devops/encoach_backend
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Updated this to the latest version of develop, got rid of most of the duplication, might be missing some packages in toml, needs testing

Browse Source
This commit is contained in:
Carlos Mesquita
2024-08-30 02:35:11 +01:00
parent 3cf9fa5cba
commit f92a803d96
73 changed files with 3642 additions and 2703 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
app/services/impl/level/__init__.py Normal file
View File

@@ -0,0 +1,5 @@
from .level import LevelService
__all__ = [
"LevelService"
]

335
app/services/impl/level/custom.py Normal file
View File

@@ -0,0 +1,335 @@
import queue
import random
from typing import Dict
from app.configs.constants import CustomLevelExerciseTypes, EducationalContent
from app.services.abc import (
ILLMService, ILevelService, IReadingService,
IWritingService, IListeningService, ISpeakingService
)
class CustomLevelModule:
def __init__(
self,
llm: ILLMService,
level: ILevelService,
reading: IReadingService,
listening: IListeningService,
writing: IWritingService,
speaking: ISpeakingService
):
self._llm = llm
self._level = level
self._reading = reading
self._listening = listening
self._writing = writing
self._speaking = speaking
# TODO: I've changed this to retrieve the args from the body request and not request query args
async def get_custom_level(self, data: Dict):
nr_exercises = int(data.get('nr_exercises'))
exercise_id = 1
response = {
"exercises": {},
"module": "level"
}
for i in range(1, nr_exercises + 1, 1):
exercise_type = data.get(f'exercise_{i}_type')
exercise_difficulty = data.get(f'exercise_{i}_difficulty', random.choice(['easy', 'medium', 'hard']))
exercise_qty = int(data.get(f'exercise_{i}_qty', -1))
exercise_topic = data.get(f'exercise_{i}_topic', random.choice(EducationalContent.TOPICS))
exercise_topic_2 = data.get(f'exercise_{i}_topic_2', random.choice(EducationalContent.TOPICS))
exercise_text_size = int(data.get(f'exercise_{i}_text_size', 700))
exercise_sa_qty = int(data.get(f'exercise_{i}_sa_qty', -1))
exercise_mc_qty = int(data.get(f'exercise_{i}_mc_qty', -1))
exercise_mc3_qty = int(data.get(f'exercise_{i}_mc3_qty', -1))
exercise_fillblanks_qty = int(data.get(f'exercise_{i}_fillblanks_qty', -1))
exercise_writeblanks_qty = int(data.get(f'exercise_{i}_writeblanks_qty', -1))
exercise_writeblanksquestions_qty = int(data.get(f'exercise_{i}_writeblanksquestions_qty', -1))
exercise_writeblanksfill_qty = int(data.get(f'exercise_{i}_writeblanksfill_qty', -1))
exercise_writeblanksform_qty = int(data.get(f'exercise_{i}_writeblanksform_qty', -1))
exercise_truefalse_qty = int(data.get(f'exercise_{i}_truefalse_qty', -1))
exercise_paragraphmatch_qty = int(data.get(f'exercise_{i}_paragraphmatch_qty', -1))
exercise_ideamatch_qty = int(data.get(f'exercise_{i}_ideamatch_qty', -1))
if exercise_type == CustomLevelExerciseTypes.MULTIPLE_CHOICE_4.value:
response["exercises"][f"exercise_{i}"] = {}
response["exercises"][f"exercise_{i}"]["questions"] = []
response["exercises"][f"exercise_{i}"]["type"] = "multipleChoice"
while exercise_qty > 0:
if exercise_qty - 15 > 0:
qty = 15
else:
qty = exercise_qty
mc_response = await self._level.gen_multiple_choice(
"normal", qty, exercise_id, utas=True,
all_exams=response["exercises"][f"exercise_{i}"]["questions"]
)
response["exercises"][f"exercise_{i}"]["questions"].extend(mc_response["questions"])
exercise_id = exercise_id + qty
exercise_qty = exercise_qty - qty
elif exercise_type == CustomLevelExerciseTypes.MULTIPLE_CHOICE_BLANK_SPACE.value:
response["exercises"][f"exercise_{i}"] = {}
response["exercises"][f"exercise_{i}"]["questions"] = []
response["exercises"][f"exercise_{i}"]["type"] = "multipleChoice"
while exercise_qty > 0:
if exercise_qty - 15 > 0:
qty = 15
else:
qty = exercise_qty
mc_response = await self._level.gen_multiple_choice(
"blank_space", qty, exercise_id, utas=True,
all_exams=response["exercises"][f"exercise_{i}"]["questions"]
)
response["exercises"][f"exercise_{i}"]["questions"].extend(mc_response["questions"])
exercise_id = exercise_id + qty
exercise_qty = exercise_qty - qty
elif exercise_type == CustomLevelExerciseTypes.MULTIPLE_CHOICE_UNDERLINED.value:
response["exercises"][f"exercise_{i}"] = {}
response["exercises"][f"exercise_{i}"]["questions"] = []
response["exercises"][f"exercise_{i}"]["type"] = "multipleChoice"
while exercise_qty > 0:
if exercise_qty - 15 > 0:
qty = 15
else:
qty = exercise_qty
mc_response = await self._level.gen_multiple_choice(
"underline", qty, exercise_id, utas=True,
all_exams=response["exercises"][f"exercise_{i}"]["questions"]
)
response["exercises"][f"exercise_{i}"]["questions"].extend(mc_response["questions"])
exercise_id = exercise_id + qty
exercise_qty = exercise_qty - qty
