cleanup

.github/workflows/build.yml

@@ -83,7 +83,7 @@ jobs:
           echo $GITHUB_WORKSPACE
           df -h
           ls -al '/c/Program Files (x86)/Microsoft Visual Studio/2019/Enterprise/VC/Redist/MSVC/'
-          ls -al '/c/Program Files (x86)/Microsoft Visual Studio/2019/Enterprise/VC/Redist/MSVC/14.26.28720'
+          # ls -al '/c/Program Files (x86)/Microsoft Visual Studio/2019/Enterprise/VC/Redist/MSVC/14.26.28720'
           ls -al '/c/Program Files (x86)/Microsoft Visual Studio/2019/Enterprise/VC/Redist/MSVC/v142'
           # ls -al '/c/Program Files (x86)/Microsoft Visual Studio/2019/Enterprise/VC/Redist/MSVC/14.16.27012/x64/Microsoft.VC141.CRT'
           ls -al '/c/Program Files (x86)/Microsoft Visual Studio/2019/Enterprise/VC/Redist/MSVC/'*/x64/Microsoft.*.CRT
@@ -163,12 +163,12 @@ jobs:
         with:
           name: wheels
           path: main/dist/*
-      - name: Copy Windows vc_redist
-        run: |
-          mkdir -p vc_redist
-          cp '/c/Program Files (x86)/Microsoft Visual Studio/2019/Enterprise/VC/Redist/MSVC/14.26.28720'/vc_redist.x64.exe vc_redist/
-          cp '/c/Program Files (x86)/Microsoft Visual Studio/2019/Enterprise/VC/Redist/MSVC/14.26.28720'/x64/Microsoft.*.CRT/* vc_redist/
-      - uses: actions/upload-artifact@v2
-        with:
-          name: vc_redist
-          path: vc_redist/*
+      # - name: Copy Windows vc_redist
+      #   run: |
+      #     mkdir -p vc_redist
+      #     cp '/c/Program Files (x86)/Microsoft Visual Studio/2019/Enterprise/VC/Redist/MSVC/14.26.28720'/vc_redist.x64.exe vc_redist/
+      #     cp '/c/Program Files (x86)/Microsoft Visual Studio/2019/Enterprise/VC/Redist/MSVC/14.26.28720'/x64/Microsoft.*.CRT/* vc_redist/
+      # - uses: actions/upload-artifact@v2
+      #   with:
+      #     name: vc_redist
+      #     path: vc_redist/*

Justfile

@@ -32,3 +32,6 @@ setup-linux-develop kaldi_root_dir:
 	ln -sr {{kaldi_root_dir}}/src/fstbin/fstaddselfloops kaldi_active_grammar/exec/linux/
 	ln -sr {{kaldi_root_dir}}/src/dragonfly/libkaldi-dragonfly.so kaldi_active_grammar/exec/linux/
 	ln -sr {{kaldi_root_dir}}/src/dragonflybin/compile-graph-agf kaldi_active_grammar/exec/linux/
+
+test_model model_dir:
+	cd {{invocation_directory()}} && rm -rf kaldi_model kaldi_model.tmp && cp -rp {{model_dir}} kaldi_model

kaldi_active_grammar/compiler.py

@@ -4,15 +4,15 @@
 # Licensed under the AGPL-3.0; see LICENSE.txt file.
 #
 
-import base64, collections, copy, logging, multiprocessing, os, re, shlex, shutil, subprocess
+import collections, copy, logging, multiprocessing, os, re, shlex, shutil, subprocess
 import concurrent.futures
 from contextlib import contextmanager
 from io import open
 
 from six.moves import range, zip
 
 from . import _log, KaldiError
-from .utils import ExternalProcess, debug_timer, lazy_readonly_property, load_symbol_table, platform, show_donation_message, symbol_table_lookup, touch_file
+from .utils import ExternalProcess, debug_timer, load_symbol_table, platform, show_donation_message, symbol_table_lookup, touch_file
 from .wfst import WFST
 from .model import Model
 import kaldi_active_grammar.alternative_dictation as alternative_dictation

kaldi_active_grammar/model.py

@@ -7,7 +7,7 @@
 import os, re, shutil
 from io import open
 
-from six import text_type
+from six import PY2, text_type
 import requests
 
 try:
@@ -383,7 +383,7 @@ def reset_user_lexicon(self):
 
 def convert_generic_model_to_agf(src_dir, model_dir):
     from .compiler import Compiler
-    if six.PY2:
+    if PY2:
         from .kaldi import augment_phones_txt_py2 as augment_phones_txt, augment_words_txt_py2 as augment_words_txt
     else:
         from .kaldi import augment_phones_txt, augment_words_txt

kaldi_active_grammar/wfst.py

@@ -36,7 +36,7 @@ def clear(self):
     def iter_arcs(self):
         return itertools.chain.from_iterable(itervalues(self._arc_table_dict))
 
-    def state_is_final(self, state):
+    def is_state_final(self, state):
         return (self._state_table[state] != 0)
 
     def add_state(self, weight=None, initial=False, final=False):
@@ -84,12 +84,14 @@ def label_is_silent(self, label):
         return ((label in self.silent_labels) or (label.startswith('#nonterm')))
 
     def scale_weights(self, factor):
+        # Unused
         factor = float(factor)
         for arcs in itervalues(self._arc_table_dict):
             for arc in arcs:
                 arc[4] = arc[4] * factor
 
     def normalize_weights(self, stochasticity=False):
+        # Unused
         for arcs in itervalues(self._arc_table_dict):
             num_weights = len(arcs)
             sum_weights = sum(arc[4] for arc in arcs)
@@ -99,6 +101,7 @@ def normalize_weights(self, stochasticity=False):
 
     def has_eps_path(self, path_src_state, path_dst_state, eps_like_labels=frozenset()):
         """ Returns True iff there is a epsilon path from src_state to dst_state. Uses BFS. Does not follow nonterminals! """
+        # Used in: dragonfly backend compiler.
         eps_like_labels = frozenset((self.eps, self.eps_disambig)) | frozenset(eps_like_labels)
         state_queue = collections.deque([path_src_state])
         queued = set(state_queue)
@@ -115,12 +118,13 @@ def has_eps_path(self, path_src_state, path_dst_state, eps_like_labels=frozenset
 
     def does_match(self, target_words, wildcard_nonterms=(), include_silent=False):
         """ Returns the olabels on a matching path if there is one, False if not. Uses BFS. Wildcard accepts zero or more words. """
+        # Used in: KaldiAG compiler.
         queue = collections.deque()  # entries: (state, path of ilabels of arcs to state, index into target_words of remaining words)
         queue.append((self.start_state, (), 0))
         while queue:
             state, path, target_word_index = queue.popleft()
             target_word = target_words[target_word_index] if target_word_index < len(target_words) else None
-            if (target_word is None) and self.state_is_final(state):
+            if (target_word is None) and self.is_state_final(state):
                 return tuple(olabel for olabel in path
                     if include_silent or not self.label_is_silent(olabel))
             for arc in self._arc_table_dict[state]: