|
| 1 | +package io.modelcontextprotocol.kotlin.sdk.utils |
| 2 | + |
| 3 | +import io.github.oshai.kotlinlogging.KotlinLogging |
| 4 | +import io.ktor.http.decodeURLPart |
| 5 | +import io.modelcontextprotocol.kotlin.sdk.types.ResourceTemplate |
| 6 | +import kotlinx.collections.immutable.toImmutableMap |
| 7 | +import kotlin.jvm.JvmOverloads |
| 8 | +import kotlin.jvm.JvmStatic |
| 9 | + |
| 10 | +// Max URL/path length to prevent DoS or unexpected large payloads. |
| 11 | +private const val MAX_URL_LENGTH = 2048 |
| 12 | + |
| 13 | +// Max template/uri depth to prevent overly nested or complex templates. |
| 14 | +private const val MAX_DEPTH = 50 |
| 15 | + |
| 16 | +// Literal segments are more specific than variable captures. |
| 17 | +private const val LITERAL_MATCH_SCORE = 2 |
| 18 | +private const val VARIABLE_MATCH_SCORE = 1 |
| 19 | + |
| 20 | +/** |
| 21 | + * A [ResourceTemplateMatcher] that matches resource URIs against an RFC 6570 Level 1 |
| 22 | + * URI template by splitting both the URI and the template on `/` and comparing each |
| 23 | + * segment in order. |
| 24 | + * |
| 25 | + * ### Supported template syntax |
| 26 | + * |
| 27 | + * Only RFC 6570 **Level 1** is supported: simple `{variable}` expressions where the |
| 28 | + * entire path segment is a single variable. Operator expressions (`{+var}`, `{#var}`, |
| 29 | + * `{.var}`, `{/var}`, etc.) and multi-variable expressions (`{a,b}`) are **not** |
| 30 | + * recognized — segments containing them are treated as literals. |
| 31 | + * |
| 32 | + * ### Matching rules |
| 33 | + * |
| 34 | + * - The URI and template must have the same number of `/`-delimited segments. |
| 35 | + * - Literal segments must match exactly (after percent-decoding the URI segment). |
| 36 | + * - `{variable}` segments capture the percent-decoded URI segment value. |
| 37 | + * - Query strings and fragments (`?`, `#`) are **not** stripped — they become part of |
| 38 | + * the captured variable value for the segment that contains them. |
| 39 | + * |
| 40 | + * ### Specificity scoring |
| 41 | + * |
| 42 | + * When multiple templates match the same URI, each matched literal segment contributes |
| 43 | + * 2 points and each variable capture contributes 1 point. The highest-scoring match wins. |
| 44 | + * |
| 45 | + * ### Safety limits |
| 46 | + * |
| 47 | + * - URIs longer than [maxUriLength] characters are rejected. |
| 48 | + * - Templates and URIs with more than [maxDepth] segments are rejected. |
| 49 | + * |
| 50 | + * ### Security contract for handler authors |
| 51 | + * |
| 52 | + * Values in [MatchResult.variables] are attacker-controlled strings extracted from |
| 53 | + * the incoming URI. They are percent-decoded (one pass only) before being returned. |
| 54 | + * Handlers **must** treat them as untrusted input and validate or sanitize them |
| 55 | + * before using them to construct file paths, database queries, downstream URLs, or |
| 56 | + * any other security-sensitive operation. |
| 57 | + * |
| 58 | + * In particular: |
| 59 | + * - A decoded value may contain `/`, `..`, null bytes (`\u0000`), `?`, or `#`. |
| 60 | + * - `%252F` in the URI becomes `%2F` in the variable — exactly one decode pass is applied. |
| 61 | + * |
| 62 | + * ### Platform limitations |
| 63 | + * |
| 64 | + * Dot-segment normalization (resolving `..` and `.` in the URI path) is performed |
| 65 | + * on JVM and JS targets using the platform-native URI parser. On native and WASM targets |
| 66 | + * no normalizer is available, so dot-segment traversal is **not** mitigated at this |
| 67 | + * layer — handlers on those targets must normalize paths themselves. |
| 68 | + * |
| 69 | + * @param resourceTemplate The resource template to match against. Must follow RFC 6570 Level 1 syntax. |
| 70 | + * @param maxUriLength Maximum allowed length for incoming URIs. Defaults to 2048. |
| 71 | + * @param maxDepth Maximum allowed segment count for the template/uri. Defaults to 50. |
| 72 | + * |
| 73 | + * @property resourceTemplate The resource template against which resource URIs will be matched. |
| 74 | + */ |
| 75 | +public class PathSegmentTemplateMatcher @JvmOverloads constructor( |
| 76 | + override val resourceTemplate: ResourceTemplate, |
| 77 | + private val maxDepth: Int = MAX_DEPTH, |
| 78 | + private val maxUriLength: Int = MAX_URL_LENGTH, |
| 79 | +) : ResourceTemplateMatcher { |
| 80 | + |
| 81 | + public companion object { |
| 82 | + /** |
| 83 | + * A [ResourceTemplateMatcherFactory] that creates [PathSegmentTemplateMatcher] instances |
| 84 | + * with default limits. Pass this to [io.modelcontextprotocol.kotlin.sdk.server.ServerOptions] |
| 85 | + * to use path-segment matching, or supply a custom factory to override the matching strategy. |
| 86 | + */ |
| 87 | + @JvmStatic |
| 88 | + public val factory: ResourceTemplateMatcherFactory = ResourceTemplateMatcherFactory { |
| 89 | + PathSegmentTemplateMatcher(it) |
| 90 | + } |
| 91 | + |
| 92 | + @JvmStatic |
| 93 | + private val logger = KotlinLogging.logger {} |
| 94 | + } |
| 95 | + |
| 96 | + private val templateParts: List<String> |
| 97 | + |
| 98 | + // Maps segment index to variable name; indices absent from this map are literal segments. |
| 99 | + private val variableIndices: Map<Int, String> |
| 100 | + |
| 101 | + init { |
| 102 | + val template = resourceTemplate.uriTemplate |
| 103 | + require(template.isNotBlank()) { "Resource template cannot be blank" } |
| 104 | + templateParts = template.trim('/').split("/") |
| 105 | + require(templateParts.size <= maxDepth) { |
| 106 | + "Template is too complex (max depth=$maxDepth)" |
| 107 | + } |
| 108 | + |
| 109 | + val vars = mutableMapOf<Int, String>() |
| 110 | + for (i in templateParts.indices) { |
| 111 | + val segment = templateParts[i] |
| 112 | + if (segment.startsWith("{") && segment.endsWith("}")) { |
| 113 | + val name = segment.removeSurrounding("{", "}").trim() |
| 114 | + require(name.isNotEmpty()) { "Invalid variable name in template: $segment" } |
| 115 | + vars[i] = name |
| 116 | + } |
| 117 | + } |
| 118 | + variableIndices = vars.toImmutableMap() |
| 119 | + } |
| 120 | + |
| 121 | + @Suppress("ReturnCount") |
| 122 | + override fun match(resourceUri: String): MatchResult? { |
| 123 | + if (resourceUri.length > maxUriLength) { |
| 124 | + logger.debug { "URL is too long (max=$maxUriLength)" } |
| 125 | + return null |
| 126 | + } |
| 127 | + |
| 128 | + // Resolve dot-segments (e.g. /a/../b → /a/b) before splitting. |
| 129 | + // Prevents traversal attacks on JVM/JS; no-op on native/WASM targets. |
| 130 | + val normalized = normalizeUri(resourceUri) |
| 131 | + |
| 132 | + val urlParts = normalized.trim('/').split("/") |
| 133 | + if (urlParts.size > maxDepth) { |
| 134 | + logger.debug { "URI has too many segments (max=$maxDepth)" } |
| 135 | + return null |
| 136 | + } |
| 137 | + if (urlParts.size != templateParts.size) return null |
| 138 | + |
| 139 | + val variables = mutableMapOf<String, String>() |
| 140 | + var score = 0 |
| 141 | + |
| 142 | + for (i in templateParts.indices) { |
| 143 | + val urlSegment = urlParts[i].decodeURLPart() |
| 144 | + val variableName = variableIndices[i] |
| 145 | + if (variableName != null) { |
| 146 | + variables[variableName] = urlSegment |
| 147 | + score += VARIABLE_MATCH_SCORE |
| 148 | + } else if (templateParts[i] == urlSegment) { |
| 149 | + score += LITERAL_MATCH_SCORE |
| 150 | + } else { |
| 151 | + return null |
| 152 | + } |
| 153 | + } |
| 154 | + |
| 155 | + return MatchResult(variables = variables.toImmutableMap(), score = score) |
| 156 | + } |
| 157 | +} |
0 commit comments