elif exercise_type == CustomLevelExerciseTypes.BLANK_SPACE_TEXT.value:
response["exercises"][f"exercise_{i}"] = await self._level.gen_blank_space_text_utas(
exercise_qty, exercise_id, exercise_text_size
)
response["exercises"][f"exercise_{i}"]["type"] = "blankSpaceText"
exercise_id = exercise_id + exercise_qty
elif exercise_type == CustomLevelExerciseTypes.READING_PASSAGE_UTAS.value:
response["exercises"][f"exercise_{i}"] = await self._level.gen_reading_passage_utas(
exercise_id, exercise_sa_qty, exercise_mc_qty, exercise_topic
)
response["exercises"][f"exercise_{i}"]["type"] = "readingExercises"
exercise_id = exercise_id + exercise_qty
elif exercise_type == CustomLevelExerciseTypes.WRITING_LETTER.value:
response["exercises"][f"exercise_{i}"] = await self._writing.get_writing_task_general_question(
1, exercise_topic, exercise_difficulty
)
response["exercises"][f"exercise_{i}"]["type"] = "writing"
exercise_id = exercise_id + 1
elif exercise_type == CustomLevelExerciseTypes.WRITING_2.value:
response["exercises"][f"exercise_{i}"] = await self._writing.get_writing_task_general_question(
2, exercise_topic, exercise_difficulty
)
response["exercises"][f"exercise_{i}"]["type"] = "writing"
exercise_id = exercise_id + 1
elif exercise_type == CustomLevelExerciseTypes.SPEAKING_1.value:
response["exercises"][f"exercise_{i}"] = await self._speaking.get_speaking_part(
1, exercise_topic, exercise_difficulty, exercise_topic_2
)
response["exercises"][f"exercise_{i}"]["type"] = "interactiveSpeaking"
exercise_id = exercise_id + 1
elif exercise_type == CustomLevelExerciseTypes.SPEAKING_2.value:
response["exercises"][f"exercise_{i}"] = await self._speaking.get_speaking_part(
2, exercise_topic, exercise_difficulty
)
response["exercises"][f"exercise_{i}"]["type"] = "speaking"
exercise_id = exercise_id + 1
elif exercise_type == CustomLevelExerciseTypes.SPEAKING_3.value:
response["exercises"][f"exercise_{i}"] = await self._speaking.get_speaking_part(
3, exercise_topic, exercise_difficulty
)
response["exercises"][f"exercise_{i}"]["type"] = "interactiveSpeaking"
exercise_id = exercise_id + 1
elif exercise_type == CustomLevelExerciseTypes.READING_1.value:
exercises = []
exercise_qty_q = queue.Queue()
total_qty = 0
if exercise_fillblanks_qty != -1:
exercises.append('fillBlanks')
exercise_qty_q.put(exercise_fillblanks_qty)
total_qty = total_qty + exercise_fillblanks_qty
if exercise_writeblanks_qty != -1:
exercises.append('writeBlanks')
exercise_qty_q.put(exercise_writeblanks_qty)
total_qty = total_qty + exercise_writeblanks_qty
if exercise_truefalse_qty != -1:
exercises.append('trueFalse')
exercise_qty_q.put(exercise_truefalse_qty)
total_qty = total_qty + exercise_truefalse_qty
if exercise_paragraphmatch_qty != -1:
exercises.append('paragraphMatch')
exercise_qty_q.put(exercise_paragraphmatch_qty)
total_qty = total_qty + exercise_paragraphmatch_qty
response["exercises"][f"exercise_{i}"] = await self._reading.gen_reading_passage(
1, exercise_topic, exercises, exercise_qty_q, exercise_difficulty, exercise_id
)
response["exercises"][f"exercise_{i}"]["type"] = "reading"
exercise_id = exercise_id + total_qty
elif exercise_type == CustomLevelExerciseTypes.READING_2.value:
exercises = []
exercise_qty_q = queue.Queue()
total_qty = 0
if exercise_fillblanks_qty != -1:
exercises.append('fillBlanks')
exercise_qty_q.put(exercise_fillblanks_qty)
total_qty = total_qty + exercise_fillblanks_qty
if exercise_writeblanks_qty != -1:
exercises.append('writeBlanks')
exercise_qty_q.put(exercise_writeblanks_qty)
total_qty = total_qty + exercise_writeblanks_qty
if exercise_truefalse_qty != -1:
exercises.append('trueFalse')
exercise_qty_q.put(exercise_truefalse_qty)
total_qty = total_qty + exercise_truefalse_qty
if exercise_paragraphmatch_qty != -1:
exercises.append('paragraphMatch')
exercise_qty_q.put(exercise_paragraphmatch_qty)
total_qty = total_qty + exercise_paragraphmatch_qty
response["exercises"][f"exercise_{i}"] = await self._reading.gen_reading_passage(
2, exercise_topic, exercises, exercise_qty_q, exercise_difficulty, exercise_id
)
response["exercises"][f"exercise_{i}"]["type"] = "reading"
exercise_id = exercise_id + total_qty
elif exercise_type == CustomLevelExerciseTypes.READING_3.value:
exercises = []
exercise_qty_q = queue.Queue()
total_qty = 0
if exercise_fillblanks_qty != -1:
exercises.append('fillBlanks')
exercise_qty_q.put(exercise_fillblanks_qty)
total_qty = total_qty + exercise_fillblanks_qty
if exercise_writeblanks_qty != -1:
exercises.append('writeBlanks')
exercise_qty_q.put(exercise_writeblanks_qty)
total_qty = total_qty + exercise_writeblanks_qty
if exercise_truefalse_qty != -1:
exercises.append('trueFalse')
exercise_qty_q.put(exercise_truefalse_qty)
