rasa-pro 3.9.18__py3-none-any.whl

rasa/studio/train.py

@@ -0,0 +1,135 @@
+import argparse
+import logging
+import sys
+from pathlib import Path
+from typing import Any, List, Optional
+
+import rasa.cli.utils
+import rasa.shared.utils.cli
+from rasa.cli.train import (
+    _model_for_finetuning,
+    extract_core_additional_arguments,
+    extract_nlu_additional_arguments,
+)
+from rasa.shared.constants import (
+    CONFIG_MANDATORY_KEYS,
+    DEFAULT_DATA_PATH,
+    DEFAULT_DOMAIN_PATH,
+)
+from rasa.shared.core.flows.yaml_flows_io import YamlFlowsWriter
+from rasa.shared.importers.importer import TrainingDataImporter
+from rasa.shared.utils.yaml import read_yaml, write_yaml
+from rasa.studio import data_handler
+from rasa.utils.common import get_temp_dir_name
+
+from rasa.studio.config import StudioConfig
+from rasa.studio.data_handler import (
+    StudioDataHandler,
+    import_data_from_studio,
+)
+
+logger = logging.getLogger(__name__)
+
+
+def handle_train(args: argparse.Namespace) -> Optional[str]:
+    from rasa import train as train_all
+
+    handler = StudioDataHandler(
+        studio_config=StudioConfig.read_config(), assistant_name=args.assistant_name[0]
+    )
+    if args.entities or args.intents:
+        handler.request_data(args.intents, args.entities)
+    else:
+        handler.request_all_data()
+
+    domain = rasa.cli.utils.get_validated_path(
+        args.domain, "domain", DEFAULT_DOMAIN_PATH, none_is_valid=True
+    )
+    config = rasa.cli.utils.get_validated_config(args.config, CONFIG_MANDATORY_KEYS)
+    data_form_studio, data_original = import_data_from_studio(
+        handler, domain, args.data
+    )
+
+    domain = data_original.get_domain().merge(data_form_studio.get_domain())  # type: ignore[assignment]
+
+    domain_file = _create_temp_file(read_yaml(domain.as_yaml()), "domain.yml")  # type: ignore[union-attr]
+
+    studio_training_files = make_training_files(
+        handler, data_form_studio, data_original
+    )
+
+    training_files = [
+        rasa.cli.utils.get_validated_path(
+            f, "data", DEFAULT_DATA_PATH, none_is_valid=True
+        )
+        for f in args.data
+    ]
+    training_files.extend(studio_training_files)
+
+    training_result = train_all(
+        domain=str(domain_file),
+        config=config,
+        training_files=args.data,
+        output=args.out,
+        dry_run=args.dry_run,
+        force_training=args.force,
+        fixed_model_name=args.fixed_model_name,
+        persist_nlu_training_data=args.persist_nlu_data,
+        core_additional_arguments={
+            **extract_core_additional_arguments(args),
+        },
+        nlu_additional_arguments=extract_nlu_additional_arguments(args),
+        model_to_finetune=_model_for_finetuning(args),
+        finetuning_epoch_fraction=args.epoch_fraction,
+    )
+
+    if training_result.code != 0:
+        sys.exit(training_result.code)
+
+    return training_result.model
+
+
+def make_training_files(
+    handler: StudioDataHandler,
+    data_form_studio: TrainingDataImporter,
+    data_original: TrainingDataImporter,
+) -> List[Path]:
+    """Create training file from studio data and original data.
