feat: add ollama as supported provider (#1543)

* feat: add ollama as supported provider

*This implementation is only working with `stream = true`*
- Uses the actual ollama api and allows for passing additional options
- Properly passes the system prompt to api

Use ollama as provider in opts like this:
opts = {
        debug = true,
        provider = "ollama",
        ollama = {
                api_key_name = "",
                endpoint = "http://127.0.0.1:11434",
                model = "qwen2.5-coder:latest",
                options = {
                        num_ctx = 32768,
                        temperature = 0,
                },
                stream = true,
        },

* fix: ollama types

---------

Co-authored-by: jtabke <25010496+jtabke@users.noreply.github.com>

README.md

@@ -683,6 +683,17 @@ return {
 
 See [highlights.lua](./lua/avante/highlights.lua) for more information
 
+## Ollama
+
+ollama is a first-class provider for avante.nvim. You can use it by setting `provider = "ollama"` in the configuration, and set the `model` field in `ollama` to the model you want to use. For example:
+
+```lua
+provider = "ollama",
+ollama = {
+  model = "qwq:32b",
+}
+```
+
 ## Custom providers
 
 Avante provides a set of default providers, but users can also create their own providers.

lua/avante/config.lua

@@ -20,7 +20,7 @@ local M = {}
 ---@field custom_tools AvanteLLMToolPublic[]
 M._defaults = {
   debug = false,
-  ---@alias ProviderName "claude" | "openai" | "azure" | "gemini" | "vertex" | "cohere" | "copilot" | "bedrock" | string
+  ---@alias ProviderName "claude" | "openai" | "azure" | "gemini" | "vertex" | "cohere" | "copilot" | "bedrock" | "ollama" | string
   provider = "claude",
   -- WARNING: Since auto-suggestions are a high-frequency operation and therefore expensive,
   -- currently designating it as `copilot` provider is dangerous because: https://github.com/yetone/avante.nvim/issues/1048
@@ -255,6 +255,15 @@ M._defaults = {
     temperature = 0,
     max_tokens = 4096,
   },
+  ---@type AvanteSupportedProvider
+  ollama = {
+    endpoint = "http://127.0.0.1:11434",
+    timeout = 30000, -- Timeout in milliseconds
+    options = {
+      temperature = 0,
+      num_ctx = 4096,
+    },
+  },
   ---To add support for custom provider, follow the format below
   ---See https://github.com/yetone/avante.nvim/wiki#custom-providers for more details
   ---@type {[string]: AvanteProvider}

lua/avante/providers/init.lua

@@ -18,6 +18,7 @@ local DressingState = { winid = nil, input_winid = nil, input_bufnr = nil }
 ---@field gemini AvanteProviderFunctor
 ---@field cohere AvanteProviderFunctor
 ---@field bedrock AvanteBedrockProviderFunctor
+---@field ollama AvanteProviderFunctor
 local M = {}
 
 ---@class EnvironmentHandler
@@ -152,8 +153,13 @@ M = setmetatable(M, {
   __index = function(t, k)
     local provider_config = M.get_config(k)
 
+    if Config.vendors[k] ~= nil and k == "ollama" then
+      Utils.warn(
+        "ollama is now a first-class provider in avante.nvim, please stop using vendors to define ollama, for migration guide please refer to: https://github.com/yetone/avante.nvim/wiki/Custom-providers#ollama"
+      )
+    end
     ---@diagnostic disable: undefined-field,no-unknown,inject-field
-    if Config.vendors[k] ~= nil then
+    if Config.vendors[k] ~= nil and k ~= "ollama" then
       if provider_config.parse_response_data ~= nil then
         Utils.error("parse_response_data is not supported for avante.nvim vendors")
       end

lua/avante/providers/ollama.lua

@@ -0,0 +1,70 @@
+local Utils = require("avante.utils")
+local P = require("avante.providers")
+
+---@class AvanteProviderFunctor
+local M = {}
+
+M.api_key_name = "" -- Ollama typically doesn't require API keys for local use
+
+M.role_map = {
+  user = "user",
+  assistant = "assistant",
+}
+
+M.parse_messages = P.openai.parse_messages
+M.is_o_series_model = P.openai.is_o_series_model
+
+function M:is_disable_stream() return false end
+
+function M:parse_stream_data(ctx, data, handler_opts)
+  local ok, json_data = pcall(vim.json.decode, data)
+  if not ok or not json_data then
+    -- Add debug logging
+    Utils.debug("Failed to parse JSON", data)
+    return
+  end
+
+  if json_data.message and json_data.message.content then
+    local content = json_data.message.content
+    if content and content ~= "" then handler_opts.on_chunk(content) end
+  end
+
+  if json_data.done then
+    handler_opts.on_stop({ reason = "complete" })
+    return
+  end
+end
+
+---@param prompt_opts AvantePromptOptions
+function M:parse_curl_args(prompt_opts)
+  local provider_conf, request_body = P.parse_config(self)
+
+  if not provider_conf.model or provider_conf.model == "" then error("Ollama model must be specified in config") end
+  if not provider_conf.endpoint then error("Ollama requires endpoint configuration") end
+
+  return {
+    url = Utils.url_join(provider_conf.endpoint, "/api/chat"),
+    headers = {
+      ["Content-Type"] = "application/json",
+      ["Accept"] = "application/json",
+    },
+    body = vim.tbl_deep_extend("force", {
+      model = provider_conf.model,
+      messages = self:parse_messages(prompt_opts),
+      stream = true,
+      system = prompt_opts.system_prompt,
+    }, request_body),
+  }
+end
+
+---@param result table
+M.on_error = function(result)
+  local error_msg = "Ollama API error"
+  if result.body then
+    local ok, body = pcall(vim.json.decode, result.body)
+    if ok and body.error then error_msg = body.error end
+  end
+  Utils.error(error_msg, { title = "Ollama" })
+end
+
+return M

lua/avante/providers/openai.lua

@@ -67,12 +67,11 @@ function M.is_o_series_model(model) return model and string.match(model, "^o%d+"
 
 function M:parse_messages(opts)
   local messages = {}
-  local provider = P[Config.provider]
+  local provider_conf, _ = P.parse_config(self)
-  local base, _ = P.parse_config(provider)
 
   -- NOTE: Handle the case where the selected model is the `o1` model
   -- "o1" models are "smart" enough to understand user prompt as a system prompt in this context
-  if self.is_o_series_model(base.model) then
+  if self.is_o_series_model(provider_conf.model) then
     table.insert(messages, { role = "user", content = opts.system_prompt })
   else
     table.insert(messages, { role = "system", content = opts.system_prompt })