brand-model parsing

[commenthol]

For device parsing a better segmentation is being introduced for Android and other Devices.

To add information on "brand" and "model" two new types brand_replacement and model_replacement have been added to the device_parser in the regexes.yaml file.

To facilitate segmentation a multi replacement function for "family", "brand" and "model" is introduced to cope with user-agent-strings which are quite different from each other but specify the same device.
Now $1 ... $9 can be used to reference up to nine matchers.

For correct matching of Generic Feature-/ Smartphones and Spiders case insensitive matching using regex_flag: 'i' has been added.

[tobie]

Isn't that just the same as saying:

regexp  = new RegExp(obj.regex, obj.regex_flag || "");

[tobie]

Think you could fix white space issues? Would make reviewing the PR a lot easier. Thanks.

[commenthol]

Do you have time to review?

[commenthol]

.trim() is equal to the previous .replace(/^\s+|\s+$/gm, '') but looks nicer.

E.g. for

Mozilla/5.0 (Linux; U; Android 4.2.1; de-de; HTC Vision Build/JOP40D) AppleWebKit/534.30 (KHTML, like Gecko) Version/4.0 Mobile Safari/534.30

  - regex: '; *(?:HTC[ ;])?(?:HTC(?:_blocked)*)(?:(?:[ _/]|; *)((?:(?![ _/;\(\)\\]|USCCHTC|HTC|Build|MIUI).)+))(?:[ _/]((?:(?![ _/;\(\)\\]|Build|MIUI|1\.0).)+))?(?:[ _/]((?:(?![ _/;\(\)\\]|Build|MIUI|1\.0).)+))?(?:[ _/]((?:(?![/;\)]|Build|MIUI|1\.0).)+))?'
    debug: '#0397'
    regex_flag: 'i'
    device_replacement: 'HTC $1 $2 $3 $4'
    brand_replacement: 'HTC'
    model_replacement: '$1 $2 $3 $4'

applies.
The optional match $4 does not apply here. The trailing whitespace from $3 $4 is corrected by trim.