furtka/tests/test_https.py

217 lines
8.1 KiB
Python
Raw Permalink Normal View History

fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
"""Tests for furtka.https — fingerprint extraction + HTTPS toggle.
Since 26.15-alpha the toggle writes/removes TWO snippets atomically:
- The top-level HTTPS listener snippet (enables :443 + tls internal)
- The :80-scoped redirect snippet (forces HTTP HTTPS)
The fingerprint case uses a throwaway self-signed EC cert with a known
reference fingerprint (computed once via `openssl x509 -fingerprint
-sha256 -noout`) so we verify the PEM DER SHA256 path without a
runtime subprocess dependency. The toggle cases stub the caddy reload
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
so we assert both snippet files are written / removed together and that
reload failures roll BOTH state back.
"""
import subprocess
import pytest
from furtka import https
# Self-signed test-only cert. Don't trust it anywhere; it's here because
# we need a real PEM whose fingerprint we can pre-compute.
_TEST_CERT_PEM = """-----BEGIN CERTIFICATE-----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-----END CERTIFICATE-----
"""
_TEST_CERT_FP_SHA256 = (
"40:A7:98:2E:8D:1F:4C:0D:9B:E6:87:ED:91:FA:6F:B1:"
"3D:8A:10:06:79:7C:08:A9:8F:AD:71:0C:B8:29:87:28"
)
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
def _paths(tmp_path):
"""Return the four paths the toggle touches, in a dict for kwargs
spreading. Keeps each test's fixture boilerplate small."""
return {
"snippet_dir": tmp_path / "furtka.d",
"snippet": tmp_path / "furtka.d" / "redirect.caddyfile",
"https_snippet_dir": tmp_path / "furtka-https.d",
"https_snippet": tmp_path / "furtka-https.d" / "https.caddyfile",
"hostname_file": tmp_path / "etc_hostname",
}
def _prepare_hostname(tmp_path, value="testbox"):
(tmp_path / "etc_hostname").write_text(f"{value}\n")
def test_ca_fingerprint_matches_openssl(tmp_path):
cert = tmp_path / "root.crt"
cert.write_text(_TEST_CERT_PEM)
fp_hex = https._ca_fingerprint(cert)
assert fp_hex is not None
assert https._format_fingerprint(fp_hex) == _TEST_CERT_FP_SHA256
def test_ca_fingerprint_missing_file(tmp_path):
assert https._ca_fingerprint(tmp_path / "nope.crt") is None
def test_ca_fingerprint_no_pem_block(tmp_path):
garbage = tmp_path / "root.crt"
garbage.write_text("not a certificate")
assert https._ca_fingerprint(garbage) is None
def test_status_no_ca_no_snippet(tmp_path):
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
s = https.status(ca_path=tmp_path / "root.crt", https_snippet=tmp_path / "https.caddyfile")
assert s == {
"ca_available": False,
"fingerprint_sha256": None,
"force_https": False,
"ca_download_url": "/rootCA.crt",
}
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
def test_status_with_ca_and_https_snippet(tmp_path):
ca = tmp_path / "root.crt"
ca.write_text(_TEST_CERT_PEM)
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
https_snip = tmp_path / "https.caddyfile"
https_snip.write_text("furtka.local, furtka {\n\ttls internal\n\timport furtka_routes\n}\n")
s = https.status(ca_path=ca, https_snippet=https_snip)
assert s["ca_available"] is True
assert s["fingerprint_sha256"] == _TEST_CERT_FP_SHA256
assert s["force_https"] is True
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
def test_status_force_reflects_https_snippet_not_redirect(tmp_path):
"""Authoritative signal for "HTTPS is on" is the listener snippet —
a lone redirect without a :443 listener wouldn't actually serve
HTTPS, so the status must NOT report it as on. Locks 26.15 semantic."""
ca = tmp_path / "root.crt"
ca.write_text(_TEST_CERT_PEM)
s = https.status(ca_path=ca, https_snippet=tmp_path / "does-not-exist.caddyfile")
assert s["force_https"] is False
def test_set_force_enable_writes_both_snippets_and_reloads(tmp_path):
_prepare_hostname(tmp_path)
p = _paths(tmp_path)
calls = []
def fake_reload():
calls.append("reload")
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
result = https.set_force_https(True, reload_caddy=fake_reload, **p)
assert result is True
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
assert p["snippet"].read_text() == https.REDIRECT_CONTENT
written = p["https_snippet"].read_text()
assert "testbox.local, testbox" in written
assert "tls internal" in written
assert "import furtka_routes" in written
assert calls == ["reload"]
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
def test_set_force_uses_fallback_hostname_when_file_missing(tmp_path):
# No /etc/hostname → fall back to 'furtka' so Caddy gets a parseable
# block instead of an empty hostname that would fail config load.
p = _paths(tmp_path)
result = https.set_force_https(True, reload_caddy=lambda: None, **p)
assert result is True
assert "furtka.local, furtka" in p["https_snippet"].read_text()
def test_set_force_disable_removes_both_snippets(tmp_path):
_prepare_hostname(tmp_path)
p = _paths(tmp_path)
p["snippet_dir"].mkdir()
p["https_snippet_dir"].mkdir()
p["snippet"].write_text(https.REDIRECT_CONTENT)
p["https_snippet"].write_text("furtka.local { tls internal }\n")
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
result = https.set_force_https(False, reload_caddy=lambda: None, **p)
assert result is False
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
assert not p["snippet"].exists()
assert not p["https_snippet"].exists()
def test_set_force_disable_is_idempotent_when_already_off(tmp_path):
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
p = _paths(tmp_path)
result = https.set_force_https(False, reload_caddy=lambda: None, **p)
assert result is False
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
assert not p["snippet"].exists()
assert not p["https_snippet"].exists()
def test_reload_failure_rolls_back_enable(tmp_path):
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
_prepare_hostname(tmp_path)
p = _paths(tmp_path)
def failing_reload():
raise subprocess.CalledProcessError(1, ["systemctl"], stderr="bad config")
with pytest.raises(https.HttpsError, match="caddy reload failed: bad config"):
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
https.set_force_https(True, reload_caddy=failing_reload, **p)
# Rollback: since neither snippet existed before, neither exists after.
assert not p["snippet"].exists()
assert not p["https_snippet"].exists()
def test_reload_failure_rolls_back_disable(tmp_path):
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
_prepare_hostname(tmp_path)
p = _paths(tmp_path)
p["snippet_dir"].mkdir()
p["https_snippet_dir"].mkdir()
original_redirect = "redir https://{host}{uri} permanent\n# marker\n"
original_https = "# old https block\nfurtka.local { tls internal }\n"
p["snippet"].write_text(original_redirect)
p["https_snippet"].write_text(original_https)
def failing_reload():
raise subprocess.CalledProcessError(1, ["systemctl"], stderr="bad config")
with pytest.raises(https.HttpsError):
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
https.set_force_https(False, reload_caddy=failing_reload, **p)
# Rollback: both snippets are restored to their exact prior contents.
assert p["snippet"].read_text() == original_redirect
assert p["https_snippet"].read_text() == original_https
def test_systemctl_missing_raises_and_rolls_back(tmp_path):
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
_prepare_hostname(tmp_path)
p = _paths(tmp_path)
def missing_systemctl():
raise FileNotFoundError(2, "No such file", "systemctl")
with pytest.raises(https.HttpsError, match="systemctl not available"):
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
https.set_force_https(True, reload_caddy=missing_systemctl, **p)
assert not p["snippet"].exists()
assert not p["https_snippet"].exists()
def test_redirect_snippet_content_is_caddy_redir_directive():