+
+    Args:
+        handler (StudioDataHandler): data handler with studio config
+        data_form_studio (TrainingDataImporter): studio data
+        data_original (TrainingDataImporter): original data
+
+    Returns:
+        List[Path]: list of training files
+    """
+    training_files = []
+    if handler.has_nlu():
+        # nlu has deduplication
+        train_data = (
+            data_original.get_nlu_data()
+            .merge(data_form_studio.get_nlu_data())
+            .nlu_as_yaml()
+        )
+        training_file = _create_temp_file(read_yaml(train_data), "nlu.yml")
+        training_files.append(training_file)
+
+    if handler.has_flows():
+        diff_flows = data_handler.create_new_flows_from_diff(
+            data_form_studio.get_flows().underlying_flows,
+            data_original.get_flows().underlying_flows,
+        )
+        tmp_dir = get_temp_dir_name()
+        training_file = Path(tmp_dir, "flows.yml")
+        YamlFlowsWriter.dump(diff_flows, training_file)
+        training_files.append(training_file)
+
+    return training_files
+
+
+def _create_temp_file(data: Any, name: str) -> Path:
+    tmp_dir = get_temp_dir_name()
+    file = Path(tmp_dir, name)
+    write_yaml(data, file)
+    return file

rasa/studio/upload.py

@@ -0,0 +1,433 @@
+import argparse
+import base64
+import sys
+from typing import Dict, Iterable, List, Set, Text, Tuple, Union, Any
+
+import questionary
+import requests
+import structlog
+
+import rasa.cli.telemetry
+import rasa.cli.utils
+import rasa.shared.utils.cli
+import rasa.shared.utils.io
+from rasa.shared.constants import (
+    DEFAULT_DOMAIN_PATHS,
+    DEFAULT_CONFIG_PATH,
+)
+from rasa.shared.core.flows.yaml_flows_io import YamlFlowsWriter
+from rasa.shared.exceptions import RasaException
+from rasa.shared.importers.importer import TrainingDataImporter, FlowSyncImporter
+from rasa.shared.nlu.training_data.formats.rasa_yaml import RasaYAMLWriter
+from rasa.shared.utils.yaml import dump_obj_as_yaml_to_string, read_yaml_file
+from rasa.studio import results_logger
+from rasa.studio.auth import KeycloakTokenReader
+from rasa.studio.config import StudioConfig
+from rasa.studio.results_logger import StudioResult, with_studio_error_handler
+
+structlogger = structlog.get_logger()
+
+
+def _get_selected_entities_and_intents(
+    args: argparse.Namespace,
+    intents_from_files: Set[Text],
+    entities_from_files: List[Text],
+) -> Tuple[List[Text], List[Text]]:
+    entities = args.entities
+
+    if entities is None or len(entities) == 0:
+        entities = entities_from_files
+        structlogger.info(
+            "rasa.studio.upload.entities_empty",
+            event_info="No entities specified. Using all entities from files.",
+        )
+
+    intents = args.intents
+
+    if intents is None or len(intents) == 0:
+        intents = intents_from_files
+        structlogger.info(
+            "rasa.studio.upload.intents_empty",
+            event_info="No intents specified. Using all intents from files.",
+        )
+
+    return list(entities), list(intents)
+
+
+def handle_upload(args: argparse.Namespace) -> None:
+    """Uploads primitives to rasa studio."""
+    endpoint = StudioConfig.read_config().studio_url
+    if not endpoint:
+        rasa.shared.utils.cli.print_error_and_exit(
+            "No GraphQL endpoint found in config. Please run `rasa studio config`."
+        )
+    else:
+        structlogger.info(
+            "rasa.studio.upload.loading_data", event_info="Loading data..."
+        )
+
+        args.domain = rasa.cli.utils.get_validated_path(
+            args.domain, "domain", DEFAULT_DOMAIN_PATHS
+        )
+
+        args.config = rasa.cli.utils.get_validated_path(
+            args.config, "config", DEFAULT_CONFIG_PATH
+        )
+
+        # check safely if args.calm is set and not fail if not
+        if hasattr(args, "calm") and args.calm:
+            upload_calm_assistant(args, endpoint)
+        else:
+            upload_nlu_assistant(args, endpoint)
+
+
+config_keys = [
+    "recipe",
+    "language",
+    "pipeline",
+    "llm",
+    "policies",
+    "model_name",
+    "assistant_id",
+]
+
+
+def extract_values(data: Dict, keys: List[Text]) -> Dict:
+    """Extracts values for given keys from a dictionary."""
+    return {key: data.get(key) for key in keys if data.get(key)}
+
+
+def _get_assistant_name(config: Dict[Text, Any]) -> str:
+    config_assistant_id = config.get("assistant_id", "")
+    assistant_name = questionary.text(
+        "Please provide the assistant name", default=config_assistant_id
+    ).ask()
+    if not assistant_name:
+        structlogger.error(
+            "rasa.studio.upload.assistant_name_empty",
+            event_info="Assistant name cannot be empty.",
+        )
+        sys.exit(1)
+
+    # if assistant_name exists and different from config assistant_id,
+    # notify user and upload with new assistant_name
+    same = assistant_name == config_assistant_id
+    if not same and config_assistant_id != "":
+        structlogger.info(
+            "rasa.studio.upload.assistant_name_mismatch",
+            event_info=(
+                f"Assistant name '{assistant_name}' is different"
+                f" from the one in the config file: '{config_assistant_id}'."
+            ),
+        )
+
+    structlogger.info(f"Uploading assistant with the name '{assistant_name}'.")
+    return assistant_name
+
+
+@with_studio_error_handler
+def upload_calm_assistant(args: argparse.Namespace, endpoint: str) -> StudioResult:
+    """Uploads the CALM assistant data to Rasa Studio.
+
+    Args:
+        args: The command line arguments
+            - data: The path to the training data
+            - domain: The path to the domain
+            - flows: The path to the flows
+            - endpoints: The path to the endpoints
+            - config: The path to the config
+        endpoint: The studio endpoint
+    Returns:
+        None
+    """
+    structlogger.info(
+        "rasa.studio.upload.loading_data", event_info="Parsing CALM assistant data..."
+    )
+
+    importer = TrainingDataImporter.load_from_dict(
+        domain_path=args.domain,
+        config_path=args.config,
+    )
+
+    # Prepare config and domain
+    config = importer.get_config()
+    domain_from_files = importer.get_domain().as_dict()
+    endpoints_from_files = read_yaml_file(args.endpoints)
+    config_from_files = read_yaml_file(args.config)
+
+    # Extract domain and config values
+    domain_keys = [
+        "version",
+        "actions",
+        "responses",
+        "slots",
+        "intents",
+        "entities",
+        "forms",
+        "session_config",
+    ]
+
+    domain = extract_values(domain_from_files, domain_keys)
+
+    assistant_name = _get_assistant_name(config)
+
+    training_data_paths = args.data
+
+    if isinstance(training_data_paths, list):
+        training_data_paths.append(args.flows)
+    elif isinstance(training_data_paths, str):
+        if isinstance(args.flows, list):
+            training_data_paths = [training_data_paths] + args.flows
+        elif isinstance(args.flows, str):
+            training_data_paths = [training_data_paths, args.flows]
+        else:
+            raise RasaException("Invalid flows path")
+
+    # Prepare flows
+    flow_importer = FlowSyncImporter.load_from_dict(
+        training_data_paths=training_data_paths
+    )
+
+    user_flows = flow_importer.get_flows().user_flows
+    flows = list(user_flows)
+
+    # We instantiate the TrainingDataImporter again on purpose to avoid
+    # adding patterns to domain's actions. More info https://t.ly/W8uuc
+    nlu_importer = TrainingDataImporter.load_from_dict(
+        domain_path=args.domain, training_data_paths=args.data
+    )
+    nlu_data = nlu_importer.get_nlu_data()
+
+    intents_from_files = nlu_data.intents
+
+    nlu_examples = nlu_data.filter_training_examples(
+        lambda ex: ex.get("intent") in intents_from_files
+    )
+
+    nlu_examples_yaml = RasaYAMLWriter().dumps(nlu_examples)
+
+    # Build GraphQL request
+    graphql_req = build_import_request(
+        assistant_name,
+        flows_yaml=YamlFlowsWriter().dumps(flows),
+        domain_yaml=dump_obj_as_yaml_to_string(domain),
+        config_yaml=dump_obj_as_yaml_to_string(config_from_files),
+        endpoints=dump_obj_as_yaml_to_string(endpoints_from_files),
+        nlu_yaml=nlu_examples_yaml,
+    )
+
+    structlogger.info("Uploading to Rasa Studio...")
