PNG  IHDRxsBIT|d pHYs+tEXtSoftwarewww.inkscape.org<,tEXtComment File Manager

File Manager

Path: /opt/cloudlinux/venv/lib/python3.11/site-packages/wmt/common/

Viewing File: report.py

#!/opt/cloudlinux/venv/bin/python3

import re
from urllib.parse import urlparse
from dataclasses import dataclass, asdict
from sqlalchemy import func
from sqlalchemy.sql.expression import literal_column
from wmt.common import cfg
from wmt.common.utils import get_domains

from wmt.db import ScrapeResult, session_scope


@dataclass
class SummaryReport:
    count_all: int
    count_successful: int
    count_failed: int
    count_undone: int
    average_time: float

    def to_template(self, *args):
        return [self.count_all,
                self.count_successful,
                self.count_failed,
                self.count_undone,
                # emails need time in ms
                int(self.average_time / 10**3)]


@dataclass
class ErrorReport:
    code: str
    count_errors: int
    url: str

    def to_template(self, alternative):
        url = url_to_domain(self.url)
        if alternative == 'html':
            url = f'<a href="{self.url}">{url}</a>'
        return [url,
                self.count_errors,
                self.code]


@dataclass
class DurationReport:
    url: str
    average_time: float

    def to_template(self, alternative):
        url = url_to_domain(self.url)
        if alternative == 'html':
            url = f'<a href="{self.url}">{url}</a>'
        return [url,
                # emails need time in ms
                int(self.average_time / 10**3)]


def url_to_domain(url):
    pattern = r'http(s)?://'
    return re.sub(pattern, '', url)


def generate_report(engine, start_date, end_date):
    with session_scope(engine) as session:
        # gets counter per status code per website -> group key: website: status_code pair
        # e.g (test.com 404 3), (test.com 500 2)
        subquery = session.query(ScrapeResult.response_code,
                                 ScrapeResult.website,
                                 func.count().label('err_count')).\
            filter(ScrapeResult.create_date >= start_date,
                   ScrapeResult.create_date <= end_date,
                   ScrapeResult.response_code != 200,
                   ScrapeResult.is_finished == True)\
            .group_by(ScrapeResult.response_code, ScrapeResult.website)\
            .subquery()

        # group previous subquery by website
        # code     count  website
        # [('451,500', 3, 'http://www.flightradar24.com'),
        #  ('404', 2, 'http://broken.com')]
        error_stats = session.query(func.group_concat(subquery.c.response_code),
                                    func.sum(subquery.c.err_count),
                                    subquery.c.website)\
            .group_by(subquery.c.website)\
            .all()
        #  website                          avg ms             count
        # [('http://www.stackoverflow.com', 538.0816599732262, 2241),
        # ('http://www.suser.com',          66.53859883980365, 2241)]
        success_stats = session.query(ScrapeResult.website,
                                      func.avg(ScrapeResult.response_time_ms).label('average_time'),
                                      func.count())\
            .filter(ScrapeResult.create_date >= start_date,
                    ScrapeResult.create_date <= end_date,
                    ScrapeResult.response_code == 200)\
            .group_by(ScrapeResult.website)\
            .order_by(literal_column('average_time').desc()) \
            .all()

        count_unsuccessful = session.query(ScrapeResult)\
            .filter(ScrapeResult.create_date >= start_date,
                    ScrapeResult.create_date <= end_date,
                    ScrapeResult.is_finished == False)\
            .count()

    # filter out not present domains using host-only comparison and tolerating www/non-www
    def extract_hostname_only(url_value: str):
        parsed = urlparse(url_value, 'http')
        hostname = parsed.netloc or parsed.path
        return hostname.lower().strip('/')

    allowed_hosts = set()
    for domain in get_domains():
        hostname = extract_hostname_only(domain)
        allowed_hosts.add(hostname)
        # include both www and non-www forms for matching
        if hostname.startswith('www.'):
            allowed_hosts.add(hostname[4:])
        else:
            allowed_hosts.add('www.' + hostname)

    success_stats = [
        (url, average_time, count)
        for url, average_time, count in success_stats
        if not cfg.is_domain_ignored(url) and extract_hostname_only(url) in allowed_hosts
    ]

    error_stats = [
        (code, count, url)
        for code, count, url in error_stats
        if not cfg.is_domain_ignored(url) and extract_hostname_only(url) in allowed_hosts
    ]

    error_report = [ErrorReport(code=code, count_errors=count_errors, url=url)
                    for code, count_errors, url in error_stats]

    # Collapse by host: prefer https entry if it exists for the host; otherwise keep the first
    # implemented because in report we do not want to show both http and https entries for the same host
    host_choice = {}
    for url, average_time, _ in success_stats:
        host = extract_hostname_only(url)
        is_https = str(url).lower().startswith('https://')
        chosen = host_choice.get(host)
        if chosen is None:
            host_choice[host] = (url, average_time, is_https)
        else:
            # upgrade to https if available
            if not chosen[2] and is_https:
                host_choice[host] = (url, average_time, True)

    duration_report = [DurationReport(url=item[0], average_time=int(round(item[1] * 1000)))
                       for item in host_choice.values()]

    successful_requests_count = sum(success_count for url, _, success_count in success_stats)
    error_requests_count = sum(errors_count for _, errors_count, url in error_stats)

    averages = [item[1] for item in success_stats]
    average_count = 0 if not averages else int(round(1000 * sum(averages) / len(averages)))
    summary_report = SummaryReport(count_all=successful_requests_count + error_requests_count + count_unsuccessful,
                                   count_successful=successful_requests_count,
                                   count_failed=error_requests_count,
                                   count_undone=count_unsuccessful,
                                   average_time=average_count)
    return {
        'summary_report': summary_report,
        'error_report': error_report,
        'duration_report': duration_report
    }


def report_dict(report):
    return {
        'summary_report': asdict(report['summary_report']),
        'error_report': [asdict(item) for item in report['error_report']],
        'duration_report': [asdict(item) for item in report['duration_report']]
    }
b IDATxytVսϓ22 A@IR :hCiZ[v*E:WũZA ^dQeQ @ !jZ'>gsV仿$|?g)&x-EIENT ;@xT.i%-X}SvS5.r/UHz^_$-W"w)Ɗ/@Z &IoX P$K}JzX:;` &, ŋui,e6mX ԵrKb1ԗ)DADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADA݀!I*]R;I2$eZ#ORZSrr6mteffu*((Pu'v{DIߔ4^pIm'77WEEE;vƎ4-$]'RI{\I&G :IHJ DWBB=\WR޽m o$K(V9ABB.}jѢv`^?IOȅ} ڶmG}T#FJ`56$-ھ}FI&v;0(h;Б38CӧOWf!;A i:F_m9s&|q%=#wZprrrla A &P\\СC[A#! {olF} `E2}MK/vV)i{4BffV\|ۭX`b@kɶ@%i$K z5zhmX[IXZ` 'b%$r5M4º/l ԃߖxhʔ)[@=} K6IM}^5k㏷݆z ΗÿO:gdGBmyT/@+Vɶ纽z񕏵l.y޴it뭷zV0[Y^>Wsqs}\/@$(T7f.InݺiR$푔n.~?H))\ZRW'Mo~v Ov6oԃxz! S,&xm/yɞԟ?'uaSѽb,8GלKboi&3t7Y,)JJ c[nzӳdE&KsZLӄ I?@&%ӟ۶mSMMњ0iؐSZ,|J+N ~,0A0!5%Q-YQQa3}$_vVrf9f?S8`zDADADADADADADADADAdqP,تmMmg1V?rSI꒟]u|l RCyEf٢9 jURbztѰ!m5~tGj2DhG*{H9)꒟ר3:(+3\?/;TUݭʴ~S6lڧUJ*i$d(#=Yݺd{,p|3B))q:vN0Y.jkק6;SɶVzHJJЀ-utѹսk>QUU\޲~]fFnK?&ߡ5b=z9)^|u_k-[y%ZNU6 7Mi:]ۦtk[n X(e6Bb."8cۭ|~teuuw|ήI-5"~Uk;ZicEmN/:]M> cQ^uiƞ??Ңpc#TUU3UakNwA`:Y_V-8.KKfRitv޲* 9S6ֿj,ՃNOMߤ]z^fOh|<>@Å5 _/Iu?{SY4hK/2]4%it5q]GGe2%iR| W&f*^]??vq[LgE_3f}Fxu~}qd-ږFxu~I N>\;͗O֊:̗WJ@BhW=y|GgwܷH_NY?)Tdi'?խwhlmQi !SUUsw4kӺe4rfxu-[nHtMFj}H_u~w>)oV}(T'ebʒv3_[+vn@Ȭ\S}ot}w=kHFnxg S 0eޢm~l}uqZfFoZuuEg `zt~? b;t%>WTkķh[2eG8LIWx,^\thrl^Ϊ{=dž<}qV@ ⠨Wy^LF_>0UkDuʫuCs$)Iv:IK;6ֲ4{^6եm+l3>݆uM 9u?>Zc }g~qhKwڭeFMM~pМuqǿz6Tb@8@Y|jx](^]gf}M"tG -w.@vOqh~/HII`S[l.6nØXL9vUcOoB\xoǤ'T&IǍQw_wpv[kmO{w~>#=P1Pɞa-we:iǏlHo׈꒟f9SzH?+shk%Fs:qVhqY`jvO'ρ?PyX3lх]˾uV{ݞ]1,MzYNW~̈́ joYn}ȚF߾׮mS]F z+EDxm/d{F{-W-4wY듏:??_gPf ^3ecg ҵs8R2מz@TANGj)}CNi/R~}c:5{!ZHӋӾ6}T]G]7W6^n 9*,YqOZj:P?Q DFL|?-^.Ɵ7}fFh׶xe2Pscz1&5\cn[=Vn[ĶE鎀uˌd3GII k;lNmشOuuRVfBE]ۣeӶu :X-[(er4~LHi6:Ѻ@ԅrST0trk%$Č0ez" *z"T/X9|8.C5Feg}CQ%͞ˣJvL/?j^h&9xF`њZ(&yF&Iݻfg#W;3^{Wo^4'vV[[K';+mӍִ]AC@W?1^{එyh +^]fm~iԵ]AB@WTk̏t uR?l.OIHiYyԶ]Aˀ7c:q}ힽaf6Z~қm(+sK4{^6}T*UUu]n.:kx{:2 _m=sAߤU@?Z-Vކеz왍Nэ{|5 pڶn b p-@sPg]0G7fy-M{GCF'%{4`=$-Ge\ eU:m+Zt'WjO!OAF@ik&t݆ϥ_ e}=]"Wz_.͜E3leWFih|t-wZۍ-uw=6YN{6|} |*={Ѽn.S.z1zjۻTH]흾 DuDvmvK.`V]yY~sI@t?/ϓ. m&["+P?MzovVЫG3-GRR[(!!\_,^%?v@ҵő m`Y)tem8GMx.))A]Y i`ViW`?^~!S#^+ѽGZj?Vģ0.))A꨷lzL*]OXrY`DBBLOj{-MH'ii-ϰ ok7^ )쭡b]UXSְmռY|5*cֽk0B7镹%ڽP#8nȎq}mJr23_>lE5$iwui+ H~F`IjƵ@q \ @#qG0".0" l`„.0! ,AQHN6qzkKJ#o;`Xv2>,tێJJ7Z/*A .@fفjMzkg @TvZH3Zxu6Ra'%O?/dQ5xYkU]Rֽkق@DaS^RSּ5|BeHNN͘p HvcYcC5:y #`οb;z2.!kr}gUWkyZn=f Pvsn3p~;4p˚=ē~NmI] ¾ 0lH[_L hsh_ғߤc_њec)g7VIZ5yrgk̞W#IjӪv>՞y睝M8[|]\շ8M6%|@PZڨI-m>=k='aiRo-x?>Q.}`Ȏ:Wsmu u > .@,&;+!!˱tﭧDQwRW\vF\~Q7>spYw$%A~;~}6¾ g&if_=j,v+UL1(tWake:@Ș>j$Gq2t7S?vL|]u/ .(0E6Mk6hiۺzښOrifޱxm/Gx> Lal%%~{lBsR4*}{0Z/tNIɚpV^#Lf:u@k#RSu =S^ZyuR/.@n&΃z~B=0eg뺆#,Þ[B/?H uUf7y Wy}Bwegל`Wh(||`l`.;Ws?V@"c:iɍL֯PGv6zctM̠':wuW;d=;EveD}9J@B(0iհ bvP1{\P&G7D޴Iy_$-Qjm~Yrr&]CDv%bh|Yzni_ˆR;kg}nJOIIwyuL}{ЌNj}:+3Y?:WJ/N+Rzd=hb;dj͒suݔ@NKMԄ jqzC5@y°hL m;*5ezᕏ=ep XL n?מ:r`۵tŤZ|1v`V뽧_csج'ߤ%oTuumk%%%h)uy]Nk[n 'b2 l.=͜E%gf$[c;s:V-͞WߤWh-j7]4=F-X]>ZLSi[Y*We;Zan(ӇW|e(HNNP5[= r4tP &0<pc#`vTNV GFqvTi*Tyam$ߏWyE*VJKMTfFw>'$-ؽ.Ho.8c"@DADADADADADADADADA~j*֘,N;Pi3599h=goضLgiJ5փy~}&Zd9p֚ e:|hL``b/d9p? fgg+%%hMgXosج, ΩOl0Zh=xdjLmhݻoO[g_l,8a]٭+ӧ0$I]c]:粹:Teꢢ"5a^Kgh,&= =՟^߶“ߢE ܹS J}I%:8 IDAT~,9/ʃPW'Mo}zNƍ쨓zPbNZ~^z=4mswg;5 Y~SVMRXUյڱRf?s:w ;6H:ºi5-maM&O3;1IKeamZh͛7+##v+c ~u~ca]GnF'ټL~PPPbn voC4R,ӟgg %hq}@#M4IÇ Oy^xMZx ) yOw@HkN˖-Sǎmb]X@n+i͖!++K3gd\$mt$^YfJ\8PRF)77Wא!Cl$i:@@_oG I{$# 8磌ŋ91A (Im7֭>}ߴJq7ޗt^ -[ԩSj*}%]&' -ɓ'ꫯVzzvB#;a 7@GxI{j޼ƌ.LÇWBB7`O"I$/@R @eee@۷>}0,ɒ2$53Xs|cS~rpTYYY} kHc %&k.], @ADADADADADADADADA@lT<%''*Lo^={رc5h %$+CnܸQ3fҥK}vUVVs9G R,_{xˇ3o߾;TTTd}馛]uuuG~iԩ@4bnvmvfϞ /Peeeq}}za I~,誫{UWW뮻}_~YƍSMMMYχ֝waw\ďcxꩧtEƍկ_?۷5@u?1kNׯWzz/wy>}zj3 k(ٺuq_Zvf̘:~ ABQ&r|!%KҥKgԞ={<_X-z !CyFUUz~ ABQIIIjݺW$UXXDٳZ~ ABQƍecW$<(~<RSSvZujjjԧOZQu@4 8m&&&jԩg$ď1h ͟?_{768@g =@`)))5o6m3)ѣƌJ;wҿUTT /KZR{~a=@0o<*狔iFɶ[ˎ;T]]OX@?K.ۈxN pppppppppppppppppPfl߾] ,{ァk۶mڿo5BTӦMӴiӴ|r DB2e|An!Dy'tkΝ[A $***t5' "!駟oaDnΝ:t֭[gDШQ06qD;@ x M6v(PiizmZ4ew"@̴ixf [~-Fٱc&IZ2|n!?$@{[HTɏ#@hȎI# _m(F /6Z3z'\r,r!;w2Z3j=~GY7"I$iI.p_"?pN`y DD?: _  Gÿab7J !Bx@0 Bo cG@`1C[@0G @`0C_u V1 aCX>W ` | `!<S `"<. `#c`?cAC4 ?c p#~@0?:08&_MQ1J h#?/`7;I  q 7a wQ A 1 Hp !#<8/#@1Ul7=S=K.4Z?E_$i@!1!E4?`P_  @Bă10#: "aU,xbFY1 [n|n #'vEH:`xb #vD4Y hi.i&EΖv#O H4IŶ}:Ikh @tZRF#(tXҙzZ ?I3l7q@õ|ۍ1,GpuY Ꮿ@hJv#xxk$ v#9 5 }_$c S#=+"K{F*m7`#%H:NRSp6I?sIՖ{Ap$I$I:QRv2$Z @UJ*$]<FO4IENDB`