstraints(
+  params: Record<string, string>,
+  constraints: Record<string, string[]> | undefined,
+): boolean {
+  if (!constraints) return true;
+  for (const name in constraints) {
+    const value = params[name];
+    if (
+      value !== undefined &&
+      value !== "" &&
+      !constraints[name].includes(value)
+    ) {
+      return false;
+    }
+  }
+  return true;
+}
+
 /**
  * Escape special regex characters in a string
  */
@@ -142,6 +248,27 @@ function escapeRegex(str: string): string {
   return str.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
 }
 
+/**
+ * Build the named-params record from a regex match. Optional segments that
+ * didn't capture leave the corresponding group `undefined`; we skip those
+ * keys so `ctx.params.<name>` reads as `undefined` rather than `""`. This
+ * keeps the runtime aligned with the `ExtractParams` type and matches the
+ * trie matcher's contract (see `trie-matching.ts:validateAndBuild`).
+ */
+function buildParamsFromMatch(
+  match: RegExpExecArray,
+  paramNames: string[],
+): Record<string, string> {
+  const params: Record<string, string> = {};
+  paramNames.forEach((name, index) => {
+    const captured = match[index + 1];
+    if (captured !== undefined) {
+      params[name] = safeDecodeURIComponent(captured);
+    }
+  });
+  return params;
+}
+
 /**
  * Extract the static prefix from a route pattern.
  * Returns everything before the first param/wildcard.
@@ -193,8 +320,10 @@ export function extractStaticPrefix(pattern: string): string {
 /**
  * Match a pathname against registered routes
  *
- * Note: Optional params that are absent in the path will have empty string value.
- * Use the pattern definition to determine if a param is optional.
+ * Note: Optional params that are absent in the path are omitted from the
+ * returned `params` (read as `undefined`), matching the trie matcher and
+ * the `ExtractParams<"/:locale?/...">` type. Use the pattern definition or
+ * `optionalParams` to determine which keys are optional.
  *
  * Trailing slash handling (priority order):
  * 1. Per-route `trailingSlash` config from route()
@@ -239,7 +368,7 @@ export interface LazyEvaluationNeeded<TEnv = any> {
  * Type guard to check if result is a lazy evaluation needed response
  */
 export function isLazyEvaluationNeeded<TEnv>(
-  result: RouteMatchResult<TEnv> | LazyEvaluationNeeded<TEnv> | null
+  result: RouteMatchResult<TEnv> | LazyEvaluationNeeded<TEnv> | null,
 ): result is LazyEvaluationNeeded<TEnv> {
   return result !== null && "lazyEntry" in result;
 }
@@ -260,22 +389,33 @@ export function enableMatchDebug(enabled: boolean): void {
 }
 
 export function getMatchDebugStats(): MatchDebugStats {
-  return { entriesChecked: debugStats.entriesChecked, entriesSkipped: debugStats.entriesSkipped, routesChecked: debugStats.routesChecked };
+  return {
+    entriesChecked: debugStats.entriesChecked,
+    entriesSkipped: debugStats.entriesSkipped,
+    routesChecked: debugStats.routesChecked,
+  };
 }
 
 export function findMatch<TEnv>(
   pathname: string,
-  routesEntries: RouteEntry<TEnv>[]
+  routesEntries: RouteEntry<TEnv>[],
 ): RouteMatchResult<TEnv> | LazyEvaluationNeeded<TEnv> | null {
-  if (debugEnabled) {
+  const effectiveDebug = debugEnabled || isRouterDebugEnabled();
+
+  if (effectiveDebug) {
     debugStats = { entriesChecked: 0, entriesSkipped: 0, routesChecked: 0 };
-    console.log(`[findMatch] pathname="${pathname}", entries=${routesEntries.length}`);
+    debugLog("findMatch", "start", { pathname, entries: routesEntries.length });
     for (const e of routesEntries) {
-      console.log(`  entry: prefix="${e.prefix}", staticPrefix="${e.staticPrefix}", routes=${Object.keys(e.routes).length}`);
+      debugLog("findMatch", "entry", {
+        prefix: e.prefix,
+        staticPrefix: e.staticPrefix,
+        routeCount: Object.keys(e.routes).length,
+      });
     }
   }
 
-  const pathnameHasTrailingSlash = pathname.length > 1 && pathname.endsWith("/");
+  const pathnameHasTrailingSlash =
+    pathname.length > 1 && pathname.endsWith("/");
   // Try alternate pathname for redirect matching
   const alternatePathname = pathnameHasTrailingSlash
     ? pathname.slice(0, -1)
@@ -285,9 +425,12 @@ export function findMatch<TEnv>(
     // Short-circuit: skip entry if pathname doesn't start with static prefix
     // staticPrefix is pre-computed at registration time, so this is O(1)
     if (entry.staticPrefix && !pathname.startsWith(entry.staticPrefix)) {
-      if (debugEnabled) {
+      if (effectiveDebug) {
         debugStats.entriesSkipped++;
-        console.log(`  SKIP entry prefix="${entry.prefix}" (staticPrefix="${entry.staticPrefix}" doesn't match)`);
+        debugLog("findMatch", "skipped entry", {
+          prefix: entry.prefix,
+          staticPrefix: entry.staticPrefix,
+        });
       }
       continue;
     }
@@ -295,20 +438,22 @@ export function findMatch<TEnv>(
     // Check if this is a lazy entry that needs evaluation
     // When staticPrefix matches but routes are not yet populated, signal caller to evaluate
     if (entry.lazy && !entry.lazyEvaluated) {
-      if (debugEnabled) {
-        console.log(`  LAZY entry needs evaluation: staticPrefix="${entry.staticPrefix}"`);
+      if (effectiveDebug) {
+        debugLog("findMatch", "lazy entry requires evaluation", {
+          staticPrefix: entry.staticPrefix,
+        });
       }
       return { lazyEntry: entry };
     }
 
-    if (debugEnabled) {
+    if (effectiveDebug) {
       debugStats.entriesChecked++;
     }
 
     const routeEntries = Object.entries(entry.routes);
 
     for (const [routeKey, pattern] of routeEntries) {
-      if (debugEnabled) {
+      if (effectiveDebug) {
         debugStats.routesChecked++;
       }
 
@@ -322,69 +467,161 @@ export function findMatch<TEnv>(
         fullPattern = entry.prefix + pattern;
       }
 
-      const { regex, paramNames, optionalParams, hasTrailingSlash } = compilePattern(fullPattern);
+      const {
+        regex,
+        paramNames,
+        optionalParams,
+        hasTrailingSlash,
+        constraints,
+      } = getCompiledPattern(fullPattern);
 
       // Get trailing slash mode for this route (per-route config or pattern-based)
-      const trailingSlashMode: TrailingSlashMode | undefined = entry.trailingSlash?.[routeKey];
-
+      const trailingSlashMode: TrailingSlashMode | undefined =
+        entry.trailingSlash?.[routeKey];
 
       // Prerender flag from entry metadata (set by urls() for prerender handlers)
-      const prFlag = entry.prerenderRouteKeys?.has(routeKey) ? { pr: true as const } : {};
+      const prFlag = entry.prerenderRouteKeys?.has(routeKey)
+        ? { pr: true as const }
+        : {};
+      const ptFlag = entry.passthroughRouteKeys?.has(routeKey)
+        ? { pt: true as const }
+        : {};
 
       // Try exact match first
       const match = regex.exec(pathname);
       if (match) {
-        const params: Record<string, string> = {};
-        paramNames.forEach((name, index) => {
-          params[name] = match[index + 1] ?? "";
-        });
+        const params = buildParamsFromMatch(match, paramNames);
+
+        // Validate constraints against decoded values; a failure falls
+        // through to the next route so other patterns can still match.
+        if (!satisfiesConstraints(params, constraints)) {
+          continue;
+        }
 
-        if (debugEnabled) {
-          console.log(`  MATCH: routeKey="${routeKey}", pattern="${fullPattern}"`);
-          console.log(`  Stats: entriesChecked=${debugStats.entriesChecked}, entriesSkipped=${debugStats.entriesSkipped}, routesChecked=${debugStats.routesChecked}`);
+        if (effectiveDebug) {
+          debugLog("findMatch", "matched route", {
+            routeKey,
+            pattern: fullPattern,
+            stats: { ...debugStats },
+          });
         }
 
         // Check if trailing slash mode requires redirect even on exact match
-        if (trailingSlashMode === "always" && !pathnameHasTrailingSlash && pathname !== "/") {
+        if (
+          trailingSlashMode === "always" &&
+          !pathnameHasTrailingSlash &&
+          pathname !== "/"
+        ) {
           // Mode says always have trailing slash, but pathname doesn't have it
-          return { entry, routeKey, params, optionalParams, redirectTo: pathname + "/", ...prFlag };
+          return {
+            entry,
+            routeKey,
+            params,
+            optionalParams,
+            redirectTo: pathname + "/",
+            ...prFlag,
+            ...ptFlag,
+          };
         } else if (trailingSlashMode === "never" && pathnameHasTrailingSlash) {
           // Mode says never have trailing slash, but pathname has it
-          return { entry, routeKey, params, optionalParams, redirectTo: pathname.slice(0, -1), ...prFlag };
+          return {
+            entry,
+            routeKey,
+            params,
+            optionalParams,
+            redirectTo: pathname.slice(0, -1),
+            ...prFlag,
+            ...ptFlag,
+          };
         }
 
-        return { entry, routeKey, params, optionalParams, ...prFlag };
+        return {
+          entry,
+          routeKey,
+          params,
+          optionalParams,
+          ...prFlag,
+          ...ptFlag,
+        };
       }
 
       // Try alternate pathname (opposite trailing slash)
       const altMatch = regex.exec(alternatePathname);
       if (altMatch) {
-        const params: Record<string, string> = {};
-        paramNames.forEach((name, index) => {
-          params[name] = altMatch[index + 1] ?? "";
-        });
+        const params = buildParamsFromMatch(altMatch, paramNames);
+
+        if (!satisfiesConstraints(params, constraints)) {
+          continue;
+        }
 
         // Determine redirect behavior based on mode
         if (trailingSlashMode === "ignore") {
           // Match without redirect
-          return { entry, routeKey, params, optionalParams, ...prFlag };
+          return {
+            entry,
+            routeKey,
+            params,
+            optionalParams,
+            ...prFlag,
+            ...ptFlag,
+          };
         } else if (trailingSlashMode === "never") {
           // Redirect to no trailing slash
           if (pathnameHasTrailingSlash) {
-            return { entry, routeKey, params, optionalParams, redirectTo: alternatePathname, ...prFlag };
+            return {
+              entry,
+              routeKey,
+              params,
+              optionalParams,
+              redirectTo: alternatePathname,
+              ...prFlag,
+              ...ptFlag,
+            };
           }
-          return { entry, routeKey, params, optionalParams, ...prFlag };
+          return {
+            entry,
+            routeKey,
+            params,
+            optionalParams,
+            ...prFlag,
+            ...ptFlag,
+          };
         } else if (trailingSlashMode === "always") {
           // Redirect to with trailing slash
           if (!pathnameHasTrailingSlash) {
-            return { entry, routeKey, params, optionalParams, redirectTo: alternatePathname, ...prFlag };
+            return {
+              entry,
+              routeKey,
+              params,
+              optionalParams,
+              redirectTo: alternatePathname,
+              ...prFlag,
+              ...ptFlag,
+            };
           }
-          return { entry, routeKey, params, optionalParams, ...prFlag };
+          return {
+            entry,
+            routeKey,
+            params,
+            optionalParams,
+            ...prFlag,
+            ...ptFlag,
+          };
         } else {
           // No explicit mode - use pattern-based detection
           // Redirect to canonical form (what the pattern defines)
-          const canonicalPath = hasTrailingSlash ? alternatePathname : pathname.slice(0, -1);
-          return { entry, routeKey, params, optionalParams, redirectTo: canonicalPath, ...prFlag };
+          const canonicalPath = hasTrailingSlash
+            ? alternatePathname
+            : pathname.slice(0, -1);
+          return {
+            entry,
+            routeKey,
+            params,
+            optionalParams,
+            redirectTo: canonicalPath,
+            ...prFlag,
+            ...ptFlag,
+          };
         }
       }
     }