@mariozechner/pi-ai 0.12.13 → 0.12.15

package/dist/utils/overflow.d.ts

@@ -0,0 +1,50 @@
+import type { AssistantMessage } from "../types.js";
+/**
+ * Check if an assistant message represents a context overflow error.
+ *
+ * This handles two cases:
+ * 1. Error-based overflow: Most providers return stopReason "error" with a
+ *    specific error message pattern.
+ * 2. Silent overflow: Some providers accept overflow requests and return
+ *    successfully. For these, we check if usage.input exceeds the context window.
+ *
+ * ## Reliability by Provider
+ *
+ * **Reliable detection (returns error with detectable message):**
+ * - Anthropic: "prompt is too long: X tokens > Y maximum"
+ * - OpenAI (Completions & Responses): "exceeds the context window"
+ * - Google Gemini: "input token count exceeds the maximum"
+ * - xAI (Grok): "maximum prompt length is X but request contains Y"
+ * - Groq: "reduce the length of the messages"
+ * - Cerebras: 400/413 status code (no body)
+ * - OpenRouter (all backends): "maximum context length is X tokens"
+ * - llama.cpp: "exceeds the available context size"
+ * - LM Studio: "greater than the context length"
+ *
+ * **Unreliable detection:**
+ * - z.ai: Sometimes accepts overflow silently (detectable via usage.input > contextWindow),
+ *   sometimes returns rate limit errors. Pass contextWindow param to detect silent overflow.
+ * - Ollama: Silently truncates input without error. Cannot be detected via this function.
+ *   The response will have usage.input < expected, but we don't know the expected value.
+ *
+ * ## Custom Providers
+ *
+ * If you've added custom models via settings.json, this function may not detect
+ * overflow errors from those providers. To add support:
+ *
+ * 1. Send a request that exceeds the model's context window
+ * 2. Check the errorMessage in the response
+ * 3. Create a regex pattern that matches the error
+ * 4. The pattern should be added to OVERFLOW_PATTERNS in this file, or
+ *    check the errorMessage yourself before calling this function
+ *
+ * @param message - The assistant message to check
+ * @param contextWindow - Optional context window size for detecting silent overflow (z.ai)
+ * @returns true if the message indicates a context overflow
+ */
+export declare function isContextOverflow(message: AssistantMessage, contextWindow?: number): boolean;
+/**
+ * Get the overflow patterns for testing purposes.
+ */
+export declare function getOverflowPatterns(): RegExp[];
+//# sourceMappingURL=overflow.d.ts.map

package/dist/utils/overflow.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"overflow.d.ts","sourceRoot":"","sources":["../../src/utils/overflow.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,aAAa,CAAC;AAoCpD;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA0CG;AACH,wBAAgB,iBAAiB,CAAC,OAAO,EAAE,gBAAgB,EAAE,aAAa,CAAC,EAAE,MAAM,GAAG,OAAO,CAuB5F;AAED;;GAEG;AACH,wBAAgB,mBAAmB,IAAI,MAAM,EAAE,CAE9C","sourcesContent":["import type { AssistantMessage } from \"../types.js\";\n\n/**\n * Regex patterns to detect context overflow errors from different providers.\n *\n * These patterns match error messages returned when the input exceeds\n * the model's context window.\n *\n * Provider-specific patterns (with example error messages):\n *\n * - Anthropic: \"prompt is too long: 213462 tokens > 200000 maximum\"\n * - OpenAI: \"Your input exceeds the context window of this model\"\n * - Google: \"The input token count (1196265) exceeds the maximum number of tokens allowed (1048575)\"\n * - xAI: \"This model's maximum prompt length is 131072 but the request contains 537812 tokens\"\n * - Groq: \"Please reduce the length of the messages or completion\"\n * - OpenRouter: \"This endpoint's maximum context length is X tokens. However, you requested about Y tokens\"\n * - llama.cpp: \"the request exceeds the available context size, try increasing it\"\n * - LM Studio: \"tokens to keep from the initial prompt is greater than the context length\"\n * - Cerebras: Returns \"400 status code (no body)\" - handled separately below\n * - z.ai: Does NOT error, accepts overflow silently - handled via usage.input > contextWindow\n * - Ollama: Silently truncates input - not detectable via error message\n */\nconst OVERFLOW_PATTERNS = [\n\t/prompt is too long/i, // Anthropic\n\t/exceeds the context window/i, // OpenAI (Completions & Responses API)\n\t/input token count.*exceeds the maximum/i, // Google (Gemini)\n\t/maximum prompt length is \\d+/i, // xAI (Grok)\n\t/reduce the length of the messages/i, // Groq\n\t/maximum context length is \\d+ tokens/i, // OpenRouter (all backends)\n\t/exceeds the available context size/i, // llama.cpp server\n\t/greater than the context length/i, // LM Studio\n\t/context length exceeded/i, // Generic fallback\n\t/too many tokens/i, // Generic fallback\n\t/token limit exceeded/i, // Generic fallback\n];\n\n/**\n * Check if an assistant message represents a context overflow error.\n *\n * This handles two cases:\n * 1. Error-based overflow: Most providers return stopReason \"error\" with a\n *    specific error message pattern.\n * 2. Silent overflow: Some providers accept overflow requests and return\n *    successfully. For these, we check if usage.input exceeds the context window.\n *\n * ## Reliability by Provider\n *\n * **Reliable detection (returns error with detectable message):**\n * - Anthropic: \"prompt is too long: X tokens > Y maximum\"\n * - OpenAI (Completions & Responses): \"exceeds the context window\"\n * - Google Gemini: \"input token count exceeds the maximum\"\n * - xAI (Grok): \"maximum prompt length is X but request contains Y\"\n * - Groq: \"reduce the length of the messages\"\n * - Cerebras: 400/413 status code (no body)\n * - OpenRouter (all backends): \"maximum context length is X tokens\"\n * - llama.cpp: \"exceeds the available context size\"\n * - LM Studio: \"greater than the context length\"\n *\n * **Unreliable detection:**\n * - z.ai: Sometimes accepts overflow silently (detectable via usage.input > contextWindow),\n *   sometimes returns rate limit errors. Pass contextWindow param to detect silent overflow.\n * - Ollama: Silently truncates input without error. Cannot be detected via this function.\n *   The response will have usage.input < expected, but we don't know the expected value.\n *\n * ## Custom Providers\n *\n * If you've added custom models via settings.json, this function may not detect\n * overflow errors from those providers. To add support:\n *\n * 1. Send a request that exceeds the model's context window\n * 2. Check the errorMessage in the response\n * 3. Create a regex pattern that matches the error\n * 4. The pattern should be added to OVERFLOW_PATTERNS in this file, or\n *    check the errorMessage yourself before calling this function\n *\n * @param message - The assistant message to check\n * @param contextWindow - Optional context window size for detecting silent overflow (z.ai)\n * @returns true if the message indicates a context overflow\n */\nexport function isContextOverflow(message: AssistantMessage, contextWindow?: number): boolean {\n\t// Case 1: Check error message patterns\n\tif (message.stopReason === \"error\" && message.errorMessage) {\n\t\t// Check known patterns\n\t\tif (OVERFLOW_PATTERNS.some((p) => p.test(message.errorMessage!))) {\n\t\t\treturn true;\n\t\t}\n\n\t\t// Cerebras returns 400/413 with no body - check for status code pattern\n\t\tif (/^4(00|13)\\s*(status code)?\\s*\\(no body\\)/i.test(message.errorMessage)) {\n\t\t\treturn true;\n\t\t}\n\t}\n\n\t// Case 2: Silent overflow (z.ai style) - successful but usage exceeds context\n\tif (contextWindow && message.stopReason === \"stop\") {\n\t\tconst inputTokens = message.usage.input + message.usage.cacheRead;\n\t\tif (inputTokens > contextWindow) {\n\t\t\treturn true;\n\t\t}\n\t}\n\n\treturn false;\n}\n\n/**\n * Get the overflow patterns for testing purposes.\n */\nexport function getOverflowPatterns(): RegExp[] {\n\treturn [...OVERFLOW_PATTERNS];\n}\n"]}

package/dist/utils/overflow.js

@@ -0,0 +1,104 @@
+/**
+ * Regex patterns to detect context overflow errors from different providers.
+ *
+ * These patterns match error messages returned when the input exceeds
+ * the model's context window.
+ *
+ * Provider-specific patterns (with example error messages):
+ *
+ * - Anthropic: "prompt is too long: 213462 tokens > 200000 maximum"
+ * - OpenAI: "Your input exceeds the context window of this model"
+ * - Google: "The input token count (1196265) exceeds the maximum number of tokens allowed (1048575)"
+ * - xAI: "This model's maximum prompt length is 131072 but the request contains 537812 tokens"
+ * - Groq: "Please reduce the length of the messages or completion"
+ * - OpenRouter: "This endpoint's maximum context length is X tokens. However, you requested about Y tokens"
+ * - llama.cpp: "the request exceeds the available context size, try increasing it"
+ * - LM Studio: "tokens to keep from the initial prompt is greater than the context length"
+ * - Cerebras: Returns "400 status code (no body)" - handled separately below
+ * - z.ai: Does NOT error, accepts overflow silently - handled via usage.input > contextWindow
+ * - Ollama: Silently truncates input - not detectable via error message
+ */
+const OVERFLOW_PATTERNS = [
+    /prompt is too long/i, // Anthropic
+    /exceeds the context window/i, // OpenAI (Completions & Responses API)
+    /input token count.*exceeds the maximum/i, // Google (Gemini)
+    /maximum prompt length is \d+/i, // xAI (Grok)
+    /reduce the length of the messages/i, // Groq
+    /maximum context length is \d+ tokens/i, // OpenRouter (all backends)
+    /exceeds the available context size/i, // llama.cpp server
+    /greater than the context length/i, // LM Studio
+    /context length exceeded/i, // Generic fallback
+    /too many tokens/i, // Generic fallback
+    /token limit exceeded/i, // Generic fallback
+];
+/**
+ * Check if an assistant message represents a context overflow error.
+ *
+ * This handles two cases:
+ * 1. Error-based overflow: Most providers return stopReason "error" with a
+ *    specific error message pattern.
+ * 2. Silent overflow: Some providers accept overflow requests and return
+ *    successfully. For these, we check if usage.input exceeds the context window.
+ *
+ * ## Reliability by Provider
+ *
+ * **Reliable detection (returns error with detectable message):**
+ * - Anthropic: "prompt is too long: X tokens > Y maximum"
+ * - OpenAI (Completions & Responses): "exceeds the context window"
+ * - Google Gemini: "input token count exceeds the maximum"
+ * - xAI (Grok): "maximum prompt length is X but request contains Y"
+ * - Groq: "reduce the length of the messages"
+ * - Cerebras: 400/413 status code (no body)
+ * - OpenRouter (all backends): "maximum context length is X tokens"
+ * - llama.cpp: "exceeds the available context size"
+ * - LM Studio: "greater than the context length"
+ *
+ * **Unreliable detection:**
+ * - z.ai: Sometimes accepts overflow silently (detectable via usage.input > contextWindow),
+ *   sometimes returns rate limit errors. Pass contextWindow param to detect silent overflow.
+ * - Ollama: Silently truncates input without error. Cannot be detected via this function.
+ *   The response will have usage.input < expected, but we don't know the expected value.
+ *
+ * ## Custom Providers
+ *
+ * If you've added custom models via settings.json, this function may not detect
+ * overflow errors from those providers. To add support:
+ *
+ * 1. Send a request that exceeds the model's context window
+ * 2. Check the errorMessage in the response
+ * 3. Create a regex pattern that matches the error
+ * 4. The pattern should be added to OVERFLOW_PATTERNS in this file, or
+ *    check the errorMessage yourself before calling this function
+ *
+ * @param message - The assistant message to check
+ * @param contextWindow - Optional context window size for detecting silent overflow (z.ai)
+ * @returns true if the message indicates a context overflow
+ */
+export function isContextOverflow(message, contextWindow) {
+    // Case 1: Check error message patterns
+    if (message.stopReason === "error" && message.errorMessage) {
+        // Check known patterns
+        if (OVERFLOW_PATTERNS.some((p) => p.test(message.errorMessage))) {
+            return true;
+        }
+        // Cerebras returns 400/413 with no body - check for status code pattern
+        if (/^4(00|13)\s*(status code)?\s*\(no body\)/i.test(message.errorMessage)) {
+            return true;
+        }
+    }
+    // Case 2: Silent overflow (z.ai style) - successful but usage exceeds context
+    if (contextWindow && message.stopReason === "stop") {
+        const inputTokens = message.usage.input + message.usage.cacheRead;
+        if (inputTokens > contextWindow) {
+            return true;
+        }
+    }
+    return false;
+}
+/**
+ * Get the overflow patterns for testing purposes.
