Accommodating pipes in programs.py

Author: cmarkello

src/toil_scripts/lib/programs.py

@@ -29,7 +29,8 @@ def docker_call(tool,
     """
     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 OR list[str] tool: Single str or list of names of the Docker images and order to be used 
+                      (e.g. quay.io/ucsc_cgl/samtools)
     :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.
@@ -83,21 +84,62 @@ 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
-
-    _log.debug("Calling docker with %s." % " ".join(base_docker_call + [tool] + parameters))
-
-    docker_call = base_docker_call + [tool] + parameters
-
+   
+    docker_call = []
+    
+    # Pipe functionality
+    shell_flag = False      # Flag for running subprocess with string command or list command
+    if '|' in parameters:
+        command_list = []
+        if isinstance(tool, list):
+            # If tool is a list of docker tool strings
+            # Parse pipes if they exist into list of commands
+            bar_indices = [idx for idx, arg in enumerate(parameters) if arg == '|']
+            command = parameters[0:bar_indices[0]]
+            command[0] = "'"+command[0]
+            command[len(command)-1] += "'"
+            command_list.append(command)
+            prev_idx = bar_indices[0]
+            for idx in bar_indices[1:]:
+                command = parameters[prev_idx+1:idx]
+                command_list.append(command)
+                prev_idx = idx
+            
+            command = parameters[prev_idx+1:]
+            command_list.append(command)
+            for i in len(tool):
+                docker_call.extend(base_docker_call + ['--entrypoint /bin/bash', tool[i], '-c'] + command_list[i])
+            
+            docker_call = " ".join(docker_call)
+            shell_flag=True
+            _log.debug("Calling docker with %s." % docker_call)
+        
+        elif isinstance(tool, str):
+            # If tool is a string representing the docker tool name
+            # Set pipes as a single command run under a single docker container
+            parameters[0] = "'"+parameters[0]
+            parameters[len(parameters)-1] += "'"
+            docker_call.extend(base_docker_call + ['--entrypoint /bin/bash', tool, '-c'] + parameters)
+            docker_call = " ".join(docker_call)
+            shell_flag=True
+            _log.debug("Calling docker with %s." % docker_call)
+             
+    else:
+        # Default behavior
+        docker_call = base_docker_call + [tool] + parameters
+        _log.debug("Calling docker with %s." % " ".join(docker_call))
+    
     try:
         if outfile:
-            subprocess.check_call(docker_call, stdout=outfile)
+            subprocess.check_call(docker_call, stdout=outfile, shell=shell_flag)
         else:
             if check_output:
-                return subprocess.check_output(docker_call)
+                return subprocess.check_output(docker_call, shell=shell_flag)
             else:
-                subprocess.check_call(docker_call)
+                subprocess.check_call(docker_call, shell=shell_flag)
     # Fix root ownership of output files
     except:
         # Panic avoids hiding the exception raised in the try block