ollama_chat 0.0.119 → 0.0.120

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: a7b876b681130c1ff2ec672460bb04aa85ed095e9358dbdaa62d744f82d68720
-  data.tar.gz: 3f85175d15231880d45e8f192f9f9af47c861cb8f1475d2fe653309ca173ab1c
+  metadata.gz: cfff20e994aa12d921fd4ce0d050fbaaae9d66ade06fe419d6dbd9ce0dad1a4e
+  data.tar.gz: 45f485ac6fc502e6634626d6e891402acf74516491af95a8e26773905924d5a0
 SHA512:
-  metadata.gz: 9cf60761b6ce28cb22873b30808d3398ffcefc9e9991fcf781621b4db8a8fed5fd6cca21de046d8bf2851dd60a7529cfa162125ca4ee567e32c09042f4dc4e81
-  data.tar.gz: a2f1e6210141a8f00c4b9428453c49e4e1673e51521f788dcec694cd2c2c9963542bd491618ec8d1a4bb3ad5788588ca3d2c7f51ed3b67b977eaac1b4a59115a
+  metadata.gz: 13a2ca128a857c8980684bf437a5c9470d92b42babed878010db2eaba75de655f5ded18f7baccf3ef27d05191c4edfad7fd1c2a190a3f098ff880e0fefe0606d
+  data.tar.gz: c6785ad30f764d56ea19c8bdaaad6dee89385317ecd4a15c6925f12f5637f8a31c4393dcca4143669f2771bd594a9a383ff24e4fdd1e1dcbd070f357128a0446

data/CHANGES.md

@@ -1,5 +1,50 @@
 # Changes
 
+## 2026-09-18 v0.0.120
+
+*   **TTS / audio.cpp integration**
+    *   Add `OllamaChat::TTS` with parallel block synthesis via worker threads,
+        ordered playback through `OrderedQueue`, RIFF-header stripping,
+        config-driven streaming, and `join_workers` for test teardown.
+    *   Add `OllamaChat::Utils::AudioPlayer` (PCM pipe to `ffplay` with silence
+        injection to prevent EOF hangs), `OllamaChat::Utils::OrderedQueue`
+        (thread-safe min-heap using `Mutex` + `ConditionVariable`), and
+        `OllamaChat::Utils::ExconLogger` (routes `Excon` log calls through
+        `chat.log(:debug)`).
+    *   Add `KramdownANSI#kramdown_markdown_remove` to strip markdown to plain
+        text for TTS playback.
+    *   Add `TTS_URL` (required, sensitive, URI-decoded) and
+        `AUDIO_PLAYER_CONFIG` (JSON: `command`, `frequency`, `bits`, `pause`)
+        to `OC`, enabling `ConstConf::JSONPlugin`.
+    *   Default `voice.handler` to `OllamaChat::TTS` in `default_config.yml`;
+        add `model: 'chatterbox'` and `stream: { enabled: false, format:
+        'audio' }`.
+    *   Simplify `state_selectors.rb` to a single
+        `ask_and_send(:voices, model:)` call with `Array()` guard.
+    *   Switch `generate_image.rb` to `Excon.new(url, logger:)` with
+        `ExconLogger`.
+    *   Add specs: `tts_spec.rb`, `audio_player_spec.rb`,
+        `excon_logger_spec.rb`, `ordered_queue_spec.rb`; migrate existing
+        specs from `chat.send` to `chat.expose`.
+
+*   **`include_hidden` in `directory_structure`**
+    *   Add `include_hidden:` keyword argument to `generate_structure` and
+        `recurse_generate_structure` in `analyze_directory.rb` with YARD docs;
+        fix the recursive call to forward the flag so dotfiles are honored at
+        all depths.
+    *   Add an `include_hidden` boolean parameter to the `directory_structure`
+        tool definition and pass-through in `execute`.
+    *   `parsing.rb` now passes `include_hidden: true` for all directory
+        references.
+    *   Add `include_hidden` tests to `analyze_directory_spec.rb`,
+        `directory_structure_spec.rb`, and `parsing_spec.rb`.
+
+*   **Housekeeping**
+    *   Bump `tins` to **~> 1.57** in `Rakefile` and gemspec (kwarg support for
+        `ask_and_send`).
+    *   Add an External Dependencies section to `README.md` with per-executable
+        tables, install instructions, and `TTS_URL` documentation.
+
 ## 2026-09-14 v0.0.119
 
 *   **Word-boundary fuzzy matching in chooser**

data/README.md

@@ -62,6 +62,7 @@ The following environment variables can be used to configure behavior:
 - `OLLAMA_URL` - Base URL for Ollama server (default: `http://localhost:11434`)
 - `OLLAMA_HOST` - Base hostname for Ollama server (default: `localhost:11434`)
 - `OLLAMA_SEARXNG_URL` - SearxNG search endpoint URL
+- `TTS_URL` - Base URL for the [audio.cpp](https://github.com/0xshug0/audio.cpp) TTS server (default: `http://localhost:8880`)
 
