nx_pov_shift.nx

nx_pov_shift.nx -- 3rd-person -> 1st-person POV transform. Takes source text (post-HTML-strip, post-text-ingest) and emits a variant where the chosen protagonist's pronouns shift to first person. The OTHER characters' pronouns stay 3rd person. Use case (from user): "the story needs to happen like a literotica story from my pov which i think may be both but the output has to go into a joy caption and future nishi captioning machine for some systems like z image, for audio like erotica that we can train to be like an erotic novel via comping to s class" So we emit TWO variants per source paragraph: 1. USER_AS_I -- protagonist pronouns become "I/me/my". Reads as literotica narrator-as-reader. 2. COMPANION_AS_HER -- companion stays "she/her" so image-prompt downstreams (z-image / JoyCaption / future Nishi captioner) get a clean scene description with the visible character described in third person. Both variants are emitted; the caller picks per downstream (audio narration uses USER_AS_I, image prompt uses COMPANION_AS_HER, video sub-system picks based on its prompt schema). Per cardinal feedback-build-intelligence-never-strip-features: every replacement preserves SOURCE BYTE LENGTH AND ORDERING outside of the substitution. No paragraph re-flow, no punctuation normalization, no quotation-mark "smartening". nx_safety_envelope: intended_use: "Transform 3rd-person source text to 1st- person POV for the chosen protagonist gender, leaving other characters in 3rd person. Emits user-as-I and companion- as-her variants per source span." sil_target: SIL2 asil_target: QM