ask-llm-providers 0.8.0 → 0.8.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 3cd9e34fd4befbdd393f4b4496b0d0e5dde254dc2ef10c3d86bb7043f81406de
-  data.tar.gz: 6de8ecfb85b0aa79f5623a0968ea8a787dc8c64ebce9da6d5d49122ef82e037c
+  metadata.gz: 47102c012ea2700606f18c21510b875399a323d5a52ee91eca0b55c176b59468
+  data.tar.gz: 9e1e42e298a7e4d4ec40c3d9a3a631fbb9605e48816a439621ac4b4129cd0abe
 SHA512:
-  metadata.gz: 10878f289f4fc79c814e839554adf05f7931e5a4eaa100cc5b584a20ee0bc16fece12c65472af23872abb30885e4e1cbdfc05310baffce4cd4f1dc4be60a622d
-  data.tar.gz: c2fbe34457dcd128b2601baf756257d6147761505150b700d988d130e580017b9e6839ad9b5ba0ea6f91ec398ec1e09ae2bf3c363c2b6508a2d44a9c16368235
+  metadata.gz: effa863afea39f4dc924a1df6d88dce2ebb8639f6d607230df47233a54ef4e124cd998d0efe1a7e63ef5c2bcb6e7dadc7814804f5b9bb35bd82e0ee3c661de00
+  data.tar.gz: 7081eb62a0db815a56cfdaa04212cf4cc96fbd88fd17c4b8f5109535186d30bfa7a2e20d642cfe8617162eb2313a0229f18ad7e4d968cb29e9b8bfec590352e9

data/CHANGELOG.md

@@ -1,3 +1,9 @@
+## [0.8.1] — 2026-07-17
+
+### Added
+
+- **Rich error categories in HTTP mapper** — `Ask::LLM::HTTP.map_error` now sets `category`, `rate_limit_type`, and `retry_after` on `RateLimitError` instances. Rate limit type is detected from error message keywords (token, budget, concurrent, requests). `retry_after` is extracted from response headers.
+
 ## [0.8.0] — 2026-07-17
 
 ### Added

data/lib/ask/llm/http.rb

@@ -14,7 +14,13 @@ module Ask
       end
 
       # Map an HTTP exception or error response to the appropriate Ask::Error.
-      def self.map_error(status, body, provider:)
+      #
+      # @param status [Integer] HTTP status code
+      # @param body [Hash, String, nil] response body
+      # @param provider [String] provider name for error messages
+      # @param headers [Hash, nil] response headers (for retry-after etc.)
+      # @return [Ask::Error]
+      def self.map_error(status, body, provider:, headers: nil)
         body = JSON.parse(body) rescue body if body.is_a?(String)
         message = extract_error_message(body, status) || "HTTP #{status} from #{provider}"
 
@@ -25,12 +31,25 @@ module Ask
         end
 
         case status
-        when 400 then Ask::ProviderError.new(message, status_code: status, response_body: body&.to_json)
-        when 401, 403 then Ask::Unauthorized.new("#{provider}: #{message}")
-        when 429 then Ask::RateLimitError.new("#{provider}: #{message}")
-        when 500 then Ask::ServerError.new("#{provider}: #{message}")
-        when 503 then Ask::ServiceUnavailable.new("#{provider}: #{message}")
-        else Ask::ProviderError.new("#{provider}: #{message}", status_code: status, response_body: body&.to_json)
+        when 400
+          Ask::ProviderError.new(message, status_code: status, response_body: body&.to_json)
+        when 401, 403
+          Ask::Unauthorized.new("#{provider}: #{message}")
+        when 429
+          retry_after = extract_retry_after(headers)
+          rate_limit_type = detect_rate_limit_type(body)
+          Ask::RateLimitError.new(
+            "#{provider}: #{message}",
+            category: Ask::RateLimitCategory::VENDOR,
+            rate_limit_type: rate_limit_type,
+            retry_after: retry_after
+          )
+        when 500
+          Ask::ServerError.new("#{provider}: #{message}")
+        when 503
+          Ask::ServiceUnavailable.new("#{provider}: #{message}")
+        else
+          Ask::ProviderError.new("#{provider}: #{message}", status_code: status, response_body: body&.to_json)
         end
       end
 
@@ -48,6 +67,37 @@ module Ask
           body.to_s
         end
       end
+
+      # Extract retry_after from response headers.
+      # Providers send Retry-After as seconds (integer) or HTTP-date.
+      # @param headers [Hash, nil]
+      # @return [Integer, nil]
+      def self.extract_retry_after(headers)
+        return nil unless headers.is_a?(Hash)
+
+        raw = headers["retry-after"] || headers["Retry-After"] || headers["retry_after"]
+        return nil unless raw
+
+        int = Integer(raw) rescue nil
+        return int if int
+
+        Time.parse(raw) - Time.now rescue nil
+      end
+
+      # Detect rate limit type from error body.
+      # @param body [Hash, nil]
+      # @return [Symbol, nil]
+      def self.detect_rate_limit_type(body)
+        return nil unless body.respond_to?(:dig)
+
+        msg = body.dig("error", "message") || body.dig("error", "code") || ""
+        msg_lower = msg.to_s.downcase
+
+        return Ask::RateLimitType::BUDGET if msg_lower.include?("budget") || msg_lower.include?("quota")
+        return Ask::RateLimitType::TOKENS if msg_lower.include?("token") || msg_lower.include?("tpm")
+        return Ask::RateLimitType::CONCURRENT if msg_lower.include?("concurrent") || msg_lower.include?("parallel")
+        Ask::RateLimitType::REQUESTS
+      end
     end
   end
 end

data/lib/ask/llm/version.rb

@@ -2,6 +2,6 @@
 
 module Ask
   module LLM
-    VERSION = "0.8.0"
+    VERSION = "0.8.1"
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: ask-llm-providers
 version: !ruby/object:Gem::Version
-  version: 0.8.0
+  version: 0.8.1
 platform: ruby
 authors:
 - Kaka Ruto