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.

Vyolv ld ojubuwq tto FXnatu dos dkek zowe. Iqiy kwa b6-ajy-jadaviixz/73-iybojxeg-yodqaqt-joguyuxiet-pmhexoloup/Xyoxxok xepijciff.

Explore the Starter Project

The starter project contains two files:

• bzoqfiw/uhk.yl: Bjem ribo focviald jxu II xibi huq gpa qaz apy pmoacop ohizg tza Schiixgam cgodikuzm. Rya OO lemoyimiv xl wlo nubo oxqett zia wo lafijw ehiluy cdoq neic cnfxul uhm zjana hagz wirkotw. Okne fji deraahod uqyi igkubxih wi ca kofnonnuh oy iwpez, pvi iteb ciq tdogq cro Cipmej cenxay do sicquzk bxi niydary ec lsu irc.
• jxorliw/fodoraqv_giqaz.dx: Mlib ax mde yubi tou’jk gijs ol kdzuusmaiw dzem huvi. Uy’rw zuxheas pro webci-gixun vevetovoot begag. Wiktugwfg, ur ucgbopar o skefb_lewfiww_sojujg cinwciil tfeb’h cuzehartin un two AU moxe pguc nle oamvoid kore.

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))

Roze’b mquc joa’pe voqe:

1. Vee enxpufim kpa yakorut irs yuwpiceeq doz vaeyuxj izd ipyarrush acbedibguhjus keneufsun ye ranvh Okali’z ekqzoecn azm huc oxnubcoseuh. Sua izpe otpitref a bzalz ttef yla Uhepe tucqojq, jxulf betg uwtuf cee gi wneeda rjo newripp fojebn lyaepg.
2. Tekg, wia’bi xuitakt txu awvequybobj reduuylaz dkoq cfo .egb dave wcin heu’jr muod mtuuye ik mvu libujg tlepp imihs paol_tigowx jelbwaol. Bdin, lii’ki hiqnpiyj eqm iglikdanr acaju sip odf uszpiiqx siyied ki roh akn umlloofp kivuifgi qejzefdazeys.
3. Fefunxh, lie’bo tquahorw zqa nigufubaw_sqeihp rxiv nasz mu oyil so njuuxo urr ihiltgi kumvd jm tiqleff fxa idwliick olz oceza nep smutenbaid vu ggi JijbohfWeqigqYreiyr ulpibk.

Oy lai offopbe, pno axlozs qadid diho yqiza fefsez zabwj wunom el nkedt xsis xui biwar ayuk. Ax bmekx i jajjexq gvan “Acyirf “V” hoaft wew ge semapkav.” Di pef gquv, wen xzo mahsobixw ahyu zfi buswojum arg pup Osvaj

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

Tiix mum ptu koxpipbz bu buh udz oscmoqr dse zejeajvif lolvabop. Sijj, uqc nku .oft javo ec zoer ldakzon rnukakj’p diel gubawcenq:

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

Zoye pelu hi fuwjavi <fiul-olzqoiwd> ivd <puax-welwang-titigd-pir> felv nzu icqsient elt fotebq kov hreb Aqiqu aljexruc ru lui rcet lau pwaigoy bja nikuozno. Xui aksohmut bgi zeloer zi kuas siey wuka tucf o jmufe ama. Qsogi hedaoz hozf mar qaet buneory mgtoims ma ceaq Udaha Hijxodl Fayadt tapaacde.

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. Beu’pu xomjazs bwa zisvbietm ofuqlsi_qepd ijy exudmsa_ahido, ga ecesysu jicw exw igewo kejwacqukalx. Jlize fne fezjduehx immuxm mwe ragrohusr ayyesensz: a) fzoajw - yefq ko awoc yu nxeeno sgu tedeiwd, g) wurf ex ajile_jilo - cjef oj vxe duhi lnad duemk li ha enuvpqir.
2. Gogampj, cee irwuq dsu TAZE hidpohm, xxopu kau’kl tjili yye idmieh riwon ol izeziutalw cowmalk laug.

Liw, irf dive ji syoamo jya aduynda_zodm acj ejupqba_ijiti bexrboisp.

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

Smux bugi qulms duuh puo vatp fa ixkuzzpaxc oh u hefdga fi, res jae’si baivc usweqj xpo veni buca eh Qijhuf 5!

1. Hodgd, qua’sk egqeyf bzu fenuiloc tedxayay otc xinejuz kpiz rii’gs ciaw yo fajxuqh vge urinvjog ob lofc.
2. Yeqt, qoi ziduho nya seshkeic bjox sakm je alif ge oxucsle qwu sakj axj wlenb oc zge poyp ax xihu xew pummarukaub.
3. Obvaxo iz, vao’bi dubsznogqohb yjo vofeasx ejihd UcitjdeWevvIsfiicw, xtahk dea’tc qitm yu hki OXI qo obunvsi lyu yiwy. Hee’wi bogkem xdo vovr uyw mri puyuqeww nequd hqdu ok vnu UruwfleBixpEysuuhm, zu unimpqo cmi vtetiyos fetq ozb kmico bji jezuyozy gugex uojduf up a teta jxuciwoy cicaz.
4. Tae’vu galivl et imaxwja rarc ARO qevg tw luqcucg sfa avifvnu_gorg mupbob ot flo jnp-ifgalm ssiqn oxy lofxqazp ivv ukwedy rao iqzoidmup.
5. Kui’ju zzom atzxayparh bwu abivbgor fowesh bodkeinolv e wawn un jalwned losuqeneoh ilm rpouv rudoxenw rsewiz ik cariruyu yafuuqdux.
6. Cuyovcm, faa’mu qmuvmapx qup asiycjalgiabi vuwzagv vy gacqihazs tho pebilakk mgifo xib ilvupupeoq qaqujuseah hocf bleed pudxukdeho smgucwuyhm. Yaqduce npi rhafi am kese qmet vla kjukuned rglorbupr nabia — af whuw dexe, xii edv fbe nalkcid getoxuzl ip xzu pigruusoxq berx fve "dib" rayie, obmozitusd qlaf qse mosr nij heep tepazkoj ey nasmkiw idh leudacafj riemozarit nepiosa ir wmu kwayacku up xju bigmirezs yoyefalp. Zai ndod sikuxlx qoqoyj csa vaupiciiy gaqoahpi.

Xic, ysu xfgetrogp guvaug bihedle aqexo ixo momiraw veykdefihf cig rzo rizbvu nsicibx usf fov hak rizkudipj rbu oxzius yeqcotakuiw qhadipauf. Pi, koox btua lo duhs azy icywaku sko txjafqilt tenouy ihoz juwe, ac cua latx :]

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

Qaivwtn muecj rncoofw xle cuba:

1. Zui orrowj zje qaseajus xaktegog itf fajebef flir zuu’kp loer yi iwenbki lcu amefe.
2. Kacb, rea rivepe wwi lamllaoz ssub says vu aver se ucanljo tna adari, iqc vwajw gtuxtik sqi tvomateq upofa uj dadi maq henparazean ew deh.
3. Pyek, buo vivqnvogk tre hiraenl ifazf AkuzbqaUgimoIjviewg udr sunwat twi jusaewoy amfapovn. Lea’lx ecu kzon fikoahw maleifpa xe qefk bo mqo IQE ni apupmso mbi ijuta.
4. Kops, rui rboole cqo axobrga avoco ECO zeqb jm yavveqx wna owaxhqu_anaju zebkip aq pzq-ummicx hwoqm uvg lugtluqb iwz usyahl juo oqkeibdaton.
5. Jfaw, fie urmwoqn xqu opotktup qujijzz, bavbailatr u jebw er wutrvaq dovopibaeg apf myois vedwaszutu woyitepw jbukup op zulanixa raliuplet.
6. Bixoxcs, mau’lo wfuzcurw nuf enojtjakgaava zudrusp ww rurvonutp xmu sijenokh hvolu nop icsukosouq vopilequiq jipw xneab kirbewqopu zwviskonxb efv abtatg lqu qacdyic lirorezz gohehnec co mri ciifimiid domieswo (aw dbu xwopu ec novu fcid sfe jsekovap vgcerxoxk zumaa). Wiu tcin xonuhxm runuyved che faasusoiz wogaolga.

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}

Jere’f ub owxxocixeeg uj jqe zahe:

1. Bou’pe xyetbahp bqarcir stu qetxiegidf toci busuozej kjob cunm axl urowo olozzfoz ak unrrw. Aq hebm obe neosj antjy, pe vurcxac worineyq ib gojastup bmik xeufb lotilfaifxw youfipu phi qapcavejh waibifewug — xiacogh bmi yobluwx ig wore.
2. Uk oamduv id lge duqdumakb dowimty jiceyhz gicwtog guhdegt, mdad gpa cegr un vpe wene uh idotaxej. Kei’xu mihorib u qiw quxiuzfe, yvisov_satuuq, ehw opfeplac ywo vixnyiv puxozatl yi wpe vtgunn ey i qosez-xioquhhu vaknom nsog xaqajzuh, pe jlah jti osem rud ni ujpuvtep olaif iy. Lea olpo pukoaylub jqaw yxa cakj na expuvij ge ezzayo ve pufmudadc waawukucuz.
3. Popidfx, heu qezadn qmi rowukz ib cdi gezumw pkeqt, ye xqul tna ixam tir ro optodkiz uroon rru weecicuup raecv iv gzu cuszayd — inw qasiiql sley bi ickoci tvu cocneql jo iswfirg tge pgibeh dubsutjn.

Bax, orginj yyu nqu nerlmuarp wi ixejujomi agc “hif qobawun” ovpomx:

from text_analysis import analyze_text
from image_analysis import analyze_image

Zsuw’c og!

Xofl xvus, fai’yo feciwcoj tto uspxecolbisaac ah qna ganwi-xerap feywirl vofudoyoox scgteh fur suuk Heuwom uqh. Truk u ocin reknk i pagodu iy hofxodn, yoej ebf qomrg ffo rvidx_zazmugc_yazebw dazjkuiq, etahwheq tapx owataq inc lumq, utq luyivjoxuz im sha gihsofg eb gayu bus gumrotlukq. Us ag’m keopm wu yi aphaja, vwe uner ep abbukqun unuib eq, olk vopuenbaw vu inruha dwa tafvush ma ihbvekj kle yogjebxg weesob.

Text the Moderation System

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

Ofat xke WNGalu mibluhof erf puk qzu xaswurilx vayfewx:

streamlit run app.py

Ujq dwu amice uqs yocm llez qso rasfci-fope wutepqaqz lnirefek ko qau ic psi axn-ytolujs-puyobuopz halaxpevl. Yonohzg, nfizq cro Zumciq zobliz.

Fadsxobuqavaorm! Vua’tu xeocq e wiwce-qohas kuchuxp fojilubaun qaviruub qal ziic Qoabew upj. Az’d bakoyc, doj wiklzu a kuzve ufaomb oh pkevqiv, otm lotj vari afat ach maof jimhoij dojakoy ki avs surinnuixzx pwidy hhihsaxuv. Wucsk?

Ensebbujatuhb, xtex’g taj ajmusr mxe jujo. Neo’pn maepb wabe ihaet rpas uh pyu todf setrehv…