# Bristlenose — Fedora / Copr package.
#
# Generated from this template by rpm/make-srpm.sh: 0.28.0 is substituted
# from bristlenose/__init__.py, which stays the single source of truth for the
# version (see CLAUDE.md — never add a version to pyproject.toml, and don't add
# a second one here either).
#
# Design, and why it is shaped like this: docs/design-fedora-packaging.md

%global bn_prefix   %{_libdir}/%{name}
%global bn_python   %{bn_prefix}/bin/python
%global marker_file     .install-method
%global spacy_model     en_core_web_sm
%global spacy_model_ver 3.8.0

# The payload is 106 manylinux wheels — binaries built by upstream against a
# different toolchain, not compiled here. Every one of rpm's post-build passes
# assumes it built the binaries itself, and each fails or does harm on wheels:
#
#   debug_package        no source to extract debuginfo from
#   strip                would strip vendored .so files we cannot rebuild
#   mangle_shebangs      would rewrite the venv's own interpreter lines
#   python_bytecompile   would compile with the SYSTEM python, not the venv's
#                        (we do it ourselves in %install, with -s so the
#                        buildroot prefix never lands in a traceback)
%global debug_package %{nil}
%global __brp_strip %{nil}
%global __brp_strip_static_archive %{nil}
%global __brp_strip_comment_note %{nil}
%global __brp_mangle_shebangs %{nil}
%global __brp_python_bytecompile %{nil}

# ...and rpm's automatic dependency generator must not look inside the venv
# either. Left on, it scans the vendored .so files and emits both unsatisfiable
# Requires (soname deps for libraries bundled inside the wheels themselves) and
# bogus Provides that would leak into the repo's metadata.
%global __requires_exclude_from ^%{bn_prefix}/.*$
%global __provides_exclude_from ^%{bn_prefix}/.*$

Name:           bristlenose
Version:        0.28.0
Release:        1%{?dist}
Summary:        User-research transcription and quote extraction engine

License:        AGPL-3.0-only
URL:            https://github.com/cassiocassio/bristlenose

# Source0 is the PyPI sdist, NOT a git archive — deliberately. The React SPA
# under bristlenose/server/static/ is gitignored but declared as a hatch
# artifact, and release.yml runs `npm run build` before `python -m build`. So
# the sdist carries the built bundle and the RPM needs no Node.js in mock.
Source0:        https://pypi.io/packages/source/b/%{name}/%{name}-%{version}.tar.gz
# Every runtime dependency, as wheels, built by rpm/make-srpm.sh at SRPM time
# where there IS network. mock has none, so %%install resolves from here alone.
Source1:        %{name}-vendor-%{version}.tar.gz

# The vendored wheels are cp3XX-specific and x86_64-specific. Fedora on laptops
# is realistically Intel/AMD (matching the snap, which has only ever published
# amd64); arm64 wheels exist for everything but an SRPM is one artefact for all
# arches, so shipping both would double the vendor tarball.
ExclusiveArch:  x86_64

BuildRequires:  python3-devel >= 3.10
BuildRequires:  python3-pip

# ffmpeg-free is genuinely sufficient: bristlenose extracts audio with -vn, so
# it never decodes a video stream to do its work, and H.264 decode is present
# anyway via libopenh264. The one gap is HEVC, which costs a thumbnail on
# iPhone-recorded .mov files and nothing else. Measurements and the full
# 26-file corpus: docs/design-fedora-packaging.md section 2.
Requires:       /usr/bin/ffmpeg
Requires:       /usr/bin/ffprobe
# The venv is built against this exact interpreter and holds cp3XX wheels.
Requires:       python(abi) = %{python3_version}

%description
Bristlenose takes a folder of interview recordings — audio, video, or existing
transcripts — and produces a browsable HTML report: extracted quotes, themes,
sentiment, friction points and user journeys, which the researcher edits,
shares, and keeps as a file they own.

The pipeline runs on your own machine. The analysis is a call to a cloud LLM —
Claude, ChatGPT, Azure OpenAI or Gemini — configured with `bristlenose
configure`. Local models via Ollama are supported as a slower alternative.

Run `bristlenose doctor` after installing to check the environment.

%prep
%setup -q -n %{name}-%{version}
# -> ./vendor/, alongside the unpacked sdist
tar -xf %{SOURCE1}
test -d vendor || { echo "vendor/ missing from %{SOURCE1}" >&2; exit 1; }

%build
# Nothing to compile. The venv is created in %%install so that every path it
# bakes in is under the buildroot and can be rewritten in one place.

%install
python3 -m venv %{buildroot}%{bn_prefix}

# --no-index is the whole point: if this ever reaches for the network, the
# build must fail rather than quietly produce a differently-resolved package.
%{buildroot}%{bn_prefix}/bin/pip install \
    --no-index --find-links=vendor --no-cache-dir --no-compile \
    "%{name}[serve]==%{version}"

# spaCy's model is not on PyPI (it ships from the spacy-models GitHub
# releases), so it is fetched into vendor/ by name at SRPM time. Installing it
# here means --redact-pii works out of the box, as it does on the snap.
%{buildroot}%{bn_prefix}/bin/pip install \
    --no-index --find-links=vendor --no-cache-dir --no-compile \
    "%{spacy_model}==%{spacy_model_ver}"

# A venv records the path it was created at, in pyvenv.cfg and in the shebang
# of every console script. Right now that path is inside the buildroot. Strip
# it so the installed venv points at its real home. Only regular files: bin/
# also holds python symlinks, and sed -i would replace a symlink with a copy.
find %{buildroot}%{bn_prefix}/bin -maxdepth 1 -type f \
    -exec sed -i 's|%{buildroot}||g' {} +
sed -i 's|%{buildroot}||g' %{buildroot}%{bn_prefix}/pyvenv.cfg

# Bytecompile with the venv's own interpreter (the system one may differ), and
# -s so the recorded source path is the installed one. Without -s, a traceback
# in front of a user would name a directory on a Copr builder.
%{buildroot}%{bn_prefix}/bin/python -m compileall -q -f \
    -s %{buildroot} %{buildroot}%{bn_prefix} || :

# Drop pip and the activate scripts. They are of no use inside a tree rpm
# owns, and `sudo %{bn_prefix}/bin/pip install ...` would desynchronise the
# file list for good — `rpm -V` dirty forever, orphans left by `dnf remove`.
rm -f %{buildroot}%{bn_prefix}/bin/pip*
rm -f %{buildroot}%{bn_prefix}/bin/activate*
rm -rf %{buildroot}%{bn_prefix}/lib/python%{python3_version}/site-packages/pip \
       %{buildroot}%{bn_prefix}/lib/python%{python3_version}/site-packages/pip-*.dist-info

# Tell doctor what installed this. Read by doctor_fixes.detect_install_method()
# from sys.prefix. Without it an RPM install reads as "pip", and every fix
# message hands the user a pip command — which on Fedora either refuses with
# EXTERNALLY-MANAGED or installs a SECOND bristlenose into ~/.local/bin that
# shadows this one on PATH.
echo rpm > %{buildroot}%{bn_prefix}/%{marker_file}

# The launcher. The venv already generated bin/bristlenose with a correct
# shebang, so this is a symlink, not a wrapper script to keep in sync.
mkdir -p %{buildroot}%{_bindir}
ln -s %{bn_prefix}/bin/%{name} %{buildroot}%{_bindir}/%{name}

# Man page, installed here and NOT in %%post. (The reason is ownership, not
# Homebrew's link-phase trap — rpm has no link phase. A file created in %%post
# is in no %%files list, so rpm does not own it: invisible to `rpm -V` and
# orphaned on `dnf remove`.)
mkdir -p %{buildroot}%{_mandir}/man1
install -p -m 0644 %{name}/data/%{name}.1 %{buildroot}%{_mandir}/man1/%{name}.1

# Nothing may point back into the buildroot by this point. Written to a file
# rather than piped into `grep -q`, which under pipefail reports the producer's
# SIGPIPE as failure — see the same note in rpm/make-srpm.sh.
grep -rIl '%{buildroot}' %{buildroot}%{bn_prefix} > %{_builddir}/leaks.txt 2>/dev/null || :
if [ -s %{_builddir}/leaks.txt ]; then
    echo "ERROR: buildroot path survives in the installed tree:" >&2
    cat %{_builddir}/leaks.txt >&2
    exit 1
fi

%check
# Prove the interpreter and the package actually work from their installed
# layout before this becomes an RPM. Runs against the buildroot copy, so the
# shebang rewrite above has to have been right.
%{buildroot}%{bn_prefix}/bin/python -c "import bristlenose; print(bristlenose.__version__)" \
    | grep -qx '%{version}'
%{buildroot}%{bn_prefix}/bin/python -c "
import importlib
for m in ('faster_whisper','ctranslate2','presidio_analyzer','presidio_anonymizer',
          'spacy','anthropic','openai','google.genai','fastapi','av','webvtt','pysrt','docx'):
    importlib.import_module(m)
print('imports OK')"
# Both halves of the SPA must be present. static/ is serve mode (a fail-loud
# 500 page without it); static-export/ is the single-file export build, which
# was missing from every published wheel until 27 Aug 2026 and made the Export
# HTML button 500 on every pip install. Assert both, not just the one we
# happen to remember.
bn_site=%{buildroot}%{bn_prefix}/lib/python%{python3_version}/site-packages/bristlenose
test -s "$bn_site/server/static/index.html"
test -s "$bn_site/server/static-export/app.js"
test -s "$bn_site/server/static-export/app.css"

# doctor must recognise this as an RPM, not guess "pip".
test "$(cat %{buildroot}%{bn_prefix}/%{marker_file})" = rpm
%{buildroot}%{bn_prefix}/bin/python -c "
import sys
from bristlenose.doctor_fixes import detect_install_method, get_fix
assert detect_install_method() == 'rpm', detect_install_method()
for k in ('ffmpeg_missing','backend_import_fail','spacy_model_missing',
          'presidio_missing','serve_deps_missing'):
    t = get_fix(k)
    assert 'pip install' not in t and 'pipx' not in t, (k, t)
print('doctor install-method OK')"

%files
%license LICENSE
%doc README.md CHANGELOG.md
%{bn_prefix}/
%{_bindir}/%{name}
%{_mandir}/man1/%{name}.1*

%changelog
* Thu Aug 27 2026 Martin Storey <martin@cassiocassio.co.uk> - 0.28.0-1
- Initial Copr package.