Merge branch 'main' into switch-to-mypy-part1

Dazu-io · Mar 11, 2021 · 768068b · 768068b
2 parents 6176569 + 4e61878
commit 768068b
Show file tree

Hide file tree

Showing 136 changed files with 2,934 additions and 2,583 deletions.
diff --git a/.github/workflows/automatic-pr-merger.yml b/.github/workflows/automatic-pr-merger.yml
diff --git a/.github/workflows/automatic-pr-update.yml b/.github/workflows/automatic-pr-update.yml
@@ -0,0 +1,19 @@
+name: Automatic PR Merger
+
+on:
+  push: {} # update PR when base branch is updated
+
+jobs:
+  # thats's all. single step is needed - if PR is mergeable according to
+  # branch protection rules it will be merged automatically
+  mergepal:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v2
+      - uses: rasahq/update-pr-branch@601ac3dadf708b14e7f0252d98230538235f99d9
+        if: github.repository == 'RasaHQ/rasa'
+        with:
+          token: ${{ secrets.RASABOT_AUTOMERGE_GITHUB_TOKEN }}
+          # required parameter by original action - 
+          # check is already done through protected branches so not needed for us
+          required_approval_count: 0
diff --git a/.github/workflows/continous-integration.yml b/.github/workflows/continous-integration.yml
@@ -192,9 +192,16 @@ jobs:
     needs: [changes]
     strategy:
       matrix:
-        os: [ubuntu-latest, windows-latest]
-        python-version: [3.6, 3.7, 3.8]
-        test: [test-non-training, test-training]
+        test:
+        - test-cli
+        - test-core-featurizers
+        - test-policies
+        - test-nlu-featurizers
+        - test-nlu-predictors
+        - test-full-model-training
+        - test-other-unit-tests
+        os: [ ubuntu-latest, windows-latest ]
+        python-version: [ 3.6, 3.7, 3.8 ]
 
     steps:
     - name: Checkout git repository 🕝
@@ -285,6 +292,7 @@ jobs:
       env:
         COVERALLS_REPO_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         COVERALLS_SERVICE_NAME: github
+        COVERALLS_PARALLEL: true
       run: poetry run coveralls
 
   integration_test:
@@ -735,20 +743,3 @@ jobs:
         pip install -U github3.py pep440-version-utils
         python3 scripts/publish_gh_release_notes.py
         ./scripts/ping_slack_about_package_release.sh
-
-  mergepal-merge: # run merge pal in the end
-    runs-on: ubuntu-latest
-    needs: [quality, test, documentation, docker]
-    steps:
-      - uses: actions/checkout@v1
-      - name: Make sure there is a github token
-        shell: bash
-        run: |
-          if [ -z "${{ secrets.RASABOT_AUTOMERGE_GITHUB_TOKEN }}" ]; then
-            echo "MERGE_TOKEN=${{ secrets.GITHUB_TOKEN }}" >> $GITHUB_ENV
-          else
-            echo "MERGE_TOKEN=${{ secrets.RASABOT_AUTOMERGE_GITHUB_TOKEN }}" >> $GITHUB_ENV
-          fi
-      - uses: rasahq/merge-pal-action@master
-        with:
-          token: ${{ env.MERGE_TOKEN }}
diff --git a/.gitignore b/.gitignore
@@ -67,6 +67,7 @@ examples/moodbot/errors.json
 examples/formbot/models*
 examples/concertbot/models*
 examples/moodbot/models*
+examples/e2ebot/results/*
 failed_stories.md
 errors.json
 pip-wheel-metadata/*
@@ -83,6 +84,6 @@ env
 .history
 rasa/segment_key
 rasa/keys
-
+/results/
 # Local Netlify folder
 .netlify
diff --git a/Makefile b/Makefile
@@ -179,13 +179,30 @@ else
 	set -o allexport; source tests_deployment/.env && OMP_NUM_THREADS=1 poetry run pytest $(INTEGRATION_TEST_FOLDER) -n $(JOBS) -m $(INTEGRATION_TEST_PYTEST_MARKERS) && set +o allexport
 endif
 
