Generate numeric distances as symbols, but correctly parse both symbol & literal

nlmaps_tools/generate_nl.py

@@ -250,7 +250,7 @@ def generate_tag_query_features(thing_table, areas, pois, around=None,
                 features['maxdist'] = choose(
                     [Symbol('DIST_INTOWN'), Symbol('DIST_OUTTOWN'),
                      Symbol('WALKING_DIST'), Symbol('DIST_DAYTRIP'),
-                     str(random.randint(1, 100)) + '000'],
+                     Symbol(str(random.randint(1, 100)) + '000')],
                     [0.3, 0.25, 0.2, 0.1, 0.15]
                 )
 
@@ -367,8 +367,8 @@ def main(areas, pois, count=100, escape=False, nl_suffix='en', noise=False,
     else:
         nl_file = mrl_file = sys.stdout
 
-    special_phrases_file = (Path(os.path.dirname(os.path.abspath(__file__)))
-                            / 'special_phrases.txt')
+    special_phrases_file = (Path(__file__)
+                            / '../data/special_phrases.txt').resolve()
     thing_table = special_phrases_table(special_phrases_file)
 
     for _ in range(count):

nlmaps_tools/nl_templates/meta/around_query_proximity_prefix.jinja2

@@ -22,9 +22,9 @@
   {% set within_or_less_than = choose(['within', 'less than']) %}
   {{ within_or_less_than }}
   {% if choose(['km', '']) %}{# kilometers #}
-    {{ int(features['maxdist']) // 1000 }}{{ choose(['km', ' km', ' kilometers', ' kilometres'], [0.25, 0.25, 0.25, 0.25]) }}
+    {{ int(features['maxdist'].string) // 1000 }}{{ choose(['km', ' km', ' kilometers', ' kilometres'], [0.25, 0.25, 0.25, 0.25]) }}
   {% else %}{# meters #}
-    {{ features['maxdist'] }}{{ choose(['m', ' m', ' meters', ' metres'], [0.25, 0.25, 0.25, 0.25]) }}
+    {{ features['maxdist'].string }}{{ choose(['m', ' m', ' meters', ' metres'], [0.25, 0.25, 0.25, 0.25]) }}
   {% endif %}
   {{ 'away' if within_or_less_than == 'less than' else '' }}
   {{ choose(['of', 'from'], [0.4, 0.6]) }}

nlmaps_tools/nl_templates/meta/around_query_proximity_standalone.jinja2

@@ -13,9 +13,9 @@
   {% set within_or_less_than = choose(['within', 'less than']) %}
   {{ within_or_less_than }}
   {% if choose(['km', '']) %}{# kilometers #}
-    {{ int(features['maxdist']) // 1000 }}{{ choose(['km', ' km', ' kilometers', ' kilometres'], [0.25, 0.25, 0.25, 0.25]) }}
+    {{ int(features['maxdist'].string) // 1000 }}{{ choose(['km', ' km', ' kilometers', ' kilometres'], [0.25, 0.25, 0.25, 0.25]) }}
   {% else %}{# meters #}
-    {{ features['maxdist'] }}{{ choose(['m', ' m', ' meters', ' metres'], [0.25, 0.25, 0.25, 0.25]) }}
+    {{ features['maxdist'].string }}{{ choose(['m', ' m', ' meters', ' metres'], [0.25, 0.25, 0.25, 0.25]) }}
   {% endif %}
   {{ 'away' if within_or_less_than == 'less than' else '' }}
 {% endif %}

nlmaps_tools/parse_mrl.py

@@ -130,7 +130,10 @@ class MrlGrammar:
         ).setParseAction(self.makeSymbol)
 
 
-        integer = pp.Word(pp.nums).setParseAction(self.makeSymbol)
+        integer = (
+            pp.Word(pp.nums)
+            ^ (pp.Suppress("'") + pp.Word(pp.nums) + pp.Suppress("'"))
+        ).setParseAction(self.makeSymbol)
 
         distance_term = distance_symbol ^ integer
 
@@ -364,6 +367,7 @@ def filetest():
     import sys
     grammar = MrlGrammar()
     testfile = '/media/data/Dauerhaft/Studium/Computerlinguistik/Master-Arbeit/data/nlmaps_v2.1/split_1_train_dev_test/nlmaps.v2.train.mrl'
+    #testfile = '/media/data/Dauerhaft/Studium/Computerlinguistik/Master-Arbeit/data/nlmaps_v3delta/v3delta.normal/nlmaps.v3delta.train.mrl'
 
     mrl_to_features = {}
     max_fails = 10
@@ -412,6 +416,7 @@ def test():
         "dist(query(around(center(area(keyval('name','Edinburgh')),nwr(keyval('name','Palace of Holyroodhouse'))),search(nwr(keyval('name','Edinburgh Waverley'),keyval('railway','station'))),maxdist(DIST_INTOWN)),qtype(latlong)))",
         "dist(query(area(keyval('name','Heidelberg')),nwr(keyval('name','Heidelberger Schloss')),qtype(latlong)),query(area(keyval('name','Heidelberg')),nwr(keyval('name','Heidelberg Hbf')),qtype(least(topx(1)))),for('walk'))",
         "query(nwr(keyval('amenity','restaurant')),qtype(latlong))",
+        "query(around(center(area(keyval('name','Heidelberg')),nwr(keyval('name','INF 325'))),search(nwr(keyval('shop','supermarket'),or(keyval('organic','only'),keyval('organic','yes')))),maxdist('5000')),qtype(latlong))",
     ]
 
     for mrl in test_mrls: