[Form Builder] Added unittests for main components

Author: meta-paul

examples/README.md

@@ -184,6 +184,8 @@ Putting it altogether, let's prepare and launch a task featuring a form containi
   mephisto review_app -h 0.0.0.0 -p 8000 -d True -f True
   ```
 
+_Note: if a package build was terminated/failed, or related source code was changed, FormComposer needs to be rebuilt with this command: `mephisto scripts form_composer rebuild_all_apps`._
+
 ---
 
 # Your Mephisto project

mephisto/client/cli.py

@@ -28,6 +28,7 @@
 import mephisto.scripts.mturk.launch_makeup_hits as launch_makeup_hits_mturk
 import mephisto.scripts.mturk.print_outstanding_hit_status as print_outstanding_hit_status_mturk
 import mephisto.scripts.mturk.print_outstanding_hit_status as soft_block_workers_by_mturk_id_mturk
+import mephisto.scripts.form_composer.rebuild_all_apps as rebuild_all_apps_form_composer
 from mephisto.client.cli_commands import get_wut_arguments
 from mephisto.generators.form_composer.config_validation.task_data_config import (
     create_extrapolated_config
@@ -223,7 +224,7 @@ def print_non_markdown_list(items: List[str]):
             res += "\n  * " + item
         return res
 
-    VALID_SCRIPT_TYPES = ["local_db", "heroku", "metrics", "mturk"]
+    VALID_SCRIPT_TYPES = ["local_db", "heroku", "metrics", "mturk", "form_composer"]
     if script_type is None or script_type.strip() not in VALID_SCRIPT_TYPES:
         print("")
         raise click.UsageError(
@@ -247,6 +248,9 @@ def print_non_markdown_list(items: List[str]):
         "print_outstanding_hit_status",
         "soft_block_workers_by_mturk_id",
     ]
+    FORM_COMPOSER_VALID_SCRIPTS_NAMES = [
+        "rebuild_all_apps",
+    ]
     script_type_to_scripts_data = {
         "local_db": {
             "valid_script_names": LOCAL_DB_VALID_SCRIPTS_NAMES,
@@ -275,10 +279,16 @@ def print_non_markdown_list(items: List[str]):
                 MTURK_VALID_SCRIPTS_NAMES[0]: cleanup_mturk.main,
                 MTURK_VALID_SCRIPTS_NAMES[1]: identify_broken_units_mturk.main,
                 MTURK_VALID_SCRIPTS_NAMES[2]: launch_makeup_hits_mturk.main,
-                MTURK_VALID_SCRIPTS_NAMES[3]: print_outstanding_hit_status_mturk.main,
+                MTURK_VALID_SCRIPTS_NAMES[3]: rebuild_all_apps_form_composer.main,
                 MTURK_VALID_SCRIPTS_NAMES[4]: soft_block_workers_by_mturk_id_mturk.main,
             },
         },
+        "form_composer": {
+            "valid_script_names": FORM_COMPOSER_VALID_SCRIPTS_NAMES,
+            "scripts": {
+                FORM_COMPOSER_VALID_SCRIPTS_NAMES[0]: rebuild_all_apps_form_composer.main,
+            },
+        },
     }
 
     if script_name is None or (

mephisto/generators/form_composer/config_validation/form_config.py

@@ -54,15 +54,15 @@ def _duplicate_values_exist(unique_names: UniqueAttrsType, errors: List[str]) ->
     return is_valid
 
 
-def validate_form_config(config_json: dict) -> Tuple[bool, List[str]]:
+def validate_form_config(config_data: dict) -> Tuple[bool, List[str]]:
     is_valid = True
     errors = []
 
-    if not isinstance(config_json, dict):
+    if not isinstance(config_data, dict):
         is_valid = False
         errors.append("Form config must be a key/value JSON Object.")
 
-    elif config_json.keys() != AVAILABLE_CONFIG_ATTRS.keys():
+    elif config_data.keys() != AVAILABLE_CONFIG_ATTRS.keys():
         is_valid = False
         errors.append(
             f"Form config must contain only these attributes: "
@@ -77,10 +77,10 @@ def validate_form_config(config_json: dict) -> Tuple[bool, List[str]]:
     unique_names: UniqueAttrsType = {}
 
     # Add main config level
-    items_to_validate.append((config_json, "Config", AVAILABLE_CONFIG_ATTRS))
+    items_to_validate.append((config_data, "Config", AVAILABLE_CONFIG_ATTRS))
 
     # Add form
-    form = config_json["form"]
+    form = config_data["form"]
     items_to_validate.append((form, "form", AVAILABLE_FORM_ATTRS))
     _collect_values_for_unique_attrs_from_item(form, unique_names)
 

mephisto/generators/form_composer/config_validation/separate_token_values_config.py

@@ -11,6 +11,7 @@
 from botocore.exceptions import BotoCoreError
 from botocore.exceptions import ClientError
 from botocore.exceptions import NoCredentialsError
+from rich import print
 
 from mephisto.generators.form_composer.constants import TOKEN_END_SYMBOLS
 from mephisto.generators.form_composer.constants import TOKEN_START_SYMBOLS
@@ -22,17 +23,17 @@
 
 
 def validate_separate_token_values_config(
-    config_json: Dict[str, List[str]],
+    config_data: Dict[str, List[str]],
 ) -> Tuple[bool, List[str]]:
     is_valid = True
     errors = []
 
-    if not isinstance(config_json, dict):
+    if not isinstance(config_data, dict):
         is_valid = False
         errors.append("Config must be a key/value JSON Object.")
         return is_valid, errors
 
-    for i, token_values in enumerate(config_json.items()):
+    for i, token_values in enumerate(config_data.items()):
         token, values = token_values
 
         if not values:
@@ -48,7 +49,7 @@ def validate_separate_token_values_config(
 def update_separate_token_values_config_with_file_urls(
     url: str,
     separate_token_values_config_path: str,
-    use_presigned_urls: bool,
+    use_presigned_urls: bool = False,
 ):
     try:
         files_locations = get_file_urls_from_s3_storage(url)

mephisto/generators/form_composer/config_validation/task_data_config.py

@@ -56,9 +56,15 @@ def _set_tokens_in_form_config_item(item: dict, tokens_values: dict):
 def _collect_form_config_items_to_extrapolate(config_data: dict) -> List[dict]:
     items_to_extrapolate = []
 
+    if not isinstance(config_data, dict):
+        return items_to_extrapolate
+
     form = config_data["form"]
     items_to_extrapolate.append(form)
 
+    submit_button = form["submit_button"]
+    items_to_extrapolate.append(submit_button)
+
     sections = form["sections"]
     for section in sections:
         items_to_extrapolate.append(section)
@@ -125,7 +131,7 @@ def _extrapolate_tokens_in_form_config(config_data: dict, tokens_values: dict) -
 
 
 def _validate_tokens_in_both_configs(
-    form_config_data, token_sets_values_config_data,
+    form_config_data: dict, token_sets_values_config_data: List[dict],
 ) -> Tuple[set, set, list]:
     tokens_from_form_config, tokens_in_unexpected_attrs_errors = (
         _collect_tokens_from_form_config(form_config_data)
@@ -245,21 +251,21 @@ def create_extrapolated_config(
         exit()
 
 
-def validate_task_data_config(config_json: List[dict]) -> Tuple[bool, List[str]]:
+def validate_task_data_config(config_data: List[dict]) -> Tuple[bool, List[str]]:
     is_valid = True
     errors = []
 
-    if not isinstance(config_json, list):
+    if not isinstance(config_data, list):
         is_valid = False
         errors.append("Config must be a JSON Array.")
 
-    if config_json:
-        if not all(config_json):
+    if config_data:
+        if not all(config_data):
             is_valid = False
             errors.append("Task data config must contain at least one non-empty item.")
 
         # Validate each form version contained in task data config
-        for item in config_json:
+        for item in config_data:
             form_config_is_valid, form_config_errors = validate_form_config(item)
             if not form_config_is_valid:
                 is_valid = False
@@ -375,11 +381,13 @@ def verify_form_composer_configs(
         print(f"\n[red]Provided Form Composer config files are invalid:[/red] {e}\n")
 
 
-def prepare_task_config_for_review_app(config: dict) -> dict:
-    config = deepcopy(config)
+def prepare_task_config_for_review_app(config_data: dict) -> dict:
+    config_data = deepcopy(config_data)
 
     procedure_code_regex = r"\s*(.+?)\s*"
-    tokens_from_inputs, _ = _collect_tokens_from_form_config(config, regex=procedure_code_regex)
+    tokens_from_inputs, _ = _collect_tokens_from_form_config(
+        config_data, regex=procedure_code_regex,
+    )
 
     url_from_rpocedure_code_regex = r"\(\"(.+?)\"\)"
     token_values = {}
@@ -396,5 +404,5 @@ def prepare_task_config_for_review_app(config: dict) -> dict:
             presigned_url = get_s3_presigned_url(url, S3_URL_EXPIRATION_MINUTES_MAX)
             token_values[token] = presigned_url
 
-    prepared_config = _extrapolate_tokens_in_form_config(config, token_values)
+    prepared_config = _extrapolate_tokens_in_form_config(config_data, token_values)
     return prepared_config

mephisto/generators/form_composer/constants.py

@@ -7,22 +7,22 @@
 
 CONTENTTYPE_BY_EXTENSION = {
     # Docs
-    'csv': 'text/csv',
-    'doc': 'application/msword',
-    'docx': 'application/vnd.openxmlformats-officedocument.wordprocessingml.document',
-    'pdf': 'application/pdf',
+    "csv": "text/csv",
+    "doc": "application/msword",
+    "docx": "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
+    "pdf": "application/pdf",
     # Images
-    'bmp': 'image/bmp',
-    'gif': 'image/gif',
-    'heic': 'image/heic',
-    'heif': 'image/heif',
-    'jpeg': 'image/jpeg',
-    'jpg': 'image/jpeg',
-    'png': 'image/png',
+    "bmp": "image/bmp",
+    "gif": "image/gif",
+    "heic": "image/heic",
+    "heif": "image/heif",
+    "jpeg": "image/jpeg",
+    "jpg": "image/jpeg",
+    "png": "image/png",
     # Videos
-    'mkv': 'video/x-matroska',
-    'mp4': 'video/mp4',
-    'webm': 'video/webm',
+    "mkv": "video/x-matroska",
+    "mp4": "video/mp4",
+    "webm": "video/webm",
 }
 
 JSON_IDENTATION = 2

mephisto/review_app/server/api/views/unit_data_static_view.py

@@ -10,6 +10,7 @@
 from flask import Response
 from flask import send_from_directory
 from flask.views import MethodView
+from werkzeug.exceptions import NotFound
 
 from mephisto.data_model.agent import Agent
 from mephisto.data_model.unit import Unit
@@ -55,5 +56,10 @@ def get(
         if filename_by_original_name:
             filename = filename_by_original_name
 
-        unit_data_folder = unit.get_assigned_agent().get_data_dir()
+        agent = unit.get_assigned_agent()
+        if not agent:
+            app.logger.debug(f"No agent found for {unit}")
+            raise NotFound("File not found")
+
+        unit_data_folder = agent.get_data_dir()
         return send_from_directory(unit_data_folder, filename)

mephisto/review_app/server/api/views/units_details_view.py

@@ -55,8 +55,8 @@ def get(self) -> dict:
             task_run: TaskRun = unit.get_task_run()
             has_task_source_review = bool(task_run.args.get("blueprint").get("task_source_review"))
 
-            inputs = unit_data.get("data", {}).get("inputs")
-            outputs = unit_data.get("data", {}).get("outputs")
+            inputs = unit_data.get("data", {}).get("inputs", {})
+            outputs = unit_data.get("data", {}).get("outputs", {})
 
             # In case if there is outdated code that returns `final_submission`
             # under `inputs` and `outputs` keys, we should use the value in side `final_submission`
@@ -67,9 +67,12 @@ def get(self) -> dict:
 
             # Perform any dynamic action on task config for current unit
             # to make it the same as it looked like for a worker
-            prepared_inputs = prepare_task_config_for_review_app(inputs)
+            prepared_inputs = inputs
+            if "form" in inputs:
+                prepared_inputs = prepare_task_config_for_review_app(inputs)
 
-            unit_data_folder = unit.get_assigned_agent().get_data_dir()
+            agent = unit.get_assigned_agent()
+            unit_data_folder = agent.get_data_dir() if agent else None
 
             units.append(
                 {

mephisto/scripts/form_composer/__init__.py

@@ -0,0 +1,4 @@
+#!/usr/bin/env python3
+# Copyright (c) Meta Platforms and its affiliates.
+# This source code is licensed under the MIT license found in the
+# LICENSE file in the root directory of this source tree.