python/bllipparser: Add Tree.visualize_nltk()

Moved NLTK tree reading and import_maybe to Utility

python/bllipparser/ParsingShell.py

@@ -26,35 +26,16 @@
 trees will be shown if you have the asciitree package."""
 import sys
 from cmd import Cmd
-import importlib
 from os.path import exists
 
 from .RerankingParser import RerankingParser
 from .ModelFetcher import list_models
-
-def import_maybe(module_name):
-    "Import a module and return it if available, otherwise returns None."
-    try:
-        return importlib.import_module(module_name)
-    except ImportError:
-        return None
-
-try:
-    import nltk.tree
-    import nltk.draw.tree
-    read_nltk_tree = nltk.tree.Tree.parse
-    have_nltk_tree_drawing = True
-except ImportError:
-    have_nltk_tree_drawing = False
-except AttributeError: # handle NLTK API changes
-    try:
-        read_nltk_tree = nltk.tree.Tree.fromstring
-        have_nltk_tree_drawing = True
-    except AttributeError:
-        have_nltk_tree_drawing = False
+from .Utility import import_maybe, get_nltk_tree_reader_maybe
 
 StanfordDependencies = import_maybe('StanfordDependencies')
 asciitree = import_maybe('asciitree')
+nltk = import_maybe('nltk')
+read_nltk_tree = get_nltk_tree_reader_maybe()
 
 class ParsingShell(Cmd):
     def __init__(self, model_dir):
@@ -68,7 +49,7 @@ def __init__(self, model_dir):
 
     def get_rrp(self, model_dir):
         if model_dir is None:
-            print "Warning: no parsing model specified."
+            print "Error: no parsing model specified."
             print "Specify with: python -mbllipparser /path/to/model/"
             print "Or one of these to download and install:"
             list_models()
@@ -105,7 +86,7 @@ def get_rrp(self, model_dir):
     def do_visual(self, text):
         """Use reranking parser to parse text.  Visualize top parses from
         parser and reranker."""
-        if not have_nltk_tree_drawing:
+        if not read_nltk_tree:
             print "Can't visualize without NLTK installation."
             return
 
@@ -153,7 +134,7 @@ def do_visualnbest(self, text):
         """Usage: visualnbest [start] stop
         Visualizes all parses from start-stop in the n-best list.
         Sentence must already be parsed."""
-        if not have_nltk_tree_drawing:
+        if not read_nltk_tree:
             print "Can't visualize without NLTK installation."
             return
 

python/bllipparser/RerankingParser.py

@@ -168,6 +168,14 @@ def dependencies(self):
                         yield (head, subhead)
             else:
                 tree_to_heads[tree.span()] = tree
+    def visualize_nltk(self):
+        from .Utility import get_nltk_tree_reader_maybe
+        read_nltk_tree = get_nltk_tree_reader_maybe()
+        if not read_nltk_tree:
+            raise ValueError("Unable to import nltk tree reading.")
+        nltk_tree = read_nltk_tree(str(self))
+        import nltk
+        nltk.draw.tree.draw_trees(nltk_tree)
 
     #
     # properties
@@ -245,6 +253,10 @@ def sd_tokens(self, sd_converter=None, conversion_kwargs=None):
                                                         **conversion_kwargs)
         return self._sd_tokens
 
+    #
+    # readers
+    #
+
     @classmethod
     def trees_from_string(this_class, text):
         """Given text containing multiple Penn Treebank trees, returns

python/bllipparser/Utility.py

@@ -11,6 +11,7 @@
 # under the License.
 
 import math
+import importlib
 
 # now part of waterworks.Tools
 class DeprecatedGetter:
@@ -37,6 +38,29 @@ def __call__(self, *args, **kwargs):
              DeprecationWarning, stacklevel=2)
         return self.__value
 
+def import_maybe(module_name):
+    "Import a module and return it if available, otherwise returns None."
+    try:
+        return importlib.import_module(module_name)
+    except ImportError:
+        return None
+
+def get_nltk_tree_reader_maybe():
+    """Attempts to find the NLTK tree reader for various versions of NLTK.
+    Returns False if it fails or a function which takes a string and
+    returns an NLTK tree object otherwise."""
+    try:
+        import nltk.tree
+        import nltk.draw.tree
+        return nltk.tree.Tree.parse
+    except ImportError:
+        return False
+    except AttributeError: # handle NLTK API changes
+        try:
+            return nltk.tree.Tree.fromstring
+        except AttributeError:
+            return False
+
 def normalize_logprobs(logprobs, exponent=1):
     """Sum probs stored as log probs in a (more) numerically stable
     fashion, see: