Workaround for Yahoo HTTPS fingerprinting

README.md

@@ -1,14 +1,44 @@
 # Financials-Extension
 
-Version 3.3.0 includes improved cookie handling and somewhat improved logic to deal with network issues.   
-
 ## Overview
 
 This is a Python based extension for LibreOffice Calc to make market data available in Calc 
 spreadsheets - currently supporting Yahoo's (FX, crypto, equities, indices, futures, options) and Financial Times' 
 (FX, equities, indices, futures) websites using old-fashioned web scraping. 
 
-Starting with version 3.1.0, we received a contribution to get crypto data directly from Coinbase
+## Latest version vs Yahoo HTTPS fingerprinting
+
+Latest version 3.8.0 was created to bypass Yahoo's recently adding crazy HTTPS fingerprinting 
+to their website. If you don't use Yahoo, no further changes are required.
+
+If you do use Yahoo as a source, here is what I had to do to get Yahoo working again on my Ubuntu system:
+- install system-wide Python module curl_cffi - on my system as root: `pip3 install curl_cffi --upgrade`
+- download latest binary of [curl-impersonate](https://github.com/lwthiker/curl-impersonate/releases) e.g. 
+libcurl-impersonate-v0.6.1.x86_64-linux-gnu.tar.gz and untar it somewhere 
+
+Now some of these bits need to be loaded/initialised before running LibreOffice: I used the below (adjust your location 
+to libcurl-impersonate-chrome.so) to run LibreOffice Calc directly from command line:
+
+```
+LD_PRELOAD=/tmp/curl-impersonate/libcurl-impersonate-chrome.so CURL_IMPERSONATE=chrome101 /usr/lib/libreoffice/program/soffice.bin --calc
+```
+
+With this I can see the below in the output from `=GETREALTIME("SUPPORT")` and the examples.ods file from this repo 
+can load data for Yahoo again.
+```
+...
+requests=curl_cffi_0.10.0
+LD_PRELOAD=/tmp/curl-impersonate/libcurl-impersonate-chrome.so
+CURL_IMPERSONATE=chrome101
+curl_version="libcurl/8.1.1 BoringSSL zlib/1.2.11 brotli/1.0.9 nghttp2/1.56.0"
+```
+
+Similar things should be possible on Windows - let me know if [this](https://stackoverflow.com/questions/1178257/ld-preload-equivalent-for-windows-to-preload-shared-libraries)
+is helpful and share your experience.
+
+Background: for a normal Python script just installing curl_cffi is enough to bypass Yahoo's HTTPS fingerprinting. 
+Because LibreOffice is loading the stock curl library before executing the extension code directly, the above hack 
+is required. Unless someone tells me otherwise...
 
 ### Feedback requested:
 
@@ -31,7 +61,8 @@ Getting data should be as simple as having this in a cell:
 Codes 21 and 90 stand for "last price" and "close" (see below), respectively. 
 Only Yahoo has historic data available.
 
-There is a file **examples.ods** there too with usage examples and possible arguments to functions.
+There is a file **examples.ods** in the Release area too with usage examples 
+and possible arguments to functions.
 
 You have to check the respective websites to work out what symbol is the right one for you. Make sure today or the date 
 requested is a trading day (exchange is not closed). If a website doesn't have 
@@ -129,18 +160,19 @@ On my system (Ubuntu) I installed packages: libreoffice-dev libreoffice-java-com
 
 cd ~/tech/IdeaProjects/Financials-Extension/
 
-python3 -m unittest discover src
+\# Assuming curl-cffi is installed, LD_PRELOAD is not required here 
+
+CURL_IMPERSONATE=chrome101 python3 -m unittest discover src
 
 \# This builds file **Financials-Extension.oxt**
 
 ./compile.sh
 
 ### Tested with:
-- Windows 10 / LibreOffice Calc 7.1.2.2 / Python 3.8.8
-- Ubuntu 22.04.1 / LibreOffice Calc 7.3.7.2 / Python 3.10.6
-- MacOS 10.15.7 / LibreOffice Calc 7.2.0.4 / Python 3.8.10
+- Ubuntu 22.04.5 / LibreOffice Calc 7.3.7.2 / Python 3.10.12
 
-(Previous versions)
+(Previously)
+- Windows 10 / LibreOffice Calc 7.1.2.2 / Python 3.8.8
+- MacOS 10.15.7 / LibreOffice Calc 7.2.0.4 / Python 3.8.10
 - Debian 10.3  / LibreOffice Calc 6.1.5.2 / Python 3.7.3
 - Ubuntu 20.04.5 / LibreOffice Calc 6.4.7.2 / Python 3.8.10
-- Ubuntu 18.04.5 / LibreOffice Calc 6 / Python 3.6.9

compile.sh

@@ -67,7 +67,7 @@ wget "https://files.pythonhosted.org/packages/ec/57/56b9bcc3c9c6a792fcbaf139543c
 unzip $TMPFILE dateutil/\* -d "${PWD}"/build/
 rm $TMPFILE
 
-wget "https://files.pythonhosted.org/packages/9c/3d/a121f284241f08268b21359bd425f7d4825cffc5ac5cd0e1b3d82ffd2b10/pytz-2024.1-py2.py3-none-any.whl" -O $TMPFILE
+wget "https://files.pythonhosted.org/packages/81/c4/34e93fe5f5429d7570ec1fa436f1986fb1f00c3e0f43a589fe2bbcd22c3f/pytz-2025.2-py2.py3-none-any.whl" -O $TMPFILE
 unzip $TMPFILE pytz/\* -d "${PWD}"/build/
 rm $TMPFILE
 
@@ -76,7 +76,7 @@ unzip $TMPFILE pyparsing.py -d "${PWD}"/build/
 rm $TMPFILE
 
 # Windows LibreOffice 7.1 Python is missing this...
-wget "https://files.pythonhosted.org/packages/d9/5a/e7c31adbe875f2abbb91bd84cf2dc52d792b5a01506781dbcf25c91daf11/six-1.16.0-py2.py3-none-any.whl" -O $TMPFILE
+wget "https://files.pythonhosted.org/packages/b7/ce/149a00dd41f10bc29e5921b496af8b574d8413afcd5e30dfa0ed46c2cc5e/six-1.17.0-py2.py3-none-any.whl" -O $TMPFILE
 unzip $TMPFILE six.py -d "${PWD}"/build/
 rm $TMPFILE
 

src/baseclient.py

@@ -8,18 +8,11 @@
 #  version 3 of the License, or (at your option) any later version.
 
 
-import codecs
-import gzip
 import logging
 import os
 import pathlib
 import random
-import select
-import urllib.request
-import urllib.parse
-from http import cookiejar
-from http.client import HTTPConnection, HTTPSConnection, HTTPException
-
+from importlib import util
 from datacode import Datacode
 
 logger = logging.getLogger(__name__)
@@ -28,12 +21,24 @@ logger = logging.getLogger(__name__)
 # logger.setLevel(logging.DEBUG)
 
 
-class RedirectException(HTTPException):
-    def __init__(self, location):
-        self.location = location
+curl_cffi_present = not util.find_spec("curl_cffi") is None
+requests_present = not util.find_spec("requests") is None
+
+if curl_cffi_present:
+    logger.debug("Importing curl_cffi...")
+    from curl_cffi import requests, __version__ as requests_version, __name__ as requests_name
+elif requests_present:
+    logger.debug("Importing requests...")
+    import requests
+    requests_version = requests.__version__
+    requests_name = requests.__name__
+else:
+    raise Exception("Neither curl_cffi nor requests found.")
+
+# import requests
 
 
-class HttpException(HTTPException):
+class HttpException(Exception):
     def __init__(self, url, response):
         self.url = url
         self.response = response
@@ -45,150 +50,63 @@ class HttpException(HTTPException):
             return f"url='{self.url}' status='{self.response}'"
         if self.response.headers:
             h = '\n'.join(sorted(self.response.headers.__str__().splitlines(), key=lambda l: l.lower()))
-            return f"url='{self.url}' status={self.response.status} reason='{self.response.reason}'{h}\n"
+            return f"url='{self.url}' status={self.response.status_code} reason='{self.response.reason}' headers={h}\n"
         else:
-            return f"url='{self.url}' status={self.response.status} reason='{self.response.reason}'"
+            return f"url='{self.url}' status={self.response.status_code} reason='{self.response.reason}'"
 
 
 class BaseClient:
     def __init__(self):
-        self.connections = {}
-        self.cookies = cookiejar.CookieJar()
         self.last_url = None
-        self.redirect_count = 0  # will be set later
+        self.redirect_count = 0
 
         self.basedir = os.path.join(str(pathlib.Path.home()), '.financials-extension')
         os.makedirs(self.basedir, exist_ok=True)
 
         user_agents = [
-            'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:128.0) Gecko/20100101 Firefox/128.0',
-            'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:129.0) Gecko/20100101 Firefox/129.0',
-            'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:130.0) Gecko/20100101 Firefox/130.0',
-            'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:131.0) Gecko/20100101 Firefox/131.0',
-            'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:132.0) Gecko/20100101 Firefox/132.0',
             'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:133.0) Gecko/20100101 Firefox/133.0',
+            'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:134.0) Gecko/20100101 Firefox/134.0',
+            'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:135.0) Gecko/20100101 Firefox/135.0',
+            'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:136.0) Gecko/20100101 Firefox/136.0',
+            'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:137.0) Gecko/20100101 Firefox/137.0',
         ]
 
-        self.default_headers = {
-            'User-Agent': random.sample(user_agents, 1)[0],
-            'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
-            'Accept-Encoding': 'gzip, deflate',
-            'Accept-Language': 'en-US,en;q=0.5',
-            'Connection': 'keep-alive',
-            'Cache-Control': 'max-age=0'
-        }
+        if curl_cffi_present:
+            self.session = requests.Session()
+            if logger.isEnabledFor(logging.DEBUG) and self.session.curl:
+                self.session.curl.debug()
+        else:
+            self.session = requests.Session()
+            self.session.headers.update({'User-Agent': random.sample(user_agents, 1)[0],
+                                         'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
+                                         'Accept-Encoding': 'gzip, deflate',
+                                         'Accept-Language': 'en-US,en;q=0.5',
+                                         'Connection': 'keep-alive',
+                                         'Cache-Control': 'max-age=0',
+                                         })
 
-        self.response = None
+        self.session.max_redirects = 5
 
-    def request(self, method: str, url: str, data=None, headers={}, **kwargs):
-
-        _headers = self.default_headers.copy()
-        if headers:
-            for key, value in headers.items():
-                _headers[key] = value
-
-        if method == 'POST' and 'Content-Type' not in _headers:
-            _headers['Content-Type'] = 'application/x-www-form-urlencoded'
-
-        connection = None
-
-        scheme, _, host, path = url.split('/', 3)
-
-        if (scheme, host) in self.connections:
-            connection = self.connections.get((scheme, host))
-
-        if connection and select.select([connection.sock], [], [], 0)[0]:
-            connection.close()
-            connection = None
-
-        if not connection:
-            logger.debug('Creating connection --------------------------------------------------')
-            connection = HTTPConnection(host, **kwargs) if scheme == 'http:' else HTTPSConnection(host, **kwargs)
-
-        logger.debug('Creating request -----------------------------------------------------')
-        logger.debug("%s %s", method, url)
-
-        self.last_url = url
-
-        # generate and add cookie headers
-        request = urllib.request.Request(url)
-
-        self.cookies.add_cookie_header(request)
-        if request.get_header('Cookie'):
-            _headers['Cookie'] = request.get_header('Cookie')
-
-        for key, value in _headers.items():
-            logger.debug('Header: %s=%s', key, value)
-
-        # request
-        connection.request(method, '/' + path, data, _headers)
-        response = connection.getresponse()
-
-        logger.debug('Processing response --------------------------------------------------')
-
-        logger.debug('response.status=%s', response.status)
-        for key, value in response.getheaders():
-            logger.debug('Header: %s=%s', key, value)
-
-        self.cookies.extract_cookies(response, request)
-        self.connections[(scheme, host)] = connection
-
-        return response
-
-    def urlopen(self, url, redirect=True, data=None, headers={}, cookies=[], **kwargs):
-
-        if cookies:
-            for c in cookies:
-                self.cookies.set_cookie(c)
+    def urlopen(self, url, data=None):
 
         self.last_url = None
 
-        self.response = self.request('POST' if data else 'GET', url, data, headers, **kwargs)
-        text = self.response.read()
+        resp = self.session.request('POST' if data else 'GET', url, data=data)
 
-        # Allow redirects - used by Yahoo for some cookie based consent
-        self.redirect_count = 5
+        if 400 <= resp.status_code < 500:
+            if resp.headers.get('X-Cache') == 'Error from cloudfront':
+                resp = self.session.request('POST' if data else 'GET', url, data=data)
 
-        # (for Yahoo) AWS CloudFront occasionally returns an incorrect, cached error responses
-        # try mitigating by re-requesting straight away
-        if 400 <= self.response.status < 500:
-            if self.response.getheader('X-Cache') == 'Error from cloudfront':
-                self.response = self.request('POST' if data else 'GET', url, data, headers, **kwargs)
-                text = self.response.read()
+        if resp.status_code >= 400:
+            logger.warning("url='%s' status=%s reason='%s' headers=%s", resp.url,
+                           resp.status_code, resp.reason,
+                           '\n'.join(sorted(resp.headers.__str__().splitlines(), key=lambda l: l.lower())))
+            raise HttpException(url, resp)
 
-        while 300 <= self.response.status < 400 and self.redirect_count >= 0:
+        self.redirect_count = len(resp.history)
+        self.last_url = resp.url
 
-            self.redirect_count -= 1
-            location = str(self.response.getheader('Location'))
-            location = location.replace(' ', '%20')  # FT bug workaround - this should not be necessary
-
-            if location and redirect:
-
-                if location.startswith('/'):
-                    scheme, _, host, path = url.split('/', 3)
-                    location = '{}//{}{}'.format(scheme, host, location)
-
-                self.response = self.request('GET', location, None, headers, **kwargs)
-                text = self.response.read()
-
-            else:
-                raise RedirectException(location)
-
-        if self.response.status >= 400:
-            logger.warning("last_url='%s' status=%s reason='%s' headers=%s", self.last_url, self.response.status,
-                           self.response.reason,
-                           '\n'.join(sorted(self.response.headers.__str__().splitlines(), key=lambda l: l.lower())))
-            raise HttpException(url, self.response)
-
-        if self.response.getheader('Content-Encoding') == 'gzip':
-            text = gzip.decompress(text)
-
-        content_type = self.response.headers.get_content_charset()
-        if content_type is None:
-            content_type = 'utf-8'
-        text = codecs.decode(text, encoding=content_type, errors='ignore')
-
-        return text
+        return resp.text
 
     def get_ticker(self):
 
@@ -397,10 +315,11 @@ class BaseClient:
 
         return None
 
+    def version(self):
+        return requests_name + "_" + requests_version
+
+    def curl(self):
+        return curl_version
+
     def close(self):
-        for connection in self.connections.values():
-            try:
-                connection.close()
-            except BaseException:
-                pass
-        self.connections = {}
+        self.session.close()

src/financials.py

@@ -240,7 +240,8 @@ class FinancialsImpl(unohelper.Base, Financials):
             if e.tag.endswith('version'):
                 version = e.attrib['value']
 
-        s = 'ctx={}\nid(self)={}\nversion={}\nfile={}\ncwd={}\nhome={}\nuname={}\npid={}\nsys.executable={}\nsys.version={}\nsys.path={}\nlocale={}\ndefaultlocale={}\ndateutil={}\npytz={}\npyparsing={}\nsix={}'.format(
+        s = ('ctx={}\nid(self)={}\nversion={}\nfile={}\ncwd={}\nhome={}\nuname={}\npid={}\nsys.executable={}\nsys.version={}\nsys.path={}\n' +
+             'locale={}\ndefaultlocale={}\ndateutil={}\npytz={}\npyparsing={}\nsix={}\nrequests={}').format(
             self.ctx,
             id(self),
             version,
@@ -258,8 +259,20 @@ class FinancialsImpl(unohelper.Base, Financials):
             pytz.__version__,
             pyparsing.__version__,
             six.__version__,
+            self.ft.version()
         )
 
+        ld_preload = os.environ.get('LD_PRELOAD')
+        if ld_preload:
+            s += f"\nLD_PRELOAD={ld_preload}"
+
+        curl_impersonate = os.environ.get('CURL_IMPERSONATE')
+        if curl_impersonate:
+            s += f"\nCURL_IMPERSONATE={curl_impersonate}"
+
+        if 'curl_cffi' in self.ft.version():
+            s += f"\ncurl_version=\"{self.ft.session.curl.version().decode()}\""
+
         if datacode:
             s = '{}\ntype(datacode)={}\nstr(datacode)={}'.format(
                 s,

src/financials_coinbase.py

@@ -59,7 +59,7 @@ class Coinbase(BaseClient):
         url = 'https://api.exchange.coinbase.com/products/{}/stats'.format(ticker)
 
         try:
-            text = self.urlopen(url, redirect=True, data=None, headers=None)
+            text = self.urlopen(url)
         except BaseException as e:
             logger.exception("BaseException ticker=%s datacode=%s last_url=%s redirect_count=%s", ticker, datacode, self.last_url, self.redirect_count)
             del self.realtime[ticker]

src/financials_ft.py

@@ -76,7 +76,7 @@ class FT(BaseClient):
         url = f'https://markets.ft.com/data/{asset_class}/tearsheet/summary?s={urllib.parse.quote_plus(ticker)}'
 
         try:
-            text = self.urlopen(url, redirect=True, data=None, headers=None)
+            text = self.urlopen(url)
         except BaseException as e:
             logger.exception("BaseException ticker=%s datacode=%s last_url=%s redirect_count=%s", ticker, datacode, self.last_url, self.redirect_count)
             del self.realtime[ticker]

src/financials_yahoo.py

@@ -119,7 +119,7 @@ class Yahoo(BaseClient):
     def handleCookiesAndConsent(self, url, ticker, datacode, html_file):
 
         try:
-            text = self.urlopen(url, redirect=True)
+            text = self.urlopen(url)
         except BaseException as e:
             logger.exception("BaseException (1) ticker=%s datacode=%s last_url=%s redirect_count=%s %s",
                              ticker, datacode, self.last_url, self.redirect_count, e)
@@ -155,7 +155,7 @@ class Yahoo(BaseClient):
                         data[d.attrib['name']] = d.attrib['value']
 
                 try:
-                    text = self.urlopen(self.last_url, redirect=True, data=urllib.parse.urlencode(data))
+                    text = self.urlopen(self.last_url, data=data)
                 except BaseException as e:
                     logger.exception("BaseException (4) ticker=%s datacode=%s last_url=%s redirect_count=%s %s",
                                      ticker, datacode, self.last_url, self.redirect_count, e)
@@ -196,7 +196,7 @@ class Yahoo(BaseClient):
 
         if not self.crumb:
 
-            url = 'https://finance.yahoo.com/quote/{}?p={}'.format(ticker, ticker)
+            url = f'https://finance.yahoo.com/quote/{ticker}'
             text = self.handleCookiesAndConsent(url, ticker, datacode, f'yahoo-{ticker}.html')
 
             if text is None:

src/generate_metainfo.py

@@ -14,7 +14,7 @@ import os
 cur_dir = os.getcwd()
 
 addin_id = "com.financials.getinfo"
-addin_version = "3.7.2"
+addin_version = "3.8.0"
 addin_displayname = "Financial Market Extension"
 addin_publisher_link = "https://github.com/cmallwitz/Financials-Extension"
 addin_publisher_name = "The Publisher"

src/test_yahoo.py

@@ -24,7 +24,7 @@ import testutils
 financials = financials.createInstance(None)
 
 
-def urlopen_fail(self, url, redirect=True, data=None, headers={}, cookies=[], **kwargs):
+def urlopen_fail(self, url, data=None):
     raise baseclient.HttpException(url, 'ERROR: simulated urlopen() failed')