From 8cbfc253abdf64a6b9beceddee3e1941a4e0fe33 Mon Sep 17 00:00:00 2001
From: Ricard Illa <rilla@monotremata.xyz>
Date: Tue, 23 Feb 2021 16:34:34 +0100
Subject: [PATCH] archivebox image

---
 Makefile                 |  5 +++-
 README.md                |  5 +++-
 archivebox/Dockerfile    | 56 ++++++++++++++++++++++++++++++++++++++++
 archivebox/Makefile      | 10 +++++++
 archivebox/entrypoint.sh |  5 ++++
 5 files changed, 79 insertions(+), 2 deletions(-)
 create mode 100644 archivebox/Dockerfile
 create mode 100644 archivebox/Makefile
 create mode 100644 archivebox/entrypoint.sh

diff --git a/Makefile b/Makefile
index dc9d77b..fcc7a91 100644
--- a/Makefile
+++ b/Makefile
@@ -1,4 +1,4 @@
-images=agate ansible backup buku ddclient git-daemon gitolite gitolite-pystagit rss-bridge sassc syncthing tasks tor vdirsyncer xandikos
+images=agate archivebox ansible backup buku ddclient git-daemon gitolite gitolite-pystagit rss-bridge sassc syncthing tasks tor vdirsyncer xandikos
 .PHONY: all $(images)
 all: $(images)
 
@@ -7,6 +7,9 @@ BUILD = make -C $@
 agate:
 	$(BUILD)
 
+archivebox:
+	$(BUILD)
+
 ansible:
 	$(BUILD)
 
diff --git a/README.md b/README.md
index 849b611..2955f9a 100644
--- a/README.md
+++ b/README.md
@@ -2,7 +2,9 @@
 
 Collection of docker images for my personal use. I usually implement my own
 docker images if I don't find one built or recommended by the original
-developer of the software I intend to run.
+developer of the software I intend to run. Or if the recommended one is too
+complex for my specific use case. Or if I don't like it for whatever reason. Or
+if I just feel like making my own image.
 
 Simply run `make $IMG_NAME` to build a specific image or run `make` to build them
 all.
@@ -16,6 +18,7 @@ useful.
 (many are self-explanatory)
 
 * agate
+* archivebox
 * ansible
 * backup: simple image used in some of my backup daemons
 * buku: buku and bukuserver
diff --git a/archivebox/Dockerfile b/archivebox/Dockerfile
new file mode 100644
index 0000000..03c9c43
--- /dev/null
+++ b/archivebox/Dockerfile
@@ -0,0 +1,56 @@
+FROM alpine:3.13
+
+COPY entrypoint.sh /usr/local/bin/entrypoint
+
+RUN apk add --no-cache --virtual .build-deps \
+        gcc \
+        python3-dev \
+        npm \
+        musl-dev && \
+    apk add --no-cache \
+        py3-pip \
+        python3 \
+        curl \
+        nodejs \
+        git \
+        chromium \
+        youtube-dl \
+        ripgrep \
+        wget && \
+    pip3 install --upgrade pip setuptools && \
+    pip3 install --upgrade archivebox && \
+    npm install -g 'git+https://github.com/ArchiveBox/ArchiveBox.git' && \
+    apk del .build-deps && \
+    addgroup --gid 10001 archivebox && \
+    adduser \
+        --uid 10000 \
+        --home /var/lib/archivebox \
+        --ingroup archivebox \
+        --disabled-password \
+        --shell /sbin/nologin \
+        archivebox && \
+    mkdir -p /data && \
+    mkdir -p /var/lib/archivebox/chromium_data/Default && \
+    touch /var/lib/archivebox/cookies.txt && \
+    chown -R archivebox /data /var/lib/archivebox && \
+    chmod +x /usr/local/bin/entrypoint
+
+ENV IN_DOCKER=True \
+    CHROME_SANDBOX=False \
+    CHROME_BINARY="/usr/bin/chromium-browser" \
+    CHROME_USER_DATA_DIR="/var/lib/archivebox/chromium_data" \
+    COOKIES_FILE="/var/lib/archivebox/cookies.txt" \
+    USE_SINGLEFILE=True \
+    SINGLEFILE_BINARY="/usr/bin/single-file" \
+    USE_READABILITY=True \
+    READABILITY_BINARY="/usr/bin/readability-extractor" \
+    USE_MERCURY=True \
+    MERCURY_BINARY="/usr/bin/mercury-parser"
+
+VOLUME /data
+EXPOSE 8000
+
+WORKDIR /data
+USER archivebox
+ENTRYPOINT ["/usr/local/bin/entrypoint"]
+CMD ["server", "0.0.0.0:8000"]
diff --git a/archivebox/Makefile b/archivebox/Makefile
new file mode 100644
index 0000000..959b564
--- /dev/null
+++ b/archivebox/Makefile
@@ -0,0 +1,10 @@
+USERNAME = rilla
+IMG_NAME = archivebox
+
+.PHONY: build build-nc
+
+build: Dockerfile
+	docker build -t $(USERNAME)/$(IMG_NAME) .
+
+build-nc: Dockerfile
+	docker build --no-cache -t $(USERNAME)/$(IMG_NAME) .
diff --git a/archivebox/entrypoint.sh b/archivebox/entrypoint.sh
new file mode 100644
index 0000000..d932701
--- /dev/null
+++ b/archivebox/entrypoint.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+archivebox init
+
+exec archivebox "$@"