Format feature aspect and profile comparison strings as table.

PiperOrigin-RevId: 652022995

nisaba/scripts/natural_translit/phonology/descriptive_features_test.py

@@ -74,79 +74,67 @@ def test_profile_p(self):
         '}\n',
     )
 
+  def test_compare_p_f_verbose(self):
+    self.assertEqual(
+        _P.p.comparison_table(_P.f, verbose=True),
+        'phonology_descriptive comparison (max distance = 23.50):\n\n'
+        '| aspect         | p              | f              |   distance |\n'
+        '|----------------|----------------|----------------|------------|\n'
+        '| ph_class       | consonant      | consonant      |      0     |\n'
+        '| airstream      | pulmonic       | pulmonic       |      0     |\n'
+        '| manner         | stop           | non_sibilant   |      1     |\n'
+        '| place          | labial         | dental         |      0.5   |\n'
+        '| articulator    | labial         | labial         |      0     |\n'
+        '| height         | not_applicable | not_applicable |      0     |\n'
+        '| backness       | not_applicable | not_applicable |      0     |\n'
+        '| breathiness    | any            | any            |      0     |\n'
+        '| voicing        | voiceless      | voiceless      |      0     |\n'
+        '| labialization  | labialized     | labialized     |      0     |\n'
+        '| lateralization | none           | none           |      0     |\n'
+        '| nasalization   | none           | none           |      0     |\n'
+        '| palatalization | none           | none           |      0     |\n'
+        '| rhoticization  | none           | none           |      0     |\n'
+        '| duration       | any            | any            |      0     |\n'
+        '| syllabicity    | none           | none           |      0     |\n'
+        '| Total distance |                |                |      1.5   |\n'
+        '| Similarity     |                |                |      0.936 |\n',
+    )
+
   def test_compare_p_f(self):
     self.assertEqual(
-        _P.p.comparison(_P.f),
-        'p - f phonology_descriptive comparison:\n'
-        '    ph_class: {consonant} vs ph_class: {consonant} = 0.00\n'
-        '    airstream: {pulmonic} vs airstream: {pulmonic} = 0.00\n'
-        '    manner: {stop} vs manner: {non_sibilant} = 1.00\n'
-        '    place: {labial} vs place: {dental} = 0.50\n'
-        '    articulator: {labial} vs articulator: {labial} = 0.00\n'
-        '    height: {not_applicable} vs height: {not_applicable} = 0.00\n'
-        '    backness: {not_applicable} vs backness: {not_applicable} = 0.00\n'
-        '    breathiness: {any} vs breathiness: {any} = 0.00\n'
-        '    voicing: {voiceless} vs voicing: {voiceless} = 0.00\n'
-        '    labialization: {labialized} vs labialization: {labialized} = 0.00'
-        '\n'
-        '    lateralization: {none} vs lateralization: {none} = 0.00\n'
-        '    nasalization: {none} vs nasalization: {none} = 0.00\n'
-        '    palatalization: {none} vs palatalization: {none} = 0.00\n'
-        '    rhoticization: {none} vs rhoticization: {none} = 0.00\n'
-        '    duration: {any} vs duration: {any} = 0.00\n'
-        '    syllabicity: {none} vs syllabicity: {none} = 0.00\n'
-        '    Total distance = 1.50/23.50\n'
-        '    Similarity = 0.936\n',
+        _P.p.comparison_table(_P.f),
+        'phonology_descriptive comparison (max distance = 23.50):\n\n'
+        '| aspect         | p      | f            |   distance |\n'
+        '|----------------|--------|--------------|------------|\n'
+        '| manner         | stop   | non_sibilant |      1     |\n'
+        '| place          | labial | dental       |      0.5   |\n'
+        '| Total distance |        |              |      1.5   |\n'
+        '| Similarity     |        |              |      0.936 |\n',
     )
 
   def test_compare_f_s(self):
     self.assertEqual(
-        _P.f.comparison(_P.s),
-        'f - s phonology_descriptive comparison:\n'
-        '    ph_class: {consonant} vs ph_class: {consonant} = 0.00\n'
-        '    airstream: {pulmonic} vs airstream: {pulmonic} = 0.00\n'
-        '    manner: {non_sibilant} vs manner: {sibilant} = 0.50\n'
-        '    place: {dental} vs place: {alveolar, dental, postalveolar} = 0.00'
-        '\n'
-        '    articulator: {labial} vs articulator: {apical, laminal} = 0.50\n'
-        '    height: {not_applicable} vs height: {not_applicable} = 0.00\n'
-        '    backness: {not_applicable} vs backness: {not_applicable} = 0.00\n'
-        '    breathiness: {any} vs breathiness: {any} = 0.00\n'
-        '    voicing: {voiceless} vs voicing: {voiceless} = 0.00\n'
-        '    labialization: {labialized} vs labialization: {none} = 1.00\n'
-        '    lateralization: {none} vs lateralization: {none} = 0.00\n'
-        '    nasalization: {none} vs nasalization: {none} = 0.00\n'
-        '    palatalization: {none} vs palatalization: {none} = 0.00\n'
-        '    rhoticization: {none} vs rhoticization: {none} = 0.00\n'
-        '    duration: {any} vs duration: {any} = 0.00\n'
-        '    syllabicity: {none} vs syllabicity: {none} = 0.00\n'
-        '    Total distance = 2.00/23.50\n'
-        '    Similarity = 0.915\n',
+        _P.f.comparison_table(_P.s),
+        'phonology_descriptive comparison (max distance = 23.50):\n\n'
+        '| aspect         | f            | s               |   distance |\n'
+        '|----------------|--------------|-----------------|------------|\n'
+        '| manner         | non_sibilant | sibilant        |      0.5   |\n'
+        '| articulator    | labial       | apical, laminal |      0.5   |\n'
+        '| labialization  | labialized   | none            |      1     |\n'
+        '| Total distance |              |                 |      2     |\n'
+        '| Similarity     |              |                 |      0.915 |\n',
     )
 
   def test_compare_f_s_low_amplitude(self):
     self.assertEqual(
-        _P.f.comparison(_P.s_low_amplitude),
-        'f - s_low_amplitude phonology_descriptive comparison:\n'
-        '    ph_class: {consonant} vs ph_class: {consonant} = 0.00\n'
-        '    airstream: {pulmonic} vs airstream: {pulmonic} = 0.00\n'
-        '    manner: {non_sibilant} vs manner: {non_sibilant} = 0.00\n'
-        '    place: {dental} vs place: {alveolar, dental, postalveolar} = 0.00'
-        '\n'
-        '    articulator: {labial} vs articulator: {apical, laminal} = 0.50\n'
-        '    height: {not_applicable} vs height: {not_applicable} = 0.00\n'
-        '    backness: {not_applicable} vs backness: {not_applicable} = 0.00\n'
-        '    breathiness: {any} vs breathiness: {any} = 0.00\n'
-        '    voicing: {voiceless} vs voicing: {voiceless} = 0.00\n'
-        '    labialization: {labialized} vs labialization: {none} = 1.00\n'
-        '    lateralization: {none} vs lateralization: {none} = 0.00\n'
-        '    nasalization: {none} vs nasalization: {none} = 0.00\n'
-        '    palatalization: {none} vs palatalization: {none} = 0.00\n'
-        '    rhoticization: {none} vs rhoticization: {none} = 0.00\n'
-        '    duration: {any} vs duration: {any} = 0.00\n'
-        '    syllabicity: {none} vs syllabicity: {none} = 0.00\n'
-        '    Total distance = 1.50/23.50\n'
-        '    Similarity = 0.936\n',
+        _P.f.comparison_table(_P.s_low_amplitude),
+        'phonology_descriptive comparison (max distance = 23.50):\n\n'
+        '| aspect         | f          | s_low_amplitude   |   distance |\n'
+        '|----------------|------------|-------------------|------------|\n'
+        '| articulator    | labial     | apical, laminal   |      0.5   |\n'
+        '| labialization  | labialized | none              |      1     |\n'
+        '| Total distance |            |                   |      1.5   |\n'
+        '| Similarity     |            |                   |      0.936 |\n',
     )
 
 if __name__ == '__main__':

nisaba/scripts/natural_translit/utils/BUILD.bazel

@@ -13,6 +13,9 @@
 # limitations under the License.
 
 
+# Required for `tabulate` below.
+load("@pip_deps//:requirements.bzl", "requirement")
+
 package(
     default_applicable_licenses = [
     ],
@@ -157,6 +160,7 @@ py_library(
         ":inventory",
         ":log_op",
         ":type_op",
+        requirement("tabulate"),
     ],
 )
 

nisaba/scripts/natural_translit/utils/feature.py

@@ -101,6 +101,9 @@
 
 import enum
 from typing import Iterable, Union
+
+import tabulate
+
 from nisaba.scripts.natural_translit.utils import inventory
 from nisaba.scripts.natural_translit.utils import type_op as ty
 
@@ -205,12 +208,13 @@ def __init__(self, *features, alias: str = ''):
       self.add(*features)
 
     def __str__(self):
-      items = [item.text for item in self._items]
-      items.sort()
       return '%s: {%s}' % (
-          self.alias, ', '.join(items)
+          self.alias, ', '.join(self.sorted_item_texts())
       )
 
