Bump version to v0.6.9 for release

Dockerfile

@@ -4,7 +4,6 @@ ENV RUNNING_IN_DOCKER=1
 
 WORKDIR /app
 
-# TODO: use custom ffmpeg builds instead of apt-get install
 RUN pip install --upgrade pip && \
 	pip install pipenv && \
 	add-apt-repository ppa:mozillateam/ppa && \
@@ -18,7 +17,6 @@ RUN pip install --upgrade pip && \
 	rm geckodriver-v*
 
 
-# TODO: avoid copying unnecessary files, including .git
 COPY Pipfile* ./
 # install from pipenv, with browsertrix-only requirements
 RUN pipenv install && \
@@ -27,11 +25,6 @@ RUN pipenv install && \
 # doing this at the end helps during development, builds are quick
 COPY ./src/ . 
 
-# TODO: figure out how to make volumes not be root, does it depend on host or dockerfile?
-# RUN useradd --system --groups sudo --shell /bin/bash archiver && chown -R archiver:sudo .
-# USER archiver
-
-
 ENTRYPOINT ["pipenv", "run", "python3", "-m", "auto_archiver"]
 
 # should be executed with 2 volumes (3 if local_storage is used)

src/auto_archiver/archivers/youtubedl_archiver.py

@@ -25,7 +25,7 @@ class YoutubeDLArchiver(Archiver):
             logger.debug('Using Facebook cookie')
             yt_dlp.utils.std_headers['cookie'] = self.facebook_cookie
 
-        ydl = yt_dlp.YoutubeDL({'outtmpl': os.path.join(ArchivingContext.get_tmp_dir(), f'%(id)s.%(ext)s'), 'quiet': False})
+        ydl = yt_dlp.YoutubeDL({'outtmpl': os.path.join(ArchivingContext.get_tmp_dir(), f'%(id)s.%(ext)s'), 'quiet': False, 'noplaylist': True})
 
         try:
             # don'd download since it can be a live stream

src/auto_archiver/enrichers/wacz_enricher.py

@@ -27,6 +27,8 @@ class WaczArchiverEnricher(Enricher, Archiver):
     def configs() -> dict:
         return {
             "profile": {"default": None, "help": "browsertrix-profile (for profile generation see https://github.com/webrecorder/browsertrix-crawler#creating-and-using-browser-profiles)."},
+            "docker_commands": {"default": None, "help":"if a custom docker invocation is needed"},
+            "browsertrix_home": {"default": None, "help": "Path to use with the custom browsertrix file locations, useful together with docker_commands"},
             "timeout": {"default": 120, "help": "timeout for WACZ generation in seconds"},
             "extract_media": {"default": True, "help": "If enabled all the images/videos/audio present in the WACZ archive will be extracted into separate Media. The .wacz file will be kept untouched."}
         }
@@ -46,9 +48,9 @@ class WaczArchiverEnricher(Enricher, Archiver):
         url = to_enrich.get_url()
 
         collection = str(uuid.uuid4())[0:8]
-        browsertrix_home = os.path.abspath(ArchivingContext.get_tmp_dir())
+        browsertrix_home = self.browsertrix_home or os.path.abspath(ArchivingContext.get_tmp_dir())
 
-        if os.getenv('RUNNING_IN_DOCKER'):
+        if os.environ.get('RUNNING_IN_DOCKER', 0) == '1':
             logger.debug(f"generating WACZ without Docker for {url=}")
 
             cmd = [
@@ -69,13 +71,10 @@ class WaczArchiverEnricher(Enricher, Archiver):
                 cmd.extend(["--profile", os.path.join("/app", str(self.profile))])
         else:
             logger.debug(f"generating WACZ in Docker for {url=}")
-
-            cmd = [
-                "docker", "run",
-                "--rm",  # delete container once it has completed running
-                "-v", f"{browsertrix_home}:/crawls/",
-                # "-it", # this leads to "the input device is not a TTY"
-                "webrecorder/browsertrix-crawler", "crawl",
+            if not self.docker_commands:
+                self.docker_commands = ["docker", "run", "--rm", "-v", f"{browsertrix_home}:/crawls/", "webrecorder/browsertrix-crawler"]
+            cmd = self.docker_commands + [
+                "crawl",
                 "--url", url,
                 "--scopeType", "page",
                 "--generateWACZ",
@@ -89,6 +88,7 @@ class WaczArchiverEnricher(Enricher, Archiver):
 
             if self.profile:
                 profile_fn = os.path.join(browsertrix_home, "profile.tar.gz")
+                logger.debug(f"copying {self.profile} to {profile_fn}")
                 shutil.copyfile(self.profile, profile_fn)
                 cmd.extend(["--profile", os.path.join("/crawls", "profile.tar.gz")])
 

src/auto_archiver/version.py

@@ -3,7 +3,7 @@ _MAJOR = "0"
 _MINOR = "6"
 # On main and in a nightly release the patch should be one ahead of the last
 # released build.
-_PATCH = "6"
+_PATCH = "9"
 # This is mainly for nightly builds which have the suffix ".dev$DATE". See
 # https://semver.org/#is-v123-a-semantic-version for the semantics.
 _SUFFIX = ""