new attempt at dind wacz

src/auto_archiver/core/orchestrator.py

@@ -90,7 +90,9 @@ class ArchivingOrchestrator:
         if cached_result:
             logger.debug("Found previously archived entry")
             for d in self.databases:
-                d.done(cached_result, cached=True)
+                try: d.done(cached_result, cached=True)
+                except Exception as e:
+                    logger.error(f"ERROR database {d.name}: {e}: {traceback.format_exc()}")
             return cached_result
 
         # 3 - call archivers until one succeeds
@@ -120,6 +122,9 @@ class ArchivingOrchestrator:
             result.status = "nothing archived"
 
         # signal completion to databases and archivers
-        for d in self.databases: d.done(result)
+        for d in self.databases:
+            try: d.done(result)
+            except Exception as e:
+                logger.error(f"ERROR database {d.name}: {e}: {traceback.format_exc()}")
 
         return result

src/auto_archiver/enrichers/wacz_enricher.py

@@ -35,6 +35,22 @@ class WaczArchiverEnricher(Enricher, Archiver):
             "socks_proxy_host": {"default": None, "help": "SOCKS proxy host for browsertrix-crawler, use in combination with socks_proxy_port. eg: user:password@host"},
             "socks_proxy_port": {"default": None, "help": "SOCKS proxy port for browsertrix-crawler, use in combination with socks_proxy_host. eg 1234"},
         }
+    
+    def setup(self) -> None:
+        self.use_docker = os.environ.get('WACZ_ENABLE_DOCKER') or not os.environ.get('RUNNING_IN_DOCKER')
+        self.docker_in_docker = os.environ.get('WACZ_ENABLE_DOCKER') and os.environ.get('RUNNING_IN_DOCKER')
+
+        self.cwd_dind = f"crawls/crawls{random_str(8)}"
+        self.browsertrix_home_host = os.environ.get('BROWSERTRIX_HOME_HOST')
+        self.browsertrix_home_container = os.environ.get('BROWSERTRIX_HOME_CONTAINER') or self.browsertrix_home_host
+        # create crawls folder if not exists, so it can be safely removed in cleanup
+        if self.docker_in_docker:
+            os.makedirs(self.cwd_dind, exist_ok=True)
+
+    def cleanup(self) -> None:
+        if self.docker_in_docker:
+            logger.debug(f"Removing {self.cwd_dind=}")
+            shutil.rmtree(self.cwd_dind, ignore_errors=True)
 
     def download(self, item: Metadata) -> Metadata:
         # this new Metadata object is required to avoid duplication
@@ -51,8 +67,8 @@ class WaczArchiverEnricher(Enricher, Archiver):
         url = to_enrich.get_url()
 
         collection = random_str(8)
-        browsertrix_home_host = os.environ.get('BROWSERTRIX_HOME_HOST') or os.path.abspath(ArchivingContext.get_tmp_dir())
-        browsertrix_home_container = os.environ.get('BROWSERTRIX_HOME_CONTAINER') or browsertrix_home_host
+        browsertrix_home_host = self.browsertrix_home_host or os.path.abspath(ArchivingContext.get_tmp_dir())
+        browsertrix_home_container = self.browsertrix_home_container or browsertrix_home_host
 
         cmd = [
             "crawl",
@@ -67,11 +83,12 @@ class WaczArchiverEnricher(Enricher, Archiver):
             "--behaviors", "autoscroll,autoplay,autofetch,siteSpecific",
             "--behaviorTimeout", str(self.timeout),
             "--timeout", str(self.timeout)]
+        
+        if self.docker_in_docker:
+            cmd.extend(["--cwd", self.cwd_dind])
 
         # call docker if explicitly enabled or we are running on the host (not in docker)
-        use_docker = os.environ.get('WACZ_ENABLE_DOCKER') or not os.environ.get('RUNNING_IN_DOCKER')
-
-        if use_docker:
+        if self.use_docker:
             logger.debug(f"generating WACZ in Docker for {url=}")
             logger.debug(f"{browsertrix_home_host=} {browsertrix_home_container=}")
             if self.docker_commands:
@@ -103,7 +120,7 @@ class WaczArchiverEnricher(Enricher, Archiver):
             logger.error(f"WACZ generation failed: {e}")
             return False
 
-        if use_docker:
+        if self.use_docker:
             wacz_fn = os.path.join(browsertrix_home_container, "collections", collection, f"{collection}.wacz")
         else:
             wacz_fn = os.path.join("collections", collection, f"{collection}.wacz")
@@ -116,7 +133,7 @@ class WaczArchiverEnricher(Enricher, Archiver):
         if self.extract_media or self.extract_screenshot:
             self.extract_media_from_wacz(to_enrich, wacz_fn)
 
-        if use_docker:
+        if self.use_docker:
             jsonl_fn = os.path.join(browsertrix_home_container, "collections", collection, "pages", "pages.jsonl")
         else:
             jsonl_fn = os.path.join("collections", collection, "pages", "pages.jsonl")

src/auto_archiver/version.py

@@ -3,7 +3,7 @@ _MAJOR = "0"
 _MINOR = "9"
 # On main and in a nightly release the patch should be one ahead of the last
 # released build.
-_PATCH = "2"
+_PATCH = "5"
 # 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 = ""