-test-non-training: clean
-	# OMP_NUM_THREADS can improve overall performance using one thread by process (on tensorflow), avoiding overload
-	RAISE_ON_TRAIN=True OMP_NUM_THREADS=1 poetry run pytest tests -n $(JOBS) --cov rasa -m "not trains_model" --ignore $(INTEGRATION_TEST_FOLDER)
+test-cli: PYTEST_MARKER=category_cli
+test-cli: test-marker
 
-test-training: clean
-	# OMP_NUM_THREADS can improve overall performance using one thread by process (on tensorflow), avoiding overload
-	OMP_NUM_THREADS=1 poetry run pytest tests -n $(JOBS) --cov rasa -m "trains_model" --ignore $(INTEGRATION_TEST_FOLDER)
+test-core-featurizers: PYTEST_MARKER=category_core_featurizers
+test-core-featurizers: test-marker
+
+test-policies: PYTEST_MARKER=category_policies
+test-policies: test-marker
+
+test-nlu-featurizers: PYTEST_MARKER=category_nlu_featurizers
+test-nlu-featurizers: test-marker
+
+test-nlu-predictors: PYTEST_MARKER=category_nlu_predictors
+test-nlu-predictors: test-marker
+
+test-full-model-training: PYTEST_MARKER=category_full_model_training
+test-full-model-training: test-marker
+
+test-other-unit-tests: PYTEST_MARKER=category_other_unit_tests
+test-other-unit-tests: test-marker
+
+test-marker: clean
+    # OMP_NUM_THREADS can improve overall performance using one thread by process (on tensorflow), avoiding overload
+	OMP_NUM_THREADS=1 poetry run pytest tests -n $(JOBS) --cov rasa -m "$(PYTEST_MARKER)" --ignore $(INTEGRATION_TEST_FOLDER)
 
 generate-pending-changelog:
 	poetry run python -c "from scripts import release; release.generate_changelog('major.minor.patch')"

diff --git a/README.md b/README.md
@@ -193,21 +193,6 @@ JOBS=[n] make test
 
 Where `[n]` is the number of jobs desired. If omitted, `[n]` will be automatically chosen by pytest.
 
-#### Tests that train
-A test that trains a model is defined as any test that explicitly or inadvertently calls any method annotated with `@rasa.shared.utils.common.raise_on_unexpected_train`.
-Currently, this is: `nlu.train`, `core.train`, `Agent.train`, and `Trainer.train`.
-
-We specify tests that train a model using the pytest mark `trains_model`.
-e.g:
-
-    @pytest.mark.trains_model
-    def test_some_training()
-        ...
-
-These test are then run separately in the CI using the make commands `make test-non-training` and `make test-training` respectively.
-
-The command `make test-non-training` will fail if training occurs.
-
 
 ### Running the Integration Tests
 

diff --git a/changelog/6484.misc.md b/changelog/6484.misc.md
@@ -0,0 +1,4 @@
+- Renamed `templates` to `responses` across Rasa code base and documentation.
+- Renamed `response_templates` to `responses` across the Rasa code base.
+- Renamed `template_name` and `response_name` to `utter_action` in Rasa Core and rasa/shared.
+- Renamed `utter_templates` to `responses` in rasa/shared.
diff --git a/changelog/6484.removal.md b/changelog/6484.removal.md
@@ -0,0 +1,19 @@
+NLG Server
+- Changed request format to send `response` as well as `template` as a field. The `template` field will be removed in Rasa Open Source 3.0.0.
+
+`rasa.core.agent`
+- The terminology `template` is deprecated and replaced by `response`. Support for `template` from the NLG response will be removed in Rasa Open Source 3.0.0. Please see [here](nlg.mdx) for more details.
+
+`rasa.core.nlg.generator`
+- `generate()` now takes in  `utter_action` as a parameter.
+- The terminology `template` is deprecated and replaced by `response`. Support for `template` in the `NaturalLanguageGenerator` will be removed in Rasa Open Source 3.0.0.
+
+`rasa.shared.core.domain`
+- The property `templates` is deprecated. Use `responses` instead. It will be removed in Rasa Open Source 3.0.0.
+- `retrieval_intent_templates` will be removed in Rasa Open Source 3.0.0. Please use `retrieval_intent_responses` instead.
+- `is_retrieval_intent_template` will be removed in Rasa Open Source 3.0.0. Please use `is_retrieval_intent_response` instead.
+- `check_missing_templates` will be removed in Rasa Open Source 3.0.0. Please use `check_missing_responses` instead.
+
+Response Selector
+- The field `template_name` will be deprecated in Rasa Open Source 3.0.0. Please use `utter_action` instead. Please see [here](components.mdx#selectors) for more details.
+- The field `response_templates` will be deprecated in Rasa Open Source 3.0.0. Please use `responses` instead. Please see [here](components.mdx#selectors) for more details.
diff --git a/changelog/7438.improvement.md b/changelog/7438.improvement.md
@@ -0,0 +1,10 @@
+Simplify our training by overwriting `train_step` instead of `fit` for our custom models.
+
+This allows us to use the build-in callbacks from Keras, such as the 
+[Tensorboard Callback](https://www.tensorflow.org/api_docs/python/tf/keras/callbacks/TensorBoard), 
+which offers more functionality compared to what we had before. 
+
+  :::warning
+  If you want to use Tensorboard for `DIETClassifier`, `ResponseSelector`, or `TEDPolicy` and log metrics after 
+  every (mini)batch, please use 'batch' instead of 'minibatch' as 'tensorboard_log_level'.
+
diff --git a/changelog/7470.bugfix.md b/changelog/7470.bugfix.md
@@ -0,0 +1,10 @@
+Allow to have slots with values that result to a dictionary under the key `slot_was_set` (in `stories.yml` file).
+
+An example would be to have the following story step in `stories.yml`:
+```yaml
+- slot_was_set:
+    - some_slot:
+        some_key: 'some_value'
+        other_key: 'other_value'
+```
+This would be allowed if the `some_slot` is also set accordingly in the `domain.yml` with type `any`.
diff --git a/data/test_domains/duplicate_templates.yml → data/test_domains/duplicate_responses.yml b/data/test_domains/duplicate_templates.yml → data/test_domains/duplicate_responses.yml
diff --git a/docs/docs/components.mdx b/docs/docs/components.mdx
@@ -1607,7 +1607,7 @@ However, additional parameters exist that can be adapted.
 +---------------------------------+------------------+--------------------------------------------------------------+
 | tensorboard_log_level           | "epoch"          | Define when training metrics for tensorboard should be       |
 |                                 |                  | logged. Either after every epoch ('epoch') or for every      |
-|                                 |                  | training step ('minibatch').                                 |
+|                                 |                  | training step ('batch').                                 |
 +---------------------------------+------------------+--------------------------------------------------------------+
 | featurizers                     | []               | List of featurizer names (alias names). Only features        |
 |                                 |                  | coming from the listed names are used. If list is empty      |
@@ -2488,7 +2488,7 @@ The `SpacyEntityExtractor` extractor does not provide a `confidence` level and w
   +---------------------------------+------------------+--------------------------------------------------------------+
   | tensorboard_log_level           | "epoch"          | Define when training metrics for tensorboard should be       |
   |                                 |                  | logged. Either after every epoch ('epoch') or for every      |
-  |                                 |                  | training step ('minibatch').                                 |
+  |                                 |                  | training step ('batch').                                 |
   +---------------------------------+------------------+--------------------------------------------------------------+
   | featurizers                     | []               | List of featurizer names (alias names). Only features        |
   |                                 |                  | coming from the listed names are used. If list is empty      |
@@ -2529,7 +2529,7 @@ Selectors predict a bot response from a set of candidate responses.
 * **Outputs**
 
