Improvements in results, stats, caching.

* Relaxer:
  - Status is now cached for future analysis!
* Stats:
  - Requires only status, can create reports.
* DirManager:
  - Added r(...) and w(...) for simple pickling
    of objects, without acting like a cache.
* CLI(config list): now shows diff between merged
    and default configuration for comparison
* Config does not save large data to cache, only
    the class for future loading. This greatly
    decreases cache sizes.

pyrelaxmapper/__init__.py

@@ -3,7 +3,7 @@
 
 from pyrelaxmapper import fileutils
 
-logging.config.fileConfig(fileutils.last_in_paths('logging.ini'))
+logging.config.fileConfig(fileutils.in_lowest_path('logging.ini'))
 logging.getLogger(__name__).addHandler(logging.NullHandler())
 
 __author__ = """Erik David Burnell"""

pyrelaxmapper/cli.py

@@ -50,7 +50,7 @@ def make(actions, conf_file):
     \b
     ACTIONS:
       clean    Clean all caches.
-      relax    Run RL algorithm.
+      relax    Run RL algorithm and save results.
       stats    Create statistics report.
     """
     if not actions:
@@ -64,15 +64,15 @@ def make(actions, conf_file):
     if Action.Clean in actions:
         commands.config_clean(config)
 
-    relaxer = commands.relaxer_load(config)
-
-    logger.info('Preloaded data.')
-
+    status = None
     if Action.Relax in actions:
-        commands.relaxer_relax(relaxer)
+        # could merge
+        relaxer = commands.relaxer_load(config)
+        status = commands.relaxer_relax(relaxer)
 
+    # Maybe should be able to pass status file!
     if Action.Stats in actions:
-        commands.relaxer_stats(relaxer, config)
+        commands.relaxer_stats(status, config)
 
     logger.info('End.')
 

pyrelaxmapper/commands.py

@@ -4,8 +4,10 @@
 from enum import Enum
 
 import click
+import sys
 
 from pyrelaxmapper import conf, fileutils, utils, relax
+from pyrelaxmapper.stats import Stats
 
 # Click Colors
 # TODO: turn into enum?
@@ -26,13 +28,34 @@ def relaxer_relax(relaxer):
     """Run RL Algorithm."""
     click.secho('Running RL algorithm.', fg=CInfo)
     relaxer.relax()
+    config = relaxer.config
+    stat_mru = config.results.path(fileutils.add_timestamp('Status.pkl'))
+    click.secho('Writing results.', fg=CInfo)
+    config.results.w(stat_mru, relaxer.status)
+    return relaxer.status
 
 
-def relaxer_stats(relaxer, config):
+# Create result selector?
+def relaxer_stats(status, config):
     """Save statistics for RL Algorithm."""
-    click.secho('Preparing statistics.', fg=CInfo)
-    with open(config.results.path(fileutils.add_timestamp('stats.csv')), 'w') as file:
-        relaxer.save_stats(file)
+    result_n = 'stats.csv'
+    status_n = 'Status.pkl'
+
+    try:
+        stat_mru = fileutils.newest(config.results.dir(), status_n)
+    except ValueError as e:
+        click.secho('No results found, run \'make relax\' to generate results.', fg=CWarn)
+        sys.exit(1)
+
+    if not status:
+        click.secho('Reading most recently saved results.')
+        status = config.results.r(stat_mru)
+    results = '{}{}'.format(stat_mru[:stat_mru.find(status_n)], result_n)
+
+    click.secho('Creating statistics report.', fg=CInfo)
+    with open(results, 'w') as file:
+        stats = Stats(status)
+        stats.create_report(file)
 
