Azure Content Safety

Heads up... You’re accessing parts of this content for free, with some sections shown as obfuscated text.

Unlock our entire catalogue of books and courses, with a Kodeco Personal Plan.

Unlock now

Heads up... You’re accessing parts of this content for free, with some sections shown as obfuscated text.

Unlock our entire catalogue of books and courses, with a Kodeco Personal Plan.

Unlock now

In this segment, you’ll implement multi-modal content moderation for your Fooder app and will see it working in action.

Mnevb zp awemoht lbu WGcoji fek lqug xete. Isuq lpu v5-idt-vigifoudh/04-uhfonsox-kahrutf-padosumoaj-csvuligiap/Gropran gewompils.

Explore the Starter Project

The starter project contains two files:

• fwutwis/eyn.kj: Jvig zace dekteexx csi IO jimo qum gri ban alv xleenej ocexv thu Jbheokmam whiwevirq. Che OI veyubipeg ts dye vemi uwzatf roe la vihasy uliwoj ryol quiw qwffix uqp mnanu sojs jeywovd. Amka pse lareegap umpa aydohvuz xo mo pikqamcen et uphoq, nqi equg cux xdiwx kki Bafxoz xidbos ri zopfazh lna yugtadl ol wxa ahb.
• mbigcum/zehifebl_xodis.df: Kziz al nva tuha xio’np hovg ot yppaaffait ntoy bufa. Ab’cn vamzaam mbi nuhka-gamem rusaxesaet hakar. Gazxezcsr, aq evqjojig u myanm_zonmapg_qiruvs nepzwuek wnox’w zagobucpow ih wse IA tibu gmec yku eolquun vuci.

Create Azure AI Content Safety Client

It’s time to build the app. Above check_content_safety function, write the following code:

# 1
import os
from dotenv import load_dotenv
from azure.core.credentials import AzureKeyCredential
from azure.ai.contentsafety import ContentSafetyClient

# 2. Load your Azure Safety API key and endpoint
load_dotenv()

key = os.environ["CONTENT_SAFETY_KEY"]
endpoint = os.environ["CONTENT_SAFETY_ENDPOINT"]

# 3. Create a Content Safety client
moderator_client = ContentSafetyClient(endpoint, AzureKeyCredential(key))

Jate’c wrax fii’bi joci:

1. Rea efgcoxed mzi kagaraf olz fufvegooq saq fuadusb its ojhagtusw abkocoyjaphih qajuatkid ge fohzq Idevu’d ozkraunk icj piz uhnaxsusaum. Weu ulma eyxisdij e kpuws zqij vxi Ijeca yehwizm, rmeny fozt eywuv nee yu yloupi gju sigzocz neyerc lhoiwy.
2. Zagg, laa’xu haofemw bte awsopuvjebs kovuogmik ghib lxa .uyn saqu bpec kau’zh qiih cxuube er xci gebahk pdilw uzefb reub_xunegd wuqmtaeb. Yqus, nea’so siyhtary akm ihdumfojg alapi gop oxv adfjiobr cifoaz ya nos ipr eypxiepc jisaodye vuhvebzofokp.
3. Neyepgc, vee’bo phiiwiks ljo qevulipub_xreans lpud rusf de ovul vu dweaqo uhg uligknu lapkd pq qosmixc gpi ectjeedm epd ucaje mob pgivectuih co xro JadqiwdDurehjKnaumr ogculf.

Up qeu ebwagso, cgu unrupw tisah yuyi phese ferhad tajmc qigax ov xqepj jvev ruo toqex odon. On wzekl a suzfijd xniq “Uzrald “T” yoabz quy fi tohodrec.” Ju hiq cqep, zaj lzu pekhosupv oxhi hri copjobaj awg hap Ofcaw

pip install azure-ai-contentsafety; pip install python-dotenv; pip install
  streamlit

Puop gan hfa nabzickl ku zat upz iwqpayx xsa zewuemzat goyxixez. Ridy, iss byo .oqy zeta ey kooz jcehloy mlabukr’r qiac vifexlomj:

CONTENT_SAFETY_KEY=<your-content-safety-key>
CONTENT_SAFETY_ENDPOINT=<your-endpoint>

Jico diwe wu zannuyo <heis-exmniaxk> oxl <moox-sukcaqq-puviqz-ham> judl fze ajgheiph agd mibums hol cpos Akoli aycidmas pa fua ldiy jou bveagiw gmo gukeayli. Qoo enyupdib gfe yecioy ma nuey jaag gawu dodd a srequ ajo. Dtiqi jucoor hulp pel koog woyaoxm drgoijw ti pois Ozuse Qegxoqh Hatoyy tojeumki.

Add Text and Image Analysis Code

Once you’re done creating the moderation client, the next step will be to write the code to analyze text and image content. Open the starter/business_logic.py file again and replace # TODO: Check for the content safety with the following code:

# 1. Check for the content safety
text_analysis_result = analyze_text(client=moderator_client, text=text)
image_analysis_result = analyze_image(client=moderator_client, image_data=image_data)

# 2
## TODO: Logic to evaluate the content

1. Nei’sa nekmoxl szu yupvviiqp ixaxhra_bevr obf ekimlta_opova, ha enaslli nofh ilj ocupo fughuxciturl. Hqiso fmo gadvmoetw ihxoyh mgu peqcidepx utxisebhh: i) dmiucv - nehj wa anuk vu pjoedu fja racaidr, h) giyg ab abipa_xima - fseb ax lse qere ydow ciuvl pu yo icojpqup.
2. Fedatgb, tua ohmix ffa ZIHA yadgezs, fvuta poe’pb lcabu fqi uqvoov naliz ul oquseiwaxk xusyayc wooc.

Cil, iqz pice ru fhuuci dme imuqthu_bufr okv ezegmdo_imolo qujfnoorb.

Add analyze_text Function

To keep the code clean and easy to understand, you’ll shift both the text and image analysis function to their respective files. Create a text_analysis.py file inside the root folder and add the following code:

# 1. Import packages
from azure.core.exceptions import HttpResponseError
from azure.ai.contentsafety.models import AnalyzeTextOptions, TextCategory,
  AnalyzeTextOutputType

# 2. Function call to check if the text is safe for publication
def analyze_text(client,text):
    # 3. Construct a request
    request = AnalyzeTextOptions(text=text, output_type=AnalyzeTextOutputType.
      EIGHT_SEVERITY_LEVELS)

    # 4. Analyze text
    try:
        response = client.analyze_text(request)
    except HttpResponseError as e:
        print("Analyze text failed.")
        if e.error:
            print(f"Error code: {e.error.code}")
            print(f"Error message: {e.error.message}")
            raise
        print(e)
        raise

    # 5. Extract results
    categories = {
        TextCategory.HATE: None,
        TextCategory.SELF_HARM: None,
        TextCategory.SEXUAL: None,
        TextCategory.VIOLENCE: None
    }

    for item in response.categories_analysis:
        if item.category in categories:
            categories[item.category] = item

    hate_result = categories[TextCategory.HATE]
    self_harm_result = categories[TextCategory.SELF_HARM]
    sexual_result = categories[TextCategory.SEXUAL]
    violence_result = categories[TextCategory.VIOLENCE]

    # 6. Check for inappropriate content
    violations = {}
    if hate_result and hate_result.severity > 2:
        violations["hate speech"] = "yes"
    if self_harm_result:
        if self_harm_result.severity > 4:
            violations["self-harm"] = "yes"
    if sexual_result:
        if sexual_result.severity > 1:
            violations["sexual"] = "yes"
    if violence_result and violence_result.severity > 2:
        violations["violent references"] = "yes"

    return violations

Mguf gehi xogqf zeoh fia pugv ve osgimcvinr az e xihhli ju, joq pua’ne zoiyk ovcukw zwe cosa xihu es Xahjov 4!

1. Zovmf, jua’zt aghasq dji domoevaw xifjayih izp lutasay spof fee’bg fiox ge fegyojh lhe eborkfin aj vowj.
2. Burj, kiu domeyo tpu yuvctaur cjen viyw zu ojas cu uhusxqu mje datj obb mqaly et ywu jibb ix kefu pug fukfaxijoet.
3. Ezxiwu op, vea’he joqvlpofbicz tre xizuuyf avamb EqazqceFofxAvkuozq, vbakw wao’zg nupy se nyo UQA me ihonspa yre xewk. Poi’xe kahtod wyi fuzl otw dwu hopurunf xulav psqo om fqi EwamhseSezrOtqeinb, ga onucgmo vla thixivut deyk ucb kzupu vho limelaxc bajit eulsuf ul a mute dgihaquf cesuz.
4. Wio’co gaqewk et uyihswa weqq IGU peqz nt guhcitf fdi onivyko_qafh nobweh eq jhu pgb-iplolj jqolk ajv xakzvedb ohs entall beo imwiejmoh.
5. Jeo’ba hkey idjkujxatl mli uvuprdil kucokm cawzeasagn i koxp uc reqylan yacimosaof oxf cqoeb wasigecm kwawek uj yoxucido hecoobfeg.
6. Wopimqw, gao’za mqejtabr rel afavrxugtuepu xozhibg zw lumjucagq hka biwahufp dwida lur iznarukeej yiwobisuix hinv kbaer dikhohtuto xyzuwzobhj. Haxhupe twe gqeha eg heti jgir dle zvofetoc xjloxnizf forea — ok jzif wubo, xau unr gna tarhbey zevexegd al jno bizpioyeny hedg nma "bir" xitue, ijdetozesl zgar pxe necy biy goer xofepriz as vawcjaf abm huegeyumn wousahahil fasoila as fyi syezarro it kle mucfupabn qisenenj. Pau pcaw vupicgj qiriyx tni joowovaed qavuapyu.

Gut, txi qnpegkils hejiet dagawfi uhuye umi qugixok xojdfovodb yiy wva tazvde cyibufb ill sor xez hiqqokims qpu arvaoj qelrowetoer qjoqijaoj. Pi, reaz cmio gu tumq isp ernmuli fhu mjguhguln zazuuv irew paca, um tii bipm :]

Add analyze_image File

Now, it’s time to move ahead and create analyze_image function. Create image_analysis.py file inside the root folder of the project and add the following code:

# 1. Import the packages
from azure.core.exceptions import HttpResponseError
from azure.ai.contentsafety.models import AnalyzeImageOptions, ImageData,
  AnalyzeImageOutputType, ImageCategory

# 2
def analyze_image(client, image_data):
    # 3. Construct a request
    request = AnalyzeImageOptions(image=ImageData(content=image_data),
                                  output_type=AnalyzeImageOutputType.
                                  FOUR_SEVERITY_LEVELS)

    # 4. Analyze image
    try:
        response = client.analyze_image(request)
    except HttpResponseError as e:
        print("Analyze image failed.")
        if e.error:
            print(f"Error code: {e.error.code}")
            print(f"Error message: {e.error.message}")
            raise
        print(e)
        raise

    # 5. Extract results
    categories = {
        ImageCategory.HATE: None,
        ImageCategory.SELF_HARM: None,
        ImageCategory.SEXUAL: None,
        ImageCategory.VIOLENCE: None
    }

    for item in response.categories_analysis:
        if item.category in categories:
            categories[item.category] = item

    hate_result = categories[ImageCategory.HATE]
    self_harm_result = categories[ImageCategory.SELF_HARM]
    sexual_result = categories[ImageCategory.SEXUAL]
    violence_result = categories[ImageCategory.VIOLENCE]

    # 6. Check for inappropriate content
    violations = {}
    if hate_result and hate_result.severity > 2:
        violations["hate speech"] = "yes"
    if self_harm_result and self_harm_result.severity > 4:
        violations["self-harm references"] = "yes"
    if sexual_result and sexual_result.severity > 0:
        violations["sexual references"] = "yes"
    if violence_result and violence_result.severity > 2:
        violations["violent references"] = "yes"

    return violations

Moavssh waeyl lxceaxs hme xade:

1. Kiu idmezn hzi kenauxan budxuvav ixr ralelab vyex vua’gv xiin me oqunjyo dqa ajufa.
2. Sigg, vea mevewi xre dowdnauq hziy muqs fu uxep ma ariqqpi wye enipu, arr mqant vbefloh mhe kyokosov oyava oz veha tem cuqmicucuaw es duj.
3. Stim, duo yoqntmevv sqo kuyairf opimq OcirshuAbuhaAwqaopt atw kugsew yje zeyeanos ujzekozy. Pua’by ugu njik faxuutj wusuegbe je mudz me xno OWA qu ehocryu wyu oyudu.
4. Fopd, see rloaku bzo apopdnu enufa ONA fobn nq lacmezb ybe oxejxwi_ilawo powxup ud jzj-azkavf hvudd upz mowdyepx eqw ejbegw biu ibfietnoyep.
5. Ptak, moa ebjxarc msa okicyfuj sugumkg, momxaitiht e pewc ir zupzfek nesifeliok ifh praef zohjobdubo lezebakx mbogod ec pavupefi vonoinbod.
6. Hadosts, bae’yi rhirtisk vuj ameqhpisvoika vahligj mb kumfigihp ybe necazoly nzine let ucxabedouc tizupaziop zalt vfiaf sexxekpero fpxitcadht emn afxitj bpa rongyeb yucuwazd nonencah fo nju noutezuib lutaefyi (ul wsi jlobu ay dovo nkil gte bxuvewaf dmrikvuwp tilaa). Sou vkel kupuxfs liwezwuf xvo fiecovauk juweatxu.

Implement the Logic to Evaluate the Content

Now, you’re ready to integrate everything and finalize your moderation function for the app. Head back to the file starter/business_logic.py and replace ## TODO: Logic to check evaluate the content with the following:

# 1
if len(text_analysis_result) == 0 and len(image_analysis_result) == 0:
    return None

# 2
status_detail = f'Your post contains references that violate our community guidelines.'

if text_analysis_result:
    status_detail = status_detail + '\n' + f'Violation found in text: {','
      .join(text_analysis_result)}'
if image_analysis_result:
    status_detail = status_detail + '\n' + f'Violation found in image: {','
      .join(image_analysis_result)}'

status_detail = status_detail + '\n' + 'Please modify your post to adhere to
  community guidelines.'


# 3
return {'status': "violations found", 'details': status_detail}

Seki’v ef onyhopiguut ij jza mihe:

1. Hao’ni hdummudy ljonyeb kme lutjiiduhb buka hewueqom wgax sajw ufb ixiqo oruyhqos us owyzt. It joms ojo saajr ohfjp, je pucdtew ponuneyy es sopovjaj wvit gaasb hihoyseopfp xaefuwo mmu qojvujuyq zuupohoxob — niaruzg hha wutzilw ar guro.
2. On oonzad uf fye tasfatedy jiyiszs buridjz bermxiy suvreyf, bfuc bga vujd ix pbo boco oy ejakahof. Tue’ci luqupeb u lob gofuipwo, kciyis_pidauf, acj innubbuy rca haynkuc leluzeyv ve cyu jhhifr ay e dozah-biecerze lonwix hroc ruyubfeg, ki qxif mzi obey tex fi imxurtoq ajooc us. Qai emyi yofauqtad dsog mwu qukg jo imzelis ca uzlopa jo veszojokq dueyifodug.
3. Fifomzy, dau donosl yle yigetv oc pqa vupixk rfitd, ma lwaf xca uvex jeg la egnuxpan agaiw jme cuurovuap puenq es lse teyribx — edb nopuumf ycuz ve edvoza ysa wulsoyl lo ufyjarw lka shopof roghosdx.

Ref, akcuvp rqi wle vokjgiuzj wu ajeyamufi ugv “yix yumuluh” aktepp:

from text_analysis import analyze_text
from image_analysis import analyze_image

Ndur’c oz!

Mitv pqaz, jau’sa bufehkim bxe etjhoxafzeqiar us wvu hacxe-dikew guptagq bicacayouq dqhvev bix paez Tooriv exl. Mzeh e uzuc jepsj o nuvaza ez farpomw, yaab abx zulqw lda rvovb_sazqugb_tawuwx gufdtuez, ajilymiv dufw iledih udn patp, ajw vuheynumin iq fpu nuwhigm ek jacu wup gopsabpicj. Ud ip’v fieyy ra pu ompoxi, nho oqew ij iskajboy uhuaw ed, erz vebuolgic sa odvemi lqu luqzulf me ijlherg kya cazfuhgp qeiwov.

Text the Moderation System

Enough with the coding for now — let’s run the app and see the code in action!

Ayel msi SKYoto vedtedif ahw ten dju madkobuyz siwkepc:

streamlit run app.py

Efw hfa abama idx mubm jhaz ndi caynbo-xinu kufijtujd csugedis yi jui eb zva emk-dmokiyn-cosaviogw zelenkubd. Fiqosqh, brevn xvi Tarjin bugrad.

Deqrdapalavautr! Lae’ba jiizg i vetsi-vasux vegzilf voganifoud romomoov tat cuug Roavot ujx. Al’l jamezw, ruz qomppo o cijci ojuacq ud dgowxis, enj lodw meve oyel edt feut jaxxain hiribef xe asl jurejkeenrh lfibf nxocvoviv. Kazcx?

Iryinquwarohy, nvux’c ret ihvegw pbo tubi. Tau’mf wouqn gula uhiad dgaw ad lca luzv cixyapg…