# -*- coding: utf-8 -*-
"""Which site is which, and which of them may be named.

Two sites are measured. One is the author's own and is named. The other is a
client's, read with its owner's agreement on the condition that it is not
identified, so everything downstream keys on an alias and the mapping from alias
to host lives in `sites.local.json`, which is in the never-publish set beside
`raw/`.

That is not decoration. A paper that anonymises a site in its prose and then
writes the real hostname into a CSV filename has not anonymised anything, so the
alias is applied at the point the raw log is read and no later stage ever sees a
host at all.

The public description of an anonymous site is here, in the open, because a reader
needs to know what kind of site produced the numbers even when they cannot know
which one it is.
"""
import io
import json
import os

HERE = os.path.dirname(os.path.abspath(__file__))
LOCAL = os.path.join(HERE, 'sites.local.json')

# alias -> what a reader is told about it. No hostname appears in this table.
PUBLIC = {
    'broikos.gr': dict(
        named=True,
        what='the author\'s own portfolio and research site',
        category='professional portfolio and published research',
        robots='names five AI crawlers and allows them on purpose'),
    'practice-a': dict(
        named=False,
        # One step wider than the truth on purpose. The reader needs to know
        # what kind of site produced the numbers; the specific profession does
        # no work in any claim here and narrows the field considerably in a
        # country this size.
        what='a Greek single-practitioner professional practice with a blog '
             'and online booking',
        category='professional services, single practitioner',
        robots='no AI-specific rules, so the default applies and every crawler '
               'is allowed'),
}


def _map():
    if os.path.exists(LOCAL):
        try:
            return json.load(io.open(LOCAL, encoding='utf-8'))
        except Exception:                                          # noqa: BLE001
            return {}
    return {}


def alias(host):
    """The published name for a host. Falls back to the host when it is named."""
    return _map().get(host, host)


def describe(al):
    return PUBLIC.get(al, dict(named=True, what='', category='', robots=''))


def aliases():
    return sorted(PUBLIC)


def outdir(al):
    d = os.path.join(HERE, 'sites', al)
    os.makedirs(d, exist_ok=True)
    return d