# syncazoom Zoom is an imporant tool for online communication. Especially in these times of social distancing This tool is made to help you manage API calls to Zoom and make it easy to store a local copy of Zoom meeting metrics. # Getting Started ### Add your keys copy the example file and rename it to `config.toml` ```toml [creds] key = "EXAMPLE" secret = "EXAMPLE" [settings] seconds_between_calls = 60 cron_interval = "0 0 3,6,9,12,15,18,21 * * Mon-Fri *" [slack] webhook = "EXAMPLE" ``` ### Install from cargo (not complete) ```bash # install rust cargo install syncazoom ``` ### Run app ```bash syncazoom -c config.toml ``` ### Query ```bash sqlite3 meetings.sql3 # SQLite version 3.28.0 2019-04-15 14:49:49 # Enter ".help" for usage hints. # sqlite> ``` Then enter the following (or more useful) SQL ```sql SELECT * FROM meetings LIMIT 10; ``` # Example Output ``` 03/04/2020 08:57:00 Start run .oooo.o oooo ooo ooo. .oo. .ooooo. .oooo. oooooooo .ooooo. .ooooo. ooo. .oo. .oo. d88( "8 `88. .8' `888P"Y88b d88' `"Y8 `P )88b d'""7d8P d88' `88b d88' `88b `888P"Y88bP"Y88b `"Y88b. `88..8' 888 888 888 .oP"888 .d8P' 888 888 888 888 888 888 888 o. )88b `888' 888 888 888 .o8 d8( 888 .d8P' .P 888 888 888 888 888 888 888 8""888P' .8' o888o o888o `Y8bod8P' `Y888""8o d8888888P `Y8bod8P' `Y8bod8P' o888o o888o o888o .o..P' `Y8P' URL | https://api.zoom.us/v2/metrics/meetings?type=past&page_size=300 Total Estimated Runtime 50 mins ─── Log ° ├── Start | 2020-04-02 ├── End | 2020-04-03 ├── Pages | 50 ├── Current | 1 ├── Per Page | 300 ├── Total | 14938 └── Remaining | 14888 ° ├── 3vjYIIsiRbOB9hjeluH6mA== ├── Alice Human's Zoom Meeting ├── Alice Human ├── 2020-04-02T00:00:01Z └── 2020-04-02T00:11:04Z ° ├── XCjX59nUS2W0rz+5zllm0g== ├── Bob Person's Zoom Meeting ├── Bob Person ├── 2020-04-02T00:40:38Z └── 2020-04-02T02:04:48Z ° └── Next Page Token: "k89o4smQsnuOGt03Z0h57EU1u3vYe0GU9a2" ``` # Developer ### Install from source ```bash # install rust cargo build --release ``` ### How it works An essential concept of syncazoom is it's heartbeat. At some interval, it will check if it is currently fetching data - and if not, it will start the process. The process makes a call to Zoom's metrics API, and then it stores the data in a local instance of sqlite3. We only "insert or replace" on each uuid so we never have duplicate entries.