From a6fc4e1bb134ff72f30e87cf3ac35324c3c8085d Mon Sep 17 00:00:00 2001
From: Miguel Sozinho Ramalho <19508417+msramalho@users.noreply.github.com>
Date: Fri, 24 Jan 2025 13:59:29 +0000
Subject: [PATCH] modifies base docker image to use browsertrix 1.4.2 (#182)

* modifies base image to newest browsertrix version

* modify browsertrix cmd args based on recent experience
---
 Dockerfile                                   | 12 +++++++-----
 src/auto_archiver/enrichers/wacz_enricher.py |  3 ++-
 2 files changed, 9 insertions(+), 6 deletions(-)

diff --git a/Dockerfile b/Dockerfile
index 0ecc7f3..cbcfdd4 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -1,4 +1,4 @@
-FROM webrecorder/browsertrix-crawler:1.0.4 AS base
+FROM webrecorder/browsertrix-crawler:1.4.2 AS base
 
 ENV RUNNING_IN_DOCKER=1 \
     LANG=C.UTF-8 \
@@ -29,21 +29,23 @@ ENV POETRY_NO_INTERACTION=1 \
     POETRY_VIRTUALENVS_CREATE=1
 
 
-RUN pip install --upgrade pip && \
-    pip install "poetry>=2.0.0,<3.0.0"
+# Create a virtual environment for poetry and install it
+RUN python3 -m venv /poetry-venv && \
+    /poetry-venv/bin/python -m pip install --upgrade pip && \
+    /poetry-venv/bin/python -m pip install "poetry>=2.0.0,<3.0.0"
 
 WORKDIR /app
 
 
 COPY pyproject.toml poetry.lock README.md ./
 # Copy dependency files and install dependencies (excluding the package itself)
-RUN poetry install --only main --no-root --no-cache
+RUN /poetry-venv/bin/poetry install --only main --no-root --no-cache
 
 
 # Copy code: This is needed for poetry to install the package itself,
 # but the environment should be cached from the previous step if toml and lock files haven't changed
 COPY ./src/ .
-RUN poetry install --only main --no-cache
+RUN /poetry-venv/bin/poetry install --only main --no-cache
 
 
 # Update PATH to include virtual environment binaries
diff --git a/src/auto_archiver/enrichers/wacz_enricher.py b/src/auto_archiver/enrichers/wacz_enricher.py
index dc38488..3c39056 100644
--- a/src/auto_archiver/enrichers/wacz_enricher.py
+++ b/src/auto_archiver/enrichers/wacz_enricher.py
@@ -84,7 +84,8 @@ class WaczArchiverEnricher(Enricher, Archiver):
             "--behaviors", "autoscroll,autoplay,autofetch,siteSpecific",
             "--behaviorTimeout", str(self.timeout),
             "--timeout", str(self.timeout),
-            "--blockAds" # TODO: test
+            "--diskUtilization", "99",
+            # "--blockAds" # note: this has been known to cause issues on cloudflare protected sites
         ]
         
         if self.docker_in_docker: