Chat Stream Integration (#52)

* rm: axios dependency * add: stream parsing for normal style * fix: empty string problem * add: stream for embedded prompts * update: version increment
2025-12-12 11:56:06 -05:00 · 2024-04-21 14:40:30 -07:00
parent bc989580a9
commit d67106c03e
10 changed files with 140 additions and 173 deletions
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -8,7 +8,7 @@ services:
    build: ./                     # find docker file in designated path
    container_name: discord
    restart: always               # rebuild container always
-    image: discord/bot:0.4.0
+    image: discord/bot:0.4.2
    environment:
      CLIENT_TOKEN: ${CLIENT_TOKEN}
      GUILD_ID: ${GUILD_ID}
--- a/package-lock.json
+++ b/package-lock.json
@@ -9,7 +9,6 @@
      "version": "0.4.0",
      "license": "ISC",
      "dependencies": {
-        "axios": "^1.6.2",
        "discord.js": "^14.14.1",
        "dotenv": "^16.3.1",
        "ollama": "^0.5.0"
@@ -661,21 +660,6 @@
      "integrity": "sha512-58S9QDqG0Xx27YwPSt9fJxivjYl432YCwfDMfZ+71RAqUrZef7LrKQZ3LHLOwCS4FLNBplP533Zx895SeOCHvA==",
      "dev": true
    },
-    "node_modules/asynckit": {
-      "version": "0.4.0",
-      "resolved": "https://registry.npmjs.org/asynckit/-/asynckit-0.4.0.tgz",
-      "integrity": "sha512-Oei9OH4tRh0YqU3GxhX79dM/mwVgvbZJaSNaRk+bshkj0S5cfHcgYakreBjrHwatXKbz+IoIdYLxrKim2MjW0Q=="
-    },
-    "node_modules/axios": {
-      "version": "1.6.8",
-      "resolved": "https://registry.npmjs.org/axios/-/axios-1.6.8.tgz",
-      "integrity": "sha512-v/ZHtJDU39mDpyBoFVkETcd/uNdxrWRrg3bKpOKzXFA6Bvqopts6ALSMU3y6ijYxbw2B+wPrIv46egTzJXCLGQ==",
-      "dependencies": {
-        "follow-redirects": "^1.15.6",
-        "form-data": "^4.0.0",
-        "proxy-from-env": "^1.1.0"
-      }
-    },
    "node_modules/balanced-match": {
      "version": "1.0.2",
      "resolved": "https://registry.npmjs.org/balanced-match/-/balanced-match-1.0.2.tgz",
@@ -740,17 +724,6 @@
        "fsevents": "~2.3.2"
      }
    },
-    "node_modules/combined-stream": {
-      "version": "1.0.8",
-      "resolved": "https://registry.npmjs.org/combined-stream/-/combined-stream-1.0.8.tgz",
-      "integrity": "sha512-FQN4MRfuJeHf7cBbBMJFXhKSDq+2kAArBlmRBvcvFE5BB1HZKXtSFASDhdlz9zOYwxh8lDdnvmMOe/+5cdoEdg==",
-      "dependencies": {
-        "delayed-stream": "~1.0.0"
-      },
-      "engines": {
-        "node": ">= 0.8"
-      }
-    },
    "node_modules/concat-map": {
      "version": "0.0.1",
      "resolved": "https://registry.npmjs.org/concat-map/-/concat-map-0.0.1.tgz",
@@ -780,14 +753,6 @@
        }
      }
    },
-    "node_modules/delayed-stream": {
-      "version": "1.0.0",
-      "resolved": "https://registry.npmjs.org/delayed-stream/-/delayed-stream-1.0.0.tgz",
-      "integrity": "sha512-ZySD7Nf91aLB0RxL4KGrKHBXl7Eds1DAmEdcoVawXnLD7SDhpNgtuII2aAkg7a7QS41jxPSZ17p4VdGnMHk3MQ==",
-      "engines": {
-        "node": ">=0.4.0"
-      }
-    },
    "node_modules/diff": {
      "version": "4.0.2",
      "resolved": "https://registry.npmjs.org/diff/-/diff-4.0.2.tgz",
@@ -892,38 +857,6 @@
        "node": ">=8"
      }
    },
-    "node_modules/follow-redirects": {
-      "version": "1.15.6",
-      "resolved": "https://registry.npmjs.org/follow-redirects/-/follow-redirects-1.15.6.tgz",
-      "integrity": "sha512-wWN62YITEaOpSK584EZXJafH1AGpO8RVgElfkuXbTOrPX4fIfOyEpW/CsiNd8JdYrAoOvafRTOEnvsO++qCqFA==",
-      "funding": [
-        {
-          "type": "individual",
-          "url": "https://github.com/sponsors/RubenVerborgh"
-        }
-      ],
-      "engines": {
-        "node": ">=4.0"
-      },
-      "peerDependenciesMeta": {
-        "debug": {
-          "optional": true
-        }
-      }
-    },
-    "node_modules/form-data": {
-      "version": "4.0.0",
-      "resolved": "https://registry.npmjs.org/form-data/-/form-data-4.0.0.tgz",
-      "integrity": "sha512-ETEklSGi5t0QMZuiXoA/Q6vcnxcLQP5vdugSpuAyi6SVGi2clPPp+xgEhuMaHC+zGgn31Kd235W35f7Hykkaww==",
-      "dependencies": {
-        "asynckit": "^0.4.0",
-        "combined-stream": "^1.0.8",
-        "mime-types": "^2.1.12"
-      },
-      "engines": {
-        "node": ">= 6"
-      }
-    },
    "node_modules/fsevents": {
      "version": "2.3.3",
      "resolved": "https://registry.npmjs.org/fsevents/-/fsevents-2.3.3.tgz",
@@ -1052,25 +985,6 @@
      "integrity": "sha512-s8UhlNe7vPKomQhC1qFelMokr/Sc3AgNbso3n74mVPA5LTZwkB9NlXf4XPamLxJE8h0gh73rM94xvwRT2CVInw==",
      "dev": true
    },
-    "node_modules/mime-db": {
-      "version": "1.52.0",
-      "resolved": "https://registry.npmjs.org/mime-db/-/mime-db-1.52.0.tgz",
-      "integrity": "sha512-sPU4uV7dYlvtWJxwwxHD0PuihVNiE7TyAbQ5SWxDCB9mUYvOgroQOwYQQOKPJ8CIbE+1ETVlOoK1UC2nU3gYvg==",
-      "engines": {
-        "node": ">= 0.6"
-      }
-    },
-    "node_modules/mime-types": {
-      "version": "2.1.35",
-      "resolved": "https://registry.npmjs.org/mime-types/-/mime-types-2.1.35.tgz",
-      "integrity": "sha512-ZDY+bPm5zTTF+YpCrAU9nK0UgICYPT0QtT1NZWFv4s++TNkcgVaT0g6+4R2uI4MjQjzysHB1zxuWL50hzaeXiw==",
-      "dependencies": {
-        "mime-db": "1.52.0"
-      },
-      "engines": {
-        "node": ">= 0.6"
-      }
-    },
    "node_modules/minimatch": {
      "version": "3.1.2",
      "resolved": "https://registry.npmjs.org/minimatch/-/minimatch-3.1.2.tgz",
@@ -1161,11 +1075,6 @@
        "url": "https://github.com/sponsors/jonschlinkert"
      }
    },
-    "node_modules/proxy-from-env": {
-      "version": "1.1.0",
-      "resolved": "https://registry.npmjs.org/proxy-from-env/-/proxy-from-env-1.1.0.tgz",
-      "integrity": "sha512-D+zkORCbA9f1tdWRK0RaCR3GPv50cMxcrz4X8k5LTSUD1Dkw47mKJEZQNunItRTkWwgtaUSo1RVFRIG9ZXiFYg=="
-    },
    "node_modules/pstree.remy": {
      "version": "1.1.8",
      "resolved": "https://registry.npmjs.org/pstree.remy/-/pstree.remy-1.1.8.tgz",
--- a/package.json
+++ b/package.json
@@ -1,6 +1,6 @@
 {
  "name": "discord-ollama",
-  "version": "0.4.0",
+  "version": "0.4.2",
  "description": "Ollama Integration into discord",
  "main": "build/index.js",
  "exports": "./build/index.js",
@@ -24,7 +24,6 @@
  "author": "Kevin Dang",
  "license": "ISC",
  "dependencies": {
-    "axios": "^1.6.2",
    "discord.js": "^14.14.1",
    "dotenv": "^16.3.1",
    "ollama": "^0.5.0"
--- a/src/events/messageCreate.ts
+++ b/src/events/messageCreate.ts
@@ -19,6 +19,8 @@ export default event(Events.MessageCreate, async ({ log, msgHist, tokens, ollama
    // Only respond if message mentions the bot
    if (!message.mentions.has(tokens.clientUid)) return

+    let shouldStream = false
+
    // Try to query and send embed     
    try {
        const config: Configuration = await new Promise((resolve, reject) => {
@@ -45,11 +47,14 @@ export default event(Events.MessageCreate, async ({ log, msgHist, tokens, ollama
                    msgHist.capacity = config.options['modify-capacity']
                }

+                // set stream state
+                shouldStream = config.options['message-stream'] as boolean || false
+
                resolve(config)
            })
        })

-        let response: ChatResponse    
+        let response: string    

        // check if we can push, if not, remove oldest
        while (msgHist.size() >= msgHist.capacity) msgHist.dequeue()
@@ -62,9 +67,9 @@ export default event(Events.MessageCreate, async ({ log, msgHist, tokens, ollama
        
        // undefined or false, use normal, otherwise use embed
        if (config.options['message-style'])
-            response = await embedMessage(message, ollama, tokens, msgHist)
+            response = await embedMessage(message, ollama, tokens, msgHist, shouldStream)
        else
-            response = await normalMessage(message, ollama, tokens, msgHist)
+            response = await normalMessage(message, ollama, tokens, msgHist, shouldStream)

        // If something bad happened, remove user query and stop
        if (response == undefined) { msgHist.pop(); return }
@@ -75,7 +80,7 @@ export default event(Events.MessageCreate, async ({ log, msgHist, tokens, ollama
        // successful query, save it in context history
        msgHist.enqueue({ 
            role: 'assistant', 
-            content: response.message.content 
+            content: response
        })
    } catch (error: any) {
        msgHist.pop() // remove message because of failure
--- a/src/utils/events.ts
+++ b/src/utils/events.ts
@@ -20,6 +20,18 @@ export type Tokens = {
    clientUid: string
 }

+/**
+ * Parameters to run the chat query
+ * @param model the model to run
+ * @param ollama ollama api client
+ * @param msgHist message history
+ */
+export type ChatParams = {
+    model: string, 
+    ollama: Ollama, 
+    msgHist: UserMessage[]
+}
+
 /** 
 * Format for the messages to be stored when communicating when the bot
 * @param role either assistant, user, or system
--- a/src/utils/index.ts
+++ b/src/utils/index.ts
@@ -3,4 +3,5 @@ export * from './env.js'
 export * from './events.js'
 export * from './messageEmbed.js'
 export * from './messageNormal.js'
-export * from './commands.js'
+export * from './commands.js'
+export * from './streamHandler.js'
--- a/src/utils/messageEmbed.ts
+++ b/src/utils/messageEmbed.ts
@@ -1,6 +1,6 @@
 import { EmbedBuilder, Message } from 'discord.js'
 import { ChatResponse, Ollama } from 'ollama'
-import { UserMessage } from './events.js'
+import { ChatParams, UserMessage, streamResponse, blockResponse } from './index.js'
 import { Queue } from '../queues/queue.js'

