o oh@sddlZddlZddlZddlZddlZddlZddlmZmZddl m Z ddl Z ddl Z ddlmZddlmZddlmZddlmZddlZddlmZdd lmZdd lmZdd lmZmZdd lm Z dd l!m"Z"m#Z#m$Z$m%Z%ddl&m'Z'ddl(m)Z)ddl*m+Z+m,Z,ddl-m.Z.m/Z/m0Z0ddl1m2Z2ddl-m3Z3m0Z0ddlm4Z4ddl5Z5ddl6m7Z7ddlm8Z8ddlm9Z9ddl:m;Z;ddlm?Z?ddl@mAZAddlBmCZCddlmZdZDdZEdZFGd d!d!e+ZGGd"d#d#e+ZHGd$d%d%e+ZIGd&d'd'e+ZJGd(d)d)e+ZKGd*d+d+eZLd,d-ZMd.d/ZNd0d1ZOd2d3ZPd4d5ZQd6d7ZRd~d9d:ZSddd?ZUd@dAZVdBdCZWdDdEZXdFdGZYdHdIZZdJdKZ[dLdMZ\dNdOZ]dPdQZ^dRdSZ_dTdUZ`ddlaZaddWdXZbdYdZZcdd\d]Zddd^d_Zedd`lfmgZgdadbZhdcddZidedGZYdfdgZjdhdiZkddkdlZlGdmdndnZmdoenfdpdqZodrdsZpdtduZqerdvkrdwZsdxesZtdyZudzZvd{Zwd|Zxd}Zyd|Zzd8Z{d8Z|eqeseweyexeze{e|dSdS)N)validateValidationError) AIMessage)OllamaEmbeddings)FAISS)ChatPromptTemplate)AsyncWebCrawler)OutputFixingParser)ChatGroq)PromptTemplate)StrOutputParserPydanticOutputParser)RunnablePassthrough) PyPDFLoader TextLoaderDocx2txtLoaderUnstructuredPowerPointLoader) PdfReader)SemanticChunker) BaseModelField)ListOptionalDict)urlparse)Unionr)RunnableLambda) Presentation) OllamaLLM) ChatOllama) PdfConverter)create_model_dict)text_from_rendered)Path)MarkdownHeaderTextSplitter8gsk_CEh3itIpUAkEkEKsUDqVWGdyb3FYoTjqmXNTBHOSxJFK3obGTzXZnomic-embed-textz"44622834-f22df6f12cf45558ee180dd8dc@szeZdZUedZeed<edddZeeed<edddZ e eed <edd dZ eed <edd dZ eeed <dS) SlideContentflashtypeNz$An optional subheading for the slide description subheading.z(List of paragraphs for the slide content paragraphszUA specific and concise suggestion for a relevant visualization or image (max 5 words)visualization_suggestionzURL of the image for the slideimage) __name__ __module__ __qualname__rr)str__annotations__r,rr-rr.r/r5r5I/var/www/eduai.edurigo.com/storigo/production/generate_storigo_content.pyr'/s r'c@s^eZdZUedZeed<edddZeed<edddZe eed<edd dZ eed <d S) MCQContentQuestionr).zThe multiple-choice questionr*questionzA list of 4 answer optionsoptionsz0The correct answer (e.g., 'a', 'b', 'c', or 'd')correct_answerN) r0r1r2rr)r3r4r9r:rr;r5r5r5r6r78s r7c@s>eZdZUedddZeeefed<edddZ e ed<dS)StorigoContent.7Dictionary of slide contents with slide numbers as keysr*slides/Total token count for all the generated content token_countN) r0r1r2rr>rr3r'r4r@intr5r5r5r6r<>s r<c@sZeZdZUedddZeeefed<edddZ eee fed<edddZ e ed<d S) StorigoContentMCQ.r=r*r>z8Dictionary of MCQs with identifiers like 'mcq_1' as keysmcqsr?r@N) r0r1r2rr>rr3r'r4rCr7r@rAr5r5r5r6rBEs rBc@sFeZdZUedddZeeeee ffe d<edddZ e e d<dS)StorigoContentMCQMid.zYDictionary of slide contents with slide numbers as keys and MCQs with MCQ numbers as keysr*r>r?r@N) r0r1r2rr>rr3rr'r7r4r@rAr5r5r5r6rDLs $rDc@seZdZddZdS)CustomMCQParserc Cst|tr d|}|dd}|dd}z t|}|j|WStj y:}z t dt |d}~wt yN}z t dt |d}~ww)N '"zHere is the MCQ:zError decoding JSON: zError parsing result: ) isinstancelistjoinreplacestripjsonloadspydantic_objectmodel_validateJSONDecodeError Exceptionr3)selfresult json_objecter5r5r6 parse_resultQs    zCustomMCQParser.parse_resultN)r0r1r2rYr5r5r5r6rEPs rEc CsVzt|}d}|jD]}||7}q |WSty*}z tdt|d}~ww)NrIz Error extracting text from PDF: )rpages extract_textrTr3)pdf_pathreadertextpagerXr5r5r6extract_text_from_pdfjs r`ct|}|dd|}td|}tgdfdd|D}|fdd|D}|dtdt|}d|S)NrF\w+ andortheaaninonattoforofwithbycg|]}|vr|qSr5r5.0word common_wordsr5r6 z)generate_search_query..crqr5r5rrsuggestion_wordsr5r6rwrx) extract_context_keywordsrLrefindalllowersetsplitminlenr. slide_contentcontext_keywordscombined_querywordsfiltered_wordsprioritized_words query_wordsr5rvrzr6generate_search_queryt   rcra)NrFrbrccrqr5r5rrrur5r6rwrxz-generate_search_query_new..crqr5r5rrryr5r6rwrxr{) extract_context_keywords_newrLr}r~rrrrrrr5rr6generate_search_query_newrrcs^|jpddd|j}td|}tgdfdd|D}tt|ddS)NrIrFrbrccrqr5r5rrrur5r6rwrxz,extract_context_keywords..)r,rLr-r}r~rrrKrr^rkeywordsr5rur6r|s  r|csf|dddd|dg}td|}tgdfdd|D}tt|dd S) Nr,rIrFr-rbrccrqr5r5rrrur5r6rwrxz0extract_context_keywords_new..r)getrLr}r~rrrKrr5rur6rs $ rc Csd}t|dddddd}z/tj||d}||}|d r2t|d d d d d }|ddWStd|WdStjyX}ztdt|WYd}~dSd}~wt ys}ztdt|WYd}~dSd}~ww)Nzhttps://pixabay.com/api/photo horizontaltrue relevance)keyq image_type orientationper_page safesearchorder)paramshitscSs|d|dS)Nlikes downloadsr5xr5r5r6z%fetch_pixabay_image..Trreverser webformatURLzNo image found for query: z#Error fetching image from Pixabay: z)Unexpected error in fetch_pixabay_image: ) PIXABAY_API_KEYrequestsrraise_for_statusrOsortedprintRequestExceptionr3rT)queryurlrresponsedata sorted_hitsrXr5r5r6fetch_pixabay_images6 rrc C|stddSt|D]S}z1t||}td|dd|t|}|r2td||WStd|tdWq ty_}ztd|ddt|WYd}~q d}~wwtd |d dS Nz%No visualization suggestion provided.zAttempt rz to fetch image for query: zValid image found: z!No image URL returned for query: z"Error in get_valid_image (attempt z): zNo valid image found after z attempts)rrangerrtimesleeprTr3r.r max_attemptsattemptr image_urlrXr5r5r6get_valid_image&   (rrc Crr)rrrrrrrTr3rr5r5r6get_valid_image_newrrcCstd|}t|S)Nrb)r}r~r)r^tokensr5r5r6 count_tokenss rc= s z|dkr tdtd}ntddd}t|jjtdtt} dWd d t fd d D} td | g} | kr]tdt j d| dt d} fdd| D} notdddD} t | fdd| D}dd|D}t |}d}|dkr|dkr||d7<|d8}n|dkr||dkr||d8<|d7}|d| }|dkst |D] \}}| |g|qddd | dDtdtd}ttd}tj||dt|}fddddfd dd!|B|BB}|d"d#}d$tfd%d&tt|jfd'dd(}t|t r4|dk}nt|trA|d)v}nd*}|r{|D]/\}}|jrmt|j|}|r`||_ qJtd+|d,d|_ qJtd-|d.d|_ qJn |D]}d|_ qd}|D]}|j!d/d/|j"d/|j} |t#| 7}q|rd0}!d1d2}"t|!}#tddd}}i}$t|$}%t%dt|%d3D]}t|$|krPg}&t%|t&|d3t|%D]}'||%|'}|&'|j!d4d/|j"qd5|&td6tz)t(fd7d|#B|Bi}(td8t|(td9td:t)|(d;r5|(j*})nt|(})|)+d<}*|*d=krd}+d=},t,|)|*d|*d>D]"\}}-|-d.estimate_slidesc3s|]}|jVqdSN page_contentrsdoc)rr5r6 sz)generate_slide_content..u.🧠 Estimated possible slides from document: u>📊 Scenario 1: Fewer slides than chunks — sampling evenly.r)numdtypecsg|]}|qSr5r5)rsi) all_chunksr5r6rw%z*generate_slide_content..uI📊 Scenario 2: More slides than chunks — distributing proportionally.cSsg|] }t|jqSr5)rrr)rscr5r5r6rw(csg|]}|qSr5r5)rsw) num_slides total_wordsr5r6rw*rxcSsg|] }tdt|qSr)rr)rsrr5r5r6rw+rz css|]}|jVqdSrrrr5r5r6r=szContext text is u= Based on the following context, generate professional and engaging content for exactly {num_slides} slides in a Storigos presentation. STRICT RULES: - ❗ ONLY use the content provided in the 'context' below. - ❌ DO NOT introduce any external knowledge, definitions, or examples not present in the context. - ⚠️ Do not assume common sense or use general facts. Stick to the exact information given. - ⚠️ Avoid generic phrases like “as we know”, “in general”, or “in this video”. Each slide must include: - A clear and concise **sub-heading** - **Exactly 2–4 concise paragraphs** derived solely from the context - A **visualization suggestion** (max 5 words, specific to the content) Important: Only output the final JSON object. No additional text, markdown, or explanation should be included. Context: {context} {format_instructions} The final output must be a valid JSON object where each slide is represented as "slide_1", "slide_2", ..., up to "slide_{num_slides}". Each slide must contain: - "subheading" - "paragraphs" - "visualization_suggestion" rQ)parserllmcsSrr5r context_textr5r6rcsz(generate_slide_content..cS|dSNrr5rr5r5r6rdcSrget_format_instructionsrrr5r6rer)contextrformat_instructionsrIrritem_keycSL|dd\}}|dkrd}n |dkrd}nd}|r t|nd}||fS)z Sort keys so that: - 'slide_1', 'slide_2', ..., 'slide_10' are in numeric order - 'mcq_1', 'mcq_2', ... come after slides in numeric order _rslidermcqrisdigitrArprefixnum_strgroupnumberr5r5r6custom_slide_sort_keypsz5generate_slide_content..custom_slide_sort_keyc |dSNrr5kvrr5r6r r)1ryes0F+Warning: No suitable image found for slide  after multiple attempts./Warning: No visualization suggestion for slide .rFus Based on the following context from the last two slides, generate one multiple-choice question (MCQ). The question should be relevant to the content and designed to test comprehension. **Context**: {context} The MCQ must include: - A **question** related to the context - Exactly **4 answer options** - A clear indication of the **correct answer** as a single letter: 'a', 'b', 'c', or 'd' ⚠️ **Critical Requirements**: - ✅ Return **only valid JSON** — no explanations, headers, or extra text. - ✅ Ensure all fields and options are enclosed in **double quotes (`"`)**. - ✅ Do **not** use letters like "A.", "B." in the options — just the plain text. - Dont give Here is the MCQ while generating MCQ - Directly follow the format given below The final output **must strictly follow** this format: ```json {{ "question": "", "options": [ "