+    def sorted_item_texts(self) -> list[str]:
+      return sorted(item.text for item in self._items)
+
     def _item_set(self) -> set['Feature']:
       return ty.type_check(self._items, set(self._items))
 
@@ -500,11 +504,17 @@ def __init__(
       self.distance_dict = {}
 
     def __str__(self):
-      """String representation of distance dicts of features for this aspect."""
-      text = '%s (%.2f):\n' % (self.text, self.max_dist)
-      for feature in self:
-        text += str(feature)
-      return text
+      """Distance matrix of this aspect as table in github format."""
+      headers = ['distances'] + [feature.text for feature in self]
+      table = [
+          [feature1.text] + [feature1.distance(feature2) for feature2 in self]
+          for feature1 in self
+      ]
+      return (
+          'aspect: %s, max_dist: %.2f\n\n' % (self.text, self.max_dist)
+          + tabulate.tabulate(table, headers, tablefmt='github')
+          + '\n'
+      )
 
     def add_feature(self, feature: 'Feature') -> None:
       self.add_item(feature)
@@ -723,50 +733,74 @@ def copy_and_update(
       return new
 
     def compare(
-        self, p: 'Feature.Profile', aspects: 'Feature.ASPECTS' = ty.UNSPECIFIED,
-    ) -> tuple[str, float]:
+        self,
+        p: 'Feature.Profile',
+        aspects: 'Feature.ASPECTS' = ty.UNSPECIFIED,
+        verbose: bool = False,
+    ) -> dict[str, Union[str, float]]:
       """Compares this Profile to another Profile.
 
       Args:
         p: Profile to be compared.
         aspects: contrastive aspects to be included while calculating the
         distance and similarity.
+        verbose: If false, aspects with zero distance are not included in the
+          output table.
 
       Returns:
-        Similarity score
+        Comparison table as string, similarity score as float
 
       """
-      text = '%s - %s %s comparison:\n' % (
-          self.alias, p.alias, self.inventory.alias
-      )
       if p.inventory != self.inventory:
-        return '    not comparable\n    Similarity = 0\n', 0
+        return {
+            'text':
+            '%s and %s profiles are not comparable\n    Similarity = 0\n'
+            % (self.inventory.alias, p.inventory.alias),
+            'similarity': 0,
+        }
       if isinstance(aspects, ty.Nothing): aspects = self.inventory
       total_dist = 0
       max_dist = 0
+      headers = ['aspect', self.text, p.text, 'distance']
+      table = []
       for aspect in aspects:
         item1 = self.get(aspect.alias)
         item2 = p.get(aspect.alias)
         dist = item1.distance(item2)
         total_dist += dist
         max_dist += aspect.max_dist
-        text += '    %s vs %s = %.2f\n' % (
-            str(item1), str(item2), dist
-        )
+        if verbose or dist:
+          table.append([
+              aspect.alias,
+              ', '.join(item1.sorted_item_texts()),
+              ', '.join(item2.sorted_item_texts()),
+              '%.2f' % dist,
+          ])
       similarity = 1 - total_dist / max_dist
-      text += '    Total distance = %.2f/%.2f\n' % (total_dist, max_dist)
-      text += '    Similarity = %.3f\n' % similarity
-      return text, similarity
+      table.extend([
+          ['Total distance', '', '', '%.2f' % total_dist],
+          ['Similarity', '', '', '%.3f' % similarity],
+      ])
+      text = (
+          self.inventory.alias
+          + ' comparison (max distance = %.2f):\n\n' % max_dist
+          + tabulate.tabulate(table, headers, tablefmt='github')
+          + '\n'
+      )
+      return {'text': text, 'similarity': similarity}
 
-    def comparison(
-        self, p: 'Feature.Profile', aspects: 'Feature.ASPECTS' = ty.UNSPECIFIED,
+    def comparison_table(
+        self,
+        p: 'Feature.Profile',
+        aspects: 'Feature.ASPECTS' = ty.UNSPECIFIED,
+        verbose: bool = False,
     ) -> str:
-      return self.compare(p, aspects)[0]
+      return self.compare(p, aspects, verbose)['text']
 
     def similarity(
         self, p: 'Feature.Profile', aspects: 'Feature.ASPECTS' = ty.UNSPECIFIED,
     ) -> float:
-      return self.compare(p, aspects)[1]
+      return self.compare(p, aspects)['similarity']
 
     def has_feature(self, value: 'Feature.Aspect.VALUES') -> bool:
       """Checks if the given value or one of its children is in this profile."""