Interactive predictions from the command-line

README.md

@@ -176,6 +176,11 @@ paragraph2actions-translate -s src-test.txt -o pred-organic.txt    -p sp_model.m
 paragraph2actions-translate -s src-test.txt -o pred-pretrained.txt -p sp_model.model -t pretrained.pt
 ```
 
+It is also possible to do interactive predictions from the command-line with the following:
+```bash
+paragraph2actions-translate -s sp_model.model -m sac.pt
+```
+
 ## Metrics
 
 To compute the metrics and produce a CSV comparing the models, execute the following:

setup.cfg

@@ -65,6 +65,7 @@ console_scripts =
     sac-augment-annotations = sac_action_extraction.create_annotation_splits:main
     sac-create-annotation-splits = sac_action_extraction.create_annotation_splits:main
     sac-concatenate-annotations = sac_action_extraction.concatenate_annotation_splits:main
+    sac-interactive-extraction = sac_action_extraction.interactive_predictions:main
     sac-metrics-grid = sac_action_extraction.metrics_grid:main
 
 [flake8]

src/sac_action_extraction/interactive_predictions.py

@@ -0,0 +1,65 @@
+import sys
+from pathlib import Path
+from typing import TextIO, Tuple
+
+import click
+from paragraph2actions.default_converters import default_action_converters
+from paragraph2actions.paragraph_translator import ParagraphTranslator
+from paragraph2actions.readable_converter import ReadableConverter
+
+from .sac_converters import default_sac_converters
+
+
+@click.command()
+@click.option(
+    "--models",
+    "-m",
+    multiple=True,
+    type=click.Path(exists=True, dir_okay=False, path_type=Path),
+    help="Translation model file(s). If multiple are given, will be an ensemble model.",
+)
+@click.option(
+    "--sentencepiece_model",
+    "-s",
+    required=True,
+    type=click.Path(exists=True, dir_okay=False, path_type=Path),
+    help="SentencePiece model file (typically ``sp_model.model``)",
+)
+@click.argument("input_file", type=click.File(mode="r"), default=sys.stdin)
+@click.argument("output_file", type=click.File(mode="w"), default=sys.stdout)
+def main(
+    models: Tuple[Path, ...],
+    sentencepiece_model: Path,
+    input_file: TextIO,
+    output_file: TextIO,
+) -> None:
+    """Interactive extraction of SAC synthesis actions.
+
+    Input and output can be either stdin/stdout, or input file and output file.
+    """
+
+    single_action_converters = default_sac_converters() + default_action_converters()
+    converter = ReadableConverter(single_action_converters=single_action_converters)
+
+    paragraph_translator = ParagraphTranslator(
+        translation_model=[str(m) for m in models],
+        sentencepiece_model=str(sentencepiece_model),
+        action_string_converter=converter,
+    )
+
+    for line in input_file:
+        line = line.strip()
+
+        # Canonicalize the SMILES, handle exception if needed
+        try:
+            result = paragraph_translator.extract_paragraph(line)
+            output_file.write("Extracted actions as Python objects:\n")
+            output_file.write(f"{result.actions}\n")
+            output_file.write("Extracted actions in readable format:\n")
+            output_file.write(f"{converter.actions_to_string(result.actions)}\n\n")
+        except Exception as e:
+            output_file.write(f"\nError: {e}\n")
+
+
+if __name__ == "__main__":
+    main()