Merge pull request #6 from EJOOSTEROP/feat_hotel-info

CHANGELOG.md

@@ -9,6 +9,14 @@ and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.
 
 - Consider downgrading Python requirement (at least to 3.10, but I dont know how low it can gowhere)
 
+## [0.8.5]
+
+### Added
+- Hotel information
+
+### Changed
+- Upgrade LLM to GPT 4o mini
+
 ## [0.8.4]
 
 ### Added

pyproject.toml

@@ -18,7 +18,7 @@ disallow_incomplete_defs = true
 
 [tool.poetry]
 name = "crewcal"
-version = "0.8.4"
+version = "0.8.5"
 description = "Convert an airline crew schedule pdf into iCalendar format."
 authors = ["Erik Oosterop <[email protected]>"]
 license = "MIT"

sample.env

@@ -0,0 +1 @@
+OPENAI_API_KEY = "YOUR-OPENAI-API-KEY"

src/crewcal/llm_extract.py

@@ -3,7 +3,7 @@
 Read the schedule from a pdf file using LLM with OpenAI
 to extract flight details in a structured format.
 
-About 0.75 US cents per call in OpenAI API costs.
+As at nov 2023 about 0.75 US cents per call in OpenAI API costs.
 """
 
 import json
@@ -37,7 +37,7 @@ class OpenAISchedule:
 
     schedule_path: str
     extracted_schedule: str = ""
-    llm_model_name: str = "gpt-3.5-turbo"
+    llm_model_name: str = "gpt-4o-mini-2024-07-18"
 
     def __init__(
         self, schedule_path: str, to_json_file: str = "", to_icalendar_file: str = ""

src/crewcal/llm_prompts.py

@@ -5,7 +5,7 @@
 
 The information contains a list of duties for an airline crew member. Duties consist of one or more flights. You will represent each duty as a calendar event.
 
-Ignore the section with Hotel Information.
+The section with Hotel Information contains no new events. Instead it contains the hotel at destination for some of the flights.
 
 Each event contains the following items:
 - Departure date in yyyy-mm-dd format
@@ -18,11 +18,12 @@
 - Destination airport code
 - Destination airport name
 - Destination timezone
-- Arrival date. A '+1' means the next day.
+- Arrival date. A '+1' means the next day. Only capture yyyy-mm-dd format.
 - Arrival time, the last occurrence of a time for each day. A '+1' means the next day. Only capture HH:mm format.
 - List of crew members
-- A list of all times found.
+- A list of all times found. Only capture HH:mm format.
 - List of all airport codes.
+- Hotel information at destination. Obtain this from the document section 'Hotel Information'. It includes the name, full address and phone number of the hotel at destination. For some flights this does not exist. In that case keep this part empty.
 
 Always include all items in your output even if they are empty.
 """

src/crewcal/schedule.py

@@ -33,6 +33,7 @@ class Event(BaseModel):
     crew_list: List[str]
     list_times: List[str]
     list_airport_codes: List[str]
+    hotel_information: str
 
     def list_airport_pairs(self):
         """Generate a formatted string of airport pairs."""
@@ -55,6 +56,7 @@ def get_description(self):
                 self.list_airport_pairs(),
                 *self.crew_list,
                 f"\nFlight ends in {self.destination_timezone[-1]} timezone.",
+                self.hotel_information,
             ]
         )