Stopped working 2 days ago with "Oops: Could not find any chords for 'coldplay'."

[makeitcount89]

Hey thanks again for maintaining freetar!
Just noticed freetar.de hasn't been working last few days. Anything I search comes up with:
"Oops: Could not find any chords for 'coldplay'." - for anything I search. Maybe UG has updated their site recently and broken freetar?
Cheers

[kmille]

Are you using freetar.de or your own instance? For me the captcha is totally broken on freetar.de

…

On February 4, 2026 12:10:01 PM GMT+01:00, makeitcount89 ***@***.***> wrote: makeitcount89 created an issue (kmille/freetar#77) Hey thanks again for maintaining freetar! Just noticed it hasn't been working last few days. Anything I search comes up with: "Oops: Could not find any chords for 'coldplay'." - for anything I search. Maybe UG has updated their site recently and broken freetar? Cheers -- Reply to this email directly or view it on GitHub: #77 You are receiving this because you are subscribed to this thread. Message ID: ***@***.***>

[makeitcount89]

Own instance works. Just the freetar.de stopped working. Previously if I entered the captcha could still look up songs but now showing the Oops error... maybe they flagged the ip address for freetar.de?

[kmille]

yea can reproduce... I will take a look the next days. Looks like only the search is broken... But maybe it's the caching that serves the favs...

[MrSprings09]

Can relate. seems all instances (except using your own) have a broken search. Looking up anything will result in "Cannot find chords"

[kmille]

Really sad. But don't have time right now to investigate

…

On February 6, 2026 6:08:27 PM GMT+01:00, MrSprings09 ***@***.***> wrote: MrSprings09 left a comment (kmille/freetar#77) Can relate. seems all instances (except using your own) have a broken search. Looking up anything will result in "Cannot find chords" -- Reply to this email directly or view it on GitHub: #77 (comment) You are receiving this because you commented. Message ID: ***@***.***>

[jpaskaruk]

Hey there, I'm running my own instance as well, via docker - it is also not working, after a fresh pull.

I have poetry installed on the same system and will try a fresh pull of the repo and running that way when I'm back from my morning school bus run.

I also owe a few of you a promised PR to split the screen into columns, which I never got to putting together, and I'll try to get that in too, assuming we haven't just been locked out for good.

[jpaskaruk]

Is there an easy way to get verbose logging in the terminal? I'm handy with Python but I've only worked on stuff like plugins and verification modules for tests, and a few very simple scripts for myself. I own a Pycharm Pro fallback license so I can probably work it out if I get the time, but time is at a premium here as well.

[birdcolour]

I've very quickly taken a look at this, and I'll add this extremely unhelpful piece of data: I can reliably get 200s if I run locally via poetry run, but if I run via docker compose I reliably get 403s. Weird! The 403s contain a Enable JavaScript and cookies to continue message. I can take a closer look later this weekend...

[birdcolour]

Downgrading the search request here from https to http seems to work both inside and outside of docker containers, for me at least.

It feels like there's some silly cloudflare shenanigans blocking the https+docker combo, despite the fact that the raw HTTP request bytes are being generated from the same code. I'll leave detecting exactly what that is to someone better versed with wireshark than I am 😅

[DangerMouse-rgb]

Downgrading the search from https to http didn't work for me as a pip install.

I just wanted to play guitar this morning without all the clutter. I'm by no means a python expert so I had Google Gemini step me through some debugging and it suggested using cloudscraper to overcome the cloudflare 403 forbidden errors it found in the debugging.

The resultant working ug.py it gave me is here main...DangerMouse-rgb:freetar:patch-2 or below, please correct me if this not allowed (first git hub commit, no idea what I'm doing):

import cloudscraper # Added for Cloudflare bypass
from bs4 import BeautifulSoup
from urllib.parse import quote, urlparse
import json
import re

from dataclasses import dataclass, field
from .utils import FreetarError

Updated to a more modern browser string

USER_AGENT = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/121.0.0.0 Safari/537.36"

Create a global scraper session to reuse cookies/connection

scraper = cloudscraper.create_scraper()

@DataClass
class SearchResult:
artist_name: str
song_name: str
tab_url: str
artist_url: str
_type: str
version: int
votes: int
rating: float

def __init__(self, data: dict):
    self.artist_name = data["artist_name"]
    self.song_name = data["song_name"]
    self.tab_url = urlparse(data["tab_url"]).path
    self.artist_url = data["artist_url"]
    self._type = data["type"]
    self.version = data["version"]
    self.votes = int(data["votes"])
    self.rating = round(data["rating"], 1)

def __repr__(self):
    return f"{self.artist_name} - {self.song_name} (ver {self.version}) ({self._type} {self.rating}/5 - {self.votes} votes)"

@DataClass
class SongDetail:
tab: str
artist_name: str
song_name: str
version: int
difficulty: str
capo: str
tuning: str
tab_url: str
alternatives: list[SearchResult] = field(default_factory=list)

def __init__(self, data: dict):
    page_data = data["store"]["page"]["data"]
    self.tab = page_data["tab_view"]["wiki_tab"]["content"]
    self.artist_name = page_data["tab"]['artist_name']
    self.song_name = page_data["tab"]["song_name"]
    self.version = int(page_data["tab"]["version"])
    self._type = page_data["tab"]["type"]
    self.rating = int(page_data["tab"]["rating"])
    self.difficulty = page_data["tab_view"]["ug_difficulty"]
    self.appliciture = page_data["tab_view"]["applicature"]
    self.chords = []
    self.fingers_for_strings = []
    
    meta = page_data["tab_view"].get("meta")
    if isinstance(meta, dict):
        self.capo = meta.get("capo")
        _tuning = meta.get("tuning")
        self.tuning = f"{_tuning['value']} ({_tuning['name']})" if _tuning else None
    else:
        self.capo = None
        self.tuning = None
        
    self.tab_url = page_data["tab"]["tab_url"]
    self.alternatives = [
        SearchResult(alt) for alt in page_data["tab_view"]["versions"] 
        if alt.get("type", "") != "Official"
    ]
    self.fix_tab()

def fix_tab(self):
    tab = self.tab.replace("\r\n", "<br/>").replace("\n", "<br/>").replace(" ", "&nbsp;")
    tab = tab.replace("[tab]", "").replace("[/tab]", "")
    tab = re.sub(r'\[ch\](?P<root>[A-Ha-h](#|b)?)(?P<quality>[^[/]+)?(?P<bass>/[A-Ha-h](#|b)?)?\[\/ch\]', self.parse_chord, tab)
    self.tab = tab

def parse_chord(self, chord):
    root = f'<span class="chord-root">{chord.group("root")}</span>'
    quality = f'<span class="chord-quality">{chord.group("quality")}</span>' if chord.group('quality') else ''
    bass = f'/<span class="chord-bass">{chord.group("bass")[1:]}</span>' if chord.group('bass') else ''
    return f'<span class="chord fw-bold">{root + quality + bass}</span>'

@DataClass
class Search:
results: list
total_pages: int
current_page: int

def __init__(self, value: str, page: int):
    try:
        url = f"https://www.ultimate-guitar.com/search.php?page={page}&search_type=title&value={quote(value)}"
        # Use scraper instead of requests
        resp = scraper.get(url, headers={'User-Agent': USER_AGENT}, timeout=15)
        
        with open('/tmp/ug_debug.html', 'w') as f:
            f.write(resp.text)
        
        resp.raise_for_status()
        bs = BeautifulSoup(resp.text, 'html.parser')
        store_div = bs.find("div", {"class": "js-store"})
        
        if not store_div:
            raise FreetarError("Could not find the data store on the page. Check /tmp/ug_debug.html")

        data = json.loads(store_div.attrs['data-content'])
        self.results = self.get_results(data)
        
        pagination = data['store']['page']['data'].get('pagination', {})
        self.total_pages = pagination.get('total', 1)
        self.current_page = pagination.get('current', 1)
        
    except Exception as e:
        raise FreetarError(f"Search failed: {str(e)}")

def get_results(self, data: dict):
    results = data['store']['page']['data'].get('results', [])
    return [SearchResult(r) for r in results if r.get("type") not in ("Pro", "Official", None)]

def get_chords(s: SongDetail) -> SongDetail:
if s.appliciture is None:
return dict(), dict()