# Lock the exact directive. A regression here silently stops the
# redirect from taking effect even though the file-swap looks fine.
assert https.REDIRECT_CONTENT.strip() == "redir https://{host}{uri} permanent"
fix(https): make HTTPS opt-in to stop the BAD_SIGNATURE trap on fresh installs Every Furtka since 26.5 shipped a Caddyfile with a `__FURTKA_HOSTNAME__.local { tls internal }` site block, so every first boot auto-generated a fresh self-signed CA + intermediate + leaf. That worked for the first-ever Furtka user, but every reinstall (or second box on the same LAN) produced a new CA whose intermediate shared the fixed CN `Caddy Local Authority - ECC Intermediate` with the previous one. Firefox caches intermediates by CN across profiles — even private windows share cert9.db — so any visitor who had trusted an older Furtka's CA got a cached intermediate with mismatched keys when they hit the new box, producing `SEC_ERROR_BAD_SIGNATURE`. Unlike UNKNOWN_ISSUER, Firefox has NO "Advanced → Accept Risk" bypass for BAD_SIGNATURE, so fresh-install boxes were effectively unreachable over HTTPS in any browser that had ever seen a previous Furtka. Validated live on the .46 test VM: fresh 26.14 ISO install → Firefox hits BAD_SIGNATURE on https://furtka.local/ (even in private mode). Chromium bypasses it via mDNS failure but the issue is the same. openssl verify on the box confirms the chain is internally valid — this is purely client-side cache pollution across boxes. Fix: - assets/Caddyfile: removed the hostname site block. Default install serves :80 only — https://furtka.local connection-refuses, which is a normal error every browser handles instead of the unbypassable crypto fault. Added top-level import of /etc/caddy/furtka-https.d/*.caddyfile so the /settings HTTPS toggle can drop a listener snippet there when a user explicitly opts in. - furtka/https.py: set_force_https now writes TWO snippets atomically — the top-level hostname + tls internal block (enables :443) and the :80-scoped redirect (forces HTTP→HTTPS). Disable removes both. Reload failure rolls both back. Added _read_hostname + _https_snippet_content helpers with `/etc/hostname` → 'furtka' fallback so a missing hostname file doesn't produce an empty site block Caddy rejects. - furtka/https.py::status: force_https now reads the listener snippet (was reading the redirect snippet). A redirect without a listener isn't actually HTTPS being served, so the listener is the authoritative "HTTPS is on" signal. - furtka/updater.py: new _maybe_migrate_preserve_https hook runs inside _refresh_caddyfile on the 26.14 → 26.15 transition. If the box had the redirect snippet on disk (user had opted into HTTPS under the old regime), it writes the new listener snippet too so HTTPS keeps working after the Caddyfile swap removes the hostname block. - webinstaller/app.py: post-install creates /etc/caddy/furtka-https.d/ alongside /etc/caddy/furtka.d/ so the glob import can't trip an older Caddy on a missing path during the first reload. Live-tested on .46: set_force_https(True) writes both snippets, Caddy reloads, :443 listener comes up with fresh CA, curl -k returns 302, HTTP 301-redirects. set_force_https(False) removes both snippets atomically, :443 goes back to connection-refused. Tests: test_https.py expanded from 13 to 15 cases. Toggle-on asserts both snippets written + hostname substituted. Toggle-off asserts both removed. Rollback cases verify BOTH snippets restore on reload failure. New test_https_snippet_content_has_tls_internal_and_routes locks the exact shape of the listener block. test_webinstaller_assets.py: updated two old asserts that assumed hostname block was in Caddyfile; new test_post_install_creates_https_snippet_dir guards the new directory. 276 tests pass, ruff check + format clean. Known remaining wart (documented in CHANGELOG): a browser that trusted a prior Furtka CA still hits BAD_SIGNATURE on this box's HTTPS after enabling it, because the fixed intermediate CN is a Caddy-side limitation. Workaround: clear cert9.db or visit in a fresh profile. Won't affect end users with one Furtka box ever. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-21 19:30:04 +02:00
def test_https_snippet_content_has_tls_internal_and_routes(tmp_path):
# Lock the shape of the opt-in HTTPS listener block. Caddy parses
# this verbatim — changing the shape without updating the test
# risks shipping a silently-broken Caddyfile import.
s = https._https_snippet_content("mybox")
assert "mybox.local, mybox {" in s
assert "\ttls internal" in s
assert "\timport furtka_routes" in s
assert s.endswith("}\n")