+    return make_request(endpoint, graphql_req)
+
+
+@with_studio_error_handler
+def upload_nlu_assistant(args: argparse.Namespace, endpoint: str) -> StudioResult:
+    """Uploads the classic (dm1) assistant data to Rasa Studio.
+
+    Args:
+        args: The command line arguments
+            - data: The path to the training data
+            - domain: The path to the domain
+            - intents: The intents to upload
+            - entities: The entities to upload
+        endpoint: The studio endpoint
+    Returns:
+        None
+    """
+    structlogger.info("Found DM1 assistant data, parsing...")
+    importer = TrainingDataImporter.load_from_dict(
+        domain_path=args.domain, training_data_paths=args.data, config_path=args.config
+    )
+
+    intents_from_files = importer.get_nlu_data().intents
+    entities_from_files = importer.get_domain().entities
+
+    entities, intents = _get_selected_entities_and_intents(
+        args, intents_from_files, entities_from_files
+    )
+
+    config_from_files = importer.get_config()
+    config = extract_values(config_from_files, config_keys)
+
+    assistant_name = _get_assistant_name(config)
+
+    structlogger.info("Validating data...")
+    _check_for_missing_primitives(
+        intents, entities, intents_from_files, entities_from_files
+    )
+
+    nlu_examples = importer.get_nlu_data().filter_training_examples(
+        lambda ex: ex.get("intent") in intents
+    )
+
+    all_entities = _add_missing_entities(nlu_examples.entities, entities)
+    nlu_examples_yaml = RasaYAMLWriter().dumps(nlu_examples)
+
+    domain = _filter_domain(all_entities, intents, importer.get_domain().as_dict())
+    domain_yaml = dump_obj_as_yaml_to_string(domain)
+
+    graphql_req = build_request(assistant_name, nlu_examples_yaml, domain_yaml)
+
+    structlogger.info("Uploading to Rasa Studio...")
+    return make_request(endpoint, graphql_req)
+
+
+def make_request(endpoint: str, graphql_req: Dict) -> StudioResult:
+    """Makes a request to the studio endpoint to upload data.
+
+    Args:
+        endpoint: The studio endpoint
+        graphql_req: The graphql request
+    """
+    token = KeycloakTokenReader().get_token()
+    res = requests.post(
+        endpoint,
+        json=graphql_req,
+        headers={
+            "Authorization": f"{token.token_type} {token.access_token}",
+            "Content-Type": "application/json",
+        },
+    )
+
+    if results_logger.response_has_errors(res.json()):
+        return StudioResult.error(res.json())
+    return StudioResult.success("Upload successful")
+
+
+def _add_missing_entities(
+    entities_from_intents: Iterable[str], entities: List[str]
+) -> List[Union[str, Dict]]:
+    all_entities: List[Union[str, Dict]] = []
+    all_entities.extend(entities)
+    for entity in entities_from_intents:
+        if entity not in entities:
+            structlogger.warning(
+                f"Adding entity '{entity}' to upload since it is used in an intent."
