From b8480cc1fbd919d034b66b7af4ab189a6f7c80a1 Mon Sep 17 00:00:00 2001 From: pierre Date: Thu, 15 May 2025 20:24:47 +0100 Subject: [PATCH] improve stash downloading script --- .../stash-pornpics-dowloader.configmap.yml | 30 +++++++++++++++ .../stash/stash-pornpics-dowloader.job.yml | 37 +++++++++++++++++++ 2 files changed, 67 insertions(+) create mode 100644 external_manifests/stash/stash-pornpics-dowloader.configmap.yml create mode 100644 external_manifests/stash/stash-pornpics-dowloader.job.yml diff --git a/external_manifests/stash/stash-pornpics-dowloader.configmap.yml b/external_manifests/stash/stash-pornpics-dowloader.configmap.yml new file mode 100644 index 0000000..b3eae7d --- /dev/null +++ b/external_manifests/stash/stash-pornpics-dowloader.configmap.yml @@ -0,0 +1,30 @@ +apiVersion: v1 +data: + download.sh: "#!/bin/bash\n\n#install required packages\napk add curl wget\n\n#named + argument generation\nskip_searching_gallery=\"${skip_searching_gallery:-false}\"\n\n\nwhile + [ $# -gt 0 ]; do\n if [[ $1 == \"--\"* ]]; then\n v=\"${1/--/}\"\n declare + \"$v\"=\"$2\"\n shift\n fi\n shift\ndone\n\n#check if we will use + a HTTP proxy or not\n\nif [[ -z $http_proxy ]]; then\n echo \"not using + proxy\"\n curl_command=\"curl\"\nelse\n echo \"using proxy\"\n curl_command=\"curl + -x $http_proxy\"\n\nfi\n\n\n#dowload from https://www.pornpics.com/pornstars/eiby-shine/\nDESTINATION_FOLDER=\"/data\"\nPROXY_HTTP=\"http://gluetun-svc.gluetun.svc:8888\"\n\n#BASE_URL=\"https://www.pornpics.com/pornstars/eiby-shine/\"\n#pornstars=\"eiby-shine\"\n\nBASE_URL=\"https://www.pornpics.com/pornstars/chloe-camilla/\"\npornstars=\"chloe-camilla\"\n\n#BASE_URL=\"https://www.pornpics.com/pornstars/becky-summer\"\n#pornstars=\"becky-summer\"\n\n\n#clear + files beforehand\n\n\nmkdir -p $DESTINATION_FOLDER/$pornstars\n\nif [[ $skip_searching_gallery + == true ]]; then\n echo \"not looping through galleries\"\n\t\techo \"using + existing file instead\"\n\t\t#because we need to keep the file $DESTINATION_FOLDER/$pornstars-urls_of_galleries.txt + with the existing URL , in this condition, \n\t\t#no deletion of this file\nelse\n + \ echo \"looping through galleries\"\n\t\t#because we will re-generate this + file, can be emptied\n\t\techo \"\" > $DESTINATION_FOLDER/$pornstars-urls_of_galleries.txt\n + \ $curl_command -s $BASE_URL | grep -Eo \"(http|https)://[a-zA-Z0-9./?=_%:-]*\" + | grep \"https://www.pornpics.com/galleries/\" |sort -u >> $DESTINATION_FOLDER/$pornstars-urls_of_galleries.txt\nfi\n\necho + \"\" > $DESTINATION_FOLDER/$pornstars-image_to_download.txt\n\n\nfor f in $(cat + $DESTINATION_FOLDER/$pornstars-urls_of_galleries.txt ); do\n\t$curl_command $f + | grep -Eo \"(http|https)://[a-zA-Z0-9./?=_%:-]*\" | grep \"https://cdni.pornpics.com/\" + >> $DESTINATION_FOLDER/$pornstars-image_to_download.txt\n\techo $f\ndone\n\n\nfor + g in $(cat $DESTINATION_FOLDER/$pornstars-image_to_download.txt ); do \n\t$curl_command + -O --output-dir $DESTINATION_FOLDER/$pornstars $g\n# wget -e use_proxy=yes + -e http_proxy=$PROXY_HTTP $g -P $DESTINATION_FOLDER/$pornstars\n echo $g + \ndone\nexit 1\n" +kind: ConfigMap +metadata: + creationTimestamp: null + name: pornpic-dowloader + namespace: stash diff --git a/external_manifests/stash/stash-pornpics-dowloader.job.yml b/external_manifests/stash/stash-pornpics-dowloader.job.yml new file mode 100644 index 0000000..936d8e1 --- /dev/null +++ b/external_manifests/stash/stash-pornpics-dowloader.job.yml @@ -0,0 +1,37 @@ +apiVersion: batch/v1 +kind: Job +metadata: + name: stash-pornpics-dowloader + namespace: stash +spec: + template: + spec: + containers: + - name: stash-pornpics-dowloader + image: bash + imagePullPolicy: Always + volumeMounts: + - mountPath: /config + name: scripts + - mountPath: /data + mountPropagation: None + name: data-pvc + command: +# - /bin/bash +# - -c +# - sleep 3600 + - bash + - /config/download.sh + - --http_proxy + - "http://gluetun-svc.gluetun.svc:8888" + restartPolicy: Never + volumes: + - configMap: + defaultMode: 0744 + name: pornpic-dowloader + name: scripts + - name: data-pvc + persistentVolumeClaim: + claimName: data-pvc + backoffLimit: 4 +