New custom level tests.

Feature/training content

.env

@@ -3,3 +3,4 @@ JWT_SECRET_KEY=6e9c124ba92e8814719dcb0f21200c8aa4d0f119a994ac5e06eb90a366c83ab2
 JWT_TEST_TOKEN=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzdWIiOiJ0ZXN0In0.Emrs2D3BmMP4b3zMjw0fJTPeyMwWEBDbxx2vvaWguO0
 GOOGLE_APPLICATION_CREDENTIALS=firebase-configs/storied-phalanx-349916.json
 HEY_GEN_TOKEN=MjY4MDE0MjdjZmNhNDFmYTlhZGRkNmI3MGFlMzYwZDItMTY5NTExNzY3MA==
+GPT_ZERO_API_KEY=0195b9bb24c5439899f71230809c74af

.gitignore

@@ -2,3 +2,4 @@ __pycache__
 .idea
 .env
 .DS_Store
+/firebase-configs/test_firebase.json

.idea/.gitignore

@@ -1,8 +0,0 @@
-# Default ignored files
-/shelf/
-/workspace.xml
-# Editor-based HTTP Client requests
-/httpRequests/
-# Datasource local storage ignored files
-/dataSources/
-/dataSources.local.xml

.idea/ielts-be.iml

@@ -1,24 +1,14 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <module type="PYTHON_MODULE" version="4">
-  <component name="Flask">
-    <option name="enabled" value="true" />
-  </component>
   <component name="NewModuleRootManager">
     <content url="file://$MODULE_DIR$">
-      <excludeFolder url="file://$MODULE_DIR$/venv" />
+      <excludeFolder url="file://$MODULE_DIR$/.venv" />
     </content>
-    <orderEntry type="jdk" jdkName="Python 3.9" jdkType="Python SDK" />
+    <orderEntry type="jdk" jdkName="Python 3.11 (ielts-be)" jdkType="Python SDK" />
     <orderEntry type="sourceFolder" forTests="false" />
   </component>
-  <component name="PackageRequirementsSettings">
+  <component name="PyDocumentationSettings">
-    <option name="versionSpecifier" value="Don't specify version" />
+    <option name="format" value="GOOGLE" />
-  </component>
+    <option name="myDocStringFormat" value="Google" />
-  <component name="TemplatesService">
-    <option name="TEMPLATE_CONFIGURATION" value="Jinja2" />
-    <option name="TEMPLATE_FOLDERS">
-      <list>
-        <option value="$MODULE_DIR$/../flaskProject\templates" />
-      </list>
-    </option>
   </component>
 </module>

.idea/misc.xml

@@ -1,4 +1,10 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <project version="4">
-  <component name="ProjectRootManager" version="2" project-jdk-name="Python 3.9" project-jdk-type="Python SDK" />
+  <component name="Black">
+    <option name="sdkName" value="Python 3.11 (ielts-be)" />
+  </component>
+  <component name="ProjectRootManager" version="2" project-jdk-name="Python 3.11 (ielts-be)" project-jdk-type="Python SDK" />
+  <component name="PyCharmProfessionalAdvertiser">
+    <option name="shown" value="true" />
+  </component>
 </project>

.idea/vcs.xml

@@ -1,6 +1,6 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <project version="4">
   <component name="VcsDirectoryMappings">
-    <mapping directory="$PROJECT_DIR$" vcs="Git" />
+    <mapping directory="" vcs="Git" />
   </component>
 </project>

helper/constants.py

@@ -18,7 +18,13 @@ GEN_FIELDS = ['topic']
 GEN_TEXT_FIELDS = ['title']
 LISTENING_GEN_FIELDS = ['transcript', 'exercise']
 READING_EXERCISE_TYPES = ['fillBlanks', 'writeBlanks', 'trueFalse', 'paragraphMatch']
+READING_3_EXERCISE_TYPES = ['fillBlanks', 'writeBlanks', 'trueFalse', 'paragraphMatch', 'ideaMatch']
 LISTENING_EXERCISE_TYPES = ['multipleChoice', 'writeBlanksQuestions', 'writeBlanksFill', 'writeBlanksForm']
+LISTENING_1_EXERCISE_TYPES = ['multipleChoice', 'writeBlanksQuestions', 'writeBlanksFill', 'writeBlanksFill',
+                              'writeBlanksForm', 'writeBlanksForm', 'writeBlanksForm', 'writeBlanksForm']
+LISTENING_2_EXERCISE_TYPES = ['multipleChoice', 'writeBlanksQuestions']
+LISTENING_3_EXERCISE_TYPES = ['multipleChoice3Options', 'writeBlanksQuestions']
+LISTENING_4_EXERCISE_TYPES = ['multipleChoice', 'writeBlanksQuestions', 'writeBlanksFill', 'writeBlanksForm']
 
 TOTAL_READING_PASSAGE_1_EXERCISES = 13
 TOTAL_READING_PASSAGE_2_EXERCISES = 13
@@ -35,7 +41,7 @@ SPEAKING_MIN_TIMER_DEFAULT = 14
 
