DataLaVista™¶
Architecture¶
This page describes how DataLaVista is put together and how data moves through it. It is for developers, maintainers, and reviewers who need to understand the app before changing it or deciding whether it fits their environment. DataLaVista is a single script that runs in the browser, with no server of its own.
The pieces¶
The built file dist/datalavista.js holds everything the app needs:
| Piece | Role |
|---|---|
Application code, from the numbered files in src/ |
The tabs, the query builder, the designer, the preview, and the report generator. |
| AlaSQL | Runs SQL queries in the browser against the loaded tables. |
| sql.js | Reads SQLite files. AlaSQL hands those files to it. |
| SheetJS | Reads Excel files. |
| ECharts, with the violin chart add-on | Draws the charts. |
| CodeMirror | The SQL editor, with highlighting and suggestions. |
| lz-string | Compresses the report configuration for links. |
| Page markup and styles | Built into the script, so one div on any page is enough. |
The same script is packed three ways: inline in dist/DataLaVista.html, loaded
by dist/DataLaVista-nojs.html, and loaded from a SharePoint site's Site Assets
library by the web part code in dist/DataLaVista-sp.html.txt. The
Repository maintenance page explains the build, and
Third-party notices lists each library's license.
Data flow¶
Data moves through six stages in the browser, so each widget stays fast and independent of the others:
- Raw table. The data as loaded from the source: a SharePoint list, a file on the web, or an uploaded file.
- Views. A virtual copy of each raw table with the extra fields DataLaVista adds for lookups, people, links, and lists, described in Connect to data.
- Query. The SQL from the Query Builder or the SQL Editor, which joins, filters, renames, and reshapes the views into one flat table.
- Materialized view. The query result, kept in memory. This is the table the dashboard uses.
- View with global filters. The materialized view with the dashboard's filter selections applied.
- Query per widget. Each widget runs its own small query against the filtered view to group and total the values it shows.
A change to a filter reruns stages 5 and 6 only. A change to the query reruns stages 3 to 6. Reconnecting a source starts again from stage 1.
Deployment¶
DataLaVista is distributed as one script with no file to install on a server. It runs:
- inside a SharePoint page, where it reads the site's lists with the signed-in person's own permissions,
- inside another web application, as an
iframeor an embedded script, - from a file on a computer, with uploaded files, or
- from any static web server, such as this website.
There is no backend service, no database of its own, and no server-side interface. Permissions come from the platform that hosts the page and the places that hold the data.
Conclusion¶
You now know the pieces inside the built script, the six stages data passes through, and where the app can run. For the build, the tests, and how the website is published, read Repository maintenance. For the security and accessibility controls, read Compliance.