total_qty = total_qty + exercise_truefalse_qty
if exercise_paragraphmatch_qty != -1:
exercises.append('paragraphMatch')
exercise_qty_q.put(exercise_paragraphmatch_qty)
total_qty = total_qty + exercise_paragraphmatch_qty
if exercise_ideamatch_qty != -1:
exercises.append('ideaMatch')
exercise_qty_q.put(exercise_ideamatch_qty)
total_qty = total_qty + exercise_ideamatch_qty
response["exercises"][f"exercise_{i}"] = await self._reading.gen_reading_passage(
3, exercise_topic, exercises, exercise_qty_q, exercise_id, exercise_difficulty
)
response["exercises"][f"exercise_{i}"]["type"] = "reading"
exercise_id = exercise_id + total_qty
elif exercise_type == CustomLevelExerciseTypes.LISTENING_1.value:
exercises = []
exercise_qty_q = queue.Queue()
total_qty = 0
if exercise_mc_qty != -1:
exercises.append('multipleChoice')
exercise_qty_q.put(exercise_mc_qty)
total_qty = total_qty + exercise_mc_qty
if exercise_writeblanksquestions_qty != -1:
exercises.append('writeBlanksQuestions')
exercise_qty_q.put(exercise_writeblanksquestions_qty)
total_qty = total_qty + exercise_writeblanksquestions_qty
if exercise_writeblanksfill_qty != -1:
exercises.append('writeBlanksFill')
exercise_qty_q.put(exercise_writeblanksfill_qty)
total_qty = total_qty + exercise_writeblanksfill_qty
if exercise_writeblanksform_qty != -1:
exercises.append('writeBlanksForm')
exercise_qty_q.put(exercise_writeblanksform_qty)
total_qty = total_qty + exercise_writeblanksform_qty
response["exercises"][f"exercise_{i}"] = await self._listening.get_listening_question(
1, exercise_topic, exercises, exercise_difficulty, exercise_qty_q, exercise_id
)
response["exercises"][f"exercise_{i}"]["type"] = "listening"
exercise_id = exercise_id + total_qty
elif exercise_type == CustomLevelExerciseTypes.LISTENING_2.value:
exercises = []
exercise_qty_q = queue.Queue()
total_qty = 0
if exercise_mc_qty != -1:
exercises.append('multipleChoice')
exercise_qty_q.put(exercise_mc_qty)
total_qty = total_qty + exercise_mc_qty
if exercise_writeblanksquestions_qty != -1:
exercises.append('writeBlanksQuestions')
exercise_qty_q.put(exercise_writeblanksquestions_qty)
total_qty = total_qty + exercise_writeblanksquestions_qty
response["exercises"][f"exercise_{i}"] = await self._listening.get_listening_question(
2, exercise_topic, exercises, exercise_difficulty, exercise_qty_q, exercise_id
)
response["exercises"][f"exercise_{i}"]["type"] = "listening"
exercise_id = exercise_id + total_qty
elif exercise_type == CustomLevelExerciseTypes.LISTENING_3.value:
exercises = []
exercise_qty_q = queue.Queue()
total_qty = 0
if exercise_mc3_qty != -1:
exercises.append('multipleChoice3Options')
exercise_qty_q.put(exercise_mc3_qty)
total_qty = total_qty + exercise_mc3_qty
if exercise_writeblanksquestions_qty != -1:
exercises.append('writeBlanksQuestions')
exercise_qty_q.put(exercise_writeblanksquestions_qty)
total_qty = total_qty + exercise_writeblanksquestions_qty
response["exercises"][f"exercise_{i}"] = await self._listening.get_listening_question(
3, exercise_topic, exercises, exercise_difficulty, exercise_qty_q, exercise_id
)
response["exercises"][f"exercise_{i}"]["type"] = "listening"
exercise_id = exercise_id + total_qty
elif exercise_type == CustomLevelExerciseTypes.LISTENING_4.value:
exercises = []
exercise_qty_q = queue.Queue()
total_qty = 0
if exercise_mc_qty != -1:
exercises.append('multipleChoice')
exercise_qty_q.put(exercise_mc_qty)
total_qty = total_qty + exercise_mc_qty
if exercise_writeblanksquestions_qty != -1:
exercises.append('writeBlanksQuestions')
exercise_qty_q.put(exercise_writeblanksquestions_qty)
total_qty = total_qty + exercise_writeblanksquestions_qty
if exercise_writeblanksfill_qty != -1:
exercises.append('writeBlanksFill')
exercise_qty_q.put(exercise_writeblanksfill_qty)
total_qty = total_qty + exercise_writeblanksfill_qty
if exercise_writeblanksform_qty != -1:
exercises.append('writeBlanksForm')
exercise_qty_q.put(exercise_writeblanksform_qty)
total_qty = total_qty + exercise_writeblanksform_qty
response["exercises"][f"exercise_{i}"] = await self._listening.get_listening_question(
4, exercise_topic, exercises, exercise_difficulty, exercise_qty_q, exercise_id
)
response["exercises"][f"exercise_{i}"]["type"] = "listening"
exercise_id = exercise_id + total_qty
return response

417
app/services/impl/level/level.py Normal file
View File

@@ -0,0 +1,417 @@
import json
import random
import uuid
from typing import Dict
from fastapi import UploadFile
from app.configs.constants import GPTModels, TemperatureSettings, EducationalContent
from app.helpers import ExercisesHelper
from app.repositories.abc import IDocumentStore
from app.services.abc import ILevelService, ILLMService, IReadingService, IWritingService, ISpeakingService, \
IListeningService
from .custom import CustomLevelModule
from .upload import UploadLevelModule
class LevelService(ILevelService):
def __init__(
self,
llm: ILLMService,
document_store: IDocumentStore,
mc_variants: Dict,
reading_service: IReadingService,
writing_service: IWritingService,
speaking_service: ISpeakingService,
listening_service: IListeningService
):
self._llm = llm
self._document_store = document_store
self._reading_service = reading_service
self._custom_module = CustomLevelModule(
llm, self, reading_service, listening_service, writing_service, speaking_service
)
self._upload_module = UploadLevelModule(llm)
# TODO: normal and blank spaces only differ on "multiple choice blank space questions" in the prompt
# mc_variants are stored in ./mc_variants.json
self._mc_variants = mc_variants
async def upload_level(self, upload: UploadFile) -> Dict:
return await self._upload_module.generate_level_from_file(upload)
async def get_custom_level(self, data: Dict):
return await self._custom_module.get_custom_level(data)
async def get_level_exam(
self, number_of_exercises: int = 25, min_timer: int = 25, diagnostic: bool = False
) -> Dict:
exercises = await self.gen_multiple_choice("normal", number_of_exercises, utas=False)
return {
"exercises": [exercises],
"isDiagnostic": diagnostic,
"minTimer": min_timer,
"module": "level"
}
async def get_level_utas(self, diagnostic: bool = False, min_timer: int = 25):
# Formats
mc = {
"id": str(uuid.uuid4()),
"prompt": "Choose the correct word or group of words that completes the sentences.",
"questions": None,
"type": "multipleChoice",
"part": 1
}
umc = {
"id": str(uuid.uuid4()),
"prompt": "Choose the underlined word or group of words that is not correct.",
"questions": None,
"type": "multipleChoice",
"part": 2
}
bs_1 = {
"id": str(uuid.uuid4()),
"prompt": "Read the text and write the correct word for each space.",
"questions": None,
"type": "blankSpaceText",
"part": 3
}
bs_2 = {
"id": str(uuid.uuid4()),
"prompt": "Read the text and write the correct word for each space.",
"questions": None,
"type": "blankSpaceText",
"part": 4
}
reading = {
"id": str(uuid.uuid4()),
"prompt": "Read the text and answer the questions below.",
"questions": None,
"type": "readingExercises",
"part": 5
}
all_mc_questions = []
# PART 1
# await self._gen_multiple_choice("normal", number_of_exercises, utas=False)
mc_exercises1 = await self.gen_multiple_choice(
"blank_space", 15, 1, utas=True, all_exams=all_mc_questions
)
print(json.dumps(mc_exercises1, indent=4))
all_mc_questions.append(mc_exercises1)
# PART 2
mc_exercises2 = await self.gen_multiple_choice(
"blank_space", 15, 16, utas=True, all_exams=all_mc_questions
)
print(json.dumps(mc_exercises2, indent=4))
all_mc_questions.append(mc_exercises2)
# PART 3
mc_exercises3 = await self.gen_multiple_choice(
"blank_space", 15, 31, utas=True, all_exams=all_mc_questions
)
print(json.dumps(mc_exercises3, indent=4))
all_mc_questions.append(mc_exercises3)
mc_exercises = mc_exercises1['questions'] + mc_exercises2['questions'] + mc_exercises3['questions']
print(json.dumps(mc_exercises, indent=4))
mc["questions"] = mc_exercises
# Underlined mc
underlined_mc = await self.gen_multiple_choice(
"underline", 15, 46, utas=True, all_exams=all_mc_questions
)
print(json.dumps(underlined_mc, indent=4))
umc["questions"] = underlined_mc
# Blank Space text 1
blank_space_text_1 = await self.gen_blank_space_text_utas(12, 61, 250)
print(json.dumps(blank_space_text_1, indent=4))
bs_1["questions"] = blank_space_text_1
# Blank Space text 2
blank_space_text_2 = await self.gen_blank_space_text_utas(14, 73, 350)
print(json.dumps(blank_space_text_2, indent=4))
bs_2["questions"] = blank_space_text_2
# Reading text
reading_text = await self.gen_reading_passage_utas(87, 10, 4)
print(json.dumps(reading_text, indent=4))
reading["questions"] = reading_text
return {
"exercises": {
"blankSpaceMultipleChoice": mc,
"underlinedMultipleChoice": umc,
"blankSpaceText1": bs_1,
"blankSpaceText2": bs_2,
"readingExercises": reading,
},
"isDiagnostic": diagnostic,
"minTimer": min_timer,
"module": "level"
}
async def gen_multiple_choice(
self, mc_variant: str, quantity: int, start_id: int = 1, *, utas: bool = False, all_exams=None
):
mc_template = self._mc_variants[mc_variant]
blank_mod = " blank space " if mc_variant == "blank_space" else " "
gen_multiple_choice_for_text: str = (
'Generate {quantity} multiple choice{blank}questions of 4 options for an english level exam, some easy '
'questions, some intermediate questions and some advanced questions. Ensure that the questions cover '
'a range of topics such as verb tense, subject-verb agreement, pronoun usage, sentence structure, and '
'punctuation. Make sure every question only has 1 correct answer.'
)
messages = [
{
"role": "system",
"content": (
f'You are a helpful assistant designed to output JSON on this format: {mc_template}'
)
},
{
"role": "user",
"content": gen_multiple_choice_for_text.format(quantity=str(quantity), blank=blank_mod)
}
]
if mc_variant == "underline":
messages.append({
"role": "user",
"content": (
'The type of multiple choice in the prompt has wrong words or group of words and the options '
'are to find the wrong word or group of words that are underlined in the prompt. \nExample:\n'
'Prompt: "I <u>complain</u> about my boss <u>all the time</u>, but my colleagues <u>thinks</u> '
'the boss <u>is</u> nice."\n'
'Options:\na: "complain"\nb: "all the time"\nc: "thinks"\nd: "is"'
)
})
question = await self._llm.prediction(
GPTModels.GPT_4_O, messages, ["questions"], TemperatureSettings.GEN_QUESTION_TEMPERATURE
)
if len(question["questions"]) != quantity:
return await self.gen_multiple_choice(mc_variant, quantity, start_id, utas=utas, all_exams=all_exams)
else:
if not utas:
all_exams = await self._document_store.get_all("level")
seen_keys = set()
for i in range(len(question["questions"])):
question["questions"][i], seen_keys = await self._replace_exercise_if_exists(
all_exams, question["questions"][i], question, seen_keys, mc_variant, utas
)
return {
"id": str(uuid.uuid4()),
"prompt": "Select the appropriate option.",
"questions": ExercisesHelper.fix_exercise_ids(question, start_id)["questions"],
"type": "multipleChoice",
}
else:
if all_exams is not None:
seen_keys = set()
for i in range(len(question["questions"])):
question["questions"][i], seen_keys = await self._replace_exercise_if_exists(
all_exams, question["questions"][i], question, seen_keys, mc_variant, utas
)
response = ExercisesHelper.fix_exercise_ids(question, start_id)
response["questions"] = ExercisesHelper.randomize_mc_options_order(response["questions"])
return response
async def _generate_single_multiple_choice(self, mc_variant: str = "normal"):
mc_template = self._mc_variants[mc_variant]["questions"][0]
blank_mod = " blank space " if mc_variant == "blank_space" else " "
messages = [
{
"role": "system",
"content": (
f'You are a helpful assistant designed to output JSON on this format: {mc_template}'
)
},
{
"role": "user",
"content": (
f'Generate 1 multiple choice {blank_mod} question of 4 options for an english level exam, '
f'it can be easy, intermediate or advanced.'
)
}
]
if mc_variant == "underline":
messages.append({
"role": "user",
"content": (
'The type of multiple choice in the prompt has wrong words or group of words and the options '
'are to find the wrong word or group of words that are underlined in the prompt. \nExample:\n'
'Prompt: "I <u>complain</u> about my boss <u>all the time</u>, but my colleagues <u>thinks</u> '
'the boss <u>is</u> nice."\n'
'Options:\na: "complain"\nb: "all the time"\nc: "thinks"\nd: "is"'
)
})
question = await self._llm.prediction(
GPTModels.GPT_4_O, messages, ["options"], TemperatureSettings.GEN_QUESTION_TEMPERATURE
)
return question
async def _replace_exercise_if_exists(
self, all_exams, current_exercise, current_exam, seen_keys, mc_variant: str, utas: bool = False
):
# Extracting relevant fields for comparison
key = (current_exercise['prompt'], tuple(sorted(option['text'] for option in current_exercise['options'])))
# Check if the key is in the set
if key in seen_keys:
return await self._replace_exercise_if_exists(
all_exams, await self._generate_single_multiple_choice(mc_variant), current_exam, seen_keys,
mc_variant, utas
)
else:
seen_keys.add(key)
if not utas:
for exam in all_exams:
exam_dict = exam.to_dict()
if len(exam_dict.get("parts", [])) > 0:
exercise_dict = exam_dict.get("parts", [])[0]
if len(exercise_dict.get("exercises", [])) > 0:
if any(
exercise["prompt"] == current_exercise["prompt"] and
any(exercise["options"][0]["text"] == current_option["text"] for current_option in
current_exercise["options"])
for exercise in exercise_dict.get("exercises", [])[0]["questions"]
):
return await self._replace_exercise_if_exists(
all_exams, await self._generate_single_multiple_choice(mc_variant), current_exam,
seen_keys, mc_variant, utas
)
else:
for exam in all_exams:
if any(
exercise["prompt"] == current_exercise["prompt"] and
any(exercise["options"][0]["text"] == current_option["text"] for current_option in
current_exercise["options"])
for exercise in exam.get("questions", [])
):
return await self._replace_exercise_if_exists(
all_exams, await self._generate_single_multiple_choice(mc_variant), current_exam,
seen_keys, mc_variant, utas
)
return current_exercise, seen_keys
async def gen_blank_space_text_utas(
self, quantity: int, start_id: int, size: int, topic=random.choice(EducationalContent.MTI_TOPICS)
):
json_template = self._mc_variants["blank_space_text"]
messages = [
{
"role": "system",
"content": f'You are a helpful assistant designed to output JSON on this format: {json_template}'
},
{
"role": "user",
"content": f'Generate a text of at least {size} words about the topic {topic}.'
},
{
"role": "user",
"content": (
f'From the generated text choose {quantity} words (cannot be sequential words) to replace '
'once with {{id}} where id starts on ' + str(start_id) + ' and is incremented for each word. '
'The ids must be ordered throughout the text and the words must be replaced only once. '
'Put the removed words and respective ids on the words array of the json in the correct order.'
)
}
]
question = await self._llm.prediction(
GPTModels.GPT_4_O, messages, ["question"], TemperatureSettings.GEN_QUESTION_TEMPERATURE
)
return question["question"]
async def gen_reading_passage_utas(
self, start_id, sa_quantity: int, mc_quantity: int, topic=random.choice(EducationalContent.MTI_TOPICS)
):
passage = await self._reading_service.generate_reading_passage(1, topic)
short_answer = await self._gen_short_answer_utas(passage["text"], start_id, sa_quantity)
mc_exercises = await self._gen_text_multiple_choice_utas(passage["text"], start_id + sa_quantity, mc_quantity)
return {
"exercises": {
"shortAnswer": short_answer,
"multipleChoice": mc_exercises,
},
"text": {
"content": passage["text"],
"title": passage["title"]
}
}
async def _gen_short_answer_utas(self, text: str, start_id: int, sa_quantity: int):
json_format = {"questions": [{"id": 1, "question": "question", "possible_answers": ["answer_1", "answer_2"]}]}
messages = [
{
"role": "system",
"content": f'You are a helpful assistant designed to output JSON on this format: {json_format}'
},
{
"role": "user",
"content": (
f'Generate {sa_quantity} short answer questions, and the possible answers, must have '
f'maximum 3 words per answer, about this text:\n"{text}"'
)
},
{
"role": "user",
"content": f'The id starts at {start_id}.'
}
]
question = await self._llm.prediction(
GPTModels.GPT_4_O, messages, ["questions"], TemperatureSettings.GEN_QUESTION_TEMPERATURE
)
return question["questions"]
async def _gen_text_multiple_choice_utas(self, text: str, start_id: int, mc_quantity: int):
json_template = self._mc_variants["text_mc_utas"]
messages = [
{
"role": "system",
"content": f'You are a helpful assistant designed to output JSON on this format: {json_template}'
},
{
"role": "user",
"content": f'Generate {mc_quantity} multiple choice questions of 4 options for this text:\n{text}'
},
{
"role": "user",
"content": 'Make sure every question only has 1 correct answer.'
}
]
question = await self._llm.prediction(
GPTModels.GPT_4_O, messages, ["questions"], TemperatureSettings.GEN_QUESTION_TEMPERATURE
)
if len(question["questions"]) != mc_quantity:
return await self._gen_text_multiple_choice_utas(text, mc_quantity, start_id)
else:
response = ExercisesHelper.fix_exercise_ids(question, start_id)
response["questions"] = ExercisesHelper.randomize_mc_options_order(response["questions"])
return response

137
app/services/impl/level/mc_variants.json Normal file
View File

@@ -0,0 +1,137 @@
{
"normal": {
"questions": [
{
"id": "9",
"options": [
{
"id": "A",
"text": "And"
},
{
"id": "B",
"text": "Cat"
},
{
"id": "C",
"text": "Happy"
},
{
"id": "D",
"text": "Jump"
}
],
"prompt": "Which of the following is a conjunction?",
"solution": "A",
"variant": "text"
}
]
},
"blank_space": {
"questions": [
{
"id": "9",
"options": [
{
"id": "A",
"text": "And"
},
{
"id": "B",
"text": "Cat"
},
{
"id": "C",
"text": "Happy"
},
{
"id": "D",
"text": "Jump"
}
],
"prompt": "Which of the following is a conjunction?",
"solution": "A",
"variant": "text"
}
]
},
"underline": {
"questions": [
{
"id": "9",
"options": [
{
"id": "A",
"text": "a"
},
{
"id": "B",
"text": "b"
},
{
"id": "C",
"text": "c"
},
{
"id": "D",
"text": "d"
}
],
"prompt": "prompt",
"solution": "A",
"variant": "text"
}
]
},
"blank_space_text": {
"question": {
"words": [
{
"id": "1",
"text": "a"
},
{
"id": "2",
"text": "b"
},
{
"id": "3",
"text": "c"
},
{
"id": "4",
"text": "d"
}
],
"text": "text"
}
},
"text_mc_utas": {
"questions": [
{
"id": "9",
"options": [
{
"id": "A",
"text": "a"
},
{
"id": "B",
"text": "b"
},
{
"id": "C",
"text": "c"
},
{
"id": "D",
"text": "d"
}
],
"prompt": "prompt",
"solution": "A",
"variant": "text"
}
]
}
}

404
app/services/impl/level/upload.py Normal file
View File

@@ -0,0 +1,404 @@
import aiofiles
import os
import uuid
from logging import getLogger
from typing import Dict, Any, Tuple, Coroutine
import pdfplumber
from fastapi import UploadFile
from app.services.abc import ILLMService
from app.helpers import LoggerHelper, FileHelper
from app.mappers import ExamMapper
from app.dtos.exam import Exam
from app.dtos.sheet import Sheet
class UploadLevelModule:
def __init__(self, openai: ILLMService):
self._logger = getLogger(__name__)
self._llm = openai
# TODO: create a doc in firestore with a status and get its id, run this in a thread and modify the doc in
# firestore, return the id right away, in generation view poll for the id
async def generate_level_from_file(self, file: UploadFile) -> Dict[str, Any] | None:
ext, path_id = await self._save_upload(file)
FileHelper.convert_file_to_pdf(
f'./tmp/{path_id}/uploaded.{ext}', f'./tmp/{path_id}/exercises.pdf'
)
file_has_images = self._check_pdf_for_images(f'./tmp/{path_id}/exercises.pdf')
if not file_has_images:
FileHelper.convert_file_to_html(f'./tmp/{path_id}/uploaded.{ext}', f'./tmp/{path_id}/exercises.html')
completion: Coroutine[Any, Any, Exam] = (
self._png_completion(path_id) if file_has_images else self._html_completion(path_id)
)
response = await completion
FileHelper.remove_directory(f'./tmp/{path_id}')
if response:
return self.fix_ids(response.dict(exclude_none=True))
return None
@staticmethod
@LoggerHelper.suppress_loggers()
def _check_pdf_for_images(pdf_path: str) -> bool:
with pdfplumber.open(pdf_path) as pdf:
for page in pdf.pages:
if page.images:
return True
return False
@staticmethod
async def _save_upload(file: UploadFile) -> Tuple[str, str]:
ext = file.filename.split('.')[-1]
path_id = str(uuid.uuid4())
os.makedirs(f'./tmp/{path_id}', exist_ok=True)
tmp_filename = f'./tmp/{path_id}/uploaded.{ext}'
file_bytes: bytes = await file.read()
async with aiofiles.open(tmp_filename, 'wb') as file:
await file.write(file_bytes)
return ext, path_id
def _level_json_schema(self):
return {
"parts": [
{
"context": "<this attribute is optional you may exclude it if not required>",
"exercises": [
self._multiple_choice_html(),
self._passage_blank_space_html()
]
}
]
}
async def _html_completion(self, path_id: str) -> Exam:
async with aiofiles.open(f'./tmp/{path_id}/exercises.html', 'r', encoding='utf-8') as f:
html = await f.read()
return await self._llm.pydantic_prediction(
[self._gpt_instructions_html(),
{
"role": "user",
"content": html
}
],
ExamMapper.map_to_exam_model,
str(self._level_json_schema())
)
def _gpt_instructions_html(self):
return {
"role": "system",
"content": (
'You are GPT Scraper and your job is to clean dirty html into clean usable JSON formatted data.'
'Your current task is to scrape html english questions sheets.\n\n'
'In the question sheet you will only see 4 types of question:\n'
'- blank space multiple choice\n'
'- underline multiple choice\n'
'- reading passage blank space multiple choice\n'
'- reading passage multiple choice\n\n'
'For the first two types of questions the template is the same but the question prompts differ, '
'whilst in the blank space multiple choice you must include in the prompt the blank spaces with '
'multiple "_", in the underline you must include in the prompt the <u></u> to '
'indicate the underline and the options a, b, c, d must be the ordered underlines in the prompt.\n\n'
'For the reading passage exercise you must handle the formatting of the passages. If it is a '
'reading passage with blank spaces you will see blanks represented with (question id) followed by a '
'line and your job is to replace the brackets with the question id and line with "{{question id}}" '
'with 2 newlines between paragraphs. For the reading passages without blanks you must remove '
'any numbers that may be there to specify paragraph numbers or line numbers, and place 2 newlines '
'between paragraphs.\n\n'
'IMPORTANT: Note that for the reading passages, the html might not reflect the actual paragraph '
'structure, don\'t format the reading passages paragraphs only by the <p></p> tags, try to figure '
'out the best paragraph separation possible.'
'You will place all the information in a single JSON: '
'{"parts": [{"exercises": [{...}], "context": ""}]}\n '
'Where {...} are the exercises templates for each part of a question sheet and the optional field '
'context.'
'IMPORTANT: The question sheet may be divided by sections but you need to only consider the parts, '
'so that you can group the exercises by the parts that are in the html, this is crucial since only '
'reading passage multiple choice require context and if the context is included in parts where it '
'is not required the UI will be messed up. Some make sure to correctly group the exercises by parts.\n'
'The templates for the exercises are the following:\n'
'- blank space multiple choice, underline multiple choice and reading passage multiple choice: '
f'{self._multiple_choice_html()}\n'
f'- reading passage blank space multiple choice: {self._passage_blank_space_html()}\n'
'IMPORTANT: For the reading passage multiple choice the context field must be set with the reading '
'passages without paragraphs or line numbers, with 2 newlines between paragraphs, for the other '
'exercises exclude the context field.'
)
}
@staticmethod
def _multiple_choice_html():
return {
"type": "multipleChoice",
"prompt": "Select the appropriate option.",
"questions": [
{
"id": "<the question id>",
"prompt": "<the question>",
"solution": "<the option id solution>",
"options": [
{
"id": "A",
"text": "<the a option>"
},
{
"id": "B",
"text": "<the b option>"
},
{
"id": "C",
"text": "<the c option>"
},
{
"id": "D",
"text": "<the d option>"
}
]
}
]
}
@staticmethod
def _passage_blank_space_html():
return {
"type": "fillBlanks",
"variant": "mc",
"prompt": "Click a blank to select the appropriate word for it.",
"text": (
"<The whole text for the exercise with replacements for blank spaces and their "
"ids with {{<question id>}} with 2 newlines between paragraphs>"
),
"solutions": [
{
"id": "<question id>",
"solution": "<the option that holds the solution>"
}
],
"words": [
{
"id": "<question id>",
"options": {
"A": "<a option>",
"B": "<b option>",
"C": "<c option>",
"D": "<d option>"
}
}
]
}
async def _png_completion(self, path_id: str) -> Exam:
FileHelper.pdf_to_png(path_id)
tmp_files = os.listdir(f'./tmp/{path_id}')
pages = [f for f in tmp_files if f.startswith('page-') and f.endswith('.png')]
pages.sort(key=lambda f: int(f.split('-')[1].split('.')[0]))
json_schema = {
"components": [
{"type": "part", "part": "<name or number of the part>"},
self._multiple_choice_png(),
{"type": "blanksPassage", "text": (
"<The whole text for the exercise with replacements for blank spaces and their "
"ids with {{<question id>}} with 2 newlines between paragraphs>"
)},
{"type": "passage", "context": (
"<reading passages without paragraphs or line numbers, with 2 newlines between paragraphs>"
)},
self._passage_blank_space_png()
]
}
components = []
for i in range(len(pages)):
current_page = pages[i]
next_page = pages[i + 1] if i + 1 < len(pages) else None
batch = [current_page, next_page] if next_page else [current_page]
sheet = await self._png_batch(path_id, batch, json_schema)
sheet.batch = i + 1
components.append(sheet.dict())
batches = {"batches": components}
return await self._batches_to_exam_completion(batches)
async def _png_batch(self, path_id: str, files: list[str], json_schema) -> Sheet:
return await self._llm.pydantic_prediction(
[self._gpt_instructions_png(),
{
"role": "user",
"content": [
*FileHelper.b64_pngs(path_id, files)
]
}
],
ExamMapper.map_to_sheet,
str(json_schema)
)
def _gpt_instructions_png(self):
return {
"role": "system",
"content": (
'You are GPT OCR and your job is to scan image text data and format it to JSON format.'
'Your current task is to scan english questions sheets.\n\n'
'You will place all the information in a single JSON: {"components": [{...}]} where {...} is a set of '
'sheet components you will retrieve from the images, the components and their corresponding JSON '
'templates are as follows:\n'
'- Part, a standalone part or part of a section of the question sheet: '
'{"type": "part", "part": "<name or number of the part>"}\n'
'- Multiple Choice Question, there are three types of multiple choice questions that differ on '
'the prompt field of the template: blanks, underlines and normal. '
'In the blanks prompt you must leave 5 underscores to represent the blank space. '
'In the underlines questions the objective is to pick the words that are incorrect in the given '
'sentence, for these questions you must wrap the answer to the question with the html tag <u></u>, '
'choose 3 other words to wrap in <u></u>, place them in the prompt field and use the underlined words '
'in the order they appear in the question for the options A to D, disreguard options that might be '
'included underneath the underlines question and use the ones you wrapped in <u></u>.'
'In normal you just leave the question as is. '
f'The template for multiple choice questions is the following: {self._multiple_choice_png()}.\n'
'- Reading Passages, there are two types of reading passages. Reading passages where you will see '
'blanks represented by a (question id) followed by a line, you must format these types of reading '
'passages to be only the text with the brackets that have the question id and line replaced with '
'"{{question id}}", also place 2 newlines between paragraphs. For the reading passages without blanks '
'you must remove any numbers that may be there to specify paragraph numbers or line numbers, '
'and place 2 newlines between paragraphs. '
'For the reading passages with blanks the template is: {"type": "blanksPassage", '
'"text": "<The whole text for the exercise with replacements for blank spaces and their '
'ids that are enclosed in brackets with {{<question id>}} also place 2 newlines between paragraphs>"}. '
'For the reading passage without blanks is: {"type": "passage", "context": "<reading passages without '
'paragraphs or line numbers, with 2 newlines between paragraphs>"}\n'
'- Blanks Options, options for a blanks reading passage exercise, this type of component is a group of '
'options with the question id and the options from a to d. The template is: '
f'{self._passage_blank_space_png()}\n'
'IMPORTANT: You must place the components in the order that they were given to you. If an exercise or '
'reading passages are cut off don\'t include them in the JSON.'
)
}
def _multiple_choice_png(self):
multiple_choice = self._multiple_choice_html()["questions"][0]
multiple_choice["type"] = "multipleChoice"
multiple_choice.pop("solution")
return multiple_choice
def _passage_blank_space_png(self):
passage_blank_space = self._passage_blank_space_html()["words"][0]
passage_blank_space["type"] = "fillBlanks"
return passage_blank_space
async def _batches_to_exam_completion(self, batches: Dict[str, Any]) -> Exam:
return await self._llm.pydantic_prediction(
[self._gpt_instructions_html(),
{
"role": "user",
"content": str(batches)
}
],
ExamMapper.map_to_exam_model,
str(self._level_json_schema())
)
def _gpt_instructions_batches(self):
return {
"role": "system",
"content": (
'You are helpfull assistant. Your task is to merge multiple batches of english question sheet '
'components and solve the questions. Each batch may contain overlapping content with the previous '
'batch, or close enough content which needs to be excluded. The components are as follows:'
'- Part, a standalone part or part of a section of the question sheet: '
'{"type": "part", "part": "<name or number of the part>"}\n'
'- Multiple Choice Question, there are three types of multiple choice questions that differ on '
'the prompt field of the template: blanks, underlines and normal. '
'In a blanks question, the prompt has underscores to represent the blank space, you must select the '
'appropriate option to solve it.'
'In a underlines question, the prompt has 4 underlines represented by the html tags <u></u>, you must '
'select the option that makes the prompt incorrect to solve it. If the options order doesn\'t reflect '
'the order in which the underlines appear in the prompt you will need to fix it.'
'In a normal question there isn\'t either blanks or underlines in the prompt, you should just '
'select the appropriate solution.'
f'The template for these questions is the same: {self._multiple_choice_png()}\n'
'- Reading Passages, there are two types of reading passages with different templates. The one with '
'type "blanksPassage" where the text field holds the passage and a blank is represented by '
'{{<some number>}} and the other one with type "passage" that has the context field with just '
'reading passages. For both of these components you will have to remove any additional data that might '
'be related to a question description and also remove some "(<question id>)" and "_" from blanksPassage'
' if there are any. These components are used in conjunction with other ones.'
'- Blanks Options, options for a blanks reading passage exercise, this type of component is a group of '
'options with the question id and the options from a to d. The template is: '
f'{self._passage_blank_space_png()}\n\n'
'Now that you know the possible components here\'s what I want you to do:\n'
'1. Remove duplicates. A batch will have duplicates of other batches and the components of '
'the next batch should always take precedence over the previous one batch, what I mean by this is that '
'if batch 1 has, for example, multiple choice question with id 10 and the next one also has id 10, '
'you pick the next one.\n'
'2. Solve the exercises. There are 4 types of exercises, the 3 multipleChoice variants + a fill blanks '
'exercise. For the multiple choice question follow the previous instruction to solve them and place '
f'them in this format: {self._multiple_choice_html()}. For the fill blanks exercises you need to match '
'the correct blanksPassage to the correct fillBlanks options and then pick the correct option. Here is '
f'the template for this exercise: {self._passage_blank_space_html()}.\n'
f'3. Restructure the JSON to match this template: {self._level_json_schema()}. '
f'You must group the exercises by the parts in the order they appear in the batches components. '
f'The context field of a part is the context of a passage component that has text relevant to normal '
f'multiple choice questions.\n'
'Do your utmost to fullfill the requisites, make sure you include all non-duplicate questions'
'in your response and correctly structure the JSON.'
)
}
@staticmethod
def fix_ids(response):
counter = 1
for part in response["parts"]:
for exercise in part["exercises"]:
if exercise["type"] == "multipleChoice":
for question in exercise["questions"]:
question["id"] = counter
counter += 1
if exercise["type"] == "fillBlanks":
for i in range(len(exercise["words"])):
exercise["words"][i]["id"] = counter
exercise["solutions"][i]["id"] = counter
counter += 1
return response