Merge branch 'main' into serve-api-docs-pr

container-images/cuda/Containerfile

@@ -22,8 +22,8 @@ RUN /usr/bin/python3 --version
 RUN pip install "huggingface_hub[cli]==${HUGGINGFACE_HUB_VERSION}"
 RUN pip install "omlmd==${OMLMD_VERSION}"
 
-# Build wouldnt complete couldnt find libcuda so made a systemlink
-# But this didnt work
+# Build wouldn't complete couldn't find libcuda so made a systemlink
+# But this didn't work
 
 # RUN ln -s /usr/local/cuda/lib64/stubs/libcuda.so /usr/local/cuda/lib64/stubs/libcuda.so.1
 # RUN LD_LIBRARY_PATH=/usr/local/cuda/lib64/stubs/
@@ -32,7 +32,7 @@ RUN pip install "omlmd==${OMLMD_VERSION}"
 
 # ENV GGML_CCACHE=0
 
-# Build wouldnt complete with cmake even with nvidia container toolkit installed
+# Build wouldn't complete with cmake even with nvidia container toolkit installed
 
 RUN git clone https://github.com/ggerganov/llama.cpp && \
     cd llama.cpp && \

docs/ramalama-containers.1.md

@@ -11,6 +11,8 @@ ramalama\-containers - list all RamaLama containers
 ## DESCRIPTION
 List all containers running AI Models
 
+Command will not work when run with --nocontainer option.
+
 ## OPTIONS
 
 #### **--format**=*format*

docs/ramalama-stop.1.md

@@ -6,6 +6,10 @@ ramalama\-stop - stop named container that is running AI Model
 ## SYNOPSIS
 **ramalama stop** [*options*] *name*
 
+Tells container engine to stop the specified container.
+
+Command will not work when run with --nocontainer option.
+
 ## OPTIONS
 
 #### **--all**, **-a**

docs/ramalama.1.md

@@ -10,16 +10,20 @@ ramalama - Simple management tool for working with AI Models
 RamaLama : The goal of RamaLama is to make AI boring.
 
 On first run RamaLama inspects your system for GPU support, falling back to CPU
-support if no GPUs are present. It then uses container engines like Podman or
-Docker to pull the appropriate OCI image with all of the software necessary to run an
-AI Model for your systems setup. This eliminates the need for the user to
-configure the system for AI themselves. After the initialization, RamaLama
+support if no GPUs are present. RamaLama uses container engines like Podman or
+Docker to pull the appropriate OCI image with all of the software necessary to
+run an AI Model for your systems setup. This eliminates the need for the user
+to configure the system for AI themselves. After the initialization, RamaLama
 will run the AI Models within a container based on the OCI image.
 
 RamaLama first pulls AI Models from model registries. It then start a chatbot
 or a service as a rest API (using llama.cpp's server) from a simple single command. 
 Models are treated similarly to the way that Podman or Docker treat container images.
 
+If you have both Podman and Docker installed, RamaLama defaults to Podman, use
+the `RAMALAMA_CONTAINER_ENGINE=docker` environment variable to override this
+behaviour.
+
 RamaLama supports multiple AI model registries types called transports. Supported transports:
 
 

pyproject.toml

@@ -3,6 +3,9 @@ name = "ramalama"
 version = "0.0.14"
 dependencies = [
   "argcomplete",
+  "tqdm",
+  "omlmd",
+  "huggingface_hub[cli]",
 ]
 requires-python = ">= 3.8"
 maintainers = [

ramalama/cli.py

@@ -6,11 +6,12 @@
 import os
 import random
 import string
+import subprocess
 import sys
 import time
 
 from ramalama.huggingface import Huggingface
-from ramalama.common import in_container, container_manager, exec_cmd, run_cmd, default_image, find_working_directory
+from ramalama.common import in_container, container_manager, exec_cmd, run_cmd, default_image, find_working_directory, perror
 from ramalama.oci import OCI
 from ramalama.ollama import Ollama
 from ramalama.shortnames import Shortnames
@@ -217,11 +218,14 @@ def _list_containers(args):
     if args.format:
         conman_args += [f"--format={args.format}"]
 
-    output = run_cmd(conman_args).stdout.decode("utf-8").strip()
-    if output == "":
-        return []
-    return output.split("\n")
-
+    try:
+        output = run_cmd(conman_args).stdout.decode("utf-8").strip()
+        if output == "":
+            return []
+        return output.split("\n")
+    except subprocess.CalledProcessError as e:
+        perror("ramalama list command requires a running container engine")
+        raise(e)
 
 def list_containers(args):
     if len(_list_containers(args)) == 0:
@@ -382,11 +386,6 @@ def serve_cli(args):
     model.serve(args)
 
 
-def stop_cli(args):
-    model = New(args.MODEL)
-    model.stop(args)
-
-
 def stop_parser(subparsers):
     parser = subparsers.add_parser("stop", help="stop named container that is running AI Model")
     parser.add_argument("--container", default=False, action="store_false", help=argparse.SUPPRESS)
@@ -405,7 +404,10 @@ def _stop_container(args, name):
     if conman == "":
         raise IndexError("no container manager (Podman, Docker) found")
 
-    conman_args = [conman, "stop", "-t=0", "--ignore=" + str(args.ignore), name]
+    conman_args = [conman, "stop", "-t=0"]
+    if args.ignore:
+        conman_args += [ "--ignore", str(args.ignore)]
+    conman_args += [name]
     run_cmd(conman_args)
 
 

ramalama/common.py

@@ -17,6 +17,10 @@ def in_container():
 
 
 def container_manager():
+    engine = os.getenv("RAMALAMA_CONTAINER_ENGINE")
+    if engine:
+        return engine
+
     if available("podman"):
         return "podman"
 

ramalama/huggingface.py

@@ -3,10 +3,10 @@
 from ramalama.model import Model
 
 missing_huggingface = """
-Huggingface models requires the huggingface-cli and tldm modules.
+Huggingface models requires the huggingface-cli and tqdm modules.
 These modules can be installed via PyPi tools like pip, pip3, pipx or via
 distribution package managers like dnf or apt. Example:
-pip install huggingface_hub[cli] tldm
+pip install huggingface_hub[cli] tqdm
 """
 
 

ramalama/oci.py

@@ -17,6 +17,12 @@ def __init__(self, model):
                 self.omlmd = f"{i}/../../../bin/omlmd"
                 if os.path.exists(self.omlmd):
                     break
+            raise """\
+OCI models requires the omlmd module.
+This module can be installed via PyPi tools like pip, pip3, pipx or via
+distribution package managers like dnf or apt. Example:
+pip install omlmd
+"""
 
     def login(self, args):
         conman_args = [self.conman, "login"]

ramalama/ollama.py

@@ -8,7 +8,17 @@
 
 
 def download_file(url, dest_path, headers=None):
-    from tqdm import tqdm
+    try:
+        from tqdm import tqdm
+    except FileNotFoundError:
+        raise NotImplementedError(
+            """\
+Ollama models requires the tqdm modules.
+This model can be installed via PyPi tools like pip, pip3, pipx or via
+distribution package managers like dnf or apt. Example:
+pip install tqdm
+"""
+        )
 
     request = urllib.request.Request(url, headers=headers or {})
     with urllib.request.urlopen(request) as response:

requirements.txt

@@ -3,4 +3,5 @@ argcomplete
 setuptools
 wheel
 omlmd
+tqdm
 huggingface_hub[cli]