+            )
+            all_entities.append(entity)
+    return all_entities
+
+
+def build_import_request(
+    assistant_name: str,
+    flows_yaml: str,
+    domain_yaml: str,
+    config_yaml: str,
+    endpoints: str,
+    nlu_yaml: str = "",
+) -> Dict:
+    # b64encode expects bytes and returns bytes so we need to decode to string
+    base64_domain = base64.b64encode(domain_yaml.encode("utf-8")).decode("utf-8")
+    base64_flows = base64.b64encode(flows_yaml.encode("utf-8")).decode("utf-8")
+    base64_config = base64.b64encode(config_yaml.encode("utf-8")).decode("utf-8")
+    base64_nlu = base64.b64encode(nlu_yaml.encode("utf-8")).decode("utf-8")
+    base64_endpoints = base64.b64encode(endpoints.encode("utf-8")).decode("utf-8")
+
+    graphql_req = {
+        "query": (
+            "mutation UploadModernAssistant($input: UploadModernAssistantInput!)"
+            "{\n  uploadModernAssistant(input: $input)\n}"
+        ),
+        "variables": {
+            "input": {
+                "assistantName": assistant_name,
+                "domain": base64_domain,
+                "flows": base64_flows,
+                "nlu": base64_nlu,
+                "config": base64_config,
+                "endpoints": base64_endpoints,
+            }
+        },
+    }
+
+    return graphql_req
+
+
+def build_request(
+    assistant_name: str, nlu_examples_yaml: str, domain_yaml: str
+) -> Dict:
+    # b64encode expects bytes and returns bytes so we need to decode to string
+    base64_domain = base64.b64encode(domain_yaml.encode("utf-8")).decode("utf-8")
+    base64_nlu_examples = base64.b64encode(nlu_examples_yaml.encode("utf-8")).decode(
+        "utf-8"
+    )
+
+    graphql_req = {
+        "query": (
+            "mutation ImportFromEncodedYaml($input: ImportFromEncodedYamlInput!)"
+            "{\n  importFromEncodedYaml(input: $input)\n}"
+        ),
+        "variables": {
+            "input": {
+                "assistantName": assistant_name,
+                "domain": base64_domain,
+                "nlu": base64_nlu_examples,
+            }
+        },
+    }
+
+    return graphql_req
+
+
+def _filter_domain(
+    entities: List[Union[str, Dict]], intents: List[str], domain_from_files: Dict
+) -> Dict:
+    """Filters the domain to only include the selected entities and intents."""
+    selected_entities = _remove_not_selected_entities(
+        entities, domain_from_files.get("entities", [])
+    )
+    return {
+        "version": domain_from_files["version"],
+        "intents": intents,
+        "entities": selected_entities,
+    }
+
+
+def _check_for_missing_primitives(
+    intents: Iterable[str],
+    entities: Iterable[str],
+    intents_found: Iterable[str],
+    entities_found: Iterable[str],
+) -> None:
+    """Checks if the data contains all intents and entities.
+
+    Args:
+        intents: Iterable of intents to check
+        entities: Iterable of entities to check
+        intents_found: Iterable of intents found in the data
+        entities_found: Iterable of entities found in the data
+    """
+    not_found_entities = [entity for entity in entities if entity not in entities_found]
+    not_found_intents = [intent for intent in intents if intent not in intents_found]
+
+    not_found_msg = ""
+    if not_found_entities:
+        not_found_msg += (
+            f"Entities were not found in the training data: {not_found_entities}\n"
+        )
+    if not_found_intents:
+        not_found_msg += (
+            f"Intents were not found in the training data: {not_found_intents}\n"
+        )
+
+    if not_found_msg:
+        raise RasaException(not_found_msg)
+
+
+def _remove_not_selected_entities(
+    entities: List[Union[Text, Dict]], domain_entities: List[Union[Text, Dict]]
+) -> List:
+    to_remove: List[Union[Text, Dict]] = []
+
+    for entity in domain_entities:
+        if isinstance(entity, str) and entity not in entities:
+            to_remove.append(entity)
+        elif isinstance(entity, dict) and len(entity) == 1:
+            entity_name = next(iter(entity))
+            if entity_name not in entities:
+                to_remove.append(entity)
+
+    for entity in to_remove:
+        domain_entities.remove(entity)
+
+    return domain_entities