A single zip for all datasets (41 datasets with not classification), first link has not been used (not so useful).
Request is made by topic (9 topics for 38 datasets as a whole) using request to https://www.bundesbank.de/cae/servlet/StatisticDownload?its_fileFormat=Archive&mode=its&tree={topic_anchor}
for each topic, we download a zip containing:
```
- list/
- time series/
- {dataset1}/
...
- {datasetn}/
- tree/
```
## Processing
- XML files located in list/ and tree/ folders are ignored by convert script
- Only XML series files found in 'time series' / {datasetdir} are used to generate time series TSV files
- series info is used to populate time series `attributes` value in dataset.json
- observation info (BBK_DIFF, ...) is reported in time series TSV files
- Topics are used to distribute datasets in categories (category_tree.json)
### To solve yet:
- No dataset info is available but dataset code (dataset dir name). Dataset names are absent
This [page](https://www.bundesbank.de/Navigation/EN/Statistics/Statistics_search/Macro_economic_time_series/statistiksuche_its_node.html?nsc=true) gives all dataset names but how to get them in one download
### Tracks to work on
Otherwise, calling https://www.bundesbank.de/cae/servlet/StatisticDownload?metaDSI={dataset_code} returns XML file with lot of information:
- Dataset name seems to be in `<structure:Name>`
- With `<structure:Codelist>`, we can perhaps populate dimensions info in dataset.json and decoding time series code, we can give a dimension value for all time series of the dataset.
