`nx_pov_shift.nx` source

↩ module page · 473 lines · 17645 B
1// nx_pov_shift.nx -- 3rd-person -> 1st-person POV transform.
2//
3// Takes source text (post-HTML-strip, post-text-ingest) and emits
4// a variant where the chosen protagonist's pronouns shift to
5// first person.  The OTHER characters' pronouns stay 3rd person.
6//
7// Use case (from user):
8//   "the story needs to happen like a literotica story from my pov
9//    which i think may be both but the output has to go into a joy
10//    caption and future nishi captioning machine for some systems
11//    like z image, for audio like erotica that we can train to be
12//    like an erotic novel via comping to s class"
13//
14// So we emit TWO variants per source paragraph:
15//   1. USER_AS_I       -- protagonist pronouns become "I/me/my".
16//                         Reads as literotica narrator-as-reader.
17//   2. COMPANION_AS_HER -- companion stays "she/her" so image-prompt
18//                          downstreams (z-image / JoyCaption / future
19//                          Nishi captioner) get a clean scene
20//                          description with the visible character
21//                          described in third person.
22//
23// Both variants are emitted; the caller picks per downstream
24// (audio narration uses USER_AS_I, image prompt uses
25// COMPANION_AS_HER, video sub-system picks based on its prompt
26// schema).
27//
28// Per cardinal feedback-build-intelligence-never-strip-features:
29// every replacement preserves SOURCE BYTE LENGTH AND ORDERING
30// outside of the substitution.  No paragraph re-flow, no
31// punctuation normalization, no quotation-mark "smartening".
32//
33// nx_safety_envelope:
34//   intended_use:       "Transform 3rd-person source text to 1st-
35//                        person POV for the chosen protagonist
36//                        gender, leaving other characters in 3rd
37//                        person.  Emits user-as-I and companion-
38//                        as-her variants per source span."
39//   sil_target:         SIL2
40//   asil_target:        QM
41//   dal_target:         DAL C
42//   iec_62304_class:    NONE
43//   evidence:           [no_floating_point,
44//                        bounded_pronoun_table_static,
45//                        word_boundary_replacement_no_substring_match,
46//                        source_preserved_outside_replacements,
47//                        capitalization_preserved]
48//   hazard_register:    [bug-tape-pronoun-in-name-falsely-matched,
49//                        bug-tape-possessive-his-her-confusion,
50//                        bug-tape-contraction-shes-hes-missed]
51//   residual_risk:      "ASCII pronoun coverage only; non-English
52//                        pronouns pass through unchanged.  Proper
53//                        nouns matching pronoun spelling (e.g.
54//                        someone literally named 'He') would mis-
55//                        substitute -- caller pins protagonist name
56//                        in V2 to suppress."
57//   verdict:            NOT_YET_EVALUATED
58
59import "nx_syscalls.nx"
60
61// ===== sealed protagonist gender ==================================
62
63const NX_POV_GENDER_UNKNOWN: i64 = 0
64const NX_POV_GENDER_HE:      i64 = 1   // male protagonist -> "I"
65const NX_POV_GENDER_SHE:     i64 = 2   // female protagonist -> "I"
66const NX_POV_GENDER_THEY:    i64 = 3   // they/them protagonist -> "I"
67
68// ===== sealed POV variant codes ===================================
69
70const NX_POV_VARIANT_SOURCE:         i64 = 0  // verbatim source
71const NX_POV_VARIANT_USER_AS_I:      i64 = 1  // protagonist -> I
72const NX_POV_VARIANT_COMPANION_3RD:  i64 = 2  // protagonist stays 3rd
73
74// ===== byte helpers ===============================================
75
76func nx_pov_load_u8(p: *u8, i: i64) -> i64 {
  let q: *u8 = ((p as i64) + i) as *u8
  return *q
79}
80
81func nx_pov_store_u8(p: *u8, i: i64, v: i64) {
  let q: *u8 = ((p as i64) + i) as *u8
  *q = v as u8
84}
85
86func nx_pov_to_lower(c: i64) -> i64 {
  if c >= 0x41 {
      if c <= 0x5A { return c | 0x20 }
  }
  return c
91}
92
93func nx_pov_is_upper(c: i64) -> i64 {
  if c >= 0x41 {
      if c <= 0x5A { return 1 }
  }
  return 0
98}
99
100func nx_pov_is_alnum(c: i64) -> i64 {
  if c >= 0x30 {
      if c <= 0x39 { return 1 }
  }
  if c >= 0x41 {
      if c <= 0x5A { return 1 }
  }
  if c >= 0x61 {
      if c <= 0x7A { return 1 }
  }
  return 0
111}
112
113// ===== pronoun substitution table =================================
114//
115// Given the protagonist gender and a lowercased source pronoun,
116// returns the replacement string (or 0-pointer if no substitution).
117//
118// Coverage:
119//   he/she/they -> i      him/her/them -> me      his/her/their -> my
120//   himself/herself/themself -> myself
121//   he's/she's/they're -> I'm      he'd/she'd/they'd -> I'd
122//   he'll/she'll/they'll -> I'll
123
124func nx_pov_replacement(gender: i64, src_word: *u8, src_len: i64,
                      out_len: *i64) -> *u8 {
  if gender == NX_POV_GENDER_UNKNOWN { *out_len = 0; return 0 as *u8 }
127
  // Single-pronoun mapping driven by gender.
  if gender == NX_POV_GENDER_HE {
      if src_len == 2 {
          // "he"
          if nx_pov_load_u8(src_word, 0) == 0x68 {
              if nx_pov_load_u8(src_word, 1) == 0x65 {
                  *out_len = 1
                  return "I" as *u8
              }
          }
      }
      if src_len == 3 {
          // "him" / "his"
          let a: i64 = nx_pov_load_u8(src_word, 0)
          let b: i64 = nx_pov_load_u8(src_word, 1)
          let c: i64 = nx_pov_load_u8(src_word, 2)
          if a == 0x68 {
              if b == 0x69 {
                  if c == 0x6D { *out_len = 2; return "me" as *u8 }  // him
                  if c == 0x73 { *out_len = 2; return "my" as *u8 }  // his
              }
          }
      }
      if src_len == 4 {
          // "he's" or "he'd"
          let a: i64 = nx_pov_load_u8(src_word, 0)
          let b: i64 = nx_pov_load_u8(src_word, 1)
          let c: i64 = nx_pov_load_u8(src_word, 2)
          let d: i64 = nx_pov_load_u8(src_word, 3)
          if a == 0x68 {
              if b == 0x65 {
                  if c == 0x27 {
                      if d == 0x73 { *out_len = 3; return "I'm" as *u8 }
                      if d == 0x64 { *out_len = 3; return "I'd" as *u8 }
                  }
              }
          }
      }
      if src_len == 5 {
          // "he'll"
          let a: i64 = nx_pov_load_u8(src_word, 0)
          let b: i64 = nx_pov_load_u8(src_word, 1)
          let c: i64 = nx_pov_load_u8(src_word, 2)
          let d: i64 = nx_pov_load_u8(src_word, 3)
          let e: i64 = nx_pov_load_u8(src_word, 4)
          if a == 0x68 {
              if b == 0x65 {
                  if c == 0x27 {
                      if d == 0x6C {
                          if e == 0x6C {
                              *out_len = 4; return "I'll" as *u8
                          }
                      }
                  }
              }
          }
      }
      if src_len == 7 {
          // "himself"
          let l_him: *u8 = "himself" as *u8
          var ok: i64 = 1
          var i: i64 = 0
          while i < 7 {
              if nx_pov_load_u8(src_word, i) != nx_pov_load_u8(l_him, i) {
                  ok = 0; i = 7
              }
              i = i + 1
          }
          if ok == 1 { *out_len = 6; return "myself" as *u8 }
      }
  }
199
  if gender == NX_POV_GENDER_SHE {
      if src_len == 3 {
          // "she" / "her"
          let a: i64 = nx_pov_load_u8(src_word, 0)
          let b: i64 = nx_pov_load_u8(src_word, 1)
          let c: i64 = nx_pov_load_u8(src_word, 2)
          if a == 0x73 {
              if b == 0x68 {
                  if c == 0x65 { *out_len = 1; return "I" as *u8 }  // she
              }
          }
          if a == 0x68 {
              if b == 0x65 {
                  if c == 0x72 {
                      // "her" is ambiguous: object or possessive.
                      // Most safe default: possessive ("her dress" -> "my dress").
                      // Caller can refine post-shift.
                      *out_len = 2; return "my" as *u8
                  }
              }
          }
      }
      if src_len == 4 {
          // "hers" / "she's"
          let a: i64 = nx_pov_load_u8(src_word, 0)
          let b: i64 = nx_pov_load_u8(src_word, 1)
          let c: i64 = nx_pov_load_u8(src_word, 2)
          let d: i64 = nx_pov_load_u8(src_word, 3)
          if a == 0x73 {
              if b == 0x68 {
                  if c == 0x65 {
                      if d == 0x27 {
                          // would need 5th char to be 's'/'d'/'l';
                          // 4 chars exactly = "she'" without close --
                          // skip.
                          *out_len = 0; return 0 as *u8
                      }
                  }
              }
          }
          if a == 0x68 {
              if b == 0x65 {
                  if c == 0x72 {
                      if d == 0x73 { *out_len = 4; return "mine" as *u8 }
                  }
              }
          }
      }
      if src_len == 5 {
          // "she's" / "she'd"
          let a: i64 = nx_pov_load_u8(src_word, 0)
          let b: i64 = nx_pov_load_u8(src_word, 1)
          let c: i64 = nx_pov_load_u8(src_word, 2)
          let d: i64 = nx_pov_load_u8(src_word, 3)
          let e: i64 = nx_pov_load_u8(src_word, 4)
          if a == 0x73 {
              if b == 0x68 {
                  if c == 0x65 {
                      if d == 0x27 {
                          if e == 0x73 { *out_len = 3; return "I'm" as *u8 }
                          if e == 0x64 { *out_len = 3; return "I'd" as *u8 }
                      }
                  }
              }
          }
      }
      if src_len == 6 {
          // "she'll"
          let l_shll: *u8 = "she'll" as *u8
          var ok2: i64 = 1
          var j: i64 = 0
          while j < 6 {
              if nx_pov_load_u8(src_word, j) != nx_pov_load_u8(l_shll, j) {
                  ok2 = 0; j = 6
              }
              j = j + 1
          }
          if ok2 == 1 { *out_len = 4; return "I'll" as *u8 }
      }
      if src_len == 7 {
          // "herself"
          let l_her: *u8 = "herself" as *u8
          var okh: i64 = 1
          var jh: i64 = 0
          while jh < 7 {
              if nx_pov_load_u8(src_word, jh) != nx_pov_load_u8(l_her, jh) {
                  okh = 0; jh = 7
              }
              jh = jh + 1
          }
          if okh == 1 { *out_len = 6; return "myself" as *u8 }
      }
  }
293
  if gender == NX_POV_GENDER_THEY {
      if src_len == 4 {
          // "they"
          let l_they: *u8 = "they" as *u8
          var oth: i64 = 1
          var ti: i64 = 0
          while ti < 4 {
              if nx_pov_load_u8(src_word, ti) != nx_pov_load_u8(l_they, ti) {
                  oth = 0; ti = 4
              }
              ti = ti + 1
          }
          if oth == 1 { *out_len = 1; return "I" as *u8 }
      }
      if src_len == 4 {
          // "them"
          let l_them: *u8 = "them" as *u8
          var otm: i64 = 1
          var tj: i64 = 0
          while tj < 4 {
              if nx_pov_load_u8(src_word, tj) != nx_pov_load_u8(l_them, tj) {
                  otm = 0; tj = 4
              }
              tj = tj + 1
          }
          if otm == 1 { *out_len = 2; return "me" as *u8 }
      }
      if src_len == 5 {
          // "their"
          let l_their: *u8 = "their" as *u8
          var oth2: i64 = 1
          var tk: i64 = 0
          while tk < 5 {
              if nx_pov_load_u8(src_word, tk) != nx_pov_load_u8(l_their, tk) {
                  oth2 = 0; tk = 5
              }
              tk = tk + 1
          }
          if oth2 == 1 { *out_len = 2; return "my" as *u8 }
      }
  }
335
  *out_len = 0
  return 0 as *u8
338}
339
340// ===== capitalize-helper ==========================================
341//
342// If src_was_capitalized, return the first byte of out in uppercase;
343// otherwise lowercase.  Returns byte to write.
344
345func nx_pov_apply_case(out_byte: i64, was_upper: i64) -> i64 {
  if was_upper == 1 {
      if out_byte >= 0x61 {
          if out_byte <= 0x7A { return out_byte - 0x20 }
      }
  }
  return out_byte
352}
353
354// ===== top-level POV-shift transform ==============================
355//
356// Walk source bytes, identify candidate word spans, attempt
357// pronoun replacement.  Write to dst buffer.  Returns dst byte
358// count.
359
360func nx_pov_shift(gender: i64, src: *u8, n: i64,
                dst: *u8, dst_cap: i64) -> i64 {
  if dst_cap <= 0 { return 0 }
  if n <= 0 { return 0 }
364
  let scratch_raw: *u8 = sys_mmap(16)
  let scratch: *i64 = scratch_raw as *i64
367
  var dpos: i64 = 0
  var p: i64 = 0
  let BUDGET: i64 = n + 2
  var iter: i64 = 0
  while p < n {
      if iter >= BUDGET { p = n }
      if p < n {
          let c: i64 = nx_pov_load_u8(src, p)
          // Word boundary: previous byte is non-alphanumeric (or BOF).
          var at_boundary: i64 = 1
          if p > 0 {
              let prev: i64 = nx_pov_load_u8(src, p - 1)
              if nx_pov_is_alnum(prev) == 1 { at_boundary = 0 }
              if prev == 0x27 { at_boundary = 0 }  // apostrophe-inside
          }
          // Word start: alpha char.
          if at_boundary == 1 {
              if nx_pov_is_alnum(c) == 1 {
                  // Find word end (alnum + apostrophe allowed inside).
                  var q: i64 = p
                  let WORD_BUDGET: i64 = 32
                  var w_iter: i64 = 0
                  var scanning: i64 = 1
                  while scanning == 1 {
                      if w_iter >= WORD_BUDGET { scanning = 0 }
                      if scanning == 1 {
                          if q >= n { scanning = 0 }
                          if scanning == 1 {
                              let qc: i64 = nx_pov_load_u8(src, q)
                              if nx_pov_is_alnum(qc) == 1 { q = q + 1 }
                              if nx_pov_is_alnum(qc) == 0 {
                                  if qc == 0x27 {
                                      // include apostrophe only if followed by alnum
                                      if q + 1 < n {
                                          let nc: i64 = nx_pov_load_u8(src, q + 1)
                                          if nx_pov_is_alnum(nc) == 1 { q = q + 1 }
                                          if nx_pov_is_alnum(nc) == 0 { scanning = 0 }
                                      }
                                      if q + 1 >= n { scanning = 0 }
                                  }
                                  if qc != 0x27 { scanning = 0 }
                              }
                          }
                      }
                      w_iter = w_iter + 1
                  }
                  let wlen: i64 = q - p
                  // Build lowercase copy in a stack buffer.
                  let lower_raw: *u8 = sys_mmap(64)
                  var li: i64 = 0
                  while li < wlen {
                      nx_pov_store_u8(lower_raw, li,
                                      nx_pov_to_lower(nx_pov_load_u8(src, p + li)))
                      li = li + 1
                  }
                  let was_upper: i64 = nx_pov_is_upper(c)
424
                  var out_len: i64 = 0
                  let rep: *u8 = nx_pov_replacement(gender, lower_raw, wlen,
                                                   &out_len)
                  if rep != (0 as *u8) {
                      if out_len > 0 {
                          // Write replacement with caps preserved on first char.
                          var ri: i64 = 0
                          while ri < out_len {
                              if dpos < dst_cap {
                                  let rb: i64 = nx_pov_load_u8(rep, ri)
                                  var wb: i64 = rb
                                  if ri == 0 {
                                      wb = nx_pov_apply_case(rb, was_upper)
                                  }
                                  nx_pov_store_u8(dst, dpos, wb)
                                  dpos = dpos + 1
                              }
                              ri = ri + 1
                          }
                          p = q
                          iter = iter + 1
                          continue
                      }
                  }
                  // No replacement: pass through verbatim.
                  var ki: i64 = 0
                  while ki < wlen {
                      if dpos < dst_cap {
                          nx_pov_store_u8(dst, dpos, nx_pov_load_u8(src, p + ki))
                          dpos = dpos + 1
                      }
                      ki = ki + 1
                  }
                  p = q
                  iter = iter + 1
                  continue
              }
          }
          // Non-word byte: pass through.
          if dpos < dst_cap {
              nx_pov_store_u8(dst, dpos, c)
              dpos = dpos + 1
          }
          p = p + 1
      }
      iter = iter + 1
  }
  return dpos
473}
nx_pov_shift.nx source

`nx_pov_shift.nx` source
