Merge pull request #97 from refgenie/bugfix

v0.9.1

.github/workflows/run-pytest.yml

@@ -17,6 +17,10 @@ jobs:
     steps:
     - uses: actions/checkout@v2
 
+    - name: install macOS-specific dependancies
+      if: startsWith(matrix.os, 'macOS')
+      run: brew install md5sha1sum
+
     - name: Set up Python ${{ matrix.python-version }}
       uses: actions/setup-python@v2
       with:

docs/changelog.md

@@ -2,6 +2,18 @@
 
 This project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html) and [Keep a Changelog](https://keepachangelog.com/en/1.0.0/) format. 
 
+## [0.9.1] - 2020-07-29
+
+### Added
+- `force_large` argument in the `pull` method, which can be used to handle large archive downloads
+- `add` method
+
+### Changed
+- `getseq` method returns the sequence string instead of printing it to the screen
+
+### Deprecated
+- `get_remote_data_str` method. Use `listr` instead
+
 ## [0.9.0] - 2020-07-01
 
 ### Changed

refgenconf/__init__.py

@@ -5,7 +5,8 @@
 from .helpers import *
 from .refgenconf import *
 
-__all__ = ["RefGenConf", "select_genome_config", "GenomeConfigFormatError",
-           "MissingAssetError", "MissingConfigDataError", "MissingGenomeError",
-           "RefgenconfError", "UnboundEnvironmentVariablesError"] + \
-          ["DEFAULT_SERVER"] + CFG_KEY_NAMES
+__all__ = ["RefGenConf", "select_genome_config", "get_dir_digest",
+           "GenomeConfigFormatError", "MissingAssetError",
+           "MissingConfigDataError", "MissingGenomeError", "RefgenconfError",
+           "UnboundEnvironmentVariablesError"] + ["DEFAULT_SERVER"] + \
+          CFG_KEY_NAMES

refgenconf/_version.py

@@ -1 +1 @@
-__version__ = "0.9.0"
+__version__ = "0.9.1"

refgenconf/helpers.py

@@ -1,11 +1,13 @@
 """ Helper functions """
 
 import os
-import yacman
-from .const import CFG_ENV_VARS
+from yacman import select_config
+from .const import CFG_ENV_VARS, BUILD_STATS_DIR
+from re import sub
+from ubiquerg import is_command_callable
 
 
-__all__ = ["select_genome_config"]
+__all__ = ["select_genome_config", "get_dir_digest"]
 
 
 def select_genome_config(filename=None, conf_env_vars=CFG_ENV_VARS, **kwargs):
@@ -17,7 +19,7 @@ def select_genome_config(filename=None, conf_env_vars=CFG_ENV_VARS, **kwargs):
         consider; basically, a prioritized search list
     :return str: path to genome configuration file
     """
-    return yacman.select_config(filename, conf_env_vars, **kwargs)
+    return select_config(filename, conf_env_vars, **kwargs)
 
 
 def unbound_env_vars(path):
@@ -38,3 +40,32 @@ def unbound_env_vars(path):
 def asciify_json_dict(json_dict):
     from ubiquerg.collection import asciify_dict
     return asciify_dict(json_dict)
+
+
+def get_dir_digest(path, pm=None):
+    """
+    Generate a MD5 digest that reflects just the contents of the
+    files in the selected directory.
+
+    :param str path: path to the directory to digest
+    :param pypiper.PipelineManager pm: a pipeline object, optional.
+    The subprocess module will be used if not provided
+    :return str: a digest, e.g. a3c46f201a3ce7831d85cf4a125aa334
+    """
+    if not is_command_callable("md5sum"):
+        raise OSError("md5sum command line tool is required for asset digest "
+                      "calculation. \n"
+                      "Install and try again, e.g on macOS: 'brew install "
+                      "md5sha1sum'")
+    cmd = "cd {}; find . -type f -not -path './" + BUILD_STATS_DIR + \
+          "*' -exec md5sum {{}} \; | sort -k 2 | awk '{{print $1}}' | md5sum"
+    try:
+        x = pm.checkprint(cmd.format(path))
+    except AttributeError:
+        try:
+            from subprocess import check_output
+            x = check_output(cmd.format(path), shell=True).decode("utf-8")
+        except Exception as e:
+
+            return
+    return str(sub(r'\W+', '', x))  # strips non-alphanumeric