Add code to properly extract tags from the incoming point in BoostSupport.cxx

[ItsGarand]

In BoostSupport.cxx queryImpl(), the response to the query (const auto response = transport->query(querty);) returns a JSON string and then loaded into a ptree.
so, all field data is represented as a string in the ptree.

The decision was made to make all strings found in fields into tags. The original tags if any are marked as such in the incoming data so they should be preserved that way in the reconstructed point. String data in fields are still fields and should not be added as tags.

so.....

at line 58 insert:

Point point{series.second.getstd::string("name", "")};
// add the tags if they exist
for (auto &tag : series.second.get_child("tags"))
{
std::string tagName = tag.first;
std::string tagValue = tag.second.data();
point.addTag(tagName, tagValue);
}
// add the field data

When adding fileds, don't assume a double - Influx is flexible - data could be int or float or string as well. The code could try to determine the type by doing some range/format checking(int/long/float/double/string) and doing the conversion here. I would suggest leaving it as a string representation and letting the user do the required conversion.
Change line 76 from:
point.addField(column, boost::lexical_cast(value));
to:
point.addField(column, value); *** See ticket 115

Don't assume string data is automatically a tag.
Remove line 80: point.addTag(column, value);
in it's place add a warning to cout saying field could not be added:
std::cout << Error adding field to point - Column: " << column << " Value: " << value << std::endl;

[offa]

Indeed, the json query handling needs some improvements.

The original tags if any are marked as such in the incoming data so they should be preserved that way in the reconstructed point.

Is this 2.x (1.x doesn't contain any tags in the response json)?

I would suggest leaving it as a string representation and letting the user do the required conversion.

Good point, this might be the safest way.

[ItsGarand]

I am using Influxdb 1.7.10. If your query contains a GROUP BY clause and you use your tags, the output JSON will contain a tags structure

So, if you have a set of points like:
time, TAG1="A", TAG2="B", DEV_NUM="1" number1=10, number2=20, number3=30

query = SELECT number1, number2 FROM my_measurement WHERE DEV_NUM=1 GROUP BY TAG1, TAG2, DEV_NUM

The response JSON will look like:
{"result":[{statement_id=0, "series":[{"name":"results", "tags":{"TAG1":"A", "TAG2":"B", DEV_NUM="1"}, "columns": ["time", "number1", "number2"], "values": [...]}]}]}

[offa]

Thanks, group by was the missing hint.

In cases where there's already a tags element it makes sense to use it. I'll have a look at it. 👍

[offa]

Merged.