o lQi- @shUdZddlmZddlZddlZddlZddlmZddlm Z ddl m Z m Z ddl mZGd d d eeZGd d d e ZGd dde ZedejZedZddddddddddd Zded<dddd d!d"d#d$d%d&d'd( ZdRdSd.d/ZdTd0d1ZdUd4d5ZdUd6d7ZdVd:d;ZdWd|-]*$))AiO))i i ))i i ))i i ))i i ))i i ))i i ))i i ))i i ))i i ) Latin DevanagariBengaliGurmukhiGujaratiOdiaTamilTeluguKannada Malayalamz&dict[str, tuple[tuple[int, int], ...]] SCRIPT_RANGESrr r!r"r#r$r%r&r'r) rAssameser r!r"r#r$r%r&r'rschemadictdefs dict | Nonereturncsdur |did|vr$|ddd}|i}tt|Si}|D]+\}}|dkr3q*t|tr@t|||<q*t|trQfdd|D||<q*|||<q*|S)Nz$defsz$ref/cs$g|]}t|trt|n|qSr) isinstancer+ _resolve_refs).0itemr,rr csz!_resolve_refs..)popsplitgetr2r+itemsr1list)r*r,ref_nameresolvedresultkeyvaluerr5rr2Ss$      r2cCs4t}t|}d|d<|ddd}d|d<|S)NFadditionalProperties propertiesrr:)rmodel_json_schemar2)r*r= item_schemarrr"get_transcript_variant_json_schemals rE language_codercCs|tvrd}t|\}}}|S)Nenr)rF_ script_namerrrget_target_script_nameusrJcCst|}t|dS)Nr)rJSCRIPT_BLOCK_BY_NAMEr9)rFrIrrrget_target_script_block|s rLtext list[str]cCs t|SN)PROTECTED_SPAN_REfindallrMrrrextract_protected_spanss rScCs td|S)N )rPsubrRrrr_strip_protected_spanss rVcharrangesIterable[tuple[int, int]]boolcst|tfdd|DS)Nc3s,|]\}}|ko|knVqdSrOr)r3startend codepointrr s*z"_char_in_ranges..)ordany)rWrXrr]r_char_in_rangessrbdict[str, int]cCs~t|}ddtD}|D]/}t|}|dddvrq d}tD]\}}t||r8||d7<d}nq#|rsz(detect_script_counts..r>LMFT)rVr( unicodedatacategoryr:rb)rMscrubbedcountsrWrimatchedrIrXrrrdetect_script_countss"  rmcs|dkrtjSt|}t||d}|dd}tfdd|D}|dkr7|dkr7|dkr7tjS|dkrF|dkrF|dkrFtjS|dkrU|dkrU|dkrUtjStj S)NrGrrc3s$|] \}}|dhvr|VqdS)rNr)r3rIcount native_blockrrr_s z(classify_input_script..) r r rmrLr9sumr:r r r)rMrFrk native_count latin_count other_countrrorclassify_input_scripts    rucCst|}t|duSrO)rVASCII_ROMAN_RE fullmatch)rMrjrrrromanized_text_is_asciisrxr: list[dict]cCsNg}|D]}||d|d|d|ddqtj|ddd}d |d S) NrrFinput_script_profilerM)rrFrzrMF) ensure_asciiindentzgProduce both variants for every input item. Return one JSON object matching the schema exactly. INPUT:  )appendjsondumps)r:normalized_itemsr4payloadrrr$build_transcript_variant_user_prompts ruROLE Convert transcript text into two deterministic script variants for the same spoken utterance. Output only the JSON object required by the schema. INPUT Each item has: id, language_code, input_script_profile, text. The caller already skipped fully_roman items, so every item here must be processed. TASK For each item return: 1. native_script_text 2. romanized_text GLOBAL RULES - Same utterance only. No translation, paraphrase, cleanup, grammar fixing, expansion, or added context. - Preserve word order, clause order, repetitions, fillers, and punctuation as closely as possible. - Never infer omitted words before or after the visible text. - Return one output item per input item in the same order. - Copy id exactly from the input. - No explanations, notes, confidence, or extra keys. NATIVE_SCRIPT_TEXT - Render the utterance in the target language's native script. - If input_script_profile is fully_native, native_script_text may be identical to the input. - If input_script_profile is mixed_native_latin, convert ordinary spoken Latin-script words into the target native script. - Do not change protected spans. - Common spoken borrowings like Google, WhatsApp, click, video, website, support, form, upload, download, meeting may be written in native script when they are ordinary spoken words, not literals. ROMANIZED_TEXT - Render the same utterance in plain ASCII Roman letters. - No diacritics, IPA, or scholarly transliteration. - Use everyday user-style Romanization. - Keep Romanization internally consistent within an item. - Well-known brand names may remain in their natural Roman form when that is the most faithful rendering. PROTECTED SPANS Preserve these byte-for-byte unchanged in both outputs: - email addresses - URLs and web domains - @handles and #hashtags - numbers, dates, times, amounts, version-like numbers - file names and literal identifier-like tokens If unsure whether a token is protected, preserve it exactly. PUNCTUATION - Keep punctuation close to the input. - Do not invent new clauses. - Do not add surrounding quotes. EXAMPLES Example 1 input: या Google पर जाकर whatsapp@support.com पर click करके भी अपने WhatsApp का access ले सकते हैं native_script_text: या गूगल पर जाकर whatsapp@support.com पर क्लिक करके भी अपने वॉट्सऐप का एक्सेस ले सकते हैं romanized_text: Ya Google par jaakar whatsapp@support.com par click karke bhi apne WhatsApp ka access le sakte hain Example 2 input: अनुराग ठाकुर जी तमिलनाडु में आकर आठ नौ जगहों पर प्रचार किए हैं native_script_text: अनुराग ठाकुर जी तमिलनाडु में आकर आठ नौ जगहों पर प्रचार किए हैं romanized_text: Anurag Thakur ji Tamilnadu mein aakar aath nau jagahon par prachaar kiye hain Example 3 input: আমাৰ যিটো website www.brainexcellencelabs.com তাত আপুনি search কৰিলে agenda খিনি জানিব পাৰিব native_script_text: আমাৰ যিটো ৱেবছাইট www.brainexcellencelabs.com তাত আপুনি চাৰ্চ কৰিলে এজেণ্ডা খিনি জানিব পাৰিব romanized_text: Amar jito website www.brainexcellencelabs.com tat apuni search korile agenda khini janib parib Example 4 input: ഈ മുഖം അങ്ങനെയല്ല എന്ന് വെച്ചാൽ they can change it. ഒന്നും കഷ്ടമേ കിട്ടത്തില്ല. native_script_text: ഈ മുഖം അങ്ങനെയല്ല എന്ന് വെച്ചാൽ ദേ കാൻ ചേഞ്ച് ഇറ്റ്. ഒന്നും കഷ്ടമേ കിട്ടത്തില്ല. romanized_text: Ee mukham anganeyalla ennu vechaal they can change it. Onnum kashtame kittathilla. Example 5 input: అందువల్ల డిపార్ట్మెంట్ టు డిపార్ట్మెంట్ యాక్షన్ తీసుకోవడం వల్ల native_script_text: అందువల్ల డిపార్ట్మెంట్ టు డిపార్ట్మెంట్ యాక్షన్ తీసుకోవడం వల్ల romanized_text: Anduvalla department tu department action teesukovadam valla Example 6 input: આ meeting 3 વાગ્યે start થશે native_script_text: આ મીટિંગ 3 વાગ્યે સ્ટાર્ટ થશે romanized_text: A meeting 3 vaagye start thashe FINAL Return a single JSON object matching the schema exactly. Never translate. Preserve protected spans exactly. cCstSrO)#CACHEABLE_TRANSCRIPT_VARIANT_PROMPTrrrr'get_cacheable_transcript_variant_prompt;srrO)r*r+r,r-r.r+)r.r+)rFrr.r)rMrr.rN)rMrr.r)rWrrXrYr.rZ)rMrr.rc)rMrrFrr.r )rMrr.rZ)r:ryr.r)r.r)'__doc__ __future__rrrerhenumrtypingrpydanticrrconfigr rr rrcompile IGNORECASErPrvr(rrKr2rErJrLrSrVrbrmrurxrrrrrrrsj               i