org.diylc.swing.plugins.cloud.sync.CloudSyncJob mirrors .diy files that users publish on their own web pages into the DIYLC cloud, on behalf of users who agreed to it. Each run scrapes the configured pages for .diy links and downloads every file. A file that is new is uploaded as a project with a rendered thumbnail. A file whose content changed replaces the files of the project it was uploaded as. Unchanged files are skipped, and files that disappear from a page are left in the cloud.

Configuration

Copy cloud-sync.example.json to cloud-sync.json (git-ignored, because it holds login tokens). For each user:

username, token, machineId: the cloud API authenticates with the username (or email), not the numeric ID, plus a login token issued for that machine ID.
userId: names the user's folder under storageDir.
sourceUrl: optional page cited as the source in project descriptions, instead of the link to each file.
category, keywords: defaults for the user's pages. Category is the name shown in the upload dialog; the leading - of subcategories is optional.
pages: page URLs. Use an object with url, category and keywords to override the defaults for one page.

The project name comes from the file name, and the description from the file's own description followed by a Source: link to the user's sourceUrl, or to the file when there is none. When the source changes, the next run rewrites the Source: line of every project it cites, keeping the rest of the description and the other project details as they are in the cloud.

Running

mvn -q -pl diylc-swing -am package -DskipTests   # builds diylc-swing/target/diylc.jar
cloud-sync/cloud-sync.sh --dry-run                # download and render only
cloud-sync/cloud-sync.sh                          # sync for real

storage/<userId>/ mirrors the downloaded files with their thumbnails (*.diy.png), and keeps sync-state.json, which maps each source URL to its cloud project ID, the hash that was last uploaded and the source its description cites. If the state is lost, the next run takes over the user's existing projects by name instead of uploading duplicates. The exit status is non-zero if any page or file failed.