Merge pull request #149 from maxibor/download

feat: add download subcommand

Author: maxibor

AMDirT/__init__.py

@@ -1 +1 @@
-__version__ = "1.5.0"
+__version__ = "1.6.0"

AMDirT/cli.py

@@ -4,9 +4,10 @@
 from AMDirT.validate import run_validation
 from AMDirT.viewer import run_app
 from AMDirT.convert import run_convert
-from AMDirT.core import get_json_path
+from AMDirT.core import get_json_path, get_amdir_tags, get_latest_tag
 from AMDirT.autofill import run_autofill
 from AMDirT.merge import merge_new_df
+from AMDirT.download import download as download_amdir
 from json import load
 
 
@@ -294,5 +295,45 @@ def merge(ctx, no_args_is_help=True, **kwargs):
     merge_new_df(**kwargs, **ctx.obj)
 
 
+@cli.command()
+@click.option(
+    "-t",
+    "--table",
+    help="AncientMetagenomeDir table to download",
+    type=click.Choice(get_table_list()),
+    default="ancientmetagenome-hostassociated",
+    show_default=True,
+)
+@click.option(
+    "-y",
+    "--table_type",
+    help="Type of table to download",
+    type=click.Choice(["samples", "libraries"]),
+    default="samples",
+    show_default=True,
+)
+@click.option(
+    "-r",
+    "--release",
+    help="Release tag to download",
+    type=click.Choice(get_amdir_tags()),
+    default=get_latest_tag(get_amdir_tags()),
+    show_default=True,
+)
+@click.option(
+    "-o",
+    "--output",
+    help="Output directory",
+    type=click.Path(writable=True),
+    default=".",
+    show_default=True,
+)
+def download(no_args_is_help=True, **kwargs):
+    """\b
+    Download a table from the AMDirT repository
+    """
+    download_amdir(**kwargs)
+
+
 if __name__ == "__main__":
     cli()

AMDirT/core/__init__.py

@@ -5,6 +5,7 @@
 import pandas as pd
 import streamlit as st
 from packaging import version
+from packaging.version import InvalidVersion
 from importlib.resources import files as get_module_dir
 import os
 import logging
@@ -65,7 +66,34 @@ def get_amdir_tags():
             if version.parse(tag["name"]) >= version.parse("v22.09")
         ]
     else:
-        return []
+        logger.warning(
+            "Could not fetch tags from AncientMetagenomeDir. Defaulting to master. Metadata may not yet be officially released."
+        )
+        return ["master"]
+
+
+@st.cache_data
+def get_latest_tag(tags):
+    try:
+        return sorted(tags, key=lambda x: version.Version(x))[-1]
+    except InvalidVersion:
+        if "master" in tags:
+            return "master"
+        else:
+            raise InvalidVersion("No valid tags found")
+
+
+def check_allowed_values(ref: list, test: str):
+    """
+    Check if test is in ref
+    Args:
+        ref(list): List of allowed values
+        test(str): value to check
+    """
+
+    if test in ref:
+        return True
+    return False
 
 
 def get_colour_chemistry(instrument: str) -> int:

AMDirT/download/__init__.py

@@ -0,0 +1,62 @@
+from AMDirT.core import (
+    logger,
+    get_amdir_tags,
+    get_remote_resources,
+    check_allowed_values,
+)
+import requests
+
+
+def download(table: str, table_type: str, release: str, output: str = ".") -> str:
+    """
+    Download a table from the AMDirT repository.
+
+    Parameters
+    ----------
+    table : str
+        The AncientMetagenomeDir table to download.
+    table_type : str
+        The type of table to download. Allowed values are ['samples', 'libraries'].
+    release : str
+        The release of the table to download. Must be a valid release tag.
+    output: str
+        The output directory to save the table. Default is the current directory.
+
+    Returns
+    -------
+    str:
+        The path to the downloaded table.
+
+    Raises
+    ------
+    ValueError
+        If an invalid table is provided.
+    ValueError
+        If an invalid table type is provided.
+    ValueError
+        If an invalid release is provided.
+    """
+
+    resources = get_remote_resources()
+    tags = get_amdir_tags()
+    if tags != ["master"]:
+        if check_allowed_values(tags, release) is False:
+            raise ValueError(f"Invalid release: {release}. Allowed values are {tags}")
+
+    tables = resources["samples"]
+    if check_allowed_values(tables, table) is False:
+        raise ValueError(f"Invalid table: {table}. Allowed values are {tables}")
+
+    if check_allowed_values(["samples", "libraries"], table_type) is False:
+        raise ValueError(
+            f"Invalid table type: {table_type}. Allowed values are ['samples', 'libraries']"
+        )
+    table_filename = f"{table}_{table_type}_{release}.tsv"
+    logger.info(
+        f"Downloading {table} {table_type} table from {release} release, saving to {output}/{table_filename}"
+    )
+    t = requests.get(resources[table_type][table].replace("master", release))
+    with open(table_filename, "w") as fh:
+        fh.write(t.text)
+
+    return table_filename

docs/source/how_to/download.md

@@ -0,0 +1,21 @@
+# download
+
+## What
+
+Download a copy of an AncientMetagenomeDir table.
+
+## When
+
+This command would be used when you want to download an AncientMetagenomeDir table locally.
+
+You typically do this if you're planning to use the `convert` command later.
+
+## How
+
+```bash
+AMDirT download --table ancientsinglegenome-hostassociated --table_type samples -r v23.12.0 -o .
+```
+
+## Output
+
+This example command above will download the `ancientsinglegenome-hostassociated` `sample` table from the `v23.12.0` AncientMetagenomeDir release, and save it locally to `ancientmetagenome-hostassociated_samples_v23.12.0.tsv`

docs/source/tutorials/convert.md

@@ -15,16 +15,15 @@ We will take use one of the previous releases of AncientMetagenomeDir as an exam
 ```bash
 mkdir amdirt-convert-tutorial
 cd amdirt-convert-tutorial
-curl -LO https://github.com/SPAAM-community/AncientMetagenomeDir/releases/download/v23.09.0/AncientMetagenomeDir_v23.09.0.zip
-unzip AncientMetagenomeDir_v23.09.0.zip
+AMDirT download --table ancientmetagenome-hostassociated --table_type samples -r v23.09.0
 ```
 
 ## Filter a sample metadata table
 
 Next we can filter the ancient metagenome 'host-associated' sample sheet for all dental calculus tables from Germany.
 
 ```bash
-cat ancientmetagenome-hostassociated/samples/ancientmetagenome-hostassociated_samples.tsv | grep -e '^project_name' -e 'dental calculus' | grep -e '^project_name' -e 'Germany' > germany_dentalcalculus.tsv
+cat ancientmetagenome-hostassociated_samples_v23.09.0.tsv | grep -e '^project_name' -e 'dental calculus' | grep -e '^project_name' -e 'Germany' > germany_dentalcalculus.tsv
 ```
 
 > ⚠ _The command above is not robust and is only used for system portability and demonstration purposes. For example the `Germany` string could be in a site name. In practice, you should use more robust filtering methods such more specific `grep` expressions or in R_.

tests/test_download.py

@@ -0,0 +1,10 @@
+from AMDirT.download import download
+
+
+def test_download():
+    table = "ancientmetagenome-hostassociated"
+    table_type = "samples"
+    release = "v23.12.0"
+
+    d = download(table, table_type, release, output=".")
+    assert d == "ancientmetagenome-hostassociated_samples_v23.12.0.tsv"