+ */
+export function getOverflowPatterns() {
+    return [...OVERFLOW_PATTERNS];
+}
+//# sourceMappingURL=overflow.js.map

package/dist/utils/overflow.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"overflow.js","sourceRoot":"","sources":["../../src/utils/overflow.ts"],"names":[],"mappings":"AAEA;;;;;;;;;;;;;;;;;;;GAmBG;AACH,MAAM,iBAAiB,GAAG;IACzB,qBAAqB,EAAE,YAAY;IACnC,6BAA6B,EAAE,uCAAuC;IACtE,yCAAyC,EAAE,kBAAkB;IAC7D,+BAA+B,EAAE,aAAa;IAC9C,oCAAoC,EAAE,OAAO;IAC7C,uCAAuC,EAAE,4BAA4B;IACrE,qCAAqC,EAAE,mBAAmB;IAC1D,kCAAkC,EAAE,YAAY;IAChD,0BAA0B,EAAE,mBAAmB;IAC/C,kBAAkB,EAAE,mBAAmB;IACvC,uBAAuB,EAAE,mBAAmB;CAC5C,CAAC;AAEF;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA0CG;AACH,MAAM,UAAU,iBAAiB,CAAC,OAAyB,EAAE,aAAsB,EAAW;IAC7F,uCAAuC;IACvC,IAAI,OAAO,CAAC,UAAU,KAAK,OAAO,IAAI,OAAO,CAAC,YAAY,EAAE,CAAC;QAC5D,uBAAuB;QACvB,IAAI,iBAAiB,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,YAAa,CAAC,CAAC,EAAE,CAAC;YAClE,OAAO,IAAI,CAAC;QACb,CAAC;QAED,wEAAwE;QACxE,IAAI,2CAA2C,CAAC,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,EAAE,CAAC;YAC5E,OAAO,IAAI,CAAC;QACb,CAAC;IACF,CAAC;IAED,8EAA8E;IAC9E,IAAI,aAAa,IAAI,OAAO,CAAC,UAAU,KAAK,MAAM,EAAE,CAAC;QACpD,MAAM,WAAW,GAAG,OAAO,CAAC,KAAK,CAAC,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,SAAS,CAAC;QAClE,IAAI,WAAW,GAAG,aAAa,EAAE,CAAC;YACjC,OAAO,IAAI,CAAC;QACb,CAAC;IACF,CAAC;IAED,OAAO,KAAK,CAAC;AAAA,CACb;AAED;;GAEG;AACH,MAAM,UAAU,mBAAmB,GAAa;IAC/C,OAAO,CAAC,GAAG,iBAAiB,CAAC,CAAC;AAAA,CAC9B","sourcesContent":["import type { AssistantMessage } from \"../types.js\";\n\n/**\n * Regex patterns to detect context overflow errors from different providers.\n *\n * These patterns match error messages returned when the input exceeds\n * the model's context window.\n *\n * Provider-specific patterns (with example error messages):\n *\n * - Anthropic: \"prompt is too long: 213462 tokens > 200000 maximum\"\n * - OpenAI: \"Your input exceeds the context window of this model\"\n * - Google: \"The input token count (1196265) exceeds the maximum number of tokens allowed (1048575)\"\n * - xAI: \"This model's maximum prompt length is 131072 but the request contains 537812 tokens\"\n * - Groq: \"Please reduce the length of the messages or completion\"\n * - OpenRouter: \"This endpoint's maximum context length is X tokens. However, you requested about Y tokens\"\n * - llama.cpp: \"the request exceeds the available context size, try increasing it\"\n * - LM Studio: \"tokens to keep from the initial prompt is greater than the context length\"\n * - Cerebras: Returns \"400 status code (no body)\" - handled separately below\n * - z.ai: Does NOT error, accepts overflow silently - handled via usage.input > contextWindow\n * - Ollama: Silently truncates input - not detectable via error message\n */\nconst OVERFLOW_PATTERNS = [\n\t/prompt is too long/i, // Anthropic\n\t/exceeds the context window/i, // OpenAI (Completions & Responses API)\n\t/input token count.*exceeds the maximum/i, // Google (Gemini)\n\t/maximum prompt length is \\d+/i, // xAI (Grok)\n\t/reduce the length of the messages/i, // Groq\n\t/maximum context length is \\d+ tokens/i, // OpenRouter (all backends)\n\t/exceeds the available context size/i, // llama.cpp server\n\t/greater than the context length/i, // LM Studio\n\t/context length exceeded/i, // Generic fallback\n\t/too many tokens/i, // Generic fallback\n\t/token limit exceeded/i, // Generic fallback\n];\n\n/**\n * Check if an assistant message represents a context overflow error.