 #### Chat Settings
 - `OLLAMA_CHAT_MODEL` - Default model to use (e.g., `llama3.1`)
@@ -479,6 +480,71 @@ low-token semantic discovery and `file_context` for a comprehensive view of
 modules or patterns. Be cautious with broad patterns in `file_context`, as
 importing too many files can exceed the LLM's context window.*
 
+## External Dependencies
+
+**ollama_chat** invokes several external executables at runtime. Most are
+optional and only required for specific features; the app degrades gracefully
+when they are absent.
+
+### Always Available
+
+| Executable | Purpose |
+| :--- | :--- |
+| `sh` | Shell executor tool (`execute_shell`) |
+| `which` | Detecting optional executables on `$PATH` |
+| `git` | Runtime info (branch, SHA, remote origin) |
+
+### Editor & Diff
+
+| Executable | Purpose | Fallback |
+| :--- | :--- | :--- |
+| `vim` | Editor integration (`/vim`, `open_file_in_editor`) | `vi` |
+| `vimdiff` | Interactive diff (`patch_file`, `/config diff`) | `$DIFF_TOOL` env var |
+| `diff` | Unified diff generation | `$DIFF_COMMAND` env var |
+| `less` / `more` | Pager fallback | `$PAGER` env var |
+
+### Optional Feature Dependencies
+
+| Executable | Feature | Notes |
+| :--- | :--- | :--- |
+| `ffplay` | TTS audio playback | Part of **FFmpeg**; required only when `voice.handler` is `OllamaChat::TTS`; the TTS backend is [audio.cpp](https://github.com/0xshug0/audio.cpp) |
+| `say` | macOS system TTS | Required only when `voice.handler` is `OllamaChat::Say` (macOS only) |
+| `ctags` | Symbol resolution | Part of **Exuberant Ctags** or **Universal Ctags**; required for `resolve_tag` tool |
+| `twg` | Jira / Teamwork Graph CLI | Required for `execute_jira_twg` and `get_jira_issue` tools |
+| `pandoc` | EPUB document parsing | Part of **Pandoc**; used when importing `.epub` files |
+| `gs` | PDF compression | Part of **Ghostscript**; used when parsing PDFs |
+| `ctc` / `pbcopy` / `wl-copy` | Clipboard copy | `ctc` (default), `pbcopy` (macOS), `wl-copy` (Wayland); configurable via `copy:` in `config.yml` |
+| `pfc` / `pbpaste` / `wl-paste` | Clipboard paste | `pfc` (default), `pbpaste` (macOS), `wl-paste` (Wayland); configurable via `paste:` in `config.yml` |
+| `open` / `xdg-open` | Open URLs in browser | macOS / Linux respectively; used by `browse` tool |
+| `rspec` | Test runner | Required for `run_tests` tool (configurable via `$OLLAMA_CHAT_TOOLS_TEST_RUNNER`) |
+| `docker` | Ruby sandbox | Required for `eval_ruby` tool |
+| `ruby` | Syntax checker | Used by `syntax_checkers.ruby` (`ruby -wc`) |
+| `node` | Syntax checker | Used by `syntax_checkers.node` (`node --check`) |
+| `python3` | Syntax checker | Used by `syntax_checkers.python` (`python3 -m py_compile`) |
+| `bash` | Syntax checker | Used by `syntax_checkers.bash` (`bash -n`) |
+
+### Example Installation (Debian / Ubuntu)
+
+```bash
+# Core (usually already present)
+sudo apt install git vim vimdiff diff
+
+# TTS (audio.cpp handler)
+sudo apt install ffmpeg
+
+# Clipboard
+# ctc is a small helper; see the project README or install from source
+
+# Document parsing
+sudo apt install pandoc ghostscript
+
+# Symbol resolution
+sudo apt install universal-ctags
+
+# eval_ruby
+sudo apt install docker.io
+```
+
 ## Download
 
