y-server-tools/config/config.yaml

565 lines
12 KiB
YAML
Raw Normal View History

2024-12-12 09:56:49 +08:00
client:
version: v3
address: "172.16.20.2:9669,172.16.20.4:9669,172.16.20.5:9669"
user: root
password: 123456
concurrencyPerAddress: 10
reconnectInitialInterval: 1s
retry: 3
retryInitialInterval: 1s
manager:
spaceName: Y24_1206
batch: 128
readerConcurrency: 50
importerConcurrency: 512
statsInterval: 10s
hooks:
before:
- statements:
- |
2024-12-12 13:02:22 +08:00
USE Y24_1206;
2024-12-12 09:56:49 +08:00
wait: 10s
log:
level: INFO
console: true
sources:
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/user-0s.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
tags:
- name: user
mode: INSERT
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 12 # id 字段作为VID
2024-12-12 09:56:49 +08:00
props:
2024-12-12 13:02:22 +08:00
- name: "friends_count"
type: "INT"
index: 0
- name: "listed_count"
type: "INT"
2024-12-12 09:56:49 +08:00
index: 1
2024-12-12 13:02:22 +08:00
- name: "favourites_count"
type: "INT"
2024-12-12 09:56:49 +08:00
index: 2
2024-12-12 13:02:22 +08:00
- name: "verified"
2024-12-12 09:56:49 +08:00
type: "STRING"
index: 3
2024-12-12 13:02:22 +08:00
- name: "label"
type: "STRING"
2024-12-12 09:56:49 +08:00
index: 4
2024-12-12 13:02:22 +08:00
- name: "platform"
type: "STRING"
2024-12-12 09:56:49 +08:00
index: 5
2024-12-12 13:02:22 +08:00
- name: "url"
type: "STRING"
2024-12-12 09:56:49 +08:00
index: 6
- name: "gather_time"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 7
- name: "screen_name"
2024-12-12 09:56:49 +08:00
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 8
- name: "followers_count"
type: "INT"
2024-12-12 09:56:49 +08:00
index: 9
2024-12-12 13:02:22 +08:00
- name: "name"
2024-12-12 09:56:49 +08:00
type: "STRING"
index: 10
2024-12-12 13:02:22 +08:00
- name: "standpoint"
type: "STRING"
index: 11
- name: "position"
type: "STRING"
index: 13
- name: "region"
type: "STRING"
index: 14
- name: "register_location"
type: "STRING"
index: 15
- name: "party"
type: "STRING"
index: 16
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/post-0s.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
tags:
- name: post
mode: INSERT
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 11 # id 字段作为VID
2024-12-12 09:56:49 +08:00
props:
2024-12-12 13:02:22 +08:00
- name: "comment_count"
type: "INT"
index: 0
- name: "author_name"
2024-12-12 09:56:49 +08:00
type: "STRING"
index: 1
2024-12-12 13:02:22 +08:00
- name: "repost_count"
type: "INT"
2024-12-12 09:56:49 +08:00
index: 2
2024-12-12 13:02:22 +08:00
- name: "like_count"
type: "INT"
2024-12-12 09:56:49 +08:00
index: 3
2024-12-12 13:02:22 +08:00
- name: "keywords"
2024-12-12 09:56:49 +08:00
type: "STRING"
index: 4
2024-12-12 13:02:22 +08:00
- name: "platform"
type: "STRING"
2024-12-12 09:56:49 +08:00
index: 5
2024-12-12 13:02:22 +08:00
- name: "url"
type: "STRING"
2024-12-12 09:56:49 +08:00
index: 6
2024-12-12 13:02:22 +08:00
- name: "gather_time"
type: "STRING"
2024-12-12 09:56:49 +08:00
index: 7
2024-12-12 13:02:22 +08:00
- name: "post_relationship"
2024-12-12 09:56:49 +08:00
type: "STRING"
index: 8
2024-12-12 13:02:22 +08:00
- name: "publish_time"
2024-12-12 09:56:49 +08:00
type: "STRING"
index: 9
2024-12-12 13:02:22 +08:00
- name: "root_mid"
2024-12-12 09:56:49 +08:00
type: "STRING"
index: 10
2024-12-12 13:02:22 +08:00
- name: "author_id"
2024-12-12 09:56:49 +08:00
type: "STRING"
index: 12
2024-12-12 13:02:22 +08:00
- name: "source_task"
2024-12-12 09:56:49 +08:00
type: "STRING"
index: 13
2024-12-12 13:02:22 +08:00
- name: "view_count"
type: "INT"
index: 14
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/group-0s.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
tags:
- name: group
mode: INSERT
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 1 # id 字段作为VID
2024-12-12 09:56:49 +08:00
props:
- name: "name"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
2024-12-12 09:56:49 +08:00
- name: "member_count"
type: "INT"
2024-12-12 13:02:22 +08:00
index: 2
- name: "platform"
type: "STRING"
2024-12-12 09:56:49 +08:00
index: 3
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/organization-0s.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
tags:
- name: organization
mode: INSERT
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # id 字段作为VID
2024-12-12 09:56:49 +08:00
props:
- name: "name"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
- name: "industry"
2024-12-12 09:56:49 +08:00
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 1
2024-12-12 09:56:49 +08:00
- name: "member_count"
type: "STRING"
index: 3
2024-12-12 13:02:22 +08:00
- name: "platform"
2024-12-12 09:56:49 +08:00
type: "STRING"
index: 4
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/task-0s.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
tags:
- name: task
mode: INSERT
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # id 字段作为VID
2024-12-12 09:56:49 +08:00
props:
- name: "keywords"
2024-12-12 13:02:22 +08:00
type: "STRING"
index: 0
- name: "name"
2024-12-12 09:56:49 +08:00
type: "STRING"
index: 1
# 边数据
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/user_follow_user-0.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
edges:
- name: user_follow_user
mode: INSERT
src:
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # src
2024-12-12 09:56:49 +08:00
dst:
id:
type: "STRING"
index: 1 # dst
props:
- name: "extra_prop"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/user_follower_user-0.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
edges:
- name: user_follower_user
mode: INSERT
src:
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # src
2024-12-12 09:56:49 +08:00
dst:
id:
type: "STRING"
index: 1 # dst
props:
- name: "extra_prop"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/user_friend_user-0.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
edges:
- name: user_friend_user
mode: INSERT
src:
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # src
2024-12-12 09:56:49 +08:00
dst:
id:
type: "STRING"
index: 1 # dst
props:
- name: "extra_prop"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/user_have_post-0.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
edges:
- name: user_have_post
mode: INSERT
src:
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # src
2024-12-12 09:56:49 +08:00
dst:
id:
type: "STRING"
index: 1 # dst
props:
- name: "extra_prop"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 15:03:13 +08:00
# - path: /data/y4/tools/data/post_repost_post-0.csv
# csv:
# delimiter: "|"
# withHeader: true
# edges:
# - name: post_repost_post
# mode: INSERT
# src:
# id:
# type: "STRING"
# index: 2 # src
# dst:
# id:
# type: "STRING"
# index: 1 # dst
# props:
# - name: "extra_prop"
# type: "STRING"
# index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/post_cite_post-0.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
edges:
- name: post_cite_post
mode: INSERT
src:
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # src
2024-12-12 09:56:49 +08:00
dst:
id:
type: "STRING"
index: 1 # dst
props:
- name: "extra_prop"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
# - path: /data/y4/tools/data/user_post_post-0.csv
# csv:
# delimiter: "|"
# withHeader: true
# edges:
# - name: user_post_post
# mode: INSERT
# src:
# id:
# type: "STRING"
# index: 2 # src
# dst:
# id:
# type: "STRING"
# index: 1 # dst
# props:
# - name: "extra_prop"
# type: "STRING"
# index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/user_comment_user-0.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
edges:
- name: user_comment_user
mode: INSERT
src:
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # src
2024-12-12 09:56:49 +08:00
dst:
id:
type: "STRING"
index: 1 # dst
props:
- name: "extra_prop"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/post_comment_post-0.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
edges:
- name: post_comment_post
mode: INSERT
src:
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # src
2024-12-12 09:56:49 +08:00
dst:
id:
type: "STRING"
index: 1 # dst
props:
- name: "extra_prop"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/post_multicomment_post-0.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
edges:
- name: post_multicomment_post
mode: INSERT
src:
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # src
2024-12-12 09:56:49 +08:00
dst:
id:
type: "STRING"
index: 1 # dst
props:
- name: "extra_prop"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/user_join_group-0.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
edges:
- name: user_join_group
mode: INSERT
src:
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # src
2024-12-12 09:56:49 +08:00
dst:
id:
type: "STRING"
index: 1 # dst
props:
- name: "extra_prop"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/group_include_user-0.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
edges:
- name: group_include_user
mode: INSERT
src:
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # src
2024-12-12 09:56:49 +08:00
dst:
id:
type: "STRING"
index: 1 # dst
props:
- name: "extra_prop"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/user_participate_task-0.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
edges:
- name: user_participate_task
mode: INSERT
src:
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # src
2024-12-12 09:56:49 +08:00
dst:
id:
type: "STRING"
index: 1 # dst
props:
- name: "extra_prop"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/task_include_user-0.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
edges:
- name: task_include_user
mode: INSERT
src:
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # src
2024-12-12 09:56:49 +08:00
dst:
id:
type: "STRING"
index: 1 # dst
props:
- name: "extra_prop"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/post_in_task-0.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
edges:
- name: post_in_task
mode: INSERT
src:
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # src
2024-12-12 09:56:49 +08:00
dst:
id:
type: "STRING"
index: 1 # dst
props:
- name: "extra_prop"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/task_include_post-0.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
edges:
- name: task_include_post
mode: INSERT
src:
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # src
2024-12-12 09:56:49 +08:00
dst:
id:
type: "STRING"
index: 1 # dst
props:
- name: "extra_prop"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
- path: /data/y4/tools/data/organization_include_user-0.csv
2024-12-12 09:56:49 +08:00
csv:
2024-12-12 13:02:22 +08:00
delimiter: "|"
2024-12-12 09:56:49 +08:00
withHeader: true
edges:
- name: organization_include_user
mode: INSERT
src:
id:
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 2 # src
2024-12-12 09:56:49 +08:00
dst:
id:
type: "STRING"
index: 1 # dst
props:
- name: "extra_prop"
type: "STRING"
2024-12-12 13:02:22 +08:00
index: 0
2024-12-12 09:56:49 +08:00
2024-12-12 13:02:22 +08:00
# - path: /data/y4/tools/data/user_join_organization-0.csv
# csv:
# delimiter: "|"
# withHeader: true
# edges:
# - name: user_join_organization
# mode: INSERT
# src:
# id:
# type: "STRING"
# index: 2 # src
# dst:
# id:
# type: "STRING"
# index: 1 # dst
# props:
# - name: "extra_prop"
# type: "STRING"
# index: 0