koishi-plugin-chatluna 1.2.0 → 1.2.2

package/lib/index.cjs

@@ -3885,6 +3885,7 @@ __name(apply40, "apply");
 
 // src/middlewares/read_chat_message.ts
 var import_koishi8 = require("koishi");
+var import_string = require("koishi-plugin-chatluna/utils/string");
 function apply41(ctx, config, chain) {
   chain.middleware("read_chat_message", async (session, context) => {
     let message = context.command != null ? context.message : session.elements;
@@ -3922,6 +3923,7 @@ function apply41(ctx, config, chain) {
     "img",
     async (session, element, message) => {
       const images = message.additional_kwargs.images ?? [];
+      const imageHashs = message.additional_kwargs.imageHashs ?? [];
       const url = element.attrs.url ?? element.attrs.src;
       logger.debug(`image url: ${url}`);
       const readImage = /* @__PURE__ */ __name(async (url2) => {
@@ -3940,6 +3942,9 @@ function apply41(ctx, config, chain) {
         const base64 = Buffer.from(buffer).toString("base64");
         images.push(`data:image/${ext ?? "jpeg"};base64,${base64}`);
       }, "readImage");
+      const imageHash = await (0, import_string.hashString)(url, 8);
+      imageHashs.push(imageHash);
+      element.attrs["imageHash"] = imageHash;
       if (url.startsWith("data:image") && url.includes("base64")) {
         images.push(url);
       } else {
@@ -3954,6 +3959,7 @@ function apply41(ctx, config, chain) {
         }
       }
       message.additional_kwargs.images = images;
+      message.additional_kwargs.imageHashs = imageHashs;
       if (message.content?.length < 1) {
         message.content = "[image]";
       }
@@ -4009,7 +4015,7 @@ var import_count_tokens3 = require("koishi-plugin-chatluna/llm-core/utils/count_
 var import_error6 = require("koishi-plugin-chatluna/utils/error");
 var import_logger5 = require("koishi-plugin-chatluna/utils/logger");
 var import_chains = require("koishi-plugin-chatluna/chains");
-var import_string = require("koishi-plugin-chatluna/utils/string");
+var import_string2 = require("koishi-plugin-chatluna/utils/string");
 
 // src/utils/buffer_text.ts
 var import_koishi9 = require("koishi");
@@ -4160,7 +4166,7 @@ function apply43(ctx, config, chain) {
       presetTemplate.config?.postHandler?.prefix,
       presetTemplate.config?.postHandler?.postfix
     );
-    const postHandler = presetTemplate.config?.postHandler ? new import_string.PresetPostHandler(
+    const postHandler = presetTemplate.config?.postHandler ? new import_string2.PresetPostHandler(
       ctx,
       config,
       presetTemplate.config?.postHandler
@@ -4343,7 +4349,7 @@ function getSystemPromptVariables(session, config, room) {
     is_group: (!session.isDirect || session.guildId != null).toString(),
     is_private: session.isDirect?.toString(),
     user_id: session.author?.user?.id ?? session.event?.user?.id ?? "0",
-    user: (0, import_string.getNotEmptyString)(
+    user: (0, import_string2.getNotEmptyString)(
       session.author?.nick,
       session.author?.name,
       session.event.user?.name,
@@ -4351,8 +4357,8 @@ function getSystemPromptVariables(session, config, room) {
     ),
     noop: "",
     time: (/* @__PURE__ */ new Date()).toLocaleTimeString(),
-    weekday: (0, import_string.getCurrentWeekday)(),
-    idle_duration: (0, import_string.getTimeDiffFormat)(
+    weekday: (0, import_string2.getCurrentWeekday)(),
+    idle_duration: (0, import_string2.getTimeDiffFormat)(
       (/* @__PURE__ */ new Date()).getTime(),
       room.updatedTime.getTime()
     )
@@ -4391,7 +4397,7 @@ async function formatUserPromptString(config, presetTemplate, session, prompt, r
     presetTemplate.formatUserPromptString,
     {
       sender_id: session.author?.user?.id ?? session.event?.user?.id ?? "0",
-      sender: (0, import_string.getNotEmptyString)(
+      sender: (0, import_string2.getNotEmptyString)(
         session.author?.nick,
         session.author?.name,
         session.event.user?.name,

package/lib/index.mjs

@@ -3879,6 +3879,7 @@ __name(apply40, "apply");
 
 // src/middlewares/read_chat_message.ts
 import { h as h4 } from "koishi";
+import { hashString } from "koishi-plugin-chatluna/utils/string";
 function apply41(ctx, config, chain) {
   chain.middleware("read_chat_message", async (session, context) => {
     let message = context.command != null ? context.message : session.elements;
@@ -3916,6 +3917,7 @@ function apply41(ctx, config, chain) {
     "img",
     async (session, element, message) => {
       const images = message.additional_kwargs.images ?? [];
+      const imageHashs = message.additional_kwargs.imageHashs ?? [];
       const url = element.attrs.url ?? element.attrs.src;
       logger.debug(`image url: ${url}`);
       const readImage = /* @__PURE__ */ __name(async (url2) => {
@@ -3934,6 +3936,9 @@ function apply41(ctx, config, chain) {
         const base64 = Buffer.from(buffer).toString("base64");
         images.push(`data:image/${ext ?? "jpeg"};base64,${base64}`);
       }, "readImage");
+      const imageHash = await hashString(url, 8);
+      imageHashs.push(imageHash);
+      element.attrs["imageHash"] = imageHash;
       if (url.startsWith("data:image") && url.includes("base64")) {
         images.push(url);
       } else {
@@ -3948,6 +3953,7 @@ function apply41(ctx, config, chain) {
         }
       }
       message.additional_kwargs.images = images;
+      message.additional_kwargs.imageHashs = imageHashs;
       if (message.content?.length < 1) {
         message.content = "[image]";
       }

package/lib/llm-core/agent/index.cjs

@@ -1139,10 +1139,10 @@ Your reasoning and thought process for the current step
 
 <tool_calling>
 [
-  {
+  {{
     "name": "tool_name",
-    "arguments": {"param1": "value1", "param2": "value2"}
-  }
+    "arguments": {{"param1": "value1", "param2": "value2"}}
+   }}
 ]
 </tool_calling>
 
@@ -1158,10 +1158,10 @@ I have gathered all the necessary information and can now provide the final answ
 
 <tool_calling>
 [
-  {
+  {{
     "name": "final_answer",
-    "arguments": {"answer": "insert your final answer here"}
-  }
+    "arguments": {{"answer": "insert your final answer here"}}
+  }}
 ]
 </tool_calling>
 
@@ -1175,10 +1175,10 @@ I need to first find out who is the oldest person in the document, then generate
 
 <tool_calling>
 [
-  {
+  {{
     "name": "document_qa",
-    "arguments": {"document": "document.pdf", "question": "Who is the oldest person mentioned?"}
-  }
+    "arguments": {{"document": "document.pdf", "question": "Who is the oldest person mentioned?"}}
+  }}
 ]
 </tool_calling>
 
@@ -1190,14 +1190,14 @@ Now I know the oldest person is John Doe, a 55-year-old lumberjack from Newfound
 
 <tool_calling>
 [
-  {
+ {{
     "name": "image_generator",
-    "arguments": {"prompt": "A portrait of John Doe, a 55-year-old lumberjack living in Newfoundland, Canada"}
-  },
-  {
+    "arguments": {{"prompt": "A portrait of John Doe, a 55-year-old lumberjack living in Newfoundland, Canada"}}
+  }},
+  {{
     "name": "search",
-    "arguments": {"query": "John Doe lumberjack Newfoundland biography"}
-  }
+    "arguments": {{"query": "John Doe lumberjack Newfoundland biography"}}
+  }}
 ]
 </tool_calling>
 
@@ -1209,10 +1209,10 @@ I have successfully generated an image and found biographical information. Now I
 
 <tool_calling>
 [
-  {
+  {{
     "name": "final_answer",
-    "arguments": {"answer": "I have generated an image (image.png) of John Doe, the oldest person in the document, and found that he is a well-known lumberjack in Newfoundland with over 30 years of experience in forestry."}
-  }
+    "arguments": {{"answer": "I have generated an image (image.png) of John Doe, the oldest person in the document, and found that he is a well-known lumberjack in Newfoundland with over 30 years of experience in forestry."}}
+  }}
 ]
 </tool_calling>
 
@@ -1225,10 +1225,10 @@ I need to calculate the sum of 5 + 3 + 1294.678. I'll use the python_interpreter
 
 <tool_calling>
 [
-  {
+  {{
     "name": "python_interpreter",
-    "arguments": {"code": "5 + 3 + 1294.678"}
-  }
+    "arguments": {{"code": "5 + 3 + 1294.678"}}
+  }}
 ]
 </tool_calling>
 
@@ -1240,10 +1240,10 @@ I have calculated the result. Now I can provide the final answer.
 
 <tool_calling>
 [
-  {
+  {{
     "name": "final_answer",
-    "arguments": {"answer": "1302.678"}
-  }
+    "arguments": {{"answer": "1302.678"}}
+  }}
 ]
 </tool_calling>
 
@@ -1256,14 +1256,14 @@ I need to search for the population data of both Guangzhou and Shanghai to compa
 
 <tool_calling>
 [
-  {
+  {{
     "name": "search",
-    "arguments": {"query": "Population Guangzhou 2023"}
-  },
-  {
+    "arguments": {{"query": "Population Guangzhou 2023"}}
+  }},
+  {{
     "name": "search",
-    "arguments": {"query": "Population Shanghai 2023"}
-  }
+    "arguments": {{"query": "Population Shanghai 2023"}}
+  }}
 ]
 </tool_calling>
 
@@ -1275,10 +1275,10 @@ Based on the search results, Shanghai has a population of 26 million while Guang
 
 <tool_calling>
 [
-  {
+  {{
     "name": "final_answer",
-    "arguments": {"answer": "Shanghai has the highest population with 26 million people, compared to Guangzhou's 15 million people."}
-  }
+    "arguments": {{"answer": "Shanghai has the highest population with 26 million people, compared to Guangzhou's 15 million people."}}
+  }}
 ]
 </tool_calling>
 

package/lib/llm-core/agent/index.mjs

@@ -1127,10 +1127,10 @@ Your reasoning and thought process for the current step
 
 <tool_calling>
 [
-  {
+  {{
     "name": "tool_name",
-    "arguments": {"param1": "value1", "param2": "value2"}
-  }
+    "arguments": {{"param1": "value1", "param2": "value2"}}
+   }}
 ]
 </tool_calling>
 
@@ -1146,10 +1146,10 @@ I have gathered all the necessary information and can now provide the final answ
 
 <tool_calling>
 [
-  {
+  {{
     "name": "final_answer",
-    "arguments": {"answer": "insert your final answer here"}
-  }
+    "arguments": {{"answer": "insert your final answer here"}}
+  }}
 ]
 </tool_calling>
 
@@ -1163,10 +1163,10 @@ I need to first find out who is the oldest person in the document, then generate
 
 <tool_calling>
 [
-  {
+  {{
     "name": "document_qa",
-    "arguments": {"document": "document.pdf", "question": "Who is the oldest person mentioned?"}
-  }
+    "arguments": {{"document": "document.pdf", "question": "Who is the oldest person mentioned?"}}
+  }}
 ]
 </tool_calling>
 
@@ -1178,14 +1178,14 @@ Now I know the oldest person is John Doe, a 55-year-old lumberjack from Newfound
 
 <tool_calling>
 [
-  {
+ {{
     "name": "image_generator",
-    "arguments": {"prompt": "A portrait of John Doe, a 55-year-old lumberjack living in Newfoundland, Canada"}
-  },
-  {
+    "arguments": {{"prompt": "A portrait of John Doe, a 55-year-old lumberjack living in Newfoundland, Canada"}}
+  }},
+  {{
     "name": "search",
-    "arguments": {"query": "John Doe lumberjack Newfoundland biography"}
-  }
+    "arguments": {{"query": "John Doe lumberjack Newfoundland biography"}}
+  }}
 ]
 </tool_calling>
 
@@ -1197,10 +1197,10 @@ I have successfully generated an image and found biographical information. Now I
 
 <tool_calling>
 [
-  {
+  {{
     "name": "final_answer",
-    "arguments": {"answer": "I have generated an image (image.png) of John Doe, the oldest person in the document, and found that he is a well-known lumberjack in Newfoundland with over 30 years of experience in forestry."}
-  }
+    "arguments": {{"answer": "I have generated an image (image.png) of John Doe, the oldest person in the document, and found that he is a well-known lumberjack in Newfoundland with over 30 years of experience in forestry."}}
+  }}
 ]
 </tool_calling>
 
@@ -1213,10 +1213,10 @@ I need to calculate the sum of 5 + 3 + 1294.678. I'll use the python_interpreter
 
 <tool_calling>
 [
-  {
+  {{
     "name": "python_interpreter",
-    "arguments": {"code": "5 + 3 + 1294.678"}
-  }
+    "arguments": {{"code": "5 + 3 + 1294.678"}}
+  }}
 ]
 </tool_calling>
 
@@ -1228,10 +1228,10 @@ I have calculated the result. Now I can provide the final answer.
 
 <tool_calling>
 [
-  {
+  {{
     "name": "final_answer",
-    "arguments": {"answer": "1302.678"}
-  }
+    "arguments": {{"answer": "1302.678"}}
+  }}
 ]
 </tool_calling>
 
@@ -1244,14 +1244,14 @@ I need to search for the population data of both Guangzhou and Shanghai to compa
 
 <tool_calling>
 [
-  {
+  {{
     "name": "search",
-    "arguments": {"query": "Population Guangzhou 2023"}
-  },
-  {
+    "arguments": {{"query": "Population Guangzhou 2023"}}
+  }},
+  {{
     "name": "search",
-    "arguments": {"query": "Population Shanghai 2023"}
-  }
+    "arguments": {{"query": "Population Shanghai 2023"}}
+  }}
 ]
 </tool_calling>
 
@@ -1263,10 +1263,10 @@ Based on the search results, Shanghai has a population of 26 million while Guang
 
 <tool_calling>
 [
-  {
+  {{
     "name": "final_answer",
-    "arguments": {"answer": "Shanghai has the highest population with 26 million people, compared to Guangzhou's 15 million people."}
-  }
+    "arguments": {{"answer": "Shanghai has the highest population with 26 million people, compared to Guangzhou's 15 million people."}}
+  }}
 ]
 </tool_calling>
 

package/lib/llm-core/agent/openai/index.d.ts

@@ -22,4 +22,4 @@ export type CreateOpenAIAgentParams = {
 };
 export declare function createOpenAIAgent({ llm, tools, prompt }: CreateOpenAIAgentParams): RunnableSequence<{
     steps: AgentStep[];
-}, AgentAction | AgentAction[] | AgentFinish>;
+}, AgentAction | AgentFinish | AgentAction[]>;

package/lib/llm-core/agent/react/index.d.ts

@@ -72,7 +72,7 @@ export type CreateReactAgentParams = {
  */
 export declare function createReactAgent({ llm, tools, prompt, streamRunnable, instructions }: CreateReactAgentParams): Promise<AgentRunnableSequence<{
     steps: AgentStep[];
-}, import("@langchain/core/agents").AgentAction[] | import("@langchain/core/agents").AgentFinish>>;
+}, import("@langchain/core/agents").AgentFinish | import("@langchain/core/agents").AgentAction[]>>;
 /**
  * Construct the scratchpad that lets the agent continue its thought process.
  * @param intermediateSteps

package/lib/llm-core/agent/react/prompt.d.ts

@@ -1 +1 @@
-export declare const FORMAT_INSTRUCTIONS = "You are an expert assistant who can solve any task using tool calls. You will be given a task to solve as best you can.\nTo do so, you have been given access to the following tools: {tool_names}\n\nThe tool calls you write are actions: after the tools are executed, you will get the results of the tool calls as \"observations\".\nAt each step, you should first explain your reasoning towards solving the task and the tools that you want to use within <thought> tags.\nThen you should write one or more valid JSON tool calls within <tool_calling> tags.\nThis Thought/Tool_calling/Observation cycle can repeat N times, you should take several steps when needed.\n\nHere are the output format:\n\n<thought>\nYour reasoning and thought process for the current step\n</thought>\n\n<tool_calling>\n[\n  {\n    \"name\": \"tool_name\",\n    \"arguments\": {\"param1\": \"value1\", \"param2\": \"value2\"}\n  }\n]\n</tool_calling>\n\nYou can call multiple tools at once by including multiple tool objects in the JSON array within the <tool_calling> tags.\n\nONLY output within the <thought> and <tool_calling> sequences. You will get the Observation from the tool calls. Do not output the Observation yourself.\n\nTo provide the final answer to the task, use a tool call with \"name\": \"final_answer\". It is the only way to complete the task, else you will be stuck on a loop. So your final output should look like this:\n\n<thought>\nI have gathered all the necessary information and can now provide the final answer.\n</thought>\n\n<tool_calling>\n[\n  {\n    \"name\": \"final_answer\",\n    \"arguments\": {\"answer\": \"insert your final answer here\"}\n  }\n]\n</tool_calling>\n\nHere are a few examples using notional tools:\n---\nTask: \"Generate an image of the oldest person in this document and also search for their biography.\"\n\n<thought>\nI need to first find out who is the oldest person in the document, then generate an image of them and search for their biography. I'll start by using the document_qa tool to find the oldest person.\n</thought>\n\n<tool_calling>\n[\n  {\n    \"name\": \"document_qa\",\n    \"arguments\": {\"document\": \"document.pdf\", \"question\": \"Who is the oldest person mentioned?\"}\n  }\n]\n</tool_calling>\n\nObservation: \"The oldest person in the document is John Doe, a 55 year old lumberjack living in Newfoundland.\"\n\n<thought>\nNow I know the oldest person is John Doe, a 55-year-old lumberjack from Newfoundland. I can generate an image of him and search for his biography at the same time.\n</thought>\n\n<tool_calling>\n[\n  {\n    \"name\": \"image_generator\",\n    \"arguments\": {\"prompt\": \"A portrait of John Doe, a 55-year-old lumberjack living in Newfoundland, Canada\"}\n  },\n  {\n    \"name\": \"search\",\n    \"arguments\": {\"query\": \"John Doe lumberjack Newfoundland biography\"}\n  }\n]\n</tool_calling>\n\nObservation: \"image.png\" and \"John Doe is a well-known lumberjack in Newfoundland with over 30 years of experience in forestry.\"\n\n<thought>\nI have successfully generated an image and found biographical information. Now I can provide the final answer with both pieces of information.\n</thought>\n\n<tool_calling>\n[\n  {\n    \"name\": \"final_answer\",\n    \"arguments\": {\"answer\": \"I have generated an image (image.png) of John Doe, the oldest person in the document, and found that he is a well-known lumberjack in Newfoundland with over 30 years of experience in forestry.\"}\n  }\n]\n</tool_calling>\n\n---\nTask: \"What is the result of the following operation: 5 + 3 + 1294.678?\"\n\n<thought>\nI need to calculate the sum of 5 + 3 + 1294.678. I'll use the python_interpreter tool to execute this calculation.\n</thought>\n\n<tool_calling>\n[\n  {\n    \"name\": \"python_interpreter\",\n    \"arguments\": {\"code\": \"5 + 3 + 1294.678\"}\n  }\n]\n</tool_calling>\n\nObservation: 1302.678\n\n<thought>\nI have calculated the result. Now I can provide the final answer.\n</thought>\n\n<tool_calling>\n[\n  {\n    \"name\": \"final_answer\",\n    \"arguments\": {\"answer\": \"1302.678\"}\n  }\n]\n</tool_calling>\n\n---\nTask: \"Which city has the highest population, Guangzhou or Shanghai?\"\n\n<thought>\nI need to search for the population data of both Guangzhou and Shanghai to compare them. I'll search for both cities' population information simultaneously.\n</thought>\n\n<tool_calling>\n[\n  {\n    \"name\": \"search\",\n    \"arguments\": {\"query\": \"Population Guangzhou 2023\"}\n  },\n  {\n    \"name\": \"search\",\n    \"arguments\": {\"query\": \"Population Shanghai 2023\"}\n  }\n]\n</tool_calling>\n\nObservation: ['Guangzhou has a population of 15 million inhabitants as of 2021.'] and ['Shanghai has a population of 26 million (2019)']\n\n<thought>\nBased on the search results, Shanghai has a population of 26 million while Guangzhou has 15 million. Therefore, Shanghai has the higher population.\n</thought>\n\n<tool_calling>\n[\n  {\n    \"name\": \"final_answer\",\n    \"arguments\": {\"answer\": \"Shanghai has the highest population with 26 million people, compared to Guangzhou's 15 million people.\"}\n  }\n]\n</tool_calling>\n\nAbove examples were using notional tools that might not exist for you. You only have access to these tools:\n\n{tool_descriptions}\n\nHere are the rules you should always follow to solve your task:\n1. ALWAYS provide tool calls within <tool_calling> tags, else you will fail.\n2. Always use the right arguments for the tools. Never use variable names as the action arguments, use the actual values instead.\n3. You can call multiple tools at once if it makes sense for efficiency.\n4. Call tools only when needed: do not call the search agent if you do not need information, try to solve the task yourself.\nIf no tool call is needed, use final_answer tool to return your answer.\n5. Never re-do a tool call that you previously did with the exact same parameters.\n6. Always include your reasoning in <thought> tags before making tool calls.\n\nNow Begin! If you solve the task correctly, you will receive a reward of $1,000,000.";
+export declare const FORMAT_INSTRUCTIONS = "You are an expert assistant who can solve any task using tool calls. You will be given a task to solve as best you can.\nTo do so, you have been given access to the following tools: {tool_names}\n\nThe tool calls you write are actions: after the tools are executed, you will get the results of the tool calls as \"observations\".\nAt each step, you should first explain your reasoning towards solving the task and the tools that you want to use within <thought> tags.\nThen you should write one or more valid JSON tool calls within <tool_calling> tags.\nThis Thought/Tool_calling/Observation cycle can repeat N times, you should take several steps when needed.\n\nHere are the output format:\n\n<thought>\nYour reasoning and thought process for the current step\n</thought>\n\n<tool_calling>\n[\n  {{\n    \"name\": \"tool_name\",\n    \"arguments\": {{\"param1\": \"value1\", \"param2\": \"value2\"}}\n   }}\n]\n</tool_calling>\n\nYou can call multiple tools at once by including multiple tool objects in the JSON array within the <tool_calling> tags.\n\nONLY output within the <thought> and <tool_calling> sequences. You will get the Observation from the tool calls. Do not output the Observation yourself.\n\nTo provide the final answer to the task, use a tool call with \"name\": \"final_answer\". It is the only way to complete the task, else you will be stuck on a loop. So your final output should look like this:\n\n<thought>\nI have gathered all the necessary information and can now provide the final answer.\n</thought>\n\n<tool_calling>\n[\n  {{\n    \"name\": \"final_answer\",\n    \"arguments\": {{\"answer\": \"insert your final answer here\"}}\n  }}\n]\n</tool_calling>\n\nHere are a few examples using notional tools:\n---\nTask: \"Generate an image of the oldest person in this document and also search for their biography.\"\n\n<thought>\nI need to first find out who is the oldest person in the document, then generate an image of them and search for their biography. I'll start by using the document_qa tool to find the oldest person.\n</thought>\n\n<tool_calling>\n[\n  {{\n    \"name\": \"document_qa\",\n    \"arguments\": {{\"document\": \"document.pdf\", \"question\": \"Who is the oldest person mentioned?\"}}\n  }}\n]\n</tool_calling>\n\nObservation: \"The oldest person in the document is John Doe, a 55 year old lumberjack living in Newfoundland.\"\n\n<thought>\nNow I know the oldest person is John Doe, a 55-year-old lumberjack from Newfoundland. I can generate an image of him and search for his biography at the same time.\n</thought>\n\n<tool_calling>\n[\n {{\n    \"name\": \"image_generator\",\n    \"arguments\": {{\"prompt\": \"A portrait of John Doe, a 55-year-old lumberjack living in Newfoundland, Canada\"}}\n  }},\n  {{\n    \"name\": \"search\",\n    \"arguments\": {{\"query\": \"John Doe lumberjack Newfoundland biography\"}}\n  }}\n]\n</tool_calling>\n\nObservation: \"image.png\" and \"John Doe is a well-known lumberjack in Newfoundland with over 30 years of experience in forestry.\"\n\n<thought>\nI have successfully generated an image and found biographical information. Now I can provide the final answer with both pieces of information.\n</thought>\n\n<tool_calling>\n[\n  {{\n    \"name\": \"final_answer\",\n    \"arguments\": {{\"answer\": \"I have generated an image (image.png) of John Doe, the oldest person in the document, and found that he is a well-known lumberjack in Newfoundland with over 30 years of experience in forestry.\"}}\n  }}\n]\n</tool_calling>\n\n---\nTask: \"What is the result of the following operation: 5 + 3 + 1294.678?\"\n\n<thought>\nI need to calculate the sum of 5 + 3 + 1294.678. I'll use the python_interpreter tool to execute this calculation.\n</thought>\n\n<tool_calling>\n[\n  {{\n    \"name\": \"python_interpreter\",\n    \"arguments\": {{\"code\": \"5 + 3 + 1294.678\"}}\n  }}\n]\n</tool_calling>\n\nObservation: 1302.678\n\n<thought>\nI have calculated the result. Now I can provide the final answer.\n</thought>\n\n<tool_calling>\n[\n  {{\n    \"name\": \"final_answer\",\n    \"arguments\": {{\"answer\": \"1302.678\"}}\n  }}\n]\n</tool_calling>\n\n---\nTask: \"Which city has the highest population, Guangzhou or Shanghai?\"\n\n<thought>\nI need to search for the population data of both Guangzhou and Shanghai to compare them. I'll search for both cities' population information simultaneously.\n</thought>\n\n<tool_calling>\n[\n  {{\n    \"name\": \"search\",\n    \"arguments\": {{\"query\": \"Population Guangzhou 2023\"}}\n  }},\n  {{\n    \"name\": \"search\",\n    \"arguments\": {{\"query\": \"Population Shanghai 2023\"}}\n  }}\n]\n</tool_calling>\n\nObservation: ['Guangzhou has a population of 15 million inhabitants as of 2021.'] and ['Shanghai has a population of 26 million (2019)']\n\n<thought>\nBased on the search results, Shanghai has a population of 26 million while Guangzhou has 15 million. Therefore, Shanghai has the higher population.\n</thought>\n\n<tool_calling>\n[\n  {{\n    \"name\": \"final_answer\",\n    \"arguments\": {{\"answer\": \"Shanghai has the highest population with 26 million people, compared to Guangzhou's 15 million people.\"}}\n  }}\n]\n</tool_calling>\n\nAbove examples were using notional tools that might not exist for you. You only have access to these tools:\n\n{tool_descriptions}\n\nHere are the rules you should always follow to solve your task:\n1. ALWAYS provide tool calls within <tool_calling> tags, else you will fail.\n2. Always use the right arguments for the tools. Never use variable names as the action arguments, use the actual values instead.\n3. You can call multiple tools at once if it makes sense for efficiency.\n4. Call tools only when needed: do not call the search agent if you do not need information, try to solve the task yourself.\nIf no tool call is needed, use final_answer tool to return your answer.\n5. Never re-do a tool call that you previously did with the exact same parameters.\n6. Always include your reasoning in <thought> tags before making tool calls.\n\nNow Begin! If you solve the task correctly, you will receive a reward of $1,000,000.";