\n *\n * This handles two cases:\n * 1. Error-based overflow: Most providers return stopReason \"error\" with a\n *    specific error message pattern.\n * 2. Silent overflow: Some providers accept overflow requests and return\n *    successfully. For these, we check if usage.input exceeds the context window.\n *\n * ## Reliability by Provider\n *\n * **Reliable detection (returns error with detectable message):**\n * - Anthropic: \"prompt is too long: X tokens > Y maximum\"\n * - OpenAI (Completions & Responses): \"exceeds the context window\"\n * - Google Gemini: \"input token count exceeds the maximum\"\n * - xAI (Grok): \"maximum prompt length is X but request contains Y\"\n * - Groq: \"reduce the length of the messages\"\n * - Cerebras: 400/413 status code (no body)\n * - OpenRouter (all backends): \"maximum context length is X tokens\"\n * - llama.cpp: \"exceeds the available context size\"\n * - LM Studio: \"greater than the context length\"\n *\n * **Unreliable detection:**\n * - z.ai: Sometimes accepts overflow silently (detectable via usage.input > contextWindow),\n *   sometimes returns rate limit errors. Pass contextWindow param to detect silent overflow.\n * - Ollama: Silently truncates input without error. Cannot be detected via this function.\n *   The response will have usage.input < expected, but we don't know the expected value.\n *\n * ## Custom Providers\n *\n * If you've added custom models via settings.json, this function may not detect\n * overflow errors from those providers. To add support:\n *\n * 1. Send a request that exceeds the model's context window\n * 2. Check the errorMessage in the response\n * 3. Create a regex pattern that matches the error\n * 4. The pattern should be added to OVERFLOW_PATTERNS in this file, or\n *    check the errorMessage yourself before calling this function\n *\n * @param message - The assistant message to check\n * @param contextWindow - Optional context window size for detecting silent overflow (z.ai)\n * @returns true if the message indicates a context overflow\n */\nexport function isContextOverflow(message: AssistantMessage, contextWindow?: number): boolean {\n\t// Case 1: Check error message patterns\n\tif (message.stopReason === \"error\" && message.errorMessage) {\n\t\t// Check known patterns\n\t\tif (OVERFLOW_PATTERNS.some((p) => p.test(message.errorMessage!))) {\n\t\t\treturn true;\n\t\t}\n\n\t\t// Cerebras returns 400/413 with no body - check for status code pattern\n\t\tif (/^4(00|13)\\s*(status code)?\\s*\\(no body\\)/i.test(message.errorMessage)) {\n\t\t\treturn true;\n\t\t}\n\t}\n\n\t// Case 2: Silent overflow (z.ai style) - successful but usage exceeds context\n\tif (contextWindow && message.stopReason === \"stop\") {\n\t\tconst inputTokens = message.usage.input + message.usage.cacheRead;\n\t\tif (inputTokens > contextWindow) {\n\t\t\treturn true;\n\t\t}\n\t}\n\n\treturn false;\n}\n\n/**\n * Get the overflow patterns for testing purposes.\n */\nexport function getOverflowPatterns(): RegExp[] {\n\treturn [...OVERFLOW_PATTERNS];\n}\n"]}

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@mariozechner/pi-ai",
-	"version": "0.12.13",
+	"version": "0.12.15",
 	"description": "Unified LLM API with automatic model discovery and provider configuration",
 	"type": "module",
 	"main": "./dist/index.js",