 The homepage of this app is located at

data/Rakefile

@@ -49,7 +49,7 @@ GemHadar do
   dependency             'reverse_markdown',      '~> 3.0'
   dependency             'kramdown-ansi',         '~> 0.5'
   dependency             'complex_config',        '~> 0.22', '>= 0.22.2'
-  dependency             'tins',                  '~> 1.55'
+  dependency             'tins',                  '~> 1.57'
   dependency             'search_ui',             '>= 0.2'
   dependency             'amatch',                '~> 0.4'
   dependency             'pdf-reader',            '~> 2.0'

data/lib/ollama_chat/chat.rb

@@ -172,6 +172,8 @@ class OllamaChat::Chat
   def voice_handler
     defined?(RSpec) and return []
     case config.voice.handler
+    when 'OllamaChat::TTS'
+      OllamaChat::TTS
     when 'OllamaChat::Say'
       OllamaChat::Say
     end

data/lib/ollama_chat/kramdown_ansi.rb

@@ -48,4 +48,19 @@ module OllamaChat::KramdownANSI
     content = content.encode(invalid: :replace)
     retry
   end
+
+  # The kramdown_markdown_remove method strips all markdown formatting from
+  # content by first rendering it to ANSI and then removing the escape
+  # sequences, yielding plain text suitable for speech synthesis.
+  #
+  # @param content [ String, nil ] the markdown content to strip.
+  #   If nil, returns an empty string.
+  #
+  # @return [ String ] the content with all markdown and ANSI
+  #   formatting removed, suitable for TTS playback
+  def kramdown_markdown_remove(content)
+    content = kramdown_ansi_parse(content)
+    content.empty? and return ''
+    OllamaChat::Utils::StripANSI.strip_ansi(content)
+  end
 end

data/lib/ollama_chat/oc.rb

@@ -14,6 +14,8 @@ require 'shellwords'
 module OC
   include ConstConf
 
+  plugin ConstConf::JSONPlugin
+
   description 'Environment config for OllamaChat'
   prefix ''
 
@@ -183,6 +185,48 @@ module OC
         default     { ENV['USER'] }
       end
 
+      TTS_URL = set do
+        description 'Base URL for the TTS (Text-to-Speech) service'
+        default     'http://localhost:8880'
+        required    true
+        sensitive   true
+        decode      { URI.parse(_1) if _1.present? }
+        check       { value.scheme =~ /\Ahttps?\z/ }
+      end
+
+      AUDIO_PLAYER_CONFIG = set do
+        description <<~EOT
+          Audio player configuration (JSON with four keys):
+
+          command   – Shell command that receives raw PCM on stdin.
+                      Opened via IO.popen(cmd, "w"); the playback thread
+                      writes audio bytes and silence chunks to its stdin.
+                      Default: ffplay s16le 24 kHz mono pipe.
+
+          frequency – Sample rate in Hz (e.g. 24000). Used to compute
+                      the byte size of silence chunks: frequency ×
+                      (bits / 8) × pause.
+
+          bits      – Bit depth per sample (e.g. 16). Determines bytes
+                      per sample for silence chunk calculation.
+
+          pause     – Seconds of silence injected when the audio queue
+                      is empty. Prevents the playback process from
+                      seeing EOF and hanging. Also the sleep duration
+                      between silence writes.
+        EOT
+        default <<~EOT
+          {
+            "command": "ffplay -autoexit -nodisp -loglevel quiet -vn -volume 80 -f s16le -ar 24000 -ch_layout mono -i -",
+            "frequency": 24000,
+            "bits": 16,
+            "pause": 0.01
+          }
+        EOT
+        decode  json
+        check   { value.command && value.frequency && value.bits && value.pause if value.present? }
+      end
+
       module TOOLS
         description 'Tool specific configuration settings'
 

data/lib/ollama_chat/ollama_chat_config/default_config.yml

@@ -445,9 +445,16 @@ prompts:
       no reference to yourself or the act of
       writing.
 voice:
-  handler: OllamaChat::Say
+  #handler: OllamaChat::Say
+  #enabled: false
+  #default: Samantha
+  handler: OllamaChat::TTS
   enabled: false
-  default: Samantha
+  default: miyu
+  model: 'chatterbox'
+  stream:
+    enabled: false # false for offline models, true for streaming
+    format: 'audio'
 roles:
   user: 👨 %{sender_name}
   assistant: 🤖 %{sender_name}

data/lib/ollama_chat/parsing.rb

@@ -280,7 +280,7 @@ module OllamaChat::Parsing
     content.scan(CONTENT_REGEXP).each { |url, file_url, quoted_file, file|
       if file && Pathname.new(file).expand_path.directory?
         max_depth = config.parsing.directory_structure_max_depth
-        contents << generate_structure(file, max_depth:).to_json
+        contents << generate_structure(file, max_depth:, include_hidden: true).to_json
         next
       end
       check_exist = false

data/lib/ollama_chat/state_selectors.rb

@@ -269,7 +269,9 @@ module OllamaChat::StateSelectors
       states:    OllamaChat::ThinkControl::THINK_MODE_STATES,
       off:       OllamaChat::ThinkControl::THINK_MODE_STATES[0, 1],
     )
-    list = voice_handler.ask_and_send(:voices) || []
+    list = Array(
+      voice_handler.ask_and_send(:voices, model: config.voice.model?)
+    )
     @voices = DatabaseStateSelector.new(
       chat:        self,
       attribute:   :current_voice,

