v3.0-dev

[AWeirdDev]

See original

I must apologize to all the people here. The original code written by me was really bad, some don't even make sense, and I'd like to deeply apologize to all contributors who take their precious time to figure all that out. (In fact, I couldn't remember how things still work after a year.)

Therefore in this update, I tried to rewrite it while keeping all the contributors' findings and suggestions.

Try it out

You can try things out in the dev branch.
To preserve the previous version, I named it fflights.

from fflights import ...

Changes

Here are the breaking changes.

Reworked source

The overall code structure has been updated. Instead of having a source file called core.py, which is a bit vague, I have split it into multiple files:

• fetcher.py - For fetching.
• parser.py - For parsing.
• querying.py - For querying (see below on "Query API").

Query API

Instead of calling it "filter" (as it overlaps with a Python utility), it's now being renamed "filter."

What's more, there are now three types of querying.

Structured

This is generally recommended, and it's pretty similar to the previous Filter API.

from fflights import FlightQuery, Passengers, query

q = query(
    flights=[
        FlightQuery(
            date="2025-12-22",
            from_airport="TPE",
            to_airport="MYJ",
        )
    ],
    seat="economy",
    trip="one-way",
    passengers=Passengers(adults=1),
    language="en-US",
    currency="",
)

Str-based

Recommended by @lookacat (ping suppressed to not disturb anyone's peace).

q = "Flights from TPE to MYJ on 2025-12-22 one way economy class"

Tuple-based

This is not really recommended, but a shortened way to write queries if you're familiar with it.

from fflights import str_query

# custom (recommended)
q = str_query((
    "Flights from TPE to MYJ on 2025-12-22 one way economy class",
    "en-US",
    "USD"
))

# simple (not really)
q = str_query((
    "TPE",  # from
    "MYJ",  # to
    "2025-12-22",  # date
    "one-way",  # trip type
    "economy",  # seat type
))

# + language & currency
q = str_query((
    "TPE",  # from
    "MYJ",  # to
    "2025-12-22",  # date
    "one-way",  # trip type
    "economy",  # seat type

    "en-US",  # language
    "USD"  # currency
))

Getting flights

Just like usual, get_flights().

from fflights import get_flights

get_flights(q)

Example result (truncated):

[JsFlights(typ='multi', airlines=['China Airlines', 'JAL'], flights=[JsSingleFlight(from_airport=Airport(name='Taiwan Taoyuan International Airport', code='TPE'), to_airport=Airport(name='Fukuoka Airport', code='FUK'), departure=SimpleDatetime(date=[2025, 12, 22], time=[16, 25]), arrival=SimpleDatetime(date=[2025, 12, 22], time=[19, 35]), duration=130, plane_type='Airbus A321neo'), JsSingleFlight(from_airport=Airport(name='Fukuoka Airport', code='FUK'), to_airport=Airport(name='Matsuyama Airport', code='MYJ'), departure=SimpleDatetime(date=[2025, 12, 23], time=[7, 50]), arrival=SimpleDatetime(date=[2025, 12, 23], time=[8, 30]), duration=40, plane_type='Embraer 170')], carbon=JsCarbonEmission(typical_on_route=146000, emission=201000)), ...]

A few notes:

1. Yes, it's typ, not a typo. Solely to avoid overlapping with Python's type.
2. It's prefixed Js because we're only using Javascript data as of now (see below on "Data source").
3. If you only want to fetch and don't want us to parse it, use fetch_flights(), which returns a str (the HTML).

Data source

It's Javascript.

Long answer: Currently, v3.0-dev only uses Javascript as the data source, as it seems to be more reliable... until Google decides to obfuscate it.

Minimum supported Python version

Bumped to Python 3.9+ as it requires several typing utilities and dataclasses.

---

That's it for this 'lil announcement. Let me know what you think (or what you hate about it, and I'll try to fix it ASAP).

Try the release candidate: https://pypi.org/project/fast-flights/3.0rc0/

[PTruscott]

I think the previous version worked pretty well, but I understand the desire to refactor although breaking changes are always sad. A couple of thoughts -

Instead of calling it "filter" (as it overlaps with a Python utility), it's now being renamed "filter."

I'm not sure what this bit is saying? The examples don't mention filter either

Yes, it's typ, not a typo. Solely to avoid overlapping with Python's type.

IMO something like trip_type or other words would be clearer and more pythonic.

[AWeirdDev]

1. There's a built-in function called filter in Python, I thought it'd be better to not overlap with it. Or perhaps we should just keep it filter for compatibility?
2. I completely agree with this. Looking at the data, it could be "multi", and also... airlines?? (Say, 7C) I'm guessing this field is about what airlines are involved? I'm uncertain.

[williamgravel]

I may be missing something, but I'm not seeing any pricing information in the get_flights response? Also I'm comparing results obtained from this project to what I would get from manually searching in Google Flights and I'm not seeing the same results, shouldn't these be 1-to-1?

[AWeirdDev]

idk maybe like today? but these changes will only be available on the dev branch, tho

[williamgravel]

That's okay, I've been working with the dev branch anyway. I've been diving into it a little bit and I've noticed a couple things which might help you, all of which from data parsed in parser.py's parse_js function:

• The pricing information seems to be located in k[1][0] as the 2nd element of a 2-element list (the first item is always null for some reason).
• I compared the outputs of data[2][0] and data[3][0] (the latter of which is the current source of information that gets iterated through with the for loop) and found a few things:
  • The first couple flights of my example search are only found in data[2][0] and not data[3][0].
  • On the other hand, data[2][0] only seems to have flights 1, 2, and 4 (in the order they appear on Google Flights). data[3][0] has every flight after and including flight 3.
  • For reference, the first 4 flights are shown in the Top flights section of the Google Flights page, while all the other 80+ flights of my query are in the Other flights section.

I can give you more details and the data I collected if this is interesting. Hope this helps!

[AWeirdDev]

noted! thanks a lot for the discovery :)

definitely goin to the todo this with this one

the current dev branch is now published on PyPi as 3.0rc0

[williamgravel]

I'm not sure what happened but it looks like there is a lot less data now being collected. When looking at the raw data collected in the parser.py file, I could see all 90+ flights for a query. Now I see less than 10 for that same query.

[AWeirdDev]

im not quite sure why..? perhaps im just stoopid
got any ideas...?