Merge pull request #357 from roedoejet/dev.misc-fixes

Improve coverage and type checking slightly

g2p/__init__.py

@@ -65,8 +65,8 @@ def make_g2p(  # noqa: C901
         NoPath: if there is path between in_lang and out_lang
     """
     # Defer expensive imports
-    from networkx import shortest_path
-    from networkx.exception import NetworkXNoPath
+    from networkx import shortest_path  # type: ignore
+    from networkx.exception import NetworkXNoPath  # type: ignore
 
     from g2p.log import LOGGER
     from g2p.mappings import Mapping

g2p/api.py

@@ -6,9 +6,9 @@
 
 from flask import Blueprint, abort
 from flask_cors import CORS
-from flask_restful import Api, Resource, inputs, reqparse
-from networkx.algorithms.dag import ancestors, descendants
-from networkx.exception import NetworkXError
+from flask_restful import Api, Resource, inputs, reqparse  # type: ignore
+from networkx.algorithms.dag import ancestors, descendants  # type: ignore
+from networkx.exception import NetworkXError  # type: ignore
 
 from g2p import make_g2p
 from g2p.exceptions import InvalidLanguageCode, NoPath

g2p/app.py

@@ -8,8 +8,8 @@
 
 from flask import Flask, render_template
 from flask_cors import CORS
-from flask_socketio import SocketIO, emit
-from networkx import shortest_path
+from flask_socketio import SocketIO, emit  # type: ignore
+from networkx import shortest_path  # type: ignore
 
 from g2p import make_g2p
 from g2p.api import g2p_api

g2p/cli.py

@@ -493,7 +493,7 @@ def convert(  # noqa: C901
     fra->fra-ipa, fra-ipa->eng-ipa and eng-ipa->eng-arpabet.
     """
     # Defer expensive imports
-    from networkx import has_path
+    from networkx import has_path  # type: ignore
 
     from g2p.log import LOGGER
     from g2p.mappings import MAPPINGS_AVAILABLE, Mapping, MappingConfig

g2p/log.py

@@ -5,7 +5,7 @@
 
 import logging
 
-import coloredlogs
+import coloredlogs  # type: ignore
 
 FIELD_STYLES = dict(
     levelname=dict(color="green"),

g2p/mappings/create_fallback_mapping.py

@@ -1,6 +1,6 @@
 import datetime as dt
 
-from text_unidecode import unidecode
+from text_unidecode import unidecode  # type: ignore
 
 from g2p import make_g2p
 from g2p.log import LOGGER

g2p/mappings/create_ipa_mapping.py

@@ -46,7 +46,7 @@ def process_character(p, is_xsampa=False):
         global _xsampa_converter
         if _xsampa_converter is None:
             # Expensive import, do it only when needed:
-            from panphon.xsampa import XSampa
+            from panphon.xsampa import XSampa  # type: ignore
 
             _xsampa_converter = XSampa()
         p = _xsampa_converter.convert(p)

g2p/mappings/langs/__init__.py

@@ -6,7 +6,7 @@
 import json
 import os
 
-import networkx
+import networkx  # type: ignore
 
 from g2p.constants import LANGS_DIR, LANGS_FILE_NAME, NETWORK_FILE_NAME
 from g2p.log import LOGGER

g2p/mappings/langs/utils.py

@@ -9,8 +9,8 @@
 from pathlib import Path
 from typing import Any, Optional
 
-import networkx
-from networkx.algorithms.dag import ancestors, descendants
+import networkx  # type: ignore
+from networkx.algorithms.dag import ancestors, descendants  # type: ignore
 
 from g2p.log import LOGGER
 from g2p.mappings import MAPPINGS_AVAILABLE, Mapping, MappingConfig
@@ -32,7 +32,8 @@
 def getPanphonDistanceSingleton():
     global _PANPHON_DISTANCE_SINGLETON
     if _PANPHON_DISTANCE_SINGLETON is None:
-        import panphon.distance  # Expensive import, only do it when actually needed
+        # Expensive import, only do it when actually needed
+        import panphon.distance  # type: ignore
 
         _PANPHON_DISTANCE_SINGLETON = panphon.distance.Distance()
     return _PANPHON_DISTANCE_SINGLETON

g2p/mappings/tokenizer.py

@@ -9,7 +9,7 @@
 import re
 from typing import List
 
-from networkx.exception import NetworkXError
+from networkx.exception import NetworkXError  # type: ignore
 
 from g2p.exceptions import MappingMissing
 from g2p.log import LOGGER

g2p/tests/test_create_mapping.py

@@ -33,12 +33,31 @@ def setUp(self):
             {"in": "ɡ", "out": "G"},
             {"in": "ʒ", "out": "ZH"},
         ]
+        self.mappings_xsampa = [
+            {"in": "A", "out": "AA"},
+            {"in": "e:", "out": "EY"},
+            {"in": "i", "out": "IY"},
+            {"in": "u", "out": "UW"},
+            {"in": "tS", "out": "CH"},
+            {"in": "p", "out": "P"},
+            {"in": "t", "out": "T"},
+            {"in": "k", "out": "K"},
+            {"in": "w", "out": "W"},
+            {"in": "g", "out": "G"},
+            {"in": "Z", "out": "ZH"},
+        ]
         self.target_mapping = Mapping(
             rules=self.mappings,
             in_lang="eng-ipa",
             out_lang="eng-arpabet",
             out_delimiter=" ",
         )
+        self.target_mapping_xsampa = Mapping(
+            rules=self.mappings_xsampa,
+            in_lang="eng-xsampa",
+            out_lang="eng-arpabet",
+            out_delimiter=" ",
+        )
 
     def test_unigram_mappings(self):
         src_mappings = [
@@ -79,6 +98,19 @@ def test_trigram_mappings(self):
         self.assertEqual(transducer("t͡ʃu").output_string, "tʃu")
         self.assertEqual(transducer("t͡ʃa").output_string, "tʃɑ")
 
+    def test_trigram_mappings_xsampa(self):
+        src_mappings = [
+            {"in": "ᒋ", "out": "tSi"},
+            {"in": "ᒍ", "out": "tSu"},
+            {"in": "ᒐ", "out": "tSa"},
+        ]
+        src_mapping = Mapping(rules=src_mappings, in_lang="crj", out_lang="crj-xsampa")
+        mapping = create_mapping(src_mapping, self.target_mapping_xsampa, quiet=True)
+        transducer = Transducer(mapping)
+        self.assertEqual(transducer("tSi").output_string, "tSi")
+        self.assertEqual(transducer("tSu").output_string, "tSu")
+        self.assertEqual(transducer("tSa").output_string, "tSA")
+
     def test_long_mappings(self):
         src_mappings = [
             {"in": "ᐧᐯ", "out": "pʷeː"},

g2p/tests/time_panphon.py

@@ -15,8 +15,8 @@
 
 """
 
-import panphon.distance
-from linetimer import CodeTimer
+import panphon.distance  # type: ignore
+from linetimer import CodeTimer  # type: ignore
 
 from g2p.mappings import Mapping
 from g2p.mappings.langs.utils import is_panphon

g2p/transducer/__init__.py

@@ -9,7 +9,7 @@
 from collections import OrderedDict, defaultdict
 from typing import DefaultDict, Dict, List, Optional, Set, Tuple, Union
 
-import text_unidecode
+import text_unidecode  # type: ignore
 
 from g2p.log import LOGGER
 from g2p.mappings import MAPPING_TYPE, Mapping