now taking into account saved details when prompting and parsing the answer of the LLM

app.py

@@ -9,14 +9,15 @@ from form.form import build_form_data_from_answers, write_pdf_form, work_categor
 from llm_manager.llm_parser import LlmParser
 from local_storage.entities import PersonalDetails, LocationDetails, ContractorDetails
 from local_storage.ls_manager import save_details, get_detail
-from prompts.prompts_manager import PromptsManager, Questions as Q
+from prompts.prompts_manager import PromptsManager
+from enums import Questions as Q
 from repository.repository import build_repo_from_environment, get_repository
 from repository import ModelRoles, Model
 from utils.parsing_utils import check_for_missing_answers
 
 user_msg = "Please describe what you need to do. To get the best results try to answer all the following questions:"
 
-find_tags_regex = re.compile(r"(@\S)")
+find_tags_regex = re.compile(r"@(\S*)")
 
 class Steps(Enum):
     INITIAL_STATE = 1
@@ -35,6 +36,7 @@ class Steps(Enum):
     def __hash__(self):
         return hash(self.value)
 
+
 class UIManager:
     def __init__(self):
         self.pm: PromptsManager = PromptsManager(work_categories=work_categories)
@@ -76,14 +78,13 @@ class UIManager:
         with st.status("parsing user input for tags"):
             tags = find_tags_regex.findall(ss["user_input"])
             details = [get_detail(t) for t in tags]
-            ss["details"] = details
         with st.status("initialising LLM"):
             self.repository.init()
         with st.status("waiting for LLM"):
             answer = self.repository.send_prompt(self.pm.verify_user_input_prompt(ss["user_input"], details))
             st.write(f"answers from LLM: {answer['content']}")
         with st.status("Checking for missing answers"):
-            answers = LlmParser.parse_verification_prompt_answers(answer['content'])
+            answers = LlmParser.parse_verification_prompt_answers(answer['content'], details)
             ss["answers"] = answers
             if len(answers) != len(Q):
                 self.set_current_step(Steps.PARSING_ERROR)

enums.py

@@ -0,0 +1,33 @@
+from enum import Enum
+
+
+class DetailsType(Enum):
+    PERSONAL_DETAILS = 1
+    LOCATION_DETAILS = 2
+    CONTRACTOR_DETAILS = 3
+
+    @classmethod
+    def values(cls):
+        return [cls.PERSONAL_DETAILS, cls.LOCATION_DETAILS, cls.CONTRACTOR_DETAILS]
+
+
+class Questions(Enum):
+    FULL_NAME = 0
+    WORK_TO_DO = 1
+    COMMUNITY = 2
+    BUILDING = 3
+    UNIT_APT_NUMBER = 4
+    OWNER_OR_TENANT = 5
+    START_DATE = 6
+    END_DATE = 7
+    CONTACT_NUMBER = 8
+    COMPANY_NAME = 9
+    COMPANY_EMAIL = 10
+    COMPANY_NUMBER = 11
+    YOUR_EMAIL = 12
+
+    @classmethod
+    def values(cls):
+        return [Questions.FULL_NAME, Questions.WORK_TO_DO, Questions.BUILDING, Questions.UNIT_APT_NUMBER,
+                Questions.OWNER_OR_TENANT, Questions.START_DATE, Questions.END_DATE, Questions.CONTACT_NUMBER,
+                Questions.COMPANY_NAME, Questions.COMPANY_EMAIL, Questions.COMPANY_NUMBER, Questions.YOUR_EMAIL]

form/form.py

@@ -2,7 +2,7 @@ from pathlib import Path
 from typing import TypedDict, Tuple
 from PyPDFForm import PdfWrapper
 
-from prompts.prompts_manager import Questions
+from enums import Questions
 from utils.date_utils import get_today_date_as_dd_mm_yyyy
 from utils.parsing_utils import find_and_parse_date, find_and_parse_phone_number
 

llm_manager/llm_parser.py

@@ -1,32 +1,41 @@
 from form.form import work_categories
-from prompts.prompts_manager import Questions
+from local_storage.entities import SavedDetails
+from enums import Questions
 
 
 class LlmParser:
 