chords = {}
fingerings = {}

for chord in s.appliciture:
    for chord_variant in s.appliciture[chord]:
        frets = chord_variant["frets"]
        min_fret = min(frets)
        max_fret = max(frets)
        possible_frets = list(range(min_fret, max_fret+1))
        variants_temp = {
            possible_fret: [1 if b == possible_fret else 0 for b in frets][::-1]
            for possible_fret
            in possible_frets
            if possible_fret > 0
        }

        variants = dict()
        found = False
        for fret, fingers in variants_temp.items():
            try:
                if not found and fingers.index(1) >= 0:
                    found = True
            except ValueError:
                ...

            if found:
                variants[fret] = fingers

        if not len(variants):
            continue
        while len(variants) < 6:
            variants[max(variants) + 1] = [0] * 6

        variant_strings_pressed = [*variants.values()]
        variant_strings_pressed = [sum(x) for x in zip(*variant_strings_pressed)]
        unstrummed_strings = [int(not bool(y)) for y in variant_strings_pressed]

        fingering_for_variant = []
        for finger, x in zip(chord_variant["fingers"][::-1], unstrummed_strings):
            fingering_for_variant.append("x" if x else finger)
        fingering_for_variant = fingering_for_variant

        if chord not in chords:
            chords[chord] = []
            fingerings[chord] = []
        chords[chord].append(variants)
        fingerings[chord].append(fingering_for_variant)

return chords, fingerings

def ug_tab(url_path: str):
try:
url = "https://tabs.ultimate-guitar.com/tab/" + url_path
resp = scraper.get(url, headers={'User-Agent': USER_AGENT}, timeout=15)
resp.raise_for_status()

    bs = BeautifulSoup(resp.text, 'html.parser')
    data_div = bs.find("div", {"class": "js-store"})
    data = json.loads(data_div.attrs['data-content'])
    
    s = SongDetail(data)
    
    # This is the missing link that populates the chord diagrams
    s.chords, s.fingers_for_strings = get_chords(s)
    
    return s
except Exception as e:
    raise FreetarError(f"Could not parse chord: {e}")

[kmille]

I looked a bit into it... They use Cloudflare in front of their web server and it seems like UG tightend their configuration which means that 403s ("you are blocked") errors come way earlier. I had a way to bypass this in the past by using a lot of different ip addresses, but this seems also not working anymore. Right now I just have the following ideas. Please add your ideas:

• find real UG ip addresses, bypass Cloudflare by modifying DNS
• use a python library for sendingn requests that uses your browser's cookies. I have not tested this yet. This could be a local fix only. I think if you solve a Cloudflare captcha, you're fine for some time. If a 403 comes up, open the UG website and solve the captcha.
• Use cloudscraper or a captcha trader like 9kw.eu to bypass the captcha.

(I don't even know if a captcha shows up, I just see the 403 directly from CF and some html...)

[kmille]

yea

[jpaskaruk]

I do not quite understand how all the web stuff fits together, but perhaps this "flaresolvarr" could help? Seems to be designed to get around this. https://docs.ultra.cc/applications/flaresolverr

…

On Sun, Feb 15, 2026 at 8:17 AM kmille ***@***.***> wrote: *kmille* left a comment (kmille/freetar#77) <#77 (comment)> https://signal.me/#eu/PWbX41fp2S_9cPmF2LZgxSjq9_-b1H_S3HY4xsU9xHUiLDagYQVFALYUfycONAh0 — Reply to this email directly, view it on GitHub <#77 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABK4XQWRBOBPPTW4Y34NUPD4MB5W5AVCNFSM6AAAAACT6FD4CGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTSMBUGUZTCNZSGY> . You are receiving this because you commented.Message ID: ***@***.***>

[MrSprings09]

I will say though, it’s not that the entire site doesn’t work. It’s specifically searches. For example, a temp fix I’ve found, is to find the chords on ultimate guitar, copy the link, then replace the “ultimateguitar.com” with “freetar.de” or whichever instance you use.

Doing this will essentially just give you the freetar variant of the chords/tabs, so if you need to access them, try that.

Apart from that, I’m not quite sure how to fix the issue.