 BLACKLISTED_WORDS = ["jesus", "sex", "gay", "lesbian", "homosexual", "god", "angel", "pornography", "beer", "wine",
                      "cocaine", "alcohol", "nudity", "lgbt", "casino", "gambling", "catholicism",
-                     "discrimination", "politics", "politic", "christianity", "islam", "christian", "christians",
+                     "discrimination", "politic", "christianity", "islam", "christian", "christians",
                      "jews", "jew", "discrimination", "discriminatory"]
 
 EN_US_VOICES = [
@@ -141,7 +147,6 @@ mti_topics = [
     "Poverty Alleviation",
     "Cybersecurity and Privacy",
     "Human Rights",
-    "Social Justice",
     "Food and Agriculture",
     "Cyberbullying and Online Safety",
     "Linguistic Diversity",
@@ -169,7 +174,6 @@ topics = [
     "Space Exploration",
     "Artificial Intelligence",
     "Climate Change",
-    "World Religions",
     "The Human Brain",
     "Renewable Energy",
     "Cultural Diversity",
@@ -232,7 +236,6 @@ topics = [
     "Meditation Practices",
     "Literary Symbolism",
     "Marine Conservation",
-    "Social Justice Movements",
     "Sustainable Tourism",
     "Ancient Philosophy",
     "Cold War Era",
@@ -656,3 +659,19 @@ academic_subjects = [
     "Ecology",
     "International Business"
 ]
+
+grammar_types = [
+    "parts of speech",
+    "parts of speech - Nouns",
+    "parts of speech - Pronouns",
+    "parts of speech - Verbs",
+    "parts of speech - Adverbs",
+    "parts of speech - Adjectives",
+    "parts of speech - Conjunctions",
+    "parts of speech - Prepositions",
+    "parts of speech - Interjections",
+    "sentence structure",
+    "types of sentences",
+    "tenses",
+    "active voice and passive voice"
+]

helper/gpt_zero.py

@@ -0,0 +1,50 @@
+from logging import getLogger
+from typing import Dict, Optional
+import requests
+
+
+class GPTZero:
+    _GPT_ZERO_ENDPOINT = 'https://api.gptzero.me/v2/predict/text'
+
+    def __init__(self, gpt_zero_key: str):
+        self._logger = getLogger(__name__)
+        if gpt_zero_key is None:
+            self._logger.warning('GPT Zero key was not included! Skipping ai detection when grading.')
+        self._gpt_zero_key = gpt_zero_key
+        self._header = {
+            'x-api-key': gpt_zero_key
+        }
+
+    def run_detection(self, text: str):
+        if self._gpt_zero_key is None:
+            return None
+        data = {
+            'document': text,
+            'version': '',
+            'multilingual': False
+        }
+        response = requests.post(self._GPT_ZERO_ENDPOINT, headers=self._header, json=data)
+        if response.status_code != 200:
+            self._logger.error(f'GPT\'s Zero Endpoint returned with {response.status_code}: {response.json()}')
+            return None
+        return self._parse_detection(response.json())
+
+    def _parse_detection(self, response: Dict) -> Optional[Dict]:
+        try:
+            text_scan = response["documents"][0]
+            filtered_sentences = [
+                {
+                    "sentence": item["sentence"],
+                    "highlight_sentence_for_ai": item["highlight_sentence_for_ai"]
+                }
+                for item in text_scan["sentences"]
+            ]
+            return {
+                "class_probabilities": text_scan["class_probabilities"],
+                "confidence_category": text_scan["confidence_category"],
+                "predicted_class": text_scan["predicted_class"],
+                "sentences": filtered_sentences
+            }
+        except Exception as e:
+            self._logger.error(f'Failed to parse GPT\'s Zero response: {str(e)}')
+            return None

helper/heygen_api.py

@@ -1,17 +1,19 @@
 import os
 import random
 import time
+from logging import getLogger
 
 import requests
 from dotenv import load_dotenv
 
-import app
 from helper.constants import *
 from helper.firebase_helper import upload_file_firebase_get_url, save_to_db_with_id
 from heygen.AvatarEnum import AvatarEnum
 
 load_dotenv()
 
+logger = getLogger(__name__)
+
 # Get HeyGen token
 TOKEN = os.getenv("HEY_GEN_TOKEN")
 FIREBASE_BUCKET = os.getenv('FIREBASE_BUCKET')
@@ -29,26 +31,32 @@ GET_HEADER = {
 
 
 def create_videos_and_save_to_db(exercises, template, id):
+    avatar = random.choice(list(AvatarEnum))
     # Speaking 1
     # Using list comprehension to find the element with the desired value in the 'type' field
     found_exercises_1 = [element for element in exercises if element.get('type') == 1]
     # Check if any elements were found
     if found_exercises_1:
         exercise_1 = found_exercises_1[0]
-        app.app.logger.info('Creating video for speaking part 1')
+        sp1_questions = []
-        sp1_result = create_video(exercise_1["question"], random.choice(list(AvatarEnum)))
+        logger.info('Creating video for speaking part 1')
-        if sp1_result is not None:
+        for question in exercise_1["questions"]:
-            sound_file_path = VIDEO_FILES_PATH + sp1_result
+            sp1_result = create_video(question, avatar)
-            firebase_file_path = FIREBASE_SPEAKING_VIDEO_FILES_PATH + sp1_result
+            if sp1_result is not None:
-            url = upload_file_firebase_get_url(FIREBASE_BUCKET, firebase_file_path, sound_file_path)
+                sound_file_path = VIDEO_FILES_PATH + sp1_result
-            sp1_video_path = firebase_file_path
+                firebase_file_path = FIREBASE_SPEAKING_VIDEO_FILES_PATH + sp1_result
-            sp1_video_url = url
+                url = upload_file_firebase_get_url(FIREBASE_BUCKET, firebase_file_path, sound_file_path)
-            template["exercises"][0]["text"] = exercise_1["question"]
+                video = {
-            template["exercises"][0]["title"] = exercise_1["topic"]
+                    "text": question,
-            template["exercises"][0]["video_url"] = sp1_video_url
+                    "video_path": firebase_file_path,
-            template["exercises"][0]["video_path"] = sp1_video_path
+                    "video_url": url
-        else:
+                }
-            app.app.logger.error("Failed to create video for part 1 question: " + exercise_1["question"])
+                sp1_questions.append(video)
+            else:
+                logger.error("Failed to create video for part 1 question: " + exercise_1["question"])
+        template["exercises"][0]["prompts"] = sp1_questions
+        template["exercises"][0]["first_title"] = exercise_1["first_topic"]
+        template["exercises"][0]["second_title"] = exercise_1["second_topic"]
 
     # Speaking 2
     # Using list comprehension to find the element with the desired value in the 'type' field
@@ -56,8 +64,8 @@ def create_videos_and_save_to_db(exercises, template, id):
     # Check if any elements were found
     if found_exercises_2:
         exercise_2 = found_exercises_2[0]
-        app.app.logger.info('Creating video for speaking part 2')
+        logger.info('Creating video for speaking part 2')
-        sp2_result = create_video(exercise_2["question"], random.choice(list(AvatarEnum)))
+        sp2_result = create_video(exercise_2["question"], avatar)
         if sp2_result is not None:
             sound_file_path = VIDEO_FILES_PATH + sp2_result
             firebase_file_path = FIREBASE_SPEAKING_VIDEO_FILES_PATH + sp2_result
@@ -70,7 +78,7 @@ def create_videos_and_save_to_db(exercises, template, id):
             template["exercises"][1]["video_url"] = sp2_video_url
             template["exercises"][1]["video_path"] = sp2_video_path
         else:
-            app.app.logger.error("Failed to create video for part 2 question: " + exercise_2["question"])
+            logger.error("Failed to create video for part 2 question: " + exercise_2["question"])
 
     # Speaking 3
     # Using list comprehension to find the element with the desired value in the 'type' field
@@ -79,8 +87,7 @@ def create_videos_and_save_to_db(exercises, template, id):
     if found_exercises_3:
         exercise_3 = found_exercises_3[0]
         sp3_questions = []
-        avatar = random.choice(list(AvatarEnum))
+        logger.info('Creating videos for speaking part 3')
-        app.app.logger.info('Creating videos for speaking part 3')
         for question in exercise_3["questions"]:
             result = create_video(question, avatar)
             if result is not None:
@@ -94,7 +101,7 @@ def create_videos_and_save_to_db(exercises, template, id):
                 }
                 sp3_questions.append(video)
             else:
-                app.app.logger.error("Failed to create video for part 3 question: " + question)
+                logger.error("Failed to create video for part 3 question: " + question)
         template["exercises"][2]["prompts"] = sp3_questions
         template["exercises"][2]["title"] = exercise_3["topic"]
 
@@ -106,7 +113,7 @@ def create_videos_and_save_to_db(exercises, template, id):
         template["exercises"].pop(0)
 
     save_to_db_with_id("speaking", template, id)
-    app.app.logger.info('Saved speaking to DB with id ' + id + " : " + str(template))
+    logger.info('Saved speaking to DB with id ' + id + " : " + str(template))
 
 
 def create_video(text, avatar):
@@ -127,8 +134,8 @@ def create_video(text, avatar):
         }
     }
     response = requests.post(create_video_url, headers=POST_HEADER, json=data)
-    app.app.logger.info(response.status_code)
+    logger.info(response.status_code)
-    app.app.logger.info(response.json())
+    logger.info(response.json())
 
     # GET TO CHECK STATUS AND GET VIDEO WHEN READY
     video_id = response.json()["data"]["video_id"]
@@ -147,11 +154,11 @@ def create_video(text, avatar):
         error = response_data["data"]["error"]
 
         if status != "completed" and error is None:
-            app.app.logger.info(f"Status: {status}")
+            logger.info(f"Status: {status}")
             time.sleep(10)  # Wait for 10 second before the next request
 
-    app.app.logger.info(response.status_code)
+    logger.info(response.status_code)
-    app.app.logger.info(response.json())
+    logger.info(response.json())
 
     # DOWNLOAD VIDEO
     download_url = response.json()['data']['video_url']
@@ -165,8 +172,8 @@ def create_video(text, avatar):
         output_path = os.path.join(output_directory, output_filename)
         with open(output_path, 'wb') as f:
             f.write(response.content)
-        app.app.logger.info(f"File '{output_filename}' downloaded successfully.")
+        logger.info(f"File '{output_filename}' downloaded successfully.")
         return output_filename
     else:
-        app.app.logger.error(f"Failed to download file. Status code: {response.status_code}")
+        logger.error(f"Failed to download file. Status code: {response.status_code}")
         return None

helper/openai_interface.py

@@ -2,8 +2,8 @@ import json
 import os
 import re
 
-from openai import OpenAI
 from dotenv import load_dotenv
+from openai import OpenAI
 
 from helper.constants import BLACKLISTED_WORDS, GPT_3_5_TURBO
 from helper.token_counter import count_tokens
@@ -54,7 +54,7 @@ def check_fields(obj, fields):
     return all(field in obj for field in fields)
 
 
-def make_openai_call(model, messages, token_count, fields_to_check, temperature):
+def make_openai_call(model, messages, token_count, fields_to_check, temperature, check_blacklisted=True):
     global try_count
     result = client.chat.completions.create(
         model=model,
@@ -65,15 +65,16 @@ def make_openai_call(model, messages, token_count, fields_to_check, temperature)
     )
     result = result.choices[0].message.content
 
-    found_blacklisted_word = get_found_blacklisted_words(result)
+    if check_blacklisted:
+        found_blacklisted_word = get_found_blacklisted_words(result)
 
-    if found_blacklisted_word is not None and try_count < TRY_LIMIT:
+        if found_blacklisted_word is not None and try_count < TRY_LIMIT:
-        from app import app
+            from app import app
-        app.logger.warning("Result contains blacklisted words: " + str(found_blacklisted_word))
+            app.logger.warning("Result contains blacklisted words: " + str(found_blacklisted_word))
-        try_count = try_count + 1
+            try_count = try_count + 1
-        return make_openai_call(model, messages, token_count, fields_to_check, temperature)
+            return make_openai_call(model, messages, token_count, fields_to_check, temperature)
-    elif found_blacklisted_word is not None and try_count >= TRY_LIMIT:
+        elif found_blacklisted_word is not None and try_count >= TRY_LIMIT:
-        return ""
+            return ""
 
     if fields_to_check is None:
         return json.loads(result)
@@ -188,7 +189,7 @@ def get_fixed_text(text):
          }
     ]
     token_count = count_total_tokens(messages)
-    response = make_openai_call(GPT_3_5_TURBO, messages, token_count, ["fixed_text"], 0.2)
+    response = make_openai_call(GPT_3_5_TURBO, messages, token_count, ["fixed_text"], 0.2, False)
     return response["fixed_text"]
 
 
@@ -203,7 +204,7 @@ def get_speaking_corrections(text):
          }
     ]
     token_count = count_total_tokens(messages)
-    response = make_openai_call(GPT_3_5_TURBO, messages, token_count, ["fixed_text"], 0.2)
+    response = make_openai_call(GPT_3_5_TURBO, messages, token_count, ["fixed_text"], 0.2, False)
     return response["fixed_text"]
 
 
@@ -211,6 +212,7 @@ def has_blacklisted_words(text: str):
     text_lower = text.lower()
     return any(word in text_lower for word in BLACKLISTED_WORDS)
 
+
 def get_found_blacklisted_words(text: str):
     text_lower = text.lower()
     for word in BLACKLISTED_WORDS:
@@ -218,6 +220,7 @@ def get_found_blacklisted_words(text: str):
             return word
     return None
 
+
 def remove_special_characters_from_beginning(string):
     cleaned_string = string.lstrip('\n')
     if string.startswith("'") or string.startswith('"'):
@@ -239,6 +242,7 @@ def replace_expression_in_object(obj, expression, replacement):
                 obj[key] = replace_expression_in_object(obj[key], expression, replacement)
     return obj
 
