Files
picpeak/backend/Dockerfile
T
Luca e111522415 feat(accounting): rasterise inbound PDFs server-side (never serve raw to browser)
Security hardening for inbound supplier-invoice previews. The admin UI no
longer renders raw PDFs — a malicious inbound PDF could otherwise run embedded
JS or phone home in the admin's session. Instead PDFs are rasterised to flat
PNGs server-side and only those images are shown.

- backend: new rasterizeService shells out to poppler `pdftoppm` (added to the
  Docker image via apk poppler-utils — an OS package, NOT a Node PDF lib, so it
  respects the pdfkit+pdf-lib "no third PDF lib" rule). pdftoppm executes no JS
  and fetches no remote resources, so it doubles as the SSRF/phone-home guard.
  Rendered pages cached under storage/business-docs/inbound/rendered/<id>/.
  - GET /inbound/:id/page/:n streams the rasterised PNG (CSP default-src 'none'
    + nosniff). GET /inbound/:id/file now serves PDFs as a DOWNLOAD only
    (Content-Disposition: attachment) — never inline; images still inline.
- frontend: triage preview switched from a raw-PDF <iframe> to rasterised page
  images (getInboundPageBlob), defaulting to the LAST page (QR-bill) with
  prev/next nav for multi-page PDFs; images stream as before.
- i18n: previewError / prevPage / nextPage / pageOf (EN + DE).

REQUIRES A BACKEND IMAGE REBUILD (Dockerfile adds poppler-utils) — a plain
`docker compose pull` of a stale image won't have pdftoppm; the route then
returns 503 RASTERIZER_UNAVAILABLE and the UI shows "preview unavailable".

Verified: node -c, a pdfkit->pdftoppm rasterise smoke test (renders + caches),
en/de JSON valid, npm run build green.
2026-06-11 00:51:10 +02:00

99 lines
4.3 KiB
Docker

FROM node:22-alpine AS builder
# Add build arguments
ARG CACHEBUST=1
ARG BUILD_DATE
ARG VCS_REF
ARG VERSION
# Add labels for GitHub Container Registry
LABEL org.opencontainers.image.source="https://github.com/the-luap/picpeak"
LABEL org.opencontainers.image.description="PicPeak Backend Service"
LABEL org.opencontainers.image.licenses="MIT"
WORKDIR /app
# Copy package files
COPY package*.json ./
# Install dependencies (--omit=dev replaces deprecated --only=production)
RUN npm ci --omit=dev
# Copy application files
COPY . .
# Production stage
FROM node:22-alpine
WORKDIR /app
# Upgrade all packages to fix security vulnerabilities (OpenSSL, libexpat, BusyBox CVEs)
RUN apk upgrade --no-cache
# Upgrade npm to fix tar, minimatch, brace-expansion CVEs in npm's own deps
# Pin to 10.x to stay compatible with Node 22 Alpine (npm 11.x has dependency issues)
RUN npm install -g npm@10
# Install dumb-init for proper signal handling, postgresql-client for database
# checks, ffmpeg for video upload support, and su-exec for the root → nodejs
# privilege drop in wait-for-db.sh (see #484: container starts as root so it
# can chown bind-mounted host volumes to UID 1001, then re-execs as nodejs
# before running the app). Alpine's ffmpeg package ships both `ffmpeg` and
# `ffprobe` built natively against musl libc — the npm
# `@ffmpeg-installer/ffmpeg` binary is glibc-built and (a) doesn't reliably
# run on Alpine and (b) only includes ffmpeg, not ffprobe (which the video
# pipeline calls via fluent-ffmpeg.ffprobe()).
# fontconfig is required so `sharp` (librsvg) can rasterise SVG logos that
# contain live <text> for the CRM PDFs. Without any font installed, librsvg
# renders text as tofu boxes (□) while the vector artwork still draws — i.e.
# a "corrupted" logo on invoices/quotes. DejaVu/Liberation provide a broad
# Unicode fallback; picpeak's own brand fonts (assets/fonts/, the same files
# PDFKit + the web UI use) are registered with fontconfig further down so the
# logo's text renders in its actual typeface, not a fallback.
# poppler-utils provides `pdftoppm`, used to rasterise inbound supplier-invoice
# PDFs to flat PNGs server-side so the admin UI NEVER renders a raw (possibly
# malicious) PDF. pdftoppm does not execute embedded JS or fetch remote
# resources, so it doubles as the SSRF/phone-home guard for untrusted inbound
# documents (see docs/accounting-inbound-invoices.md).
RUN apk add --no-cache dumb-init postgresql-client ffmpeg su-exec \
fontconfig ttf-dejavu ttf-liberation poppler-utils && \
fc-cache -f
# Create non-root user
RUN addgroup -g 1001 -S nodejs && adduser -S nodejs -u 1001
# Copy from builder
COPY --from=builder --chown=nodejs:nodejs /app/node_modules ./node_modules
COPY --chown=nodejs:nodejs . .
# Ensure all source files are readable and wait script is executable
RUN chmod -R a+r /app && chmod +x wait-for-db.sh
# Register picpeak's bundled brand fonts (assets/fonts/<Family>/*.ttf — the
# same files PDFKit and the web UI use) with fontconfig, so when sharp/librsvg
# rasterises an SVG logo its <text> renders in the actual brand typeface
# rather than a DejaVu/Liberation fallback. fontconfig indexes by each font's
# internal family name and recurses into the per-family subdirectories.
RUN printf '<?xml version="1.0"?>\n<!DOCTYPE fontconfig SYSTEM "fonts.dtd">\n<fontconfig>\n <dir>/app/assets/fonts</dir>\n</fontconfig>\n' > /etc/fonts/conf.d/99-picpeak-fonts.conf && \
fc-cache -f /app/assets/fonts
# Create necessary directories
RUN mkdir -p storage/events/active storage/events/archived storage/thumbnails data logs && \
chown -R nodejs:nodejs storage data logs
# No USER directive — the container starts as root so wait-for-db.sh can
# chown bind-mounted host directories to UID 1001 before dropping privs
# via su-exec. See #484 for the fresh-install restart loop this avoids.
EXPOSE 3000
# Healthcheck hits the same /health endpoint already used by the e2e
# runner and by the docker-compose `depends_on: condition: service_healthy`
# checks. wget is part of the Alpine base image. Long start-period covers
# the wait-for-db.sh delay before the Node process starts listening.
HEALTHCHECK --interval=30s --timeout=5s --start-period=60s --retries=3 \
CMD wget --no-verbose --tries=1 --spider http://localhost:3000/health || exit 1
ENTRYPOINT ["dumb-init", "--"]
CMD ["./wait-for-db.sh", "node", "server.js"]