 
 #######################################################################
@@ -73,14 +96,27 @@ def config_list():
         click.echo(''.join([path, ': (exists)' if os.path.exists(path) else '']))
 
     click.secho('Merged configuration:', fg=CInfo)
-    parser = fileutils.conf_merge()
-    for section in parser.sections():
+    default = fileutils.conf_merge([fileutils.dir_pkg_conf()])
+    merged = fileutils.conf_merge()
+    sections = list(default.keys())
+    for section in sections:
         click.echo(section)
-        for key in parser[section].keys():
-            value = parser[section][key]
-            value = os.path.expanduser(value)
-            exists = ': (exists)' if os.path.exists(value) else ''
-            click.echo(''.join(['\t', key, ': ', value, exists]))
+        keys = list(default[section].keys())
+        keys.extend(key for key in merged[section].keys() if key not in keys)
+        for key in keys:
+            v_merged = merged.get(section, key, fallback='')
+            v_default = default.get(section, key, fallback='')
+            diff = ' '
+            if v_merged and not v_default:
+                diff = '+'
+            elif not v_merged and v_default:
+                diff = '-'
+            elif v_merged != v_default:
+                diff = '~'
+            diff += ' '
+            value = os.path.expanduser(v_merged)
+            exists = ' <-(exists)' if os.path.exists(value) else ''
+            click.echo(''.join(['\t', diff, key, ': ', value, exists]))
 
 
 def config_exists():
@@ -108,7 +144,7 @@ def logger_list(debug=True):
     with open(logger_file(), 'r') as file:
         for line in file:
             if 'level=' in line:
-                level = line[line.find('=')+1:-1]
+                level = line[line.find('=') + 1:-1]
                 break
     color = CWarn if level == 'ERROR' else CWarn
     click.secho('Logger level: {}'.format(level), fg=color)
@@ -132,7 +168,7 @@ def logger_file():
 
 def logger_edit():
     """Edit logger config file."""
-    click.edit(filename=fileutils.last_in_paths('logging.ini'))
+    click.edit(filename=fileutils.in_lowest_path('logging.ini'))
 
 
 #######################################################################

pyrelaxmapper/conf.py

@@ -127,6 +127,20 @@ def __init__(self, parser, wn_classes=None, constrainer=None, translater=None):
         self.cleaner = clean
         self.translater = translater if translater else Translater()
 
+    def __getstate__(self):
+        # Save only source and target classes, not the data itself.
+        parser = self._parser
+        sections = ['source', 'target']
+        source_wn, target_wn, = _select_wordnets(parser, sections, self._wn_classes)
+        return (self._parser, self._wn_classes, source_wn, target_wn, self.data, self.results,
+                self.cache, self.pos, self.constraints, self.constr_weights, self.cleaner,
+                self.translater)
+
+    def __setstate__(self, state):
+        (self._parser, self._wn_classes, self._source_wn, self._target_wn, self.data, self.results,
+         self.cache, self.pos, self.constraints, self.constr_weights, self.cleaner,
+         self.translater) = state
+
     def map_name(self):
         """Mapping name for folder organization."""
         return '{} -> {}'.format(self._source_wn.name(), self._target_wn.name())

pyrelaxmapper/constraints/hh.py

@@ -13,11 +13,13 @@
 
 
 # TODO: Init from config file.
+# TODO: Use only HHType to speed up hashing!
 class HHConstraint(Constraint):
     """Constraints which utilize hyper/hyponym connections."""
 
     def __init__(self, orig, dest, weights):
         super().__init__(orig, dest, weights)
+        # codes = ['ii', 'aa', 'ai', 'ia']
         codes = ['ii']
         # Anything better?
         self.rel_weight = {HHDirection.hyper: 1.0, HHDirection.hypo: 0.93}

pyrelaxmapper/dirmanager.py

@@ -11,6 +11,7 @@
 
 
 # TODO: Better extension handling
+# TODO: Default name is type name
 class DirManager:
     """Directory Manager, simplifying many common tasks.
 
@@ -45,12 +46,32 @@ def r(self, name, main_group=False, group=None):
             File group (folder)
             If main_group is set, will be placed inside it.
         """
-        path = self.path(name, main_group, group)
+        path = ensure_ext(self.path(name, main_group, group), self._extension)
+        try:
+            return self._load_obj(path)
+        except FileNotFoundError as e:
+            logger.debug('Reading error "{}". File: {}.'.format(e, path))
+
+    def w(self, name, obj, main_group=False, group=None):
+        """Load file inside managed directory.
+
+        Parameters
+        ----------
+        name : str
+            Name
+        obj : any
+            Object to save if file does not exist.
+        main_group : bool
+            If should use main context group.
+        group : str
+            File group (folder)
+            If main_group is set, will be placed inside it.
+        """
+        path = ensure_ext(self.path(name, main_group, group), self._extension)
         try:
-            return self._load_file(self.path(name, main_group, group))
+            self._save_obj(obj, path)
         except FileNotFoundError as e:
-            logger.debug('Loading error "{}". File: {}.'.format(e, path))
-            raise e
+            logger.debug('Writing error "{}". File: {}.'.format(e, path))
 
     def rw(self, name, obj, main_group=False, group=None, force=False):
         """Load or write object instance inside managed directory.
@@ -109,7 +130,7 @@ def rw_lazy(self, name, call, args=None, main_group=False, group=None, force=Fal
             try:
                 data = self._load_obj(path)
             except FileNotFoundError as e:
-                logger.debug('Loading error "{}". File: {}.'.format(e, path))
+                logger.debug('Reading error "{}". File: {}.'.format(e, path))
         if not data:
             data = call(*args)
             self._save_obj(data, path)
@@ -157,6 +178,7 @@ def remove_all(self):
         shutil.rmtree(self._directory)
         os.makedirs(self._directory)
 
+    # Extension
     def path(self, filename='', main_group=False, group=None, ensure=True):
         """Path to file inside managed directory."""
         directory = self._directory
@@ -170,28 +192,8 @@ def path(self, filename='', main_group=False, group=None, ensure=True):
         else:
             return os.path.join(directory, filename)
 
-    @staticmethod
-    def _save_file(lines, filename):
-        """Pickle object to filename.
-
-        Parameters
-        ----------
-        obj : any
-        filename : str
-        """
-        with open(filename, 'w') as f:
-            f.write(lines)
-
-    @staticmethod
-    def _load_file(filename):
-        """Loads pickled object from filename.
-
-        Parameters
-        ----------
-        filename : str
-        """
-        with open(filename, 'w') as f:
-            return f.read()
+    def dir(self):
+        return self._directory
 
     @staticmethod
     def _save_obj(obj, filename):

pyrelaxmapper/fileutils.py

@@ -1,6 +1,7 @@
 # -*- coding: utf-8 -*-
 """Configuration utilities."""
 import datetime
+import glob
 import logging
 import os
 import shutil
@@ -50,7 +51,7 @@ def conf_merge(paths=search_paths()):
     ConfigParser
         Single config parser with merged settings.
     """
-    file_paths = last_in_paths('conf.ini')
+    file_paths = in_lowest_path('conf.ini', paths)
     parser = ConfigParser()
     parser.read(file_paths)
     return parser
@@ -79,7 +80,7 @@ def cp_data_app_data(source_name, target_name=None):
     cp(dir_pkg_data(), source_name, dir_app_data(), target_name)
 
 
-def last_in_paths(filename, paths=search_paths()):
+def in_lowest_path(filename, paths=search_paths()):
     """Find last file with pattern in paths.
 
     Parameters
@@ -98,6 +99,12 @@ def last_in_paths(filename, paths=search_paths()):
     return file_paths[-1] if file_paths else file_paths
 
 
+def newest(directory, extension=''):
+    """Newest file in directory with extension."""
+    pattern = os.path.join(directory, '*{}'.format(extension))
+    return max(glob.iglob(pattern), key=os.path.getctime)
+
+
 def find_in_paths(filename, paths=search_paths()):
     """Find last file with pattern in paths.
 

pyrelaxmapper/relax.py

@@ -60,7 +60,8 @@ def _relax_loop(self, constrainer):
         writer.writerows(self.stats.stat_total().items())
 
     def _relax(self, constrainer):
-        with click.progressbar(self.status.remaining.values(), label='Constraining') as nodes:
+        todo = self.status.remaining.values()
+        with click.progressbar(todo, label='Constraining nodes:') as nodes:
             # for node in nodes:
             for idx, node in enumerate(nodes):
                 # if idx > 0 and idx % 50 == 0: