ffdn-db/ffdnispdb/crawler.py

import io
import cgi
import pytz
from datetime import datetime, timedelta
from werkzeug.http import parse_date
from flask import escape, json
import requests

from ispformat.validator import validate_isp
from .models import ISP
from . import app


def get_encoding(content_type):
    content_type, params = cgi.parse_header(content_type)

    if 'charset' in params:
        return params['charset'].strip("'\"")


class Crawler(object):

    MAX_JSON_SIZE=1*1024*1024

    escape=staticmethod(lambda x: unicode(str(x), 'utf8') if type(x) != unicode else x)

    def __init__(self):
        self.success=False
        self.modified=True
        self.jdict={}
        self.cache_info=None
        self.jdict_max_age=self.config('DEFAULT_CACHE_TIME')

    def m(self, msg, evt=None):
        if not evt:
            return u'%s\n'%msg
        else:
            return u''

    def err(self, msg, *args):
        return self.m(u'! %s'%msg, *args)

    def warn(self, msg):
        return self.m(u'@ %s'%msg)

    def info(self, msg):
        return self.m(u'\u2013 %s'%msg)

    def abort(self, msg):
        raise NotImplemented

    def color(self, color, msg):
        return msg

    def bold(self, msg):
        return msg

    def italics(self, msg):
        return msg

    def nl(self):
        return self.m('')

    def format_validation_errors(self, errs):
        r=[]
        for e in errs:
            r.append(u'    %s: %s'%('.'.join(list(e.schema_path)[1:]), e.message))

        return u'\n'.join(r)

    def pre_done_cb(self, *args):
        pass

    def done_cb(self):
        pass

    def config(self, name):
        return app.config.get('CRAWLER_'+name)

    def parse_cache_control(self, _cachectl):
        cachectl={}
        for cc in _cachectl.split(','):
            cc=cc.strip()
            if not cc:
                continue
            cc=cc.split('=')
            if cc[0] not in ('max-age', 's-maxage'):
                continue
            try:
                cachectl[cc[0]]=cc[1]
            except IndexError:
                cachectl[cc[0]]=True
        return cachectl

    def __call__(self, url, cache_info={}):
        esc=self.escape
        yield self.m('Starting the validation process...')
        r=None
        try:
            yield self.m('* Attempting to retreive %s'%self.bold(url))
            headers={'User-Agent': 'FFDN DB validator'}
            if cache_info.get('etag'):
                headers['If-None-Match'] = cache_info['etag']
            if cache_info.get('last-modified'):
                headers['If-Modified-Since'] = cache_info['last-modified']
            r=requests.get(url, verify='/etc/ssl/certs/ca-certificates.crt',
                           headers=headers, stream=True, timeout=10)
        except requests.exceptions.SSLError as e:
            yield self.err('Unable to connect, SSL Error: '+self.color('#dd1144', esc(e)))
        except requests.exceptions.ConnectionError as e:
            yield self.err('Unable to connect: '+self.color('#dd1144', esc(e)))
        except requests.exceptions.Timeout as e:
            yield self.err('Connection timeout')
        except requests.exceptions.TooManyRedirects as e:
            yield self.err('Too many redirects')
        except requests.exceptions.RequestException as e:
            yield self.err('Internal request exception')
#        except Exception as e:
#            yield self.err('Unexpected request exception')

        if r is None:
            yield self.abort('Connection could not be established, aborting')
            return

        yield self.info('Connection established')

        yield self.info('Response code: '+self.bold(str(r.status_code)+' '+esc(r.reason)))
        try:
            r.raise_for_status()
        except requests.exceptions.HTTPError as e:
            yield self.err('Response code indicates an error')
            yield self.abort('Invalid response code')
            return

        _cachecontrol=r.headers.get('cache-control')
        cachecontrol=self.parse_cache_control(_cachecontrol) if _cachecontrol else None
        max_age=None
        if cachecontrol:
            try:
                _maxage=cachecontrol.get('max-age')
                _maxage=cachecontrol.get('s-maxage', _maxage) # s-maxage takes precedence
                max_age=int(_maxage)
            except ValueError:
                yield self.warn('Invalid max-age '+esc(_maxage))

            yield self.info('Cache control: '+self.bold(esc(
                ', '.join([k+'='+v if type(v) != bool else k for k, v in cachecontrol.iteritems()]))
            ))

        _expires=r.headers.get('expires')
        expires=parse_date(_expires)
        if expires:
            _now=r.headers.get('date')
            if _now: # use server date when possible
                now=parse_date(_now)
            else:
                now=datetime.utcnow()

            if expires > now:
                expires=(expires-now).total_seconds()
                yield self.info('Expires: '+self.bold(esc(_expires)))
            else:
                yield self.warn('Invalid Expires header. Expiry date must be in the future.')
                expires=None
        elif _expires and not expires:
            yield self.warn('Invalid Expires header %r'%esc(_expires))

        if not max_age and not expires:
            yield self.warn('No valid expiration time provided ! Please provide it either '
                             'with a Cache-Control or Expires header.')
            max_age=self.config('DEFAULT_CACHE_TIME')
            yield self.info('Using default expiration time of %d seconds'%(max_age))

        self.jdict_max_age = max_age if max_age else expires
        self.jdict_max_age = min(
            self.config('MAX_CACHE_TIME'),
            max(self.config('MIN_CACHE_TIME'), self.jdict_max_age)
        )
        yield self.info('Next update will be in %s'%(timedelta(seconds=self.jdict_max_age)))


        etag=r.headers.get('etag')
        last_modified=r.headers.get('last-modified')
        if not etag and not last_modified:
            yield self.warn('Please, provide at an ETag or Last-Modified header for '
                            'conditional requests')

        self.cache_info={}
        if etag:
            self.cache_info['etag']=etag
        if last_modified:
            self.cache_info['last-modified']=last_modified

        if cache_info and r.status_code == 304: # not modified
            self.m('== '+self.color('forestgreen', 'Response not modified. All good !'))
            self.modified=False
            self.success=True
            self.done_cb()
            return


        yield self.info('Content type: '+self.bold(esc(r.headers.get('content-type', 'not defined'))))
        if not r.headers.get('content-type'):
            yield self.err('Content-type '+self.bold('MUST')+' be defined')
            yield self.abort('The file must have a proper content-type to continue')
            return
        elif r.headers.get('content-type').lower() != 'application/json':
            yield self.warn('Content-type '+self.italics('SHOULD')+' be application/json')

        encoding=get_encoding(r.headers.get('content-type'))
        if not encoding:
            yield self.warn('Encoding not set. Assuming it\'s unicode, as per RFC4627 section 3')

        yield self.info('Content length: %s'%(self.bold(esc(r.headers.get('content-length', 'not set')))))

        cl=r.headers.get('content-length')
        if not cl:
            yield self.warn('No content-length. Note that we will not process a file whose size exceed 1MiB')
        elif int(cl) > self.MAX_JSON_SIZE:
            yield self.abort('File too big ! File size must be less then 1MiB')
            return


        yield self.info('Reading response into memory...')
        b=io.BytesIO()
        for d in r.iter_content(requests.models.CONTENT_CHUNK_SIZE):
            b.write(d)
            if b.tell() > self.MAX_JSON_SIZE:
                yield self.abort('File too big ! File size must be less then 1MiB')
                return
        r._content=b.getvalue()
        del b
        yield self.info('Successfully read %d bytes'%len(r.content))

        yield self.nl()+self.m('* Parsing the JSON file')
        if not encoding:
            charset=requests.utils.guess_json_utf(r.content)
            if not charset:
                yield self.err('Unable to guess unicode charset')
                yield self.abort('The file MUST be unicode-encoded when no explicit charset is in the content-type')
                return

            yield self.info('Guessed charset: '+self.bold(charset))

        try:
            txt=r.content.decode(encoding or charset)
            yield self.info('Successfully decoded file as %s'%esc(encoding or charset))
        except LookupError as e:
            yield self.err('Invalid/unknown charset: %s'%esc(e))
            yield self.abort('Charset error, Cannot continue')
            return
        except UnicodeDecodeError as e:
            yield self.err('Unicode decode error: %s'%e)
            yield self.abort('Charset error, cannot continue')
            return
        except Exception:
            yield self.abort('Unexpected charset error')
            return

        jdict=None
        try:
            jdict=json.loads(txt)
        except ValueError as e:
            yield self.err('Error while parsing JSON: %s'%esc(e))
        except Exception as e:
            yield self.err('Unexpected error while parsing JSON: %s'%esc(e))

        if not jdict:
            yield self.abort('Could not parse JSON')
            return

        yield self.info('JSON parsed successfully')

        yield self.nl()+self.m('* Validating the JSON against the schema')

        v=list(validate_isp(jdict))
        if v:
            yield self.err('Validation errors:')+self.format_validation_errors(v)
            yield self.abort('Your JSON file does not follow the schema, please fix it')
            return
        else:
            yield self.info('Done. No errors encountered \o')

        ret=self.pre_done_cb(jdict)
        if ret:
            yield ret
            return

        yield (self.nl()+self.m('== '+self.color('forestgreen', 'All good ! You can click on Confirm now'))+
               self.m(json.dumps({'passed': 1}), 'control'))

        self.jdict=jdict
        self.success=True
        self.done_cb()



class PrettyValidator(Crawler):

    def __init__(self, session=None, sesskey=None, *args, **kwargs):
        super(PrettyValidator, self).__init__(*args, **kwargs)
        self.session=session
        self.sesskey=sesskey
        self.escape=lambda x: escape(unicode(str(x), 'utf8') if type(x) != unicode else x)

    def m(self, msg, evt=None):
        return u'%sdata: %s\n\n'%(u'event: %s\n'%evt if evt else '', msg)

    def err(self, msg, *args):
        return self.m(u'<strong style="color: crimson">!</strong> %s'%msg, *args)

    def warn(self, msg):
        return self.m(u'<strong style="color: dodgerblue">@</strong> %s'%msg)

    def info(self, msg):
        return self.m(u'&ndash; %s'%msg)

    def abort(self, msg):
        return (self.m(u'<br />== <span style="color: crimson">%s</span>'%msg)+
                self.m(json.dumps({'closed': 1}), 'control'))

    def bold(self, msg):
        return u'<strong>%s</strong>'%msg

    def italics(self, msg):
        return u'<em>%s</em>'%msg

    def color(self, color, msg):
        return u'<span style="color: %s">%s</span>'%(color, msg)

    def format_validation_errors(self, errs):
        lns=super(PrettyValidator, self).format_validation_errors(errs)
        buf=u''
        for l in lns.split('\n'):
            buf+=self.m(self.escape(l))
        return buf

    def done_cb(self):
        self.session[self.sesskey]['validated']=True
        self.session[self.sesskey]['jdict']=self.jdict
        self.session[self.sesskey]['cache_info']=self.cache_info
        self.session[self.sesskey]['last_update']=datetime.now()
        self.session[self.sesskey]['next_update']=datetime.now()+timedelta(seconds=self.jdict_max_age)
        self.session.save()


class WebValidator(PrettyValidator):
    def pre_done_cb(self, jdict):
        # check name uniqueness
        where = (ISP.name == jdict['name'])
        if 'shortname' in jdict and jdict['shortname']:
            where |= (ISP.shortname == jdict.get('shortname'))
        if ISP.query.filter(where).count() > 0:
            ret = self.nl()
            ret += self.err('An ISP named "%s" already exist in our database'%self.escape(
                jdict['name']+(' ('+jdict['shortname']+')' if jdict.get('shortname') else '')
            ))
            ret += self.abort('The name of your ISP must be unique')
            return ret


class TextValidator(Crawler):
    def abort(self, msg):
        res=u'FATAL ERROR: %s\n'%msg
        pad=u'='*(len(res)-1)+'\n'
        return self.m(pad+res+pad)