   A dictionary with the key as the retrieval intent of the response selector
-   and value containing predicted response templates, confidence and the response key under the retrieval intent
+   and value containing predicted responses, confidence and the response key under the retrieval intent
 
 
 
@@ -2547,7 +2547,7 @@ Selectors predict a bot response from a set of candidate responses.
     and stores two properties:
 
     * `response`: The predicted response key under the corresponding retrieval intent,
-    prediction's confidence and the associated response templates.
+    prediction's confidence and the associated responses.
 
     * `ranking`: Ranking with confidences of top 10 candidate response keys.
 
@@ -2561,7 +2561,7 @@ Selectors predict a bot response from a set of candidate responses.
               "id": 1388783286124361986,
               "confidence": 0.7,
               "intent_response_key": "chitchat/ask_weather",
-              "response_templates": [
+              "responses": [
                 {
                   "text": "It's sunny in Berlin today",
                   "image": "https://i.imgur.com/nGF1K8f.jpg"
@@ -2570,7 +2570,7 @@ Selectors predict a bot response from a set of candidate responses.
                   "text": "I think it's about to rain."
                 }
               ],
-              "template_name": "utter_chitchat/ask_weather"
+              "utter_action": "utter_chitchat/ask_weather"
              },
             "ranking": [
               {
@@ -2600,7 +2600,7 @@ Selectors predict a bot response from a set of candidate responses.
               "id": 1388783286124361986,
               "confidence": 0.7,
               "intent_response_key": "chitchat/ask_weather",
-              "response_templates": [
+              "responses": [
                 {
                   "text": "It's sunny in Berlin today",
                   "image": "https://i.imgur.com/nGF1K8f.jpg"
@@ -2609,7 +2609,7 @@ Selectors predict a bot response from a set of candidate responses.
                   "text": "I think it's about to rain."
                 }
               ],
-              "template_name": "utter_chitchat/ask_weather"
+              "utter_action": "utter_chitchat/ask_weather"
              },
             "ranking": [
               {
@@ -2715,9 +2715,8 @@ Selectors predict a bot response from a set of candidate responses.
   Default is `None`, i.e. the model is trained for all retrieval intents.
 
   In its default configuration, the component uses the retrieval intent with the response key(e.g. `faq/ask_name`) as the label for training.
-  Alternatively, it can also be configured to use the text of the response templates as the training label
-  by switching `use_text_as_label` to `True`. In this mode, the component will use the first available response
-  template which has a text attribute for training. If none are found, it falls back to using the retrieval intent
+  Alternatively, it can also be configured to use the text of the responses as the training label
+  by switching `use_text_as_label` to `True`. In this mode, the component will use the first available response which has a text attribute for training. If none are found, it falls back to using the retrieval intent
   combined with the response key as the label.
 
   :::note examples and tutorials
@@ -2846,7 +2845,7 @@ However, additional parameters exist that can be adapted.
 +---------------------------------+-------------------+--------------------------------------------------------------+
 | tensorboard_log_level           | "epoch"           | Define when training metrics for tensorboard should be       |
 |                                 |                   | logged. Either after every epoch ("epoch") or for every      |
-|                                 |                   | training step ("minibatch").                                 |
+|                                 |                   | training step ("batch").                                 |
 +---------------------------------+-------------------+--------------------------------------------------------------+
 | featurizers                     | []                | List of featurizer names (alias names). Only features        |
 |                                 |                   | coming from the listed names are used. If list is empty      |

diff --git a/docs/docs/connectors/facebook-messenger.mdx b/docs/docs/connectors/facebook-messenger.mdx
@@ -72,7 +72,7 @@ to make the new channel endpoint available for Facebook Messenger to send messag
 ## Supported response attachments
 
 In addition to typical text, image, and custom responses, the Facebook Messenger
-channel supports the following additional response template attachments:
+channel supports the following additional response attachments:
 
 * [Buttons](https://developers.facebook.com/docs/messenger-platform/send-messages/buttons)
   are structured the same as other Rasa buttons. Facebook API limits the amount of

diff --git a/docs/docs/migration-guide.mdx b/docs/docs/migration-guide.mdx
@@ -10,6 +10,28 @@ description: |
 This page contains information about changes between major versions and
 how you can migrate from one version to another.
 
+## Rasa 2.3 to Rasa 2.4
+### Deprecating `template` for `response`
+NLG Server
+- Changed request format to send `response` as well as `template` as a field. The `template` field will be removed in Rasa Open Source 3.0.0.
+
+`rasa.core.agent`
+- The terminology `template` is deprecated and replaced by `response`. Support for `template` from the NLG response will be removed in Rasa Open Source 3.0.0. Please see [here](./nlg.mdx) for more details.
+
+`rasa.core.nlg.generator`
+- `generate()` now takes in  `utter_action` as a parameter.
+- The terminology `template` is deprecated and replaced by `response`. Support for `template` in the `NaturalLanguageGenerator` will be removed in Rasa Open Source 3.0.0.
+
+`rasa.shared.core.domain`
+- The property `templates` is deprecated. Use `responses` instead. It will be removed in Rasa Open Source 3.0.0.
+- `retrieval_intent_templates` will be removed in Rasa Open Source 3.0.0. Please use `retrieval_intent_responses` instead.
+- `is_retrieval_intent_template` will be removed in Rasa Open Source 3.0.0. Please use `is_retrieval_intent_response` instead.
+- `check_missing_templates` will be removed in Rasa Open Source 3.0.0. Please use `check_missing_responses` instead.
+
+Response Selector
+- The field `template_name` will be deprecated in Rasa Open Source 3.0.0. Please use `utter_action` instead. Please see [here](./components.mdx#selectors) for more details.
+- The field `response_templates` will be deprecated in Rasa Open Source 3.0.0. Please use `responses` instead. Please see [here](./components.mdx#selectors) for more details.
+
 ## Rasa 2.3.3 to Rasa 2.3.4
 
 :::caution
@@ -38,6 +60,12 @@ If you trained a model with `model_confidence=cosine` or `model_confidence=inner
 
 ## Rasa 2.2 to Rasa 2.3
 
+### General
+
+If you want to use Tensorboard for `DIETClassifier`, `ResponseSelector`, or `TEDPolicy` and log metrics after
+every (mini)batch, please use 'batch' instead of 'minibatch' as 'tensorboard_log_level'.
+
+
 ### Machine Learning Components
 
 A few changes have been made to the loss function inside machine learning (ML)

diff --git a/docs/docs/nlg.mdx b/docs/docs/nlg.mdx
@@ -17,7 +17,7 @@ external HTTP server that you define.
 
 ## Responding to Requests
 
-### Response Format
+### Request Format
 
 When your model predicts that your bot should send a response to the user,
 it will send a request to your server, giving you the information
@@ -28,7 +28,7 @@ like this:
 
 ```json
 {
-  "template":"utter_what_can_do",
+  "response":"utter_what_can_do",
   "arguments":{
 
   },
@@ -170,7 +170,7 @@ Here is an overview of the high-level keys in the post request:
 
 Key | Description
 ---|---
-`template` | The name of the response predicted by Rasa Open Source.
+`response` | The name of the response predicted by Rasa Open Source.
 `arguments` | Optional keyword arguments that can be provided by custom actions.
 `tracker` | A dictionary containing the entire conversation history.
 `channel` | The output channel this message will be sent to.