Add MT tutorial

.vscode/settings.json

@@ -1,24 +1,15 @@
 {
   "editor.formatOnSave": true,
   "editor.codeActionsOnSave": {
-    "source.organizeImports": "explicit",
+    "source.organizeImports": "explicit"
   },
   "python.testing.unittestEnabled": false,
   "python.testing.pytestEnabled": true,
-  "python.analysis.extraPaths": [
-    "tests"
-  ],
+  "python.analysis.extraPaths": ["tests"],
   "python.analysis.importFormat": "relative",
   "[python]": {
     "editor.defaultFormatter": "ms-python.black-formatter",
     "editor.formatOnSave": true
   },
-  "black-formatter.path": [
-    "poetry",
-    "run",
-    "black"
-  ],
-  "python.analysis.extraPaths": [
-    "./tests"
-  ]
-}
+  "black-formatter.path": ["poetry", "run", "black"]
+}

machine/translation/translation_suggester.py

@@ -1,8 +1,10 @@
 from abc import ABC, abstractmethod
-from typing import Iterable, Sequence
+from typing import Iterable, Optional, Sequence
 
+from .interactive_translator import InteractiveTranslator
 from .translation_result import TranslationResult
 from .translation_suggestion import TranslationSuggestion
+from .truecaser import Truecaser
 
 
 class TranslationSuggester(ABC):
@@ -14,3 +16,18 @@ def __init__(self, confidence_threshold: float = 0, break_on_punctuation: bool =
     def get_suggestions(
         self, n: int, prefix_count: int, is_last_word_complete: bool, results: Iterable[TranslationResult]
     ) -> Sequence[TranslationSuggestion]: ...
+
+    def get_suggestions_from_translator(
+        self, n: int, translator: InteractiveTranslator, truecaser: Optional[Truecaser] = None
+    ) -> Sequence[TranslationSuggestion]:
+        results = translator.get_current_results()
+        if truecaser is not None:
+            results = (
+                truecaser.truecase_translation_result(result, translator.target_detokenizer) for result in results
+            )
+        return self.get_suggestions(
+            n,
+            len(translator.prefix_word_ranges),
+            translator.is_last_word_complete,
+            results
+        )

machine/translation/truecaser.py

@@ -1,8 +1,11 @@
 from abc import ABC, abstractmethod
-from typing import Sequence
+from typing import Optional, Sequence
 
 from ..corpora.text_corpus import TextCorpus
+from ..tokenization.detokenizer import Detokenizer
+from ..tokenization.whitespace_detokenizer import WHITESPACE_DETOKENIZER
 from .trainer import Trainer
+from .translation_result import TranslationResult
 
 
 class Truecaser(ABC):
@@ -15,5 +18,21 @@ def train_segment(self, segment: Sequence[str], sentence_start: bool = True) ->
     @abstractmethod
     def truecase(self, segment: Sequence[str]) -> Sequence[str]: ...
 
+    def truecase_translation_result(
+        self, result: TranslationResult, detokenizer: Optional[Detokenizer] = None
+    ) -> TranslationResult:
+        if detokenizer is None:
+            detokenizer = WHITESPACE_DETOKENIZER
+        target_tokens = self.truecase(result.target_tokens)
+        return TranslationResult(
+            detokenizer.detokenize(target_tokens),
+            result.source_tokens,
+            target_tokens,
+            result.confidences,
+            result.sources,
+            result.alignment,
+            result.phrases,
+        )
+
     @abstractmethod
     def save(self) -> None: ...