     @classmethod
-    def parse_verification_prompt_answers(cls, llm_answer) -> dict[Questions, str | None]:
+    def parse_verification_prompt_answers(cls, llm_answer:str, details: list[SavedDetails] | None)\
+            -> dict[Questions, str | None]:
         print(f"llm answer: {llm_answer}")
         answers = {}
+        for d in details:
+            answers.update(d.to_answers())
+
+        unanswered_questions = sorted([k.value for k in Questions.values() if k not in answers])
+        unanswered_questions = iter(unanswered_questions)
         i = 0
-        question_id = 0
+        question_id = next(unanswered_questions)
         lines = [l for l in llm_answer.split("\n") if len(l.strip()) > 0 and not l.strip().endswith("?")]
         while i < len(lines):
+            if question_id is None:
+                break
             line = lines[i].strip()
             if len(line) == 0:
                 i += 1
             elif "null" in lines[i]:
                 answers[Questions(question_id)] = None
                 i += 1
-                question_id += 1
+                question_id = next(unanswered_questions, None)
             elif ":" in lines[i]:
                 answers[Questions(question_id)] = line.split(":")[1]
                 i += 1
-                question_id += 1
+                question_id = next(unanswered_questions, None)
             else:
                 answers[Questions(question_id)] = line
                 i += 1
-                question_id += 1
+                question_id = next(unanswered_questions, None)
         return answers
 
     @classmethod
@@ -35,3 +44,4 @@ class LlmParser:
         for category in category_answer.split(";"):
             categories.extend([k for k, v in work_categories.items() if category.lower() in v.lower()])
         return categories
+

local_storage/entities.py

@@ -1,15 +1,6 @@
 import abc
-from enum import Enum
 
-
-class DetailsType(Enum):
-    PERSONAL_DETAILS = 1
-    LOCATION_DETAILS = 2
-    CONTRACTOR_DETAILS = 3
-
-    @classmethod
-    def values(cls):
-        return [cls.PERSONAL_DETAILS, cls.LOCATION_DETAILS, cls.CONTRACTOR_DETAILS]
+from enums import DetailsType, Questions
 
 
 class SavedDetails(abc.ABC):
@@ -28,6 +19,9 @@ class SavedDetails(abc.ABC):
     def to_json(self):
         return {k:v for k,v in self.__dict__.items() if k not in self.excluded_fields}
 
+    def to_answers(self):
+        pass
+
 class PersonalDetails(SavedDetails):
     type_ = DetailsType.PERSONAL_DETAILS
 
@@ -37,6 +31,15 @@ class PersonalDetails(SavedDetails):
         self.email = email
         self.contact_number = contact_number
 
+    def to_answers(self) -> dict[Questions, str]:
+        answers = {}
+        if self.full_name is not None:
+            answers[Questions.FULL_NAME] = self.full_name
+        if self.email is not None:
+            answers[Questions.YOUR_EMAIL] = self.email
+        if self.contact_number is not None:
+            answers[Questions.CONTACT_NUMBER] = self.contact_number
+        return answers
 
 class LocationDetails(SavedDetails):
     type_ = DetailsType.LOCATION_DETAILS
@@ -48,6 +51,17 @@ class LocationDetails(SavedDetails):
         self.building = building
         self.unit_number = unit_number
 
+    def to_answers(self) -> dict[Questions, str]:
+        answers = {}
+        if self.owner_or_tenant is not None:
+            answers[Questions.OWNER_OR_TENANT] = self.owner_or_tenant
+        if self.building is not None:
+            answers[Questions.BUILDING] = self.building
+        if self.community is not None:
+            answers[Questions.COMMUNITY] = self.community
+        if self.unit_number is not None:
+            answers[Questions.UNIT_APT_NUMBER] = self.unit_number
+        return answers
 
 class ContractorDetails(SavedDetails):
     type_ = DetailsType.CONTRACTOR_DETAILS
@@ -57,3 +71,13 @@ class ContractorDetails(SavedDetails):
         self.contractor_name = contractor_name
         self.contractor_contact_number = contractor_contact_number
         self.contractor_email = contractor_email
+
+    def to_answers(self) -> dict[Questions, str]:
+        answers = {}
+        if self.contractor_email is not None:
+            answers[Questions.COMPANY_EMAIL] = self.contractor_email
+        if self.contractor_contact_number is not None:
+            answers[Questions.COMPANY_NUMBER] = self.contractor_contact_number
+        if self.contractor_name is not None:
+            answers[Questions.COMPANY_NAME] = self.contractor_name
+        return answers

local_storage/ls_manager.py

@@ -2,21 +2,19 @@ import json
 
 from streamlit_local_storage import LocalStorage
 
-from local_storage.entities import DetailsType, SavedDetails, PersonalDetails, LocationDetails, ContractorDetails
-
-ls = LocalStorage()
+from local_storage.entities import SavedDetails, PersonalDetails, LocationDetails, ContractorDetails
+from enums import DetailsType
 
 
 def get_detail(key: str):
     for detail_type in DetailsType.values():
-        detail = (ls.getItem(detail_type.name) or {}).get(key)
+        detail = LocalStorage(key=detail_type.name).getItem(detail_type.name).get(key)
         if detail:
-            type_ = detail["type"]
-            if type_ == DetailsType.PERSONAL_DETAILS.name:
+            if detail_type.value == DetailsType.PERSONAL_DETAILS.value:
                 return PersonalDetails(**detail)
-            elif type_ == DetailsType.LOCATION_DETAILS.name:
+            elif detail_type.value == DetailsType.LOCATION_DETAILS.value:
                 return LocationDetails(**detail)
-            elif type_ == DetailsType.CONTRACTOR_DETAILS.name:
+            elif detail_type.value == DetailsType.CONTRACTOR_DETAILS.value:
                 return ContractorDetails(**detail)
             return None
     return None

prompts/prompts_manager.py

@@ -1,26 +1,10 @@
-import datetime
-from enum import Enum
 from pathlib import Path
 
-from local_storage.entities import DetailsType, SavedDetails
+from enums import Questions
+from local_storage.entities import SavedDetails, PersonalDetails, ContractorDetails, LocationDetails
 from utils.date_utils import get_today_date_as_dd_mm_yyyy
 
 
-class Questions(Enum):
-    FULL_NAME = 0
-    WORK_TO_DO = 1
-    COMMUNITY = 2
-    BUILDING = 3
-    UNIT_APT_NUMBER = 4
-    OWNER_OR_TENANT = 5
-    START_DATE = 6
-    END_DATE = 7
-    CONTACT_NUMBER = 8
-    COMPANY_NAME = 9
-    COMPANY_EMAIL = 10
-    COMPANY_NUMBER = 11
-    YOUR_EMAIL = 12
-
 class PromptsManager:
     def __init__(self, work_categories: dict[str, str] = None):
         self.work_categories = work_categories
@@ -43,7 +27,7 @@ class PromptsManager:
             and maintain the order in which the questions are asked. Do not add any preamble:
         """
 
-        skip_questions = [self.get_questions(d) for d in exclude_questions_group]
+        skip_questions = self.get_questions(exclude_questions_group)
         questions = [q for idx, q in enumerate(self.verification_prompt_questions) if idx not in skip_questions]
         return prompt + "\n".join(questions)
 
@@ -57,17 +41,21 @@ class PromptsManager:
     def questions_to_field_labels():
         return {
             Questions.FULL_NAME: "Full name", Questions.WORK_TO_DO: "Work to do", Questions.COMMUNITY: "Community",
-            Questions.BUILDING:"Building name", Questions.UNIT_APT_NUMBER: "Unit/apartment number",
+            Questions.BUILDING: "Building name", Questions.UNIT_APT_NUMBER: "Unit/apartment number",
             Questions.OWNER_OR_TENANT: "Owner/Tenant", Questions.START_DATE: "Start date",
             Questions.END_DATE: "End date", Questions.CONTACT_NUMBER: "Your contact number",
             Questions.COMPANY_NAME: "Contractor company name", Questions.COMPANY_EMAIL: "Contracting company email",
             Questions.COMPANY_NUMBER: "Contracting company contact number", Questions.YOUR_EMAIL: "Your email"
         }
-    def get_questions(self, type_:DetailsType):
-        if type_ == DetailsType.PERSONAL_DETAILS:
-            return [Questions.FULL_NAME, Questions.CONTACT_NUMBER, Questions.YOUR_EMAIL]
-        if type_ == DetailsType.CONTRACTOR_DETAILS:
-            return [Questions.COMPANY_NAME, Questions.COMPANY_NUMBER, Questions.COMPANY_EMAIL]
-        if type_ == DetailsType.LOCATION_DETAILS:
-            return [Questions.OWNER_OR_TENANT, Questions.COMMUNITY, Questions.BUILDING, Questions.UNIT_APT_NUMBER]
 
+    @staticmethod
+    def get_questions(details: list[int]):
+        to_skip: list[int] = []
+        for d in details:
+            if isinstance(d, PersonalDetails):
+                to_skip.extend([Questions.FULL_NAME.value, Questions.CONTACT_NUMBER.value, Questions.YOUR_EMAIL.value])
+            if isinstance(d, ContractorDetails):
+                to_skip.extend([Questions.COMPANY_NAME.value, Questions.COMPANY_NUMBER.value, Questions.COMPANY_EMAIL.value])
+            if isinstance(d, LocationDetails):
+                to_skip.extend([Questions.OWNER_OR_TENANT.value, Questions.COMMUNITY.value, Questions.BUILDING.value, Questions.UNIT_APT_NUMBER.value])
+        return to_skip

ui_manager.py

@@ -7,7 +7,8 @@ from streamlit_local_storage import LocalStorage
 from form.form import build_form_data_from_answers, write_pdf_form
 from llm_manager.llm_parser import LlmParser
 from local_storage.entities import PersonalDetails, LocationDetails, ContractorDetails
-from prompts.prompts_manager import PromptsManager, Questions as Q
+from prompts.prompts_manager import PromptsManager
+from enums import Questions as Q
 from repository.repository import get_repository
 from repository import ModelRoles, Model
 from utils.parsing_utils import check_for_missing_answers