remove nlp_tools

7 files changed, 0 insertions, 208 deletions

diff --git a/.gitmodules b/.gitmodules
index 404b718..e69de29 100644
--- a/.gitmodules
+++ b/.gitmodules
@@ -1,3 +0,0 @@
-[submodule "nlp_tools"]
-	path = nlp_tools
-	url = https://github.com/jacobandreas/nlp_tools.git
diff --git a/nlp_tools/dict_utils.py b/nlp_tools/dict_utils.py
deleted file mode 100644
index 8b9b94b..0000000
--- a/nlp_tools/dict_utils.py
+++ /dev/null
@@ -1,101 +0,0 @@
-"""
-Utilities for doing math on sparse vectors indexed by arbitrary objects.
-(These will usually be feature vectors.)
-"""
-
-import math_utils as mu
-import math
-
-def d_elt_op_keep(op, zero, args):
-  """
-  Applies op to arguments elementwise, keeping entries that don't occur in
-  every argument (i.e. behaves like a sum).
-  """
-  ret = {}
-  for d in args:
-    for key in d:
-      if key not in ret:
-        ret[key] = d[key]
-      else:
-        ret[key] = op([ret[key], d[key]])
-  for key in ret.keys():
-    if ret[key] == zero:
-      del ret[key]
-  return ret
-
-def d_elt_op_drop(op, args):
-  """
-  Applies op to arguments elementwise, discarding entries that don't occur in
-  every argument (i.e. behaves like a product).
-  """
-  # avoid querying lots of nonexistent keys
-  smallest = min(args, key=len)
-  sindex = args.index(smallest)
-  ret = dict(smallest)
-  for i in range(len(args)):
-    if i == sindex:
-      continue
-    d = args[i]
-    for key in ret.keys():
-      if key in d:
-        ret[key] = op([ret[key], d[key]])
-      else:
-        del ret[key]
-  return ret
-
-def d_sum(args):
-  """
-  Computes a sum of vectors.
-  """
-  return d_elt_op_keep(sum, 0, args)
-
-def d_logspace_sum(args):
-  """
-  Computes a sum of vectors whose elements are represented in logspace.
-  """
-  return d_elt_op_keep(mu.logspace_sum, -float('inf'), args)
-
-def d_elt_prod(args):
-  """
-  Computes an elementwise product of vectors.
-  """
-  return d_elt_op_drop(lambda l: reduce(lambda a,b: a*b, l), args)
-
-def d_dot_prod(d1, d2):
-  """
-  Takes the dot product of the two arguments.
-  """
-  # avoid querying lots of nonexistent keys
-  if len(d2) < len(d1):
-    d1, d2 = d2, d1
-  dot_prod = 0
-  for key in d1:
-    if key in d2:
-      dot_prod += d1[key] * d2[key]
-  return dot_prod
-
-def d_logspace_scalar_prod(c, d):
-  """
-  Multiplies every element of d by c, where c and d are both represented in
-  logspace.
-  """
-  ret = {}
-  for key in d:
-    ret[key] = c + d[key]
-  return ret
-
-def d_op(op, d):
-  """
-  Applies op to every element of the dictionary.
-  """
-  ret = {}
-  for key in d:
-    ret[key] = op(d[key])
-  return ret
-
-# convenience methods
-def d_log(d):
-  return d_op(math.log, d)
-
-def d_exp(d):
-  return d_op(math.exp, d)
diff --git a/nlp_tools/dict_utils.pyc b/nlp_tools/dict_utils.pyc
deleted file mode 100644
index ada4c58..0000000
--- a/nlp_tools/dict_utils.pyc
+++ /dev/null
diff --git a/nlp_tools/feature.pyc b/nlp_tools/feature.pyc
deleted file mode 100644
index 9c96271..0000000
--- a/nlp_tools/feature.pyc
+++ /dev/null
diff --git a/nlp_tools/vocabulary.py b/nlp_tools/vocabulary.py
deleted file mode 100644
index ed200f5..0000000
--- a/nlp_tools/vocabulary.py
+++ /dev/null
@@ -1,49 +0,0 @@
-import cPickle
-
-class Vocabulary:
-
-  OOV_VAL = -1
-
-  def __init__(self):
-    self.str_to_tok = {}
-    self.tok_to_str = {}
-
-  def put(self, string):
-    if string in self.str_to_tok:
-      raise ValueError("%s is already in this vocabulary (token %d)" % \
-          (string, self.str_to_tok[string]))
-    return self.ensure(string)
-
-  def ensure(self, string):
-    if string in self.str_to_tok:
-      return
-    tok = len(self)
-    self.str_to_tok[string] = tok
-    self.tok_to_str[tok] = string
-    return tok
-
-  def gett(self, string):
-    if string not in self.str_to_tok:
-      return self.OOV_VAL
-    return self.str_to_tok[string]
-
-  def gets(self, tok):
-    return self.tok_to_str[tok]
-
-  def strs(self):
-    return self.str_to_tok.keys()
-
-  def toks(self):
-    return self.tok_to_str.keys()
-
-  def __len__(self):
-    return len(self.str_to_tok)
-
-  def save(self, path):
-    with open(path, 'w') as f:
-      cPickle.dump(self, f)
-
-  @classmethod
-  def load(cls, path):
-    with open(path) as f:
-      return cPickle.load(f)
diff --git a/nlp_tools/vocabulary.pyc b/nlp_tools/vocabulary.pyc
deleted file mode 100644
index 952b7fd..0000000
--- a/nlp_tools/vocabulary.pyc
+++ /dev/null
diff --git a/run2.py b/run2.py
deleted file mode 100755
index a92c118..0000000
--- a/run2.py
+++ /dev/null
@@ -1,55 +0,0 @@
-#!/usr/bin/env python2
-
-import os
-import datetime
-import logging
-from src.evaluator import Evaluator
-from src.smt_semparse_config import SMTSemparseConfig
-from src.smt_semparse_experiment import SMTSemparseExperiment
-
-LOGFILE_NAME = 'run.log'
-
-def run_one(config):
-  # create work dir for this run
-  # moses can't handle paths with colons
-  timestamp = datetime.datetime.now().strftime('%Y-%m-%dT%H.%M.%S')
-  run_work_dir = os.path.join(base_work_dir, timestamp)
-  assert not os.path.exists(run_work_dir)
-  os.makedirs(run_work_dir)
-  config.put('work_dir', run_work_dir)
-  if os.path.exists('latest'):
-    os.remove('latest')
-  os.symlink(run_work_dir, 'latest')
-
-  # set up logging
-  if config.run == 'debug':
-    logging.basicConfig(level=logging.DEBUG)
-  else:
-    log_path = os.path.join(run_work_dir, LOGFILE_NAME)
-    logging.basicConfig(filename=log_path, level=logging.INFO)
-
-  experiment = SMTSemparseExperiment(config)
-  if config.run == 'debug':
-    experiment.run_fold(1)
-  elif config.run == 'dev':
-    for i in range(10):
-      experiment.run_fold(i)
-  elif config.run == 'test':
-    experiment.run_split()
-  else:
-    assert False
-
-  logging.info('evaluating')
-  Evaluator(config).run()
-
-if __name__ == '__main__':
-
-  # load config
-  config = SMTSemparseConfig('settings.yaml', 'dependencies.yaml')
-
-  # create base work dir if it doesn't exist
-  base_work_dir = os.path.join(config.smt_semparse, config.workdir)
-  if not os.path.exists(base_work_dir):
-    os.makedirs(base_work_dir)
-
-  run_one(config)