 /**
@@ -16,10 +16,12 @@ export async function embedMessage(
        channel: string,
        model: string
    },
-    msgHist: Queue<UserMessage>
-) {
+    msgHist: Queue<UserMessage>,
+    stream: boolean
+): Promise<string> {
    // bot response
-    let response: ChatResponse
+    let response: ChatResponse | AsyncGenerator<ChatResponse, any, unknown>
+    let result: string = ''

    // initial message to client
    const botMessage = new EmbedBuilder()
@@ -30,28 +32,43 @@ export async function embedMessage(
    // send the message
    const sentMessage = await message.channel.send({ embeds: [botMessage] })

-    try {
-        // Attempt to query model for message
-        response = await ollama.chat({
-            model: tokens.model,
-            messages: msgHist.getItems(),
-            options: {
-                num_thread: 8, // remove if optimization needed further
-                mirostat: 1,
-                mirostat_tau: 2.0,
-                top_k: 70
-            },
-            stream: false
-        })
+    // create params
+    const params: ChatParams = {
+        model: tokens.model,
+        ollama: ollama,
+        msgHist: msgHist.getItems()
+    }

-        // dummy message to let user know that query is underway
-        const newEmbed = new EmbedBuilder()
+    try {
+        // check if embed needs to stream
+        if (stream) {
+            response = await streamResponse(params)
+
+            for await (const portion of response) {
+                result += portion.message.content
+
+                // new embed per token...
+                const newEmbed = new EmbedBuilder()
+                .setTitle(`Responding to ${message.author.tag}`)
+                .setDescription(result || 'No Content Yet...')
+                .setColor('#00FF00')
+
+                // edit the message
+                sentMessage.edit({ embeds: [newEmbed] })
+            }
+        } else {
+            response = await blockResponse(params)
+            result = response.message.content
+
+            // only need to create 1 embed again
+            const newEmbed = new EmbedBuilder()
            .setTitle(`Responding to ${message.author.tag}`)
-            .setDescription(response.message.content || 'No Content to Provide...')
+            .setDescription(result || 'No Content to Provide...')
            .setColor('#00FF00')

-        // edit the message
-        sentMessage.edit({ embeds: [newEmbed] })
+            // edit the message
+            sentMessage.edit({ embeds: [newEmbed] })
+        }
    } catch(error: any) {
        console.log(`[Util: messageEmbed] Error creating message: ${error.message}`)
        const errorEmbed = new EmbedBuilder()
@@ -64,5 +81,5 @@ export async function embedMessage(
    } 

    // Hope there is a response! undefined otherwie
-    return response!!
+    return result
 }
--- a/src/utils/messageNormal.ts
+++ b/src/utils/messageNormal.ts
@@ -1,6 +1,6 @@
 import { Message } from 'discord.js'
 import { ChatResponse, Ollama } from 'ollama'
-import { UserMessage } from './events.js'
+import { ChatParams, UserMessage, streamResponse, blockResponse } from './index.js'
 import { Queue } from '../queues/queue.js'

 /**
@@ -16,38 +16,49 @@ export async function normalMessage(
        channel: string,
        model: string
    },
-    msgHist: Queue<UserMessage>
-) {
+    msgHist: Queue<UserMessage>,
+    stream: boolean
+): Promise<string> {
    // bot's respnse
-    let response: ChatResponse
+    let response: ChatResponse | AsyncGenerator<ChatResponse, any, unknown>
+    let result: string = ''

    await message.channel.send('Generating Response . . .').then(async sentMessage => {
        try {
-            // Attempt to query model for message
-            response = await ollama.chat({
+            const params: ChatParams = {
                model: tokens.model,
-                messages: msgHist.getItems(),
-                options: {
-                    num_thread: 8, // remove if optimization needed further
-                    mirostat: 1,
-                    mirostat_tau: 2.0,
-                    top_k: 70
-                },
-                stream: false
-            })
+                ollama: ollama,
+                msgHist: msgHist.getItems()
+            } 

-            // check if message length > discord max for normal messages
-            if (response.message.content.length > 2000) {
-                sentMessage.edit(response.message.content.slice(0, 2000))
-                message.channel.send(response.message.content.slice(2000))
-            } else // edit the 'generic' response to new message
-                sentMessage.edit(response.message.content)            
+            // run query based on stream preference, true = stream, false = block
+            if (stream) {
+                response = await streamResponse(params)
+                for await (const portion of response) {
+                    // append token to message
+                    result += portion.message.content
+
+                    // resent current output, THIS WILL BE SLOW due to discord limits!
+                    sentMessage.edit(result || 'No Content Yet...')
+                }
+            }
+            else {
+                response = await blockResponse(params)
+                result = response.message.content
+
+                // check if message length > discord max for normal messages
+                if (result.length > 2000) {
+                    sentMessage.edit(result.slice(0, 2000))
+                    message.channel.send(result.slice(2000))
+                } else // edit the 'generic' response to new message
+                    sentMessage.edit(result)
+            }            
        } catch(error: any) {
            console.log(`[Util: messageNormal] Error creating message: ${error.message}`)
            sentMessage.edit(`**Response generation failed.**\n\nReason: ${error.message}`)
        }
    })

-    // Hope there is a response, force client to believe
-    return response!!
-}
+    // return the string representation of response
+    return result
+}
--- a/src/utils/streamHandler.ts
+++ b/src/utils/streamHandler.ts
@@ -0,0 +1,40 @@
+import { ChatResponse } from "ollama"
+import { ChatParams } from "./index.js"
+
+/**
+ * Method to query the Ollama client for async generation
+ * @param params 
+ * @returns Asyn
+ */
+export async function streamResponse(params: ChatParams): Promise<AsyncGenerator<ChatResponse, any, unknown>> {
+    return await params.ollama.chat({
+        model: params.model,
+        messages: params.msgHist,
+        options: {
+            num_thread: 8, // remove if optimization needed further
+            mirostat: 1,
+            mirostat_tau: 2.0,
+            top_k: 70
+        },
+        stream: true
+    })
+}
+
+/**
+ * Method to query the Ollama client for a block response 
+ * @param params parameters to query the client
+ * @returns ChatResponse generated by the Ollama client
+ */
+export async function blockResponse(params: ChatParams): Promise<ChatResponse> {
+    return await params.ollama.chat({
+        model: params.model,
+        messages: params.msgHist,
+        options: {
+            num_thread: 8, // remove if optimization needed further
+            mirostat: 1,
+            mirostat_tau: 2.0,
+            top_k: 70
+        },
+        stream: false
+    })
+}
--- a/src/utils/streamParse.ts
+++ b/src/utils/streamParse.ts
@@ -1,27 +0,0 @@
-import { AxiosResponse } from 'axios'
-
-/**
- * When running a /api/chat stream, the output needs to be parsed into an array of objects
- *  This method is used for development purposes and testing
- * 
- * This will not work as intended with the inclusion of ollama-js, needs to be modified to work with it
- * 
- * @param stream Axios response to from Ollama
- */
-export function parseStream(stream: AxiosResponse<any, any>) {
-    // split string by newline
-    const keywordObjects: string[] = stream.data.trim().split('\n')
-
-    // parse string and load them into objects
-    const keywordsArray: {
-        model: string,
-        created_at: string,
-        message: {
-            role: string,
-            content: string
-        },
-        done: boolean
-    }[] = keywordObjects.map((keywordString) => JSON.parse(keywordString))
-
-    return keywordsArray
-}