package/lib/llm-core/chain/prompt.cjs

@@ -239,7 +239,6 @@ Your goal is to craft an insightful, engaging response that seamlessly integrate
         [await loreBooksPrompt.format({ input: array.join("\n") })],
         variables
       ).then((value) => value[0]);
-      console.log(message);
       if (position === "default") {
         if (hasLongMemory) {
           const index = result.findIndex(

package/lib/llm-core/chain/prompt.mjs

@@ -226,7 +226,6 @@ Your goal is to craft an insightful, engaging response that seamlessly integrate
         [await loreBooksPrompt.format({ input: array.join("\n") })],
         variables
       ).then((value) => value[0]);
-      console.log(message);
       if (position === "default") {
         if (hasLongMemory) {
           const index = result.findIndex(

package/lib/services/chat.cjs

@@ -726,7 +726,7 @@ var MessageTransformer = class {
     if (session.quote && !quote && this._config.includeQuoteReply) {
       const quoteMessage = await this.transform(
         session,
-        session.quote.elements,
+        session.quote.elements ?? [],
         {
           content: "",
           additional_kwargs: {}
@@ -742,10 +742,15 @@ Please consider this quote when generating your response. User's message: ${mess
       }
       if (quoteMessage.additional_kwargs["images"]) {
         const currentImages = message.additional_kwargs["images"] ?? [];
+        const currentImageHashs = message.additional_kwargs["imageHashs"] ?? [];
         message.additional_kwargs["images"] = [
           ...currentImages,
           ...quoteMessage.additional_kwargs["images"]
         ];
+        message.additional_kwargs["imageHashs"] = [
+          ...currentImageHashs,
+          ...quoteMessage.additional_kwargs["imageHashs"]
+        ];
       }
     }
     return message;

package/lib/services/chat.mjs

@@ -713,7 +713,7 @@ var MessageTransformer = class {
     if (session.quote && !quote && this._config.includeQuoteReply) {
       const quoteMessage = await this.transform(
         session,
-        session.quote.elements,
+        session.quote.elements ?? [],
         {
           content: "",
           additional_kwargs: {}
@@ -729,10 +729,15 @@ Please consider this quote when generating your response. User's message: ${mess
       }
       if (quoteMessage.additional_kwargs["images"]) {
         const currentImages = message.additional_kwargs["images"] ?? [];
+        const currentImageHashs = message.additional_kwargs["imageHashs"] ?? [];
         message.additional_kwargs["images"] = [
           ...currentImages,
           ...quoteMessage.additional_kwargs["images"]
         ];
+        message.additional_kwargs["imageHashs"] = [
+          ...currentImageHashs,
+          ...quoteMessage.additional_kwargs["imageHashs"]
+        ];
       }
     }
     return message;

package/lib/utils/lock.cjs

@@ -44,6 +44,12 @@ var ObjectLock = class {
       }
     }, "unlock");
     if (this._lock) {
+      let error = null;
+      try {
+        throw new Error(`Lock timeout after ${this._timeout}ms`);
+      } catch (e) {
+        error = e;
+      }
       return new Promise((resolve, reject) => {
         const timeoutId = setTimeout(() => {
           const index = this._queue.findIndex(
@@ -52,7 +58,9 @@ var ObjectLock = class {
           if (index !== -1) {
             this._queue.splice(index, 1);
           }
-          reject(new Error(`Lock timeout after ${this._timeout}ms`));
+          reject(
+            error ?? new Error(`Lock timeout after ${this._timeout}ms`)
+          );
         }, this._timeout);
         this._queue.push({
           resolve: /* @__PURE__ */ __name((unlockFn) => {

package/lib/utils/lock.mjs

@@ -23,6 +23,12 @@ var ObjectLock = class {
       }
     }, "unlock");
     if (this._lock) {
+      let error = null;
+      try {
+        throw new Error(`Lock timeout after ${this._timeout}ms`);
+      } catch (e) {
+        error = e;
+      }
       return new Promise((resolve, reject) => {
         const timeoutId = setTimeout(() => {
           const index = this._queue.findIndex(
@@ -31,7 +37,9 @@ var ObjectLock = class {
           if (index !== -1) {
             this._queue.splice(index, 1);
           }
-          reject(new Error(`Lock timeout after ${this._timeout}ms`));
+          reject(
+            error ?? new Error(`Lock timeout after ${this._timeout}ms`)
+          );
         }, this._timeout);
         this._queue.push({
           resolve: /* @__PURE__ */ __name((unlockFn) => {

package/lib/utils/string.cjs

@@ -32,6 +32,7 @@ __export(string_exports, {
   getTimeInUTC: () => getTimeInUTC,
   gzipDecode: () => gzipDecode,
   gzipEncode: () => gzipEncode,
+  hashString: () => hashString,
   rollDice: () => rollDice,
   selectFromList: () => selectFromList
 });
@@ -224,6 +225,16 @@ function bufferToArrayBuffer(buffer) {
   return arrayBuffer;
 }
 __name(bufferToArrayBuffer, "bufferToArrayBuffer");
+async function hashString(text, length = 8) {
+  const hash = await crypto.subtle.digest(
+    "SHA-256",
+    new TextEncoder().encode(text)
+  );
+  const hashArray = Array.from(new Uint8Array(hash));
+  const hashString2 = hashArray.map((b) => b.toString(16).padStart(2, "0")).join("");
+  return hashString2.substring(0, length);
+}
+__name(hashString, "hashString");
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
   PresetPostHandler,
@@ -238,6 +249,7 @@ __name(bufferToArrayBuffer, "bufferToArrayBuffer");
   getTimeInUTC,
   gzipDecode,
   gzipEncode,
+  hashString,
   rollDice,
   selectFromList
 });

package/lib/utils/string.d.ts

@@ -29,4 +29,5 @@ export declare class PresetPostHandler implements PostHandler {
 export declare function gzipEncode<T extends Encoding = 'buffer'>(text: string, encoding?: T): Promise<BufferType<T>>;
 export declare function gzipDecode(data: ArrayBuffer | Buffer | string, inputEncoding?: Encoding): Promise<string>;
 export declare function bufferToArrayBuffer(buffer: Buffer): ArrayBuffer;
+export declare function hashString(text: string, length?: number): Promise<string>;
 export {};

package/lib/utils/string.mjs

@@ -190,6 +190,16 @@ function bufferToArrayBuffer(buffer) {
   return arrayBuffer;
 }
 __name(bufferToArrayBuffer, "bufferToArrayBuffer");
+async function hashString(text, length = 8) {
+  const hash = await crypto.subtle.digest(
+    "SHA-256",
+    new TextEncoder().encode(text)
+  );
+  const hashArray = Array.from(new Uint8Array(hash));
+  const hashString2 = hashArray.map((b) => b.toString(16).padStart(2, "0")).join("");
+  return hashString2.substring(0, length);
+}
+__name(hashString, "hashString");
 export {
   PresetPostHandler,
   bufferToArrayBuffer,
@@ -203,6 +213,7 @@ export {
   getTimeInUTC,
   gzipDecode,
   gzipEncode,
+  hashString,
   rollDice,
   selectFromList
 };

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "koishi-plugin-chatluna",
   "description": "chatluna for koishi",
-  "version": "1.2.0",
+  "version": "1.2.2",
   "main": "lib/index.cjs",
   "module": "lib/index.mjs",
   "typings": "lib/index.d.ts",