laminate-cli 0.1.2__tar.gz → 0.1.4__tar.gz

{laminate_cli-0.1.2 → laminate_cli-0.1.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: laminate-cli
-Version: 0.1.2
+Version: 0.1.4
 Summary: Generate release-ready changelogs from git history using a local LLM
 Author-email: Apleut <me@apleut.dev>
 License-Expression: MIT
@@ -20,6 +20,7 @@ Requires-Dist: llama-cpp-python
 Requires-Dist: pydantic
 Requires-Dist: typer
 Requires-Dist: rich
+Requires-Dist: platformdirs
 Dynamic: license-file
 
 <div style="text-align: center;">
@@ -97,25 +98,35 @@ pip install llama-cpp-python --extra-index-url https://abetlen.github.io/llama-c
 Then install Laminate:
 
 ```bash
-pip install .
+pip install laminate-cli
 ```
 
 ## Model
 
 Laminate runs a **local GGUF model** using [`llama-cpp-python`](https://github.com/abetlen/llama-cpp-python).
 
-Place exactly one `.gguf` model file in:
+[Phi-4-mini-instruct](https://huggingface.co/unsloth/Phi-4-mini-instruct-GGUF) is a good pick, and is the model used for Laminate's development.
 
-```text
-models/
-└── your-model.gguf
-```
+Laminate looks for exactly one `.gguf` file in two places, in this order:
 
-[Phi-4-mini-instruct](https://huggingface.co/unsloth/Phi-4-mini-instruct-GGUF) is a good pick, and is the model used for Laminate's development.
+1. **A `models/` folder in your current directory**, which the recommended location if you're running Laminate from a cloned repo:
+
+   ```text
+   models/
+   └── your-model.gguf
+   ```
+
+2. **Your OS's standard application data folder**, under `Laminate/models/`, used automatically if no local `models/` folder is found. This is the normal location if you installed via `pip install laminate-cli`:
+
+   | OS | Location |
+   |---|---|
+   | Windows | `%LOCALAPPDATA%\Apleut\Laminate\models\` |
+   | macOS | `~/Library/Application Support/Laminate/models/` |
+   | Linux | `~/.local/share/Laminate/models/` |
 
-Laminate automatically finds and loads the model when it runs.
+Create the folder if it doesn't already exist, and place your `.gguf` file inside.
 
-If the `models/` directory does not exist, or contains zero or multiple `.gguf` files, Laminate will stop and explain what needs to be fixed.
+If Laminate can't find a model, or finds more than one `.gguf` file in whichever folder it checks, it will stop and tell you exactly which locations it looked in and what to fix.
 
 The model is intentionally kept separate from the package because GGUF models can be large and have different licensing requirements.
 

{laminate_cli-0.1.2 → laminate_cli-0.1.4}/README.md

@@ -73,25 +73,35 @@ pip install llama-cpp-python --extra-index-url https://abetlen.github.io/llama-c
 Then install Laminate:
 
 ```bash
-pip install .
+pip install laminate-cli
 ```
 
 ## Model
 
 Laminate runs a **local GGUF model** using [`llama-cpp-python`](https://github.com/abetlen/llama-cpp-python).
 
-Place exactly one `.gguf` model file in:
+[Phi-4-mini-instruct](https://huggingface.co/unsloth/Phi-4-mini-instruct-GGUF) is a good pick, and is the model used for Laminate's development.
 
-```text
-models/
-└── your-model.gguf
-```
+Laminate looks for exactly one `.gguf` file in two places, in this order:
 
-[Phi-4-mini-instruct](https://huggingface.co/unsloth/Phi-4-mini-instruct-GGUF) is a good pick, and is the model used for Laminate's development.
+1. **A `models/` folder in your current directory**, which the recommended location if you're running Laminate from a cloned repo:
+
+   ```text
+   models/
+   └── your-model.gguf
+   ```
+
+2. **Your OS's standard application data folder**, under `Laminate/models/`, used automatically if no local `models/` folder is found. This is the normal location if you installed via `pip install laminate-cli`:
+
+   | OS | Location |
+   |---|---|
+   | Windows | `%LOCALAPPDATA%\Apleut\Laminate\models\` |
+   | macOS | `~/Library/Application Support/Laminate/models/` |
+   | Linux | `~/.local/share/Laminate/models/` |
 
-Laminate automatically finds and loads the model when it runs.
+Create the folder if it doesn't already exist, and place your `.gguf` file inside.
 
-If the `models/` directory does not exist, or contains zero or multiple `.gguf` files, Laminate will stop and explain what needs to be fixed.
+If Laminate can't find a model, or finds more than one `.gguf` file in whichever folder it checks, it will stop and tell you exactly which locations it looked in and what to fix.
 
 The model is intentionally kept separate from the package because GGUF models can be large and have different licensing requirements.
 
@@ -287,4 +297,4 @@ For changes to Laminate itself, pull requests are welcome.
 
 ## License
 
-License information will be added before the first stable release.
+License information will be added before the first stable release.

{laminate_cli-0.1.2 → laminate_cli-0.1.4}/laminate/cli.py

@@ -1,6 +1,7 @@
 import typer
 
 from concurrent.futures import ThreadPoolExecutor
+from importlib.metadata import version as get_version, PackageNotFoundError
 
 import sys
 
@@ -17,6 +18,17 @@ app = typer.Typer(add_completion=False)
 console = Console()
 
 
+def _version_callback(value: bool):
+    if not value:
+        return
+    try:
+        v = get_version("laminate-cli")
+    except PackageNotFoundError:
+        v = "unknown (not installed as a package)"
+    typer.echo(f"laminate {v}")
+    raise typer.Exit()
+
+
 @app.command()
 def generate(
     repo: str = typer.Option(".", help="Path to the git repository."),
@@ -26,6 +38,13 @@ def generate(
     output: str = typer.Option("CHANGELOG.md", help="Path to write the generated changelog to."),
     batch_size: int = typer.Option(15, "--batch-size", help="Number of commits sent to the LLM per batch."),
     release: str = typer.Option(None, "--release", help="Version number for this release, e.g. 1.1.0. If omitted, the changelog is headed [Unreleased]."),
+    version: bool = typer.Option(
+        False,
+        "--version",
+        callback=_version_callback,
+        is_eager=True,
+        help="Show Laminate's version and exit.",
+    ),
 ):
     if repo != ".":
         typer.echo(f"Reading commits from '{repo}'...")

{laminate_cli-0.1.2 → laminate_cli-0.1.4}/laminate/llm.py

@@ -1,9 +1,13 @@
 from pathlib import Path
 from llama_cpp import Llama
+from platformdirs import user_data_dir
 
-MODELS_DIR = Path(__file__).resolve().parent.parent / "models"
+_CANDIDATE_DIRS = [
+    Path.cwd() / "models",
+    Path(user_data_dir("Laminate", "Apleut")) / "models",
+]
 
-DEFAULT_N_CTX = 4096  
+DEFAULT_N_CTX = 4096
 _llm: Llama | None = None
 
 
@@ -12,29 +16,33 @@ class LLMError(Exception):
 
 
 def _find_model_file() -> Path:
-    if not MODELS_DIR.exists():
-        raise LLMError(
-            f"Models folder not found at {MODELS_DIR}. Create it and "
-            "place a .gguf model file inside (see README for a "
-            "recommended model and download link)."
-        )
-
-    gguf_files = list(MODELS_DIR.glob("*.gguf"))
-
-    if not gguf_files:
-        raise LLMError(
-            f"No .gguf model file found in {MODELS_DIR}. Download a "
-            "compatible model (see README) and place it in that folder."
-        )
-
-    if len(gguf_files) > 1:
-        names = ", ".join(f.name for f in gguf_files)
-        raise LLMError(
-            f"Multiple .gguf files found in {MODELS_DIR} ({names}) - "
-            "keep only one so Laminate knows which to load."
-        )
-
-    return gguf_files[0]
+    checked = []
+
+    for candidate_dir in _CANDIDATE_DIRS:
+        checked.append(str(candidate_dir))
+
+        if not candidate_dir.exists():
+            continue
+
+        gguf_files = list(candidate_dir.glob("*.gguf"))
+
+        if len(gguf_files) == 1:
+            return gguf_files[0]
+
+        if len(gguf_files) > 1:
+            names = ", ".join(f.name for f in gguf_files)
+            raise LLMError(
+                f"Multiple .gguf files found in {candidate_dir} ({names}) - "
+                "keep only one so Laminate knows which to load."
+            )
+
+    checked_str = "\n  - ".join(checked)
+    raise LLMError(
+        "No .gguf model file found. Laminate checked:\n  - "
+        f"{checked_str}\n"
+        "Download a compatible model (see README) and place it in one "
+        "of those folders (create the folder if it doesn't exist)."
+    )
 
 
 def get_llm(n_ctx: int = DEFAULT_N_CTX) -> Llama:

{laminate_cli-0.1.2 → laminate_cli-0.1.4}/laminate_cli.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: laminate-cli
-Version: 0.1.2
+Version: 0.1.4
 Summary: Generate release-ready changelogs from git history using a local LLM
 Author-email: Apleut <me@apleut.dev>
 License-Expression: MIT
@@ -20,6 +20,7 @@ Requires-Dist: llama-cpp-python
 Requires-Dist: pydantic
 Requires-Dist: typer
 Requires-Dist: rich
+Requires-Dist: platformdirs
 Dynamic: license-file
 
 <div style="text-align: center;">
@@ -97,25 +98,35 @@ pip install llama-cpp-python --extra-index-url https://abetlen.github.io/llama-c
 Then install Laminate:
 
 ```bash
-pip install .
+pip install laminate-cli
 ```
 
 ## Model
 
 Laminate runs a **local GGUF model** using [`llama-cpp-python`](https://github.com/abetlen/llama-cpp-python).
 
-Place exactly one `.gguf` model file in:
+[Phi-4-mini-instruct](https://huggingface.co/unsloth/Phi-4-mini-instruct-GGUF) is a good pick, and is the model used for Laminate's development.
 
-```text
-models/
-└── your-model.gguf
-```
+Laminate looks for exactly one `.gguf` file in two places, in this order:
 
-[Phi-4-mini-instruct](https://huggingface.co/unsloth/Phi-4-mini-instruct-GGUF) is a good pick, and is the model used for Laminate's development.
+1. **A `models/` folder in your current directory**, which the recommended location if you're running Laminate from a cloned repo:
+
+   ```text
+   models/
+   └── your-model.gguf
+   ```
+
+2. **Your OS's standard application data folder**, under `Laminate/models/`, used automatically if no local `models/` folder is found. This is the normal location if you installed via `pip install laminate-cli`:
+
+   | OS | Location |
+   |---|---|
+   | Windows | `%LOCALAPPDATA%\Apleut\Laminate\models\` |
+   | macOS | `~/Library/Application Support/Laminate/models/` |
+   | Linux | `~/.local/share/Laminate/models/` |
 
-Laminate automatically finds and loads the model when it runs.
+Create the folder if it doesn't already exist, and place your `.gguf` file inside.
 
-If the `models/` directory does not exist, or contains zero or multiple `.gguf` files, Laminate will stop and explain what needs to be fixed.
+If Laminate can't find a model, or finds more than one `.gguf` file in whichever folder it checks, it will stop and tell you exactly which locations it looked in and what to fix.
 
 The model is intentionally kept separate from the package because GGUF models can be large and have different licensing requirements.
 

{laminate_cli-0.1.2 → laminate_cli-0.1.4}/laminate_cli.egg-info/requires.txt

@@ -2,3 +2,4 @@ llama-cpp-python
 pydantic
 typer
 rich
+platformdirs

{laminate_cli-0.1.2 → laminate_cli-0.1.4}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "laminate-cli"
-version = "0.1.2"
+version = "0.1.4"
 authors = [
   { name="Apleut", email="me@apleut.dev" },
 ]
@@ -22,6 +22,7 @@ dependencies = [
     "pydantic",
     "typer",
     "rich",
+    "platformdirs",
 ]
 
 license = "MIT"
@@ -42,4 +43,4 @@ requires = ["setuptools>=68"]
 build-backend = "setuptools.build_meta"
 
 [tool.setuptools.packages.find]
-include = ["laminate*"]
+include = ["laminate*"]