+
 def count_total_tokens(messages):
     total_tokens = 0
     for message in messages:

helper/question_templates.py

@@ -1136,12 +1136,11 @@ def getSpeakingTemplate():
         "exercises": [
             {
                 "id": str(uuid.uuid4()),
-                "prompts": [],
+                "prompts": ["questions"],
-                "text": "text",
+                "text": "Listen carefully and respond.",
-                "title": "topic",
+                "first_title": "first_topic",
-                "video_url": "sp1_video_url",
+                "second_title": "second_topic",
-                "video_path": "sp1_video_path",
+                "type": "interactiveSpeaking"
-                "type": "speaking"
             },
             {
                 "id": str(uuid.uuid4()),

helper/speech_to_text_helper.py

@@ -95,17 +95,26 @@ def conversation_text_to_speech(conversation: list, file_name: str):
 
 
 def has_words(text: str):
+    if not has_common_words(text):
+        return False
     english_words = set(words.words())
     words_in_input = text.split()
     return any(word.lower() in english_words for word in words_in_input)
 
 
 def has_x_words(text: str, quantity):
+    if not has_common_words(text):
+        return False
     english_words = set(words.words())
     words_in_input = text.split()
     english_word_count = sum(1 for word in words_in_input if word.lower() in english_words)
     return english_word_count >= quantity
 
+def has_common_words(text: str):
+    english_words = {"the", "be", "to", "of", "and", "a", "in", "that", "have", "i"}
+    words_in_input = text.split()
+    english_word_count = sum(1 for word in words_in_input if word.lower() in english_words)
+    return english_word_count >= 10
 
 def divide_text(text, max_length=3000):
     if len(text) <= max_length:

training_content/__init__.py

@@ -0,0 +1,9 @@
+from .kb import TrainingContentKnowledgeBase
+from .service import TrainingContentService
+from .gpt import GPT
+
+__all__ = [
+    "TrainingContentService",
+    "TrainingContentKnowledgeBase",
+    "GPT"
+]

training_content/dtos.py

@@ -0,0 +1,29 @@
+from pydantic import BaseModel
+from typing import List
+
+
+class QueryDTO(BaseModel):
+    category: str
+    text: str
+
+
+class DetailsDTO(BaseModel):
+    exam_id: str
+    date: int
+    performance_comment: str
+    detailed_summary: str
+
+
+class WeakAreaDTO(BaseModel):
+    area: str
+    comment: str
+
+
+class TrainingContentDTO(BaseModel):
+    details: List[DetailsDTO]
+    weak_areas: List[WeakAreaDTO]
+    queries: List[QueryDTO]
+
+
+class TipsDTO(BaseModel):
+    tip_ids: List[str]

training_content/gpt.py

@@ -0,0 +1,64 @@
+import json
+from logging import getLogger
+
+from typing import List, Optional, Callable
+
+from openai.types.chat import ChatCompletionMessageParam
+from pydantic import BaseModel
+
+
+class GPT:
+
+    def __init__(self, openai_client):
+        self._client = openai_client
+        self._default_model = "gpt-4o"
+        self._logger = getLogger(__name__)
+
+    def prediction(
+            self,
+            messages: List[ChatCompletionMessageParam],
+            map_to_model: Callable,
+            json_scheme: str,
+            *,
+            model: Optional[str] = None,
+            temperature: Optional[float] = None,
+            max_retries: int = 3
+    ) -> List[BaseModel] | BaseModel | str | None:
+        params = {
+            "messages": messages,
+            "response_format": {"type": "json_object"},
+            "model": model if model else self._default_model
+        }
+
+        if temperature:
+            params["temperature"] = temperature
+
+        attempt = 0
+        while attempt < max_retries:
+            result = self._client.chat.completions.create(**params)
+            result_content = result.choices[0].message.content
+            try:
+                result_json = json.loads(result_content)
+                return map_to_model(result_json)
+            except Exception as e:
+                attempt += 1
+                self._logger.info(f"GPT returned malformed response: {result_content}\n {str(e)}")
+                params["messages"] = [
+                    {
+                        "role": "user",
+                        "content": (
+                            "Your previous response wasn't in the json format I've explicitly told you to output. "
+                            f"In your next response, you will fix it and return me just the json I've asked."
+                        )
+                    },
+                    {
+                        "role": "user",
+                        "content": (
+                            f"Previous response: {result_content}\n"
+                            f"JSON format: {json_scheme}"
+                        )
+                    }
+                ]
+                if attempt >= max_retries:
+                    self._logger.error(f"Max retries exceeded!")
+                    return None

training_content/kb.py

@@ -0,0 +1,85 @@
+import json
+import os
+from logging import getLogger
+from typing import Dict, List
+
+import faiss
+import pickle
+
+
+class TrainingContentKnowledgeBase:
+
+    def __init__(self, embeddings, path: str = 'pathways_2_rw_with_ids.json'):
+        self._embedding_model = embeddings
+        self._tips = None  # self._read_json(path)
+        self._category_metadata = None
+        self._indices = None
+        self._logger = getLogger(__name__)
+
+    @staticmethod
+    def _read_json(path: str) -> Dict[str, any]:
+        with open(path, 'r', encoding="utf-8") as json_file:
+            return json.loads(json_file.read())
+
+    def print_category_count(self):
+        category_tips = {}
+        for unit in self._tips['units']:
+            for page in unit['pages']:
+                for tip in page['tips']:
+                    category = tip['category'].lower().replace(" ", "_")
+                    if category not in category_tips:
+                        category_tips[category] = 0
+                    else:
+                        category_tips[category] = category_tips[category] + 1
+        print(category_tips)
+
+    def create_embeddings_and_save_them(self) -> None:
+        category_embeddings = {}
+        category_metadata = {}
+
+        for unit in self._tips['units']:
+            for page in unit['pages']:
+                for tip in page['tips']:
+                    category = tip['category'].lower().replace(" ", "_")
+                    if category not in category_embeddings:
+                        category_embeddings[category] = []
+                        category_metadata[category] = []
+
+                    category_embeddings[category].append(tip['embedding'])
+                    category_metadata[category].append({"id": tip['id'], "text": tip['text']})
+
+        category_indices = {}
+        for category, embeddings in category_embeddings.items():
+            embeddings_array = self._embedding_model.encode(embeddings)
+            index = faiss.IndexFlatL2(embeddings_array.shape[1])
+            index.add(embeddings_array)
+            category_indices[category] = index
+
+            faiss.write_index(index, f"./faiss/{category}_tips_index.faiss")
+
+        with open("./faiss/tips_metadata.pkl", "wb") as f:
+            pickle.dump(category_metadata, f)
+
+    def load_indices_and_metadata(
+            self,
+            directory: str = './faiss',
+            suffix: str = '_tips_index.faiss',
+            metadata_path: str = './faiss/tips_metadata.pkl'
+    ):
+        files = os.listdir(directory)
+        self._indices = {}
+        for file in files:
+            if file.endswith(suffix):
+                self._indices[file[:-len(suffix)]] = faiss.read_index(f'{directory}/{file}')
+                self._logger.info(f'Loaded embeddings for {file[:-len(suffix)]} category.')
+
+        with open(metadata_path, 'rb') as f:
+            self._category_metadata = pickle.load(f)
+        self._logger.info("Loaded tips metadata")
+
+    def query_knowledge_base(self, query: str, category: str, top_k: int = 5) -> List[Dict[str, str]]:
+        query_embedding = self._embedding_model.encode([query])
+        index = self._indices[category]
+        D, I = index.search(query_embedding, top_k)
+        results = [self._category_metadata[category][i] for i in I[0]]
+        return results

training_content/service.py

@@ -0,0 +1,341 @@
+import json
+from datetime import datetime
+from logging import getLogger
+
+from typing import Dict, List
+
+from training_content.dtos import TrainingContentDTO, WeakAreaDTO, QueryDTO, DetailsDTO, TipsDTO
+
+
+class TrainingContentService:
+
+    TOOLS = [
+        'critical_thinking',
+        'language_for_writing',
+        'reading_skills',
+        'strategy',
+        'words',
+        'writing_skills'
+    ]
+    # strategy word_link ct_focus reading_skill word_partners writing_skill language_for_writing
+
+    def __init__(self, kb, openai, firestore):
+        self._training_content_module = kb
+        self._db = firestore
+        self._logger = getLogger(__name__)
+        self._llm = openai
+
+    def get_tips(self, stats):
+        exam_data, exam_map = self._sort_out_solutions(stats)
+        training_content = self._get_exam_details_and_tips(exam_data)
+        tips = self._query_kb(training_content.queries)
+        usefull_tips = self._get_usefull_tips(exam_data, tips)
+        exam_map = self._merge_exam_map_with_details(exam_map, training_content.details)
+
+        weak_areas = {"weak_areas": []}
+        for area in training_content.weak_areas:
+            weak_areas["weak_areas"].append(area.dict())
+
+        training_doc = {
+            'created_at': int(datetime.now().timestamp() * 1000),
+            **exam_map,
+            **usefull_tips.dict(),
+            **weak_areas
+        }
+        doc_ref = self._db.collection('training').add(training_doc)
+
+        return {
+            "id": doc_ref[1].id
+        }
+
+    @staticmethod
+    def _merge_exam_map_with_details(exam_map: Dict[str, any], details: List[DetailsDTO]):
+        new_exam_map = {"exams": []}
+        for detail in details:
+            new_exam_map["exams"].append({
+                "id": detail.exam_id,
+                "date": detail.date,
+                "performance_comment": detail.performance_comment,
+                "detailed_summary": detail.detailed_summary,
+                **exam_map[detail.exam_id]
+            })
+        return new_exam_map
+
+    def _query_kb(self, queries: List[QueryDTO]):
+        map_categories = {
+            "critical_thinking": "ct_focus",
+            "language_for_writing": "language_for_writing",
+            "reading_skills": "reading_skill",
+            "strategy": "strategy",
+            "writing_skills": "writing_skill"
+        }
+
+        tips = {"tips": []}
+        for query in queries:
+            if query.category == "words":
+                tips["tips"].extend(
+                    self._training_content_module.query_knowledge_base(query.text, "word_link")
+                )
+                tips["tips"].extend(
+                    self._training_content_module.query_knowledge_base(query.text, "word_partners")
+                )
+            else:
+                if query.category in map_categories:
+                    tips["tips"].extend(
+                        self._training_content_module.query_knowledge_base(query.text, map_categories[query.category])
+                    )
+                else:
+                    self._logger.info(f"GTP tried to query knowledge base for {query.category} and it doesn't exist.")
+        return tips
+
+    def _get_exam_details_and_tips(self, exam_data: Dict[str, any]) -> TrainingContentDTO:
+        json_schema = (
+            '{ "details": [{"exam_id": "", "date": 0, "performance_comment": "", "detailed_summary": ""}],'
+            ' "weak_areas": [{"area": "", "comment": ""}], "queries": [{"text": "", "category": ""}] }'
+        )
+        messages = [
+            {
+                "role": "user",
+                "content": (
+                    f"I'm going to provide you with exam data, you will take the exam data and fill this json "
+                    f'schema : {json_schema}. "performance_comment" is a short sentence that describes the '
+                    'students\'s performance and main mistakes in a single exam, "detailed_summary" is a detailed '
+                    'summary of the student\'s performance, "weak_areas" are identified areas'
+                    ' across all exams which need to be improved upon, for example, area "Grammar and Syntax" comment "Issues'
+                    ' with sentence structure and punctuation.", the "queries" field is where you will write queries '
+                    'for tips that will be displayed to the student, the category attribute is a collection of '
+                    'embeddings and the text will be the text used to query the knowledge base. The categories are '
+                    f'the following [{", ".join(self.TOOLS)}]. The exam data will be a json where the key of the field '
+                    '"exams" is the exam id, an exam can be composed of multiple modules or single modules.'
+                )
+            },
+            {
+                "role": "user",
+                "content": f'Exam Data: {str(exam_data)}'
+            }
+        ]
+        return self._llm.prediction(messages, self._map_gpt_response, json_schema)
+
+    def _get_usefull_tips(self, exam_data: Dict[str, any], tips: Dict[str, any]) -> TipsDTO:
+        json_schema = (
+            '{ "tip_ids": [] }'
+        )
+        messages = [
+            {
+                "role": "user",
+                "content": (
+                    f"I'm going to provide you with tips and I want you to return to me the tips that "
+                    f"can be usefull for the student that made the exam that I'm going to send you, return "
+                    f"me the tip ids in this json format {json_schema}."
+                )
+            },
+            {
+                "role": "user",
+                "content": f'Exam Data: {str(exam_data)}'
+            },
+            {
+                "role": "user",
+                "content": f'Tips: {str(tips)}'
+            }
+        ]
+        return self._llm.prediction(messages, lambda response: TipsDTO(**response), json_schema)
+
+    @staticmethod
+    def _map_gpt_response(response: Dict[str, any]) -> TrainingContentDTO:
+        parsed_response = {
+            "details": [DetailsDTO(**detail) for detail in response["details"]],
+            "weak_areas": [WeakAreaDTO(**area) for area in response["weak_areas"]],
+            "queries": [QueryDTO(**query) for query in response["queries"]]
+        }
+        return TrainingContentDTO(**parsed_response)
+
+    def _sort_out_solutions(self, stats):
+        grouped_stats = {}
+        for stat in stats:
+            session_key = f'{str(stat["date"])}-{stat["user"]}'
+            module = stat["module"]
+            exam_id = stat["exam"]
+
+            if session_key not in grouped_stats:
+                grouped_stats[session_key] = {}
+            if module not in grouped_stats[session_key]:
+                grouped_stats[session_key][module] = {
+                    "stats": [],
+                    "exam_id": exam_id
+                }
+            grouped_stats[session_key][module]["stats"].append(stat)
+
+        exercises = {}
+        exam_map = {}
+        for session_key, modules in grouped_stats.items():
+            exercises[session_key] = {}
+            for module, module_stats in modules.items():
+                exercises[session_key][module] = {}
+
+                exam_id = module_stats["exam_id"]
+                if exam_id not in exercises[session_key][module]:
+                    exercises[session_key][module][exam_id] = {"date": None, "exercises": []}
+
+                exam_total_questions = 0
+                exam_total_correct = 0
+
+                for stat in module_stats["stats"]:
+                    exam_total_questions += stat["score"]["total"]
+                    exam_total_correct += stat["score"]["correct"]
+                    exercises[session_key][module][exam_id]["date"] = stat["date"]
+
+                    if session_key not in exam_map:
+                        exam_map[session_key] = {"stat_ids": [], "score": 0}
+                    exam_map[session_key]["stat_ids"].append(stat["id"])
+
+                    exam = self._get_doc_by_id(module, exam_id)
+                    if module == "listening":
+                        exercises[session_key][module][exam_id]["exercises"].extend(
+                            self._get_listening_solutions(stat, exam))
+                    elif module == "reading":
+                        exercises[session_key][module][exam_id]["exercises"].extend(
+                            self._get_reading_solutions(stat, exam))
+                    elif module == "writing":
+                        exercises[session_key][module][exam_id]["exercises"].extend(
+                            self._get_writing_prompts_and_answers(stat, exam)
+                        )
+                    elif module == "speaking":
+                        exercises[session_key][module][exam_id]["exercises"].extend(
+                            self._get_speaking_solutions(stat, exam)
+                        )
+                    elif module == "level":  # same structure as listening
+                        exercises[session_key][module][exam_id]["exercises"].extend(
+                            self._get_listening_solutions(stat, exam)
+                        )
+
+                exam_map[session_key]["score"] = round((exam_total_correct / exam_total_questions) * 100)
+                exam_map[session_key]["module"] = module
+
+        return {"exams": exercises}, exam_map
+
+    def _get_writing_prompts_and_answers(self, stat, exam):
+        result = []
+        try:
+            exercises = []
+            for solution in stat['solutions']:
+                answer = solution['solution']
+                exercise_id = solution['id']
+                exercises.append({
+                    "exercise_id": exercise_id,
+                    "answer": answer
+                })
+            for exercise in exercises:
+                for exam_exercise in exam["exercises"]:
+                    if exam_exercise["id"] == exercise["exercise_id"]:
+                        result.append({
+                            "exercise": exam_exercise["prompt"],
+                            "answer": exercise["answer"]
+                        })
+
+        except KeyError as e:
+            self._logger.warning(f"Malformed stat object: {str(e)}")
+
+        return result
+
+    def _get_listening_solutions(self, stat, exam):
+        result = []
+        try:
+            for part in exam["parts"]:
+                for exercise in part["exercises"]:
+                    if exercise["id"] == stat["exercise"]:
+                        if stat["type"] == "writeBlanks":
+                            result.append({
+                                "question": exercise["prompt"],
+                                "template": exercise["text"],
+                                "solution": exercise["solutions"],
+                                "answer": stat["solutions"]
+                            })
+                        elif stat["type"] == "multipleChoice":
+                            result.append({
+                                "question": exercise["prompt"],
+                                "exercise": exercise["questions"],
+                                "answer": stat["solutions"]
+                            })
+        except KeyError as e:
+            self._logger.warning(f"Malformed stat object: {str(e)}")
+        return result
+
+    def _get_speaking_solutions(self, stat, exam):
+        result = {}
+        try:
+            result = {
+                "comments": {
+                    key: value['comment'] for key, value in stat['solutions'][0]['evaluation']['task_response'].items()}
+                ,
+                "exercises": {}
+            }
+
+            for exercise in exam["exercises"]:
+                if exercise["id"] == stat["exercise"]:
+                    if stat["type"] == "interactiveSpeaking":
+                        for i in range(len(exercise["prompts"])):
+                            result["exercises"][f"exercise_{i+1}"] = {
+                                "question": exercise["prompts"][i]["text"]
+                            }
+                        for i in range(len(exercise["prompts"])):
+                            answer = stat['solutions'][0]["evaluation"].get(f'transcript_{i+1}', '')
+                            result["exercises"][f"exercise_{i+1}"]["answer"] = answer
+                    elif stat["type"] == "speaking":
+                        result["exercises"]["exercise_1"] = {
+                            "question": exercise["text"],
+                            "answer": stat['solutions'][0]["evaluation"].get(f'transcript', '')
+                        }
+        except KeyError as e:
+            self._logger.warning(f"Malformed stat object: {str(e)}")
+        return [result]
+
+    def _get_reading_solutions(self, stat, exam):
+        result = []
+        try:
+            for part in exam["parts"]:
+                text = part["text"]
+                for exercise in part["exercises"]:
+                    if exercise["id"] == stat["exercise"]:
+                        if stat["type"] == "fillBlanks":
+                            result.append({
+                                "text": text,
+                                "question": exercise["prompt"],
+                                "template": exercise["text"],
+                                "words": exercise["words"],
+                                "solutions": exercise["solutions"],
+                                "answer": stat["solutions"]
+                            })
+                        elif stat["type"] == "writeBlanks":
+                            result.append({
+                                "text": text,
+                                "question": exercise["prompt"],
+                                "template": exercise["text"],
+                                "solutions": exercise["solutions"],
+                                "answer": stat["solutions"]
+                            })
+                        elif stat["type"] == "trueFalse":
+                            result.append({
+                                "text": text,
+                                "questions": exercise["questions"],
+                                "answer": stat["solutions"]
+                            })
+                        elif stat["type"] == "matchSentences":
+                            result.append({
+                                "text": text,
+                                "question": exercise["prompt"],
+                                "sentences": exercise["sentences"],
+                                "options": exercise["options"],
+                                "answer": stat["solutions"]
+                            })
+        except KeyError as e:
+            self._logger.warning(f"Malformed stat object: {str(e)}")
+        return result
+
+    def _get_doc_by_id(self, collection: str, doc_id: str):
+        collection_ref = self._db.collection(collection)
+        doc_ref = collection_ref.document(doc_id)
+        doc = doc_ref.get()
+
+        if doc.exists:
+            return doc.to_dict()
+        return None