data/lib/ollama_chat/tools/directory_structure.rb

@@ -55,6 +55,10 @@ class OllamaChat::Tools::DirectoryStructure
                 unlimited (defaults to nil)
               EOT
             ),
+            include_hidden: Tool::Function::Parameters::Property.new(
+              type: 'boolean',
+              description: 'Include hidden files and directories (dotfiles), (default: false)'
+            ),
           },
           required: []
         )
@@ -75,16 +79,18 @@ class OllamaChat::Tools::DirectoryStructure
   # @raise [StandardError] if there's an issue with directory traversal or JSON
   #   serialization
   def execute(tool_call, **opts)
-    chat      = opts[:chat]
-    config    = chat.config
-    path      = Pathname.new(tool_call.function.arguments.path || '.')
-    suffix    = tool_call.function.arguments.suffix.full?
-    max_depth = tool_call.function.arguments.max_depth.full?
+    chat           = opts[:chat]
+    config         = chat.config
+    path           = Pathname.new(tool_call.function.arguments.path || '.')
+    suffix         = tool_call.function.arguments.suffix.full?
+    max_depth      = tool_call.function.arguments.max_depth.full?
+    include_hidden = tool_call.function.arguments.include_hidden
 
     structure = generate_structure(
       path,
       max_depth:,
       suffix:,
+      include_hidden:,
       exclude: config.tools.functions.directory_structure.exclude?,
     )
     structure.to_json

data/lib/ollama_chat/tools/generate_image.rb

@@ -77,13 +77,13 @@ class OllamaChat::Tools::GenerateImage
     url        = service_url + '/prompt'
     payload    = { prompt: workflow }
     started    = Time.now
-    prompt_id  = post_url(url, payload).prompt_id
+    prompt_id  = post_url(url, payload, chat).prompt_id
 
     prompt_id.nil? and raise OllamaChat::OllamaChatError,
       "failed to trigger ComfyUI with #{prompt_id}"
 
     # 3. Poll for completion
-    filename = poll_for_image(service_url, prompt_id, config)
+    filename = poll_for_image(service_url, prompt_id, config, chat)
 
     filename.nil? and raise OllamaChat::OllamaChatError,
        'Image generation took too long or failed'
@@ -122,9 +122,9 @@ class OllamaChat::Tools::GenerateImage
   # @param url [URI] the target URL
   # @param payload [Hash] the data to be sent as JSON
   # @return [JSON::GenericObject] the parsed JSON response
-  def post_url(url, payload)
-    response = Excon.post(
-      url,
+  def post_url(url, payload, chat)
+    logger = OllamaChat::Utils::ExconLogger.new(chat)
+    response = Excon.new(url, logger:).post(
       body: JSON.dump(payload),
       headers: { 'Content-Type' => 'application/json' },
       expects: 200
@@ -136,9 +136,9 @@ class OllamaChat::Tools::GenerateImage
   #
   # @param url [URI] the target URL
   # @return [JSON::GenericObject] the parsed JSON response
-  def get_url(url)
-    response = Excon.get(
-      url,
+  def get_url(url, chat)
+    logger = OllamaChat::Utils::ExconLogger.new(chat)
+    response = Excon.new(url, logger:).get(
       headers: { 'Accept' => 'application/json' },
       expects: 200
     )
@@ -151,7 +151,7 @@ class OllamaChat::Tools::GenerateImage
   # @param prompt_id [String] the ID of the prompt to track
   # @param config [ComplexConfig::Settings] the configuration settings for timeout
   # @return [String, nil] the filename of the generated image, or nil if it timed out
-  def poll_for_image(service_url, prompt_id, config)
+  def poll_for_image(service_url, prompt_id, config, chat)
     history_url = service_url + '/history'
     filename = nil
 
@@ -159,7 +159,7 @@ class OllamaChat::Tools::GenerateImage
     sleep    = -(config.tools.functions.generate_image.timeout_duration? || 60)
 
     attempt attempts:, sleep:, exception_class: nil do
-      response = get_url(history_url)
+      response = get_url(history_url, chat)
       output_data = response[prompt_id]
 
       if filename = output_data&.outputs&.each_pair&.first&.last&.images&.first&.filename

data/lib/ollama_chat/tools/patch_file.rb

@@ -190,7 +190,7 @@ class OllamaChat::Tools::PatchFile
       message:      ,
       mtime:        ,
       checksum:     ,
-      syntax_check: syntax_check,
+      syntax_check: ,
     }).compact.to_json
 
   rescue => e