Accommodating pipes in programs.py

Adding in docker pipes unit test

Added docs to docker_call and fixed docker_call unit test

Changed datasize for docker_call pipes unit test to 1GB

Polished docker_call functionality

Added stderr to file handle support in docker_call

Factored out pipe_of_containers in docker_call

Author: cmarkello

src/toil_scripts/lib/programs.py

@@ -2,6 +2,7 @@
 import subprocess
 import logging
 from bd2k.util.exceptions import panic
+from toil_scripts.lib import require
 
 _log = logging.getLogger(__name__)
 
@@ -15,34 +16,55 @@ def mock_mode():
     return True if int(os.environ.get('TOIL_SCRIPTS_MOCK_MODE', '0')) else False
 
 
-def docker_call(tool,
+def docker_call(tool=None,
+                tools=None,
                 parameters=None,
                 work_dir='.',
                 rm=True,
                 env=None,
                 outfile=None,
+                errfile=None,
                 inputs=None,
                 outputs=None,
                 docker_parameters=None,
                 check_output=False,
+                return_stderr=False,
                 mock=None):
     """
     Calls Docker, passing along parameters and tool.
 
-    :param str tool: Name of the Docker image to be used (e.g. quay.io/ucsc_cgl/samtools)
+    :param (str tool | str tools): str tool name of the Docker image to be used (e.g. tool='quay.io/ucsc_cgl/samtools')
+                     OR str tools of the Docker images and order to be used when piping commands to
+                     Docker. (e.g. 'quay.io/ucsc_cgl/samtools'). Both tool and tools are mutually
+                     exclusive parameters to docker_call.
     :param list[str] parameters: Command line arguments to be passed to the tool
     :param str work_dir: Directory to mount into the container via `-v`. Destination convention is /data
     :param bool rm: Set to True to pass `--rm` flag.
     :param dict[str,str] env: Environment variables to be added (e.g. dict(JAVA_OPTS='-Xmx15G'))
     :param bool sudo: If True, prepends `sudo` to the docker call
-    :param file outfile: Pipe output of Docker call to file handle
+    :param file outfile: Pipe stdout of Docker call to file handle
+    :param file errfile: Pipe stderr of Docker call to file handle
     :param list[str] inputs: A list of the input files.
     :param dict[str,str] outputs: A dictionary containing the outputs files as keys with either None
                                   or a url. The value is only used if mock=True
     :param dict[str,str] docker_parameters: Parameters to pass to docker
     :param bool check_output: When True, this function returns docker's output
+    :param bool return_stderr: When True, this function includes stderr in docker's output
     :param bool mock: Whether to run in mock mode. If this variable is unset, its value will be determined by
                       the environment variable.
+
+    Pipes in docker commands:
+    Running a pipe in docker in 'pipe-in-single-container' mode produces command structure
+        docker '... | ... | ...' where each '...' command corresponds to each element in the 'parameters'
+        argument that uses a docker container. This is the most efficient method if you want to run a pipe of
+        commands where each command uses the same docker container.
+    
+    Example for running command 'head -c 1M /dev/urandom | gzip | gunzip | md5sum 1>&2':
+        Running 'pipe-in-single-container' mode:
+            command= ['head -c 1M /dev/urandom', 'gzip', 'gunzip', 'md5sum 1>&2']
+            docker_work_dir=curr_work_dir
+            docker_tools='ubuntu'
+            stdout = docker_call(work_dir=docker_work_dir, parameters=command, tools=docker_tools, check_output=True)
     """
     from toil_scripts.lib.urls import download_url
 
@@ -83,37 +105,56 @@ def docker_call(tool,
     if env:
         for e, v in env.iteritems():
             base_docker_call.extend(['-e', '{}={}'.format(e, v)])
+    
     if docker_parameters:
         base_docker_call += docker_parameters
+   
+    docker_call = []
+    
+    require(bool(tools) != bool(tool), 'Either "tool" or "tools" must contain a value, but not both')
+    
+    # Pipe functionality
+    #   each element in the parameters list must represent a sub-pipe command
+    if bool(tools):
+        # If tools is set then format the docker call in the 'pipe-in-single-container' mode
+        docker_call = " ".join(base_docker_call + ['--entrypoint /bin/bash', tools, '-c \'{}\''.format(" | ".join(parameters))])
+        _log.debug("Calling docker with %s." % docker_call)
+        
+    else:        
+        docker_call = " ".join(base_docker_call + [tool] + parameters)
+        _log.debug("Calling docker with %s." % docker_call)
 
-    _log.debug("Calling docker with %s." % " ".join(base_docker_call + [tool] + parameters))
-
-    docker_call = base_docker_call + [tool] + parameters
-
+    
     try:
         if outfile:
-            subprocess.check_call(docker_call, stdout=outfile)
+            if errfile:
+                subprocess.check_call(docker_call, stdout=outfile, stderr=errfile, shell=True)
+            else:
+                subprocess.check_call(docker_call, stdout=outfile, shell=True)
         else:
             if check_output:
-                return subprocess.check_output(docker_call)
+                if return_stderr:
+                    return subprocess.check_output(docker_call, shell=True, stderr=subprocess.STDOUT)
+                else:
+                    return subprocess.check_output(docker_call, shell=True)
             else:
-                subprocess.check_call(docker_call)
+                subprocess.check_call(docker_call, shell=True)
     # Fix root ownership of output files
     except:
         # Panic avoids hiding the exception raised in the try block
         with panic():
-            _fix_permissions(base_docker_call, tool, work_dir)
+            _fix_permissions(base_docker_call, tool, tools, work_dir)
     else:
-        _fix_permissions(base_docker_call, tool, work_dir)
+        _fix_permissions(base_docker_call, tool, tools, work_dir)
 
     for filename in outputs.keys():
         if not os.path.isabs(filename):
             filename = os.path.join(work_dir, filename)
         assert(os.path.isfile(filename))
 
 
-def _fix_permissions(base_docker_call, tool, work_dir):
-    """
+def _fix_permissions(base_docker_call, tool, tools, work_dir):
+    """ 
     Fix permission of a mounted Docker directory by reusing the tool
 
     :param list base_docker_call: Docker run parameters
@@ -122,5 +163,13 @@ def _fix_permissions(base_docker_call, tool, work_dir):
     """
     base_docker_call.append('--entrypoint=chown')
     stat = os.stat(work_dir)
-    command = base_docker_call + [tool] + ['-R', '{}:{}'.format(stat.st_uid, stat.st_gid), '/data']
-    subprocess.check_call(command)
+    if tools:
+        command = base_docker_call + [tools] + ['-R', '{}:{}'.format(stat.st_uid, stat.st_gid), '/data']
+        subprocess.check_call(command)
+    else:
+        command = base_docker_call + [tool] + ['-R', '{}:{}'.format(stat.st_uid, stat.st_gid), '/data']
+        subprocess.check_call(command)
+
+
+
+

src/toil_scripts/lib/test/test_programs.py

@@ -1,5 +1,5 @@
 import os
-
+import re
 
 def test_docker_call(tmpdir):
     from toil_scripts.lib.programs import docker_call
@@ -12,3 +12,15 @@ def test_docker_call(tmpdir):
     with open(fpath, 'w') as f:
         docker_call(tool='ubuntu', env=dict(foo='bar'), parameters=['printenv', 'foo'], outfile=f)
     assert open(fpath).read() == 'bar\n'
+
+    # Test pipe functionality
+    # download ubuntu docker image
+    docker_call(work_dir=work_dir, tool="ubuntu")
+    command1 = ['head -c 1G /dev/urandom | tee /data/first', 'gzip', 'gunzip', 'md5sum 1>&2']
+    command2 = ['md5sum /data/first 1>&2']
+    # Test 'pipe-in-single-container' mode
+    stdout1 = docker_call(work_dir=work_dir, parameters=command1, tools='ubuntu', check_output=True, return_stderr=True)
+    stdout2 = docker_call(work_dir=work_dir, parameters=command2, tool='ubuntu', check_output=True, return_stderr=True)
+    test1 = re.findall(r"([a-fA-F\d]{32})", stdout1)
+    test2 = re.findall(r"([a-fA-F\d]{32})", stdout2)
+    assert test1[0] == test2[0]