h=ddlZddlZddlZddlZddlmZddlmZejddZ dZ da de de efd Zd Zdd e d ed efdZy)N)dict_row)OpenAI EMBED_MODELztext-embedding-3-smallc tjtjddtjddtjddtjddtjd d t S) NDB_HOST localhostDB_PORT5432DB_NAME dynbot_dbDB_USER chatbot_user DB_PASSWORDz"tLgPX5dQJ^LgXttV3Q3PfJ7V3YJ*X9Q7JL)hostportdbnameuserpassword row_factory)psycopgconnectosgetenvr*/var/www/html/dynbot/referenz_retrieval.py get_db_connr s] ?? YYy+ . YYy& )yyK0 YYy. 1=*NO  rtextreturnctttjdatjj t |xsdj}|jdjS)NOPENAI_API_KEY)api_key)modelinputr) _openairrr embeddingscreaterstripdata embedding)rresps r embed_queryr-sW+;!<=    $ $;tzr>P>P>R $ SD 99Q< ! !!rcd}d}d}t||D]\}}|||zz }|||zz }|||zz }|dk(s|dk(ry|tj|tj|zz S)Nr)zipmathsqrt)abdotnanbxys r cosine_simr:s} CCrcAq )1 qs B!A#IBrQqSyr) Qw"' $))B-$))B-/ 00rchallenge_text tenant_idtop_kcNt|}d}d}d}t5}|j5}|j||f|j } i} | D]} g} | dr t j | d} | r t|| nd} | d}| j|}|s| d| | dk\rd nd d }n$t|d | |d <| dk\r |d xxd z cc<|| |<| sgcdddcdddSg}| jD]\}}|j|||f|j}d}|r*|dr% t j |d}t||}|j||d|d ||d dt|d |zd|zzd|jdd|dtd |}g}|D]}|j|||d|dd|j }|Dcgc]+}|djdk7s|ddk(r |d|dd-}}|j|d|dt!|d|d|cdddcdddS#t$rg} YwxYw#t$rY!wxYwcc}w#1swYnxYwdddy#1swYyxYw)a Python-side similarity: 1) Fetch all Einordnung bullets for tenant 2) Score with cosine_sim 3) Roll-up by story_id (take max), bonus for multiple close hits 4) Re-rank by full Einordnung similarity as tie-break 5) Return top_k with all sections (HTML) z SELECT id, story_id, title, content_text, embedding_json FROM documents WHERE tenant_id = %s AND section ILIKE 'Einordnung' AND bullet_index >= 0 z SELECT embedding_json FROM documents WHERE tenant_id = %s AND section ILIKE 'Einordnung' AND bullet_index = -1 AND story_id = %s LIMIT 1 a SELECT section, bullet_index, content_html FROM documents WHERE tenant_id = %(tenant_id)s AND title = %(title)s AND story_id = %(story_id)s AND NOT (section ILIKE 'Einordnung' AND bullet_index >= 0) ORDER BY CASE WHEN section ILIKE 'Titel' THEN 0 WHEN section ILIKE 'Subtitel' THEN 1 WHEN section ILIKE 'Branche' THEN 2 WHEN section ILIKE 'Beschreibung-Titel' THEN 3 WHEN section ILIKE 'Beschreibung' THEN 4 WHEN section ILIKE 'Beschreibung-Highlight' THEN 5 WHEN section ILIKE 'Beschreibung-Finish' THEN 6 WHEN section ILIKE 'Einordnung' THEN 7 WHEN section ILIKE 'Beitrag' THEN 8 WHEN section ILIKE 'Zitat' THEN 9 ELSE 99 END, section; embedding_jsonr/story_idtitleg333333?r)rAbest close_hitsrCrDNgffffff?g333333?)r@rA bullet_simfull_simrDfinalc|d|dfS)NrGrDr)rs rz/find_matching_referenzstories..s1W:q"?rT)keyreverse)r<rAr@section einordnung bullet_index content_html)rMrQrG)r@rAscoresections)r-rcursorexecutefetchalljsonloads Exceptionr:getmaxitemsfetchoneappendsortlowerfloat)r;r<r=qvecFETCH_BULLETS_SQLFETCH_FULL_SQLFETCH_SECTIONS_SQLconncurbullets per_storyrowembsimsidrecrankedfullrFfvectopresultsrIsecss clean_secss rfind_matching_referenzstoriesrw$s? ~ &DN. H$ H % |4,,.  !CC#$**S)9%:;C,/*T3'CCj/C--$C #G ccUYk_`a!#f+s3F $; %*% IcN! !&5HHH:!) HC KKC(8 9<<>DH-.::d+;&<=D)$5H MMW!&k$!,/CK ::S8^K   &  ? N^c!Um$ A KK&"+ajaPZm\ <<>DY<%%'<71^;LPR;RiL!N:KLJ NNjM7qz*&   $QHHH!C6!,yHHHHHsJ2J I8A'J J2AJ8$I1BJ/0J -J J I. *J-I. .J1 I> :J=I> >JJ JJ$))rrWr1r psycopg.rowsropenairrrrr&strlistrar-r:intrwrrrr~sm!bii '?@  "c"d5k"1{#{#{c{r