Adding a first simple grammar to parse OpenCCG's output into digestible JSON objects.

Dockerfile

@@ -23,7 +23,7 @@ RUN curl -o openccg-0.9.5.tgz https://datapacket.dl.sourceforge.net/project/open
 # Server software: python 3, nginx, uwsgi
     && apt-get update \
     && apt-get install -y python3 python3-pip nginx \
-    && pip3 install flask uwsgi \
+    && pip3 install flask uwsgi tatsu \
 # Configure nginx
     && ln -s /etc/nginx/sites-available/occg /etc/nginx/sites-enabled/occg \
     && rm /etc/nginx/sites-enabled/default

app/ccgapp.py

@@ -24,10 +24,20 @@ def is_non_gui_agent(ua_string):
 
 
 def create_response(sentence):
+    """Prepares a response.
+
+    Parses the sentence using wccg, then enriches the result with
+    some meta data.
+
+    Args:
+        sentence: The sentence to parse.
+    Returns:
+        The response as a dictionary.
+    """
     content = wccg.parse(sentence)
 
     response = {
-        'version': '1.0.0',
+        'version': '1.1.0',
         'application': 'web-openccg',
         'uuid': str(uuid.uuid4())
     }

app/ccgparser.py

+from tatsu import parse
+from tatsu.util import asjson
+
+
+GRAMMAR = r"""
+
+start
+    = semspec $
+    ;
+
+semspec
+    = nominal
+    | term
+    | conjunction
+    | expression
+    ;
+
+conjunction
+    = @:expression '^' ~ @:conjunction
+    | @:expression
+    ;
+
+expression
+    = role_expression
+    | nominal_expression
+    | variable_expression
+    | atom_expression
+    ;
+
+role_expression
+    = role:role
+    ;
+
+variable_expression
+    = variable:variable
+    ;
+
+nominal_expression
+    = nominal:nominal
+    ;
+
+atom_expression
+    = atom:atom
+    ;
+
+term
+    = '(' ~ @:conjunction ')'
+    ;
+
+nominal
+    = '@' ~ nominal:variable roles:term
+    ;
+
+role
+    = '<' type:atom '>' target:variable
+    | '<' type:atom '>' target:term
+    | '<' type:atom '>' target:atom
+    ;
+
+variable
+    = /[a-z]\d+(:[a-zA-Z\-]+)*/
+    ;
+
+atom
+    = /[a-zA-Z\-\.]+/
+    ;
+"""
+
+
+def ccg_to_json(to_parse):
+    """Parses an OpenCCG string into a more easily digestiable JSON format.
+
+    Args:
+        to_parse: The OpenCCG string.
+    Returns:
+        A JSON representation of the OpenCCG string.
+    """
+    return asjson(parse(GRAMMAR, to_parse))

app/wccg.py

@@ -3,6 +3,8 @@ import re
 import string
 import subprocess
 
+from ccgparser import ccg_to_json
+
 
 def parse(sentence):
     """Parses a sentence using OpenCCG's command line tool wccg.
@@ -27,7 +29,11 @@ def parse(sentence):
                                  universal_newlines=True)
     sentence = re.sub(f'[{re.escape(string.punctuation)}]', '', sentence).lower()
     response = wccg_proc.communicate(input=sentence)[0]
-    return _as_dict(response or f'"{sentence}": Unable to parse. wccg returned an empty response.')
+
+    wccg_response = _as_dict(response or f'"{sentence}": Unable to parse. wccg returned an empty response.')
+    wccg_response = jsonify_parses(wccg_response)
+
+    return wccg_response
 
 
 def ensure_unique_key(new_key, dictionary):
@@ -94,3 +100,20 @@ def _as_dict(response):
     return dict(sentence=sentence[1:-1],
                 parses=parses,
                 http_status=200)
+
+
+def jsonify_parses(wccg_response):
+    """Converts the OpenCCG responses to proper JSON objects and adds them
+    to the supplied dictionary.
+
+    Args:
+        wccg_response: The dictionary as returned by _as_dict.
+    Returns:
+        A copy of the dictionary, additionally containing parses_json.
+    """
+    json_parses = {}
+    for key, parse in wccg_response.get('parses', {}).items():
+        json_parses[key] = ccg_to_json(parse)
+    copy = wccg_response.copy()
+    copy['json_parses'] = json_parses
+    return copy