devcommit 0.1.5.5__tar.gz → 0.1.5.7__tar.gz

{devcommit-0.1.5.5 → devcommit-0.1.5.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: devcommit
-Version: 0.1.5.5
+Version: 0.1.5.7
 Summary: AI-powered git commit message generator
 License: GNU GENERAL PUBLIC LICENSE
                                 Version 3, 29 June 2007
@@ -782,6 +782,7 @@ A command-line AI tool for autocommits.
    # GEMINI_MODEL / OPENAI_MODEL / GROQ_MODEL / OPENROUTER_MODEL / ANTHROPIC_MODEL / OLLAMA_MODEL / CUSTOM_MODEL
    MODEL_NAME = gemini-2.5-flash
    COMMIT_MODE = auto
+   CHANGELOG_MODE = timestamped
    EOF
    ```
 
@@ -796,6 +797,7 @@ A command-line AI tool for autocommits.
    COMMIT_TYPE = conventional
    MODEL_NAME = gemini-2.0-flash-exp
    COMMIT_MODE = auto
+   CHANGELOG_MODE = timestamped
    EOF
    ```
 
@@ -1150,8 +1152,11 @@ devcommit --stageAll --changelog --files src/
 
 - **With `--stageAll`**: Changelog is generated from unstaged changes **before** staging
 - **Without `--stageAll`**: Changelog is generated from the last commit **after** committing
-- Changelogs are saved as markdown files inside `branch/year/month/day` folders with time-based names (e.g., `changelogs/staging/2026/01/28/00-55-30.md`)
-- Default base directory: `changelogs/` (configurable via `CHANGELOG_DIR` in `.dcommit`)
+- Storage mode is controlled by `CHANGELOG_MODE`:
+  - `timestamped` (default): new file per run in `changelogs/<branch>/<year>/<month>/<day>/<time>.md`
+  - `branch`: one file per branch, updated in place at `changelogs/<branch>.md`
+- Default base directory is `changelogs/` (configurable via `CHANGELOG_DIR` in `.dcommit`)
+- Empty sections are automatically removed (including empty `## [Unreleased]`)
 - Uses Keep a Changelog format with AI-generated content
 
 **Example workflow:**
@@ -1163,7 +1168,9 @@ devcommit --stageAll --changelog --files src/
 # Stage all changes and generate changelog before committing
 devcommit --stageAll --changelog
 
-# The changelog file is created in changelogs/<branch>/<year>/<month>/<day>/ directory
+# The changelog file is created based on CHANGELOG_MODE
+# timestamped: changelogs/<branch>/<year>/<month>/<day>/<time>.md
+# branch:      changelogs/<branch>.md
 # Then changes are staged and committed
 ```
 
@@ -1177,6 +1184,8 @@ devcommit --stageAll --changelog
 - `--files` or `-f`: Stage and commit specific files or folders (can specify multiple)
 - `--push` or `-p`: Push commits to remote after committing
 - `--changelog` or `-c`: Generate changelog file from changes
+- `--models`: Fetch and list models from the configured `AI_PROVIDER`
+- `--free` / `--paid`: With `--models`, limit OpenRouter results to free or paid models
 
 ### Examples
 
@@ -1255,6 +1264,19 @@ export OPENROUTER_MODEL='meta-llama/llama-3.3-70b-instruct:free'
 devcommit
 ```
 
+To fetch the live model catalogue for your configured provider without opening a browser:
+
+```bash
+AI_PROVIDER=openrouter devcommit --models
+
+# Only free OpenRouter models
+AI_PROVIDER=openrouter devcommit --models --free
+```
+
+For OpenRouter, the table identifies free and paid models and shows their live token prices. For other providers, it lists the models accessible with that provider's configured API key; Ollama lists models installed locally. Copy an ID into the matching provider model setting, such as `OPENROUTER_MODEL`.
+
+Set `REASONING_EFFORT=none` to disable reasoning where the selected model supports it and favor faster commit messages. The default is `auto`; `low`, `medium`, `high`, `xhigh`, and `max` are also accepted where supported.
+
 **Popular free models on OpenRouter (add `:free` suffix):**
 
 **Recommended Models:**
@@ -1356,9 +1378,11 @@ All configuration can be set via **environment variables** or **`.dcommit` file*
 | `MAX_NO`        | Number of commit message suggestions | `1`                                                    | Any positive integer                     |
 | `COMMIT_TYPE`   | Style of commit messages             | `general`                                              | `general`, `conventional`, etc.          |
 | `COMMIT_MODE`   | Default commit strategy              | `auto`                                                 | `auto`, `directory`, `global`, `related` |
-| `EXCLUDE_FILES` | Files to exclude from diff           | `package-lock.json, pnpm-lock.yaml, yarn.lock, *.lock` | Comma-separated file patterns            |
+| `REASONING_EFFORT` | Reasoning effort for supported models | `auto`                                              | `auto`, `none`, `low`, `medium`, `high`, `xhigh`, `max` |
+| `EXCLUDE_FILES` | Files to exclude from diff           | `package-lock.json, pnpm-lock.yaml, yarn.lock, *.lock` | Comma-separated patterns or absolute paths within one repo |
 | `MAX_TOKENS`    | Maximum tokens for AI response       | `8192`                                                 | Any positive integer                     |
 | `CHANGELOG_DIR` | Directory for changelog files        | `changelogs`                                           | Any directory path                       |
+| `CHANGELOG_MODE` | Changelog file strategy             | `timestamped`                                          | `timestamped`, `branch`                  |
 
 ### Configuration Priority
 

{devcommit-0.1.5.5 → devcommit-0.1.5.7}/README.md

@@ -77,6 +77,7 @@ A command-line AI tool for autocommits.
    # GEMINI_MODEL / OPENAI_MODEL / GROQ_MODEL / OPENROUTER_MODEL / ANTHROPIC_MODEL / OLLAMA_MODEL / CUSTOM_MODEL
    MODEL_NAME = gemini-2.5-flash
    COMMIT_MODE = auto
+   CHANGELOG_MODE = timestamped
    EOF
    ```
 
@@ -91,6 +92,7 @@ A command-line AI tool for autocommits.
    COMMIT_TYPE = conventional
    MODEL_NAME = gemini-2.0-flash-exp
    COMMIT_MODE = auto
+   CHANGELOG_MODE = timestamped
    EOF
    ```
 
@@ -445,8 +447,11 @@ devcommit --stageAll --changelog --files src/
 
 - **With `--stageAll`**: Changelog is generated from unstaged changes **before** staging
 - **Without `--stageAll`**: Changelog is generated from the last commit **after** committing
-- Changelogs are saved as markdown files inside `branch/year/month/day` folders with time-based names (e.g., `changelogs/staging/2026/01/28/00-55-30.md`)
-- Default base directory: `changelogs/` (configurable via `CHANGELOG_DIR` in `.dcommit`)
+- Storage mode is controlled by `CHANGELOG_MODE`:
+  - `timestamped` (default): new file per run in `changelogs/<branch>/<year>/<month>/<day>/<time>.md`
+  - `branch`: one file per branch, updated in place at `changelogs/<branch>.md`
+- Default base directory is `changelogs/` (configurable via `CHANGELOG_DIR` in `.dcommit`)
+- Empty sections are automatically removed (including empty `## [Unreleased]`)
 - Uses Keep a Changelog format with AI-generated content
 
 **Example workflow:**
@@ -458,7 +463,9 @@ devcommit --stageAll --changelog --files src/
 # Stage all changes and generate changelog before committing
 devcommit --stageAll --changelog
 
-# The changelog file is created in changelogs/<branch>/<year>/<month>/<day>/ directory
+# The changelog file is created based on CHANGELOG_MODE
+# timestamped: changelogs/<branch>/<year>/<month>/<day>/<time>.md
+# branch:      changelogs/<branch>.md
 # Then changes are staged and committed
 ```
 
@@ -472,6 +479,8 @@ devcommit --stageAll --changelog
 - `--files` or `-f`: Stage and commit specific files or folders (can specify multiple)
 - `--push` or `-p`: Push commits to remote after committing
 - `--changelog` or `-c`: Generate changelog file from changes
+- `--models`: Fetch and list models from the configured `AI_PROVIDER`
+- `--free` / `--paid`: With `--models`, limit OpenRouter results to free or paid models
 
 ### Examples
 
@@ -550,6 +559,19 @@ export OPENROUTER_MODEL='meta-llama/llama-3.3-70b-instruct:free'
 devcommit
 ```
 
+To fetch the live model catalogue for your configured provider without opening a browser:
+
+```bash
+AI_PROVIDER=openrouter devcommit --models
+
+# Only free OpenRouter models
+AI_PROVIDER=openrouter devcommit --models --free
+```
+
+For OpenRouter, the table identifies free and paid models and shows their live token prices. For other providers, it lists the models accessible with that provider's configured API key; Ollama lists models installed locally. Copy an ID into the matching provider model setting, such as `OPENROUTER_MODEL`.
+
+Set `REASONING_EFFORT=none` to disable reasoning where the selected model supports it and favor faster commit messages. The default is `auto`; `low`, `medium`, `high`, `xhigh`, and `max` are also accepted where supported.
+
 **Popular free models on OpenRouter (add `:free` suffix):**
 
 **Recommended Models:**
@@ -651,9 +673,11 @@ All configuration can be set via **environment variables** or **`.dcommit` file*
 | `MAX_NO`        | Number of commit message suggestions | `1`                                                    | Any positive integer                     |
 | `COMMIT_TYPE`   | Style of commit messages             | `general`                                              | `general`, `conventional`, etc.          |
 | `COMMIT_MODE`   | Default commit strategy              | `auto`                                                 | `auto`, `directory`, `global`, `related` |
-| `EXCLUDE_FILES` | Files to exclude from diff           | `package-lock.json, pnpm-lock.yaml, yarn.lock, *.lock` | Comma-separated file patterns            |
+| `REASONING_EFFORT` | Reasoning effort for supported models | `auto`                                              | `auto`, `none`, `low`, `medium`, `high`, `xhigh`, `max` |
+| `EXCLUDE_FILES` | Files to exclude from diff           | `package-lock.json, pnpm-lock.yaml, yarn.lock, *.lock` | Comma-separated patterns or absolute paths within one repo |
 | `MAX_TOKENS`    | Maximum tokens for AI response       | `8192`                                                 | Any positive integer                     |
 | `CHANGELOG_DIR` | Directory for changelog files        | `changelogs`                                           | Any directory path                       |
+| `CHANGELOG_MODE` | Changelog file strategy             | `timestamped`                                          | `timestamped`, `branch`                  |
 
 ### Configuration Priority
 

{devcommit-0.1.5.5 → devcommit-0.1.5.7}/devcommit/app/ai_providers.py

@@ -34,6 +34,19 @@ except ImportError:
 sys.stderr = _stderr
 _devnull.close()
 
+REASONING_EFFORTS = {"auto", "none", "low", "medium", "high", "xhigh", "max"}
+
+
+def normalize_reasoning_effort(value: str) -> str:
+    effort = value.lower()
+    if effort not in REASONING_EFFORTS:
+        raise ValueError("REASONING_EFFORT must be auto, none, low, medium, high, xhigh, or max")
+    return effort
+
+
+def reasoning_extra_body(effort: str) -> dict:
+    return {} if effort == "auto" else {"reasoning_effort": effort}
+
 
 class AIProvider(ABC):
     """Base class for AI providers"""
@@ -106,11 +119,12 @@ class GeminiProvider(AIProvider):
 class OpenAIProvider(AIProvider):
     """OpenAI GPT provider"""
 
-    def __init__(self, api_key: str, model: str = "gpt-4o-mini"):
+    def __init__(self, api_key: str, model: str = "gpt-4o-mini", reasoning_effort: str = "auto"):
         if not openai:
             raise ImportError("openai not installed. Run: pip install openai")
         self.client = openai.OpenAI(api_key=api_key)
         self.model = model
+        self.reasoning_effort = normalize_reasoning_effort(reasoning_effort)
 
     def generate_commit_message(
         self, diff: str, prompt: str, max_tokens: int
@@ -123,6 +137,7 @@ class OpenAIProvider(AIProvider):
             ],
             max_tokens=max_tokens,
             temperature=0.7,
+            extra_body=reasoning_extra_body(self.reasoning_effort),
         )
         if not response or not response.choices:
             raise ValueError("OpenAI returned empty response (no choices).")
@@ -132,13 +147,14 @@ class OpenAIProvider(AIProvider):
 class GroqProvider(AIProvider):
     """Groq AI provider (OpenAI-compatible)"""
 
-    def __init__(self, api_key: str, model: str = "llama-3.3-70b-versatile"):
+    def __init__(self, api_key: str, model: str = "llama-3.3-70b-versatile", reasoning_effort: str = "auto"):
         if not openai:
             raise ImportError("openai not installed. Run: pip install openai")
         self.client = openai.OpenAI(
             api_key=api_key, base_url="https://api.groq.com/openai/v1"
         )
         self.model = model
+        self.reasoning_effort = normalize_reasoning_effort(reasoning_effort)
 
     def generate_commit_message(
         self, diff: str, prompt: str, max_tokens: int
@@ -151,6 +167,7 @@ class GroqProvider(AIProvider):
             ],
             max_tokens=max_tokens,
             temperature=0.7,
+            extra_body=reasoning_extra_body(self.reasoning_effort),
         )
         if not response or not response.choices:
             raise ValueError("Groq returned empty response (no choices).")
@@ -164,6 +181,7 @@ class OpenRouterProvider(AIProvider):
         self,
         api_key: str,
         model: str = "meta-llama/llama-3.3-70b-instruct:free",
+        reasoning_effort: str = "auto",
     ):
         if not openai:
             raise ImportError("openai not installed. Run: pip install openai")
@@ -176,6 +194,7 @@ class OpenRouterProvider(AIProvider):
             },
         )
         self.model = model
+        self.reasoning_effort = normalize_reasoning_effort(reasoning_effort)
 
     def generate_commit_message(
         self, diff: str, prompt: str, max_tokens: int
@@ -188,7 +207,7 @@ class OpenRouterProvider(AIProvider):
             ],
             max_tokens=max_tokens,
             temperature=0.7,
-            extra_body={"transforms": ["middle-out"]},
+            extra_body={"transforms": ["middle-out"], **reasoning_extra_body(self.reasoning_effort)},
             extra_headers={
                 "HTTP-Referer": "https://github.com/hordunlarmy/DevCommit",
                 "X-Title": "DevCommit",
@@ -207,23 +226,30 @@ class OpenRouterProvider(AIProvider):
 class AnthropicProvider(AIProvider):
     """Anthropic Claude provider"""
 
-    def __init__(self, api_key: str, model: str = "claude-3-haiku-20240307"):
+    def __init__(self, api_key: str, model: str = "claude-3-haiku-20240307", reasoning_effort: str = "auto"):
         if not anthropic:
             raise ImportError(
                 "anthropic not installed. Run: pip install anthropic"
             )
         self.client = anthropic.Anthropic(api_key=api_key)
         self.model = model
+        self.reasoning_effort = normalize_reasoning_effort(reasoning_effort)
 
     def generate_commit_message(
         self, diff: str, prompt: str, max_tokens: int
     ) -> str:
-        message = self.client.messages.create(
+        request = dict(
             model=self.model,
             max_tokens=max_tokens,
             system=prompt,
             messages=[{"role": "user", "content": diff}],
         )
+        if self.reasoning_effort == "none":
+            request["thinking"] = {"type": "disabled"}
+        elif self.reasoning_effort != "auto":
+            request["thinking"] = {"type": "adaptive"}
+            request["output_config"] = {"effort": self.reasoning_effort}
+        message = self.client.messages.create(**request)
         return message.content[0].text.strip()
 
 
@@ -231,7 +257,7 @@ class OllamaProvider(AIProvider):
     """Ollama local model provider"""
 
     def __init__(
-        self, base_url: str = "http://localhost:11434", model: str = "llama3"
+        self, base_url: str = "http://localhost:11434", model: str = "llama3", reasoning_effort: str = "auto"
     ):
         if not requests:
             raise ImportError(
@@ -239,6 +265,7 @@ class OllamaProvider(AIProvider):
             )
         self.base_url = base_url.rstrip("/")
         self.model = model
+        self.reasoning_effort = normalize_reasoning_effort(reasoning_effort)
 
     def generate_commit_message(
         self, diff: str, prompt: str, max_tokens: int
@@ -250,6 +277,8 @@ class OllamaProvider(AIProvider):
             "stream": False,
             "options": {"temperature": 0.7, "num_predict": max_tokens},
         }
+        if self.reasoning_effort != "auto":
+            data["think"] = False if self.reasoning_effort == "none" else self.reasoning_effort
 
         response = requests.post(url, json=data, timeout=60)
         response.raise_for_status()
@@ -267,6 +296,7 @@ class CustomProvider(AIProvider):
         api_url: str,
         api_key: Optional[str] = None,
         model: str = "default",
+        reasoning_effort: str = "auto",
     ):
         if not openai:
             raise ImportError("openai not installed. Run: pip install openai")
@@ -282,6 +312,7 @@ class CustomProvider(AIProvider):
             api_key=api_key or "dummy-key", base_url=base_url
         )
         self.model = model
+        self.reasoning_effort = normalize_reasoning_effort(reasoning_effort)
 
     def generate_commit_message(
         self, diff: str, prompt: str, max_tokens: int
@@ -294,6 +325,7 @@ class CustomProvider(AIProvider):
             ],
             max_tokens=max_tokens,
             temperature=0.7,
+            extra_body=reasoning_extra_body(self.reasoning_effort),
         )
         if not response or not response.choices:
             raise ValueError("Custom API returned empty response (no choices).")
@@ -304,6 +336,7 @@ def get_ai_provider(config) -> AIProvider:
     """Factory function to get the appropriate AI provider based on config"""
 
     provider_name = config("AI_PROVIDER", default="gemini").lower()
+    reasoning_effort = config("REASONING_EFFORT", default="auto")
 
     if provider_name == "gemini":
         api_key = config("GEMINI_API_KEY", default=None)
@@ -313,6 +346,8 @@ def get_ai_provider(config) -> AIProvider:
         model = config("GEMINI_MODEL", default=None) or config(
             "MODEL_NAME", default="gemini-2.0-flash-exp"
         )
+        if normalize_reasoning_effort(reasoning_effort) != "auto":
+            raise ValueError("REASONING_EFFORT is unsupported by the legacy Gemini SDK; use auto")
         return GeminiProvider(api_key, model)
 
     elif provider_name == "openai":
@@ -322,7 +357,7 @@ def get_ai_provider(config) -> AIProvider:
         model = config("OPENAI_MODEL", default=None) or config(
             "MODEL_NAME", default="gpt-4o-mini"
         )
-        return OpenAIProvider(api_key, model)
+        return OpenAIProvider(api_key, model, reasoning_effort)
 
     elif provider_name == "groq":
         api_key = config("GROQ_API_KEY", default=None)
@@ -331,7 +366,7 @@ def get_ai_provider(config) -> AIProvider:
         model = config("GROQ_MODEL", default=None) or config(
             "MODEL_NAME", default="llama-3.3-70b-versatile"
         )
-        return GroqProvider(api_key, model)
+        return GroqProvider(api_key, model, reasoning_effort)
 
     elif provider_name == "openrouter":
         api_key = config("OPENROUTER_API_KEY", default=None)
@@ -340,7 +375,7 @@ def get_ai_provider(config) -> AIProvider:
         model = config("OPENROUTER_MODEL", default=None) or config(
             "MODEL_NAME", default="meta-llama/llama-3.3-70b-instruct:free"
         )
-        return OpenRouterProvider(api_key, model)
+        return OpenRouterProvider(api_key, model, reasoning_effort)
 
     elif provider_name == "anthropic":
         api_key = config("ANTHROPIC_API_KEY", default=None)
@@ -349,14 +384,14 @@ def get_ai_provider(config) -> AIProvider:
         model = config("ANTHROPIC_MODEL", default=None) or config(
             "MODEL_NAME", default="claude-3-haiku-20240307"
         )
-        return AnthropicProvider(api_key, model)
+        return AnthropicProvider(api_key, model, reasoning_effort)
 
     elif provider_name == "ollama":
         base_url = config("OLLAMA_BASE_URL", default="http://localhost:11434")
         model = config("OLLAMA_MODEL", default=None) or config(
             "MODEL_NAME", default="llama3"
         )
-        return OllamaProvider(base_url, model)
+        return OllamaProvider(base_url, model, reasoning_effort)
 
     elif provider_name == "custom":
         api_url = config("CUSTOM_API_URL", default=None)
@@ -366,7 +401,7 @@ def get_ai_provider(config) -> AIProvider:
         model = config("CUSTOM_MODEL", default=None) or config(
             "MODEL_NAME", default="default"
         )
-        return CustomProvider(api_url, api_key, model)
+        return CustomProvider(api_url, api_key, model, reasoning_effort)
 
     else:
         raise ValueError(