Improved the makefile a bit for fetching OpenNLP dependencies. Update…

…d the setup guide accordingly.

Improved output from dependency checker.

examples/setup_guide.md

@@ -138,12 +138,10 @@ Beautiful Soup.
 Simple as that.
 
 OpenNLP is a little trickier. To make things simple, we just get all the OpenNLP tools at once. There are also a 
-couple of other libraries required to run the OpenNLP wrappers and we get them in the same way.
+couple of other libraries required to run the OpenNLP wrappers. The `opennlp` make target gets all of these at once.
 
     cd ~/myproject/pimlico/lib/java
     make opennlp
-    make py4j
-    make argparse4j.jar
 
 At the moment, it's also necessary to build the Java wrappers around OpenNLP that are provided as part of Pimlico. For 
 this, you'll need a Java compiler installed on your system.

lib/java/Makefile

@@ -10,8 +10,13 @@ else
 endif
 
 
+########## Meta targets ##########
+# These are collections of targets for whole modules, including all the dependencies for the module
+opennlp : opennlp-all py4j argparse4j guava
+
+
 ########## OPEN NLP ##############
-opennlp : opennlp-maxent-3.0.3.jar opennlp-tools-1.5.3.jar opennlp-uima-1.5.3.jar jwnl-1.3.3.jar
+opennlp-all : opennlp-maxent-3.0.3.jar opennlp-tools-1.5.3.jar opennlp-uima-1.5.3.jar jwnl-1.3.3.jar
 	# Clear up downloaded archive
 	rm -f apache-opennlp-1.5.3-bin.tar.gz
 	$(MAKE) argparse4j.jar
@@ -47,6 +52,10 @@ apache-opennlp-1.5.3-bin.tar.gz :
 	# Fetch OpenNlp archive
 	$(FETCH) http://apache.mesi.com.ar//opennlp/opennlp-1.5.3/apache-opennlp-1.5.3-bin.tar.gz
 
+##############################
+
+argparse4j : argparse4j.jar
+
 argparse4j.jar :
 	$(FETCH_STDOUT) http://sourceforge.net/projects/argparse4j/files/latest/download?source=files >argparse4j.jar
 
@@ -81,6 +90,8 @@ stanford-postagger-2014-01-04.zip :
 
 ####################################
 
+guava : guava.jar
+
 guava.jar :
 	@echo "Fetching Guava"
 	$(FETCH_STDOUT) http://search.maven.org/remotecontent?filepath=com/google/guava/guava/15.0/guava-15.0.jar >guava.jar

src/python/pimlico/cli/check.py

@@ -2,6 +2,7 @@
 
 from pimlico.core.config import check_for_cycles, PipelineStructureError
 from pimlico.core.modules.base import ModuleInfoLoadError
+from pimlico.utils.format import multiline_tablate
 
 
 def check_cmd(pipeline, opts):
@@ -46,8 +47,8 @@ def check_cmd(pipeline, opts):
             missing_dependencies.extend(pipeline[module_name].check_runtime_dependencies())
 
         if len(missing_dependencies):
-            print "Runtime dependencies not satisfied:\n%s" % (
-                "\n".join("- %s for '%s' (%s)" % (name, module, desc) for (name, module, desc) in missing_dependencies)
-            )
+            print "\nRuntime dependencies not satisfied:\n%s" % \
+                  multiline_tablate(missing_dependencies, [30, 30, 150],
+                                    tablefmt="orgtbl", headers=["Dependency", "Module", "Description"])
         else:
-            print "Runtime dependencies all satisfied"
+            print "\nRuntime dependencies all satisfied"

src/python/pimlico/core/external/java.py

@@ -1,12 +1,8 @@
-import Queue
-from collections import deque
-import os
 import time
 from subprocess import Popen, PIPE, check_output, STDOUT, CalledProcessError
 
 from pimlico import JAVA_LIB_DIR, JAVA_BUILD_DIR
 from pimlico.core.modules.base import DependencyError
-from py4j.java_gateway import JavaGateway, GatewayParameters, OutputConsumer, ProcessConsumer
 
 CLASSPATH = ":".join(["%s/*" % JAVA_LIB_DIR, JAVA_BUILD_DIR])
 
@@ -83,6 +79,8 @@ def start(self):
         Likewise with python_port and a --python-port option.
 
         """
+        from py4j.java_gateway import JavaGateway, GatewayParameters
+
         args = list(self.gateway_args)
         gateway_kwargs = {}
 
@@ -117,6 +115,8 @@ def launch_gateway(gateway_class="py4j.GatewayServer", args=[],
     """
     Our own more flexble version of Py4J's launch_gateway.
     """
+    from py4j.java_gateway import OutputConsumer, ProcessConsumer
+
     # Launch the server in a subprocess.
     command = ["java", "-classpath", CLASSPATH] + javaopts + [gateway_class] + args
     proc = Popen(command, stdout=PIPE, stdin=PIPE, stderr=PIPE)

src/python/pimlico/utils/format.py

@@ -0,0 +1,14 @@
+from tabulate import tabulate
+from textwrap import wrap
+
+def multiline_tablate(table, widths, **kwargs):
+    # Wrap columns
+    table = [[wrap(cell, width=width) for (cell, width) in zip(row, widths)] for row in table]
+    table_split = []
+    for row in table:
+        subrows = max(len(cell) for cell in row)
+        new_row = [cell + [""] * (subrows - len(cell)) for cell in row]
+        table_split.extend(zip(*new_row))
+        # Add a blank line
+        table_split.append([""] * len(table_split[0]))
+    return tabulate(table_split[:-1], **kwargs)