`nx_proof.nx` source

↩ module page · 299 lines · 12759 B
1// nx_proof.nx -- Checks for grammatical and stylistic errors in prose without relying on external dictionaries.
2﻿// nx_proof.nx -- GRAMMAR + STYLE PROOFING for the Nishi Office suite (office axis "Grammar and style
3// proofing", sym pf_check). DELIBERATELY DICTIONARY-FREE: every rule here is decidable from the text
4// itself, so the organ has no lexicon to ship, stale, or lie about -- spelling is a SEPARATE axis with a
5// separate derived lexicon (nx_spell). What that buys: these findings are never wrong because a word was
6// missing from a word list; a flag here is a real structural defect in the prose.
7//
8// RULES (each independently gate-proven, each with a negative control so the rule is proven able to FIRE
9// and able to STAY SILENT):
10//   1 doubled word          "the the"            -- case-insensitive, word-boundary exact
11//   2 double space          "a  b"               -- inside a sentence, not leading indentation
12//   3 lowercase start       ". foo"              -- a sentence opening lowercase
13//   4 no terminal punct     text ending bare     -- last non-space char is not . ! ? : ; or a closer
14//   5 unbalanced bracket    "( [ {"              -- depth tracked, reports the unclosed count
15//   6 unbalanced quote      '"'                  -- odd count of double quotes
16//   7 trailing whitespace   "foo   \n"           -- space/tab immediately before a newline
17//
18// MARKUP IS SKIPPED: a run of <...> is not prose, so rules never fire inside a tag (an HTML/XML doc would
19// otherwise drown in false "lowercase start" and "no terminal punctuation" findings).
20//   nx_proof check <file>     -- report every finding as  LINE:COL RULE detail
21//   nx_proof (argless)        -- selftest
22// expect_exit: 0  license_tier: ORIGINAL
23import "nx_syscalls.nx"
24
25const PF_CAP:      i64 = 1048576
26const PF_RULE_DBL: i64 = 1
27const PF_RULE_SP2: i64 = 2
28const PF_RULE_LOW: i64 = 3
29const PF_RULE_END: i64 = 4
30const PF_RULE_BRK: i64 = 5
31const PF_RULE_QUO: i64 = 6
32const PF_RULE_TWS: i64 = 7
33
34func pf_puts(s: *u8) -> i64 { var n: i64 = 0; while s[n] != (0 as u8) { n = n + 1 } sys_write(1, s, n); return 0 }
35func pf_num(v: i64) -> i64 {
  if v == 0 { pf_puts("0" as *u8); return 0 }
  let t: *u8 = sys_mmap(32)
  var m: i64 = v
  var k: i64 = 0
  if m < 0 { pf_puts("-" as *u8); m = 0 - m }
  while m > 0 { t[k] = (48 + (m - (m / 10) * 10)) as u8; m = m / 10; k = k + 1 }
  let r: *u8 = sys_mmap(32)
  var i: i64 = 0
  while k > 0 { k = k - 1; r[i] = t[k]; i = i + 1 }
  sys_write(1, r, i)
  return 0
47}
48
49func pf_is_alpha(c: i64) -> i64 {
  if c >= 65 { if c <= 90 { return 1 } }
  if c >= 97 { if c <= 122 { return 1 } }
  return 0
53}
54func pf_lower(c: i64) -> i64 { if c >= 65 { if c <= 90 { return c + 32 } } return c }
55func pf_is_upper(c: i64) -> i64 { if c >= 65 { if c <= 90 { return 1 } } return 0 }
56func pf_is_space(c: i64) -> i64 { if c == 32 { return 1 } if c == 9 { return 1 } return 0 }
57
58// a sentence terminator, or a closer that legitimately ends a block
59func pf_is_term(c: i64) -> i64 {
  if c == 46 { return 1 }
  if c == 33 { return 1 }
  if c == 63 { return 1 }
  if c == 58 { return 1 }
  if c == 59 { return 1 }
  if c == 41 { return 1 }
  if c == 93 { return 1 }
  if c == 125 { return 1 }
  if c == 34 { return 1 }
  return 0
70}
71
72func pf_rule_name(r: i64) -> *u8 {
  if r == PF_RULE_DBL { return "doubled-word" as *u8 }
  if r == PF_RULE_SP2 { return "double-space" as *u8 }
  if r == PF_RULE_LOW { return "lowercase-sentence-start" as *u8 }
  if r == PF_RULE_END { return "no-terminal-punctuation" as *u8 }
  if r == PF_RULE_BRK { return "unbalanced-bracket" as *u8 }
  if r == PF_RULE_QUO { return "unbalanced-quote" as *u8 }
  if r == PF_RULE_TWS { return "trailing-whitespace" as *u8 }
  return "unknown" as *u8
81}
82
83// THE CHECKER. Returns the number of findings; counts[r] gets the per-rule tally (r = 1..7).
84// emit=1 prints each finding as LINE:COL rule.
85func pf_check(text: *u8, n: i64, counts: *i64, emit: i64) -> i64 {
  var total: i64 = 0
  var line: i64 = 1
  var col: i64 = 1
  var depth: i64 = 0
  var quotes: i64 = 0
  var in_markup: i64 = 0
  var i: i64 = 0
  // start-of-sentence tracking: we are AT a sentence start until the first prose char of it
  var want_upper: i64 = 1
  var last_prose: i64 = 0
  var any_prose: i64 = 0
97
  while i < n {
      let c: i64 = text[i] as i64
100
      if c == 60 { in_markup = 1 }
      if in_markup == 1 {
          if c == 62 { in_markup = 0 }
          if c == 10 { line = line + 1; col = 0 }
          i = i + 1
          col = col + 1
          continue
      }
109
      // rule 7: whitespace immediately before a newline
      if c == 10 {
          if i > 0 { if pf_is_space(text[i - 1] as i64) == 1 {
              total = total + 1
              counts[PF_RULE_TWS] = counts[PF_RULE_TWS] + 1
              if emit == 1 { pf_num(line); pf_puts(":" as *u8); pf_num(col); pf_puts(" " as *u8); pf_puts(pf_rule_name(PF_RULE_TWS)); pf_puts("\n" as *u8) }
          } }
          line = line + 1
          col = 0
          i = i + 1
          col = col + 1
          continue
      }
123
      // rule 2: two spaces inside a line that has already started (leading indent is not a defect)
      if c == 32 { if i + 1 < n { if (text[i + 1] as i64) == 32 { if any_prose == 1 {
          total = total + 1
          counts[PF_RULE_SP2] = counts[PF_RULE_SP2] + 1
          if emit == 1 { pf_num(line); pf_puts(":" as *u8); pf_num(col); pf_puts(" " as *u8); pf_puts(pf_rule_name(PF_RULE_SP2)); pf_puts("\n" as *u8) }
      } } } }
130
      // rules 5/6: bracket depth + quote parity
      if c == 40 { depth = depth + 1 }
      if c == 91 { depth = depth + 1 }
      if c == 123 { depth = depth + 1 }
      if c == 41 { depth = depth - 1 }
      if c == 93 { depth = depth - 1 }
      if c == 125 { depth = depth - 1 }
      if c == 34 { quotes = quotes + 1 }
139
      if pf_is_alpha(c) == 1 {
          any_prose = 1
          // rule 3: the first letter after a terminator must be uppercase
          if want_upper == 1 {
              if pf_is_upper(c) == 0 {
                  total = total + 1
                  counts[PF_RULE_LOW] = counts[PF_RULE_LOW] + 1
                  if emit == 1 { pf_num(line); pf_puts(":" as *u8); pf_num(col); pf_puts(" " as *u8); pf_puts(pf_rule_name(PF_RULE_LOW)); pf_puts("\n" as *u8) }
              }
              want_upper = 0
          }
          // rule 1: doubled word -- compare this word with the previous one
          var ws: i64 = i
          var we: i64 = i
          while we < n { if pf_is_alpha(text[we] as i64) == 0 { break } we = we + 1 }
          // walk back over exactly one space to the previous word
          var ps: i64 = 0
          var pe: i64 = 0
          if ws >= 2 { if (text[ws - 1] as i64) == 32 {
              pe = ws - 1
              ps = pe
              while ps > 0 { if pf_is_alpha(text[ps - 1] as i64) == 0 { break } ps = ps - 1 }
          } }
          if pe > ps {
              let la: i64 = we - ws
              let lb: i64 = pe - ps
              if la == lb { if la > 1 {
                  var same: i64 = 1
                  var k: i64 = 0
                  while k < la { if pf_lower(text[ws + k] as i64) != pf_lower(text[ps + k] as i64) { same = 0; k = la } else { k = k + 1 } }
                  if same == 1 {
                      total = total + 1
                      counts[PF_RULE_DBL] = counts[PF_RULE_DBL] + 1
                      if emit == 1 { pf_num(line); pf_puts(":" as *u8); pf_num(col); pf_puts(" " as *u8); pf_puts(pf_rule_name(PF_RULE_DBL)); pf_puts("\n" as *u8) }
                  }
              } }
          }
          last_prose = c
          i = we
          col = col + (we - ws)
          continue
      }
182
      if pf_is_term(c) == 1 { if c != 34 { want_upper = 1 } }
      if pf_is_space(c) == 0 { if c != 10 { last_prose = c; any_prose = 1 } }
185
      i = i + 1
      col = col + 1
  }
189
  // rule 4: the document must end on a terminator
  if any_prose == 1 { if pf_is_term(last_prose) == 0 {
      total = total + 1
      counts[PF_RULE_END] = counts[PF_RULE_END] + 1
      if emit == 1 { pf_num(line); pf_puts(":0 " as *u8); pf_puts(pf_rule_name(PF_RULE_END)); pf_puts("\n" as *u8) }
  } }
  // rule 5 tail
  if depth != 0 {
      total = total + 1
      counts[PF_RULE_BRK] = counts[PF_RULE_BRK] + 1
      if emit == 1 { pf_puts("0:0 " as *u8); pf_puts(pf_rule_name(PF_RULE_BRK)); pf_puts(" depth=" as *u8); pf_num(depth); pf_puts("\n" as *u8) }
  }
  // rule 6 tail
  if quotes - (quotes / 2) * 2 != 0 {
      total = total + 1
      counts[PF_RULE_QUO] = counts[PF_RULE_QUO] + 1
      if emit == 1 { pf_puts("0:0 " as *u8); pf_puts(pf_rule_name(PF_RULE_QUO)); pf_puts("\n" as *u8) }
  }
  return total
209}
210
211func pf_readfile(path: *u8, buf: *u8, cap: i64) -> i64 {
  let fd: i64 = sys_openat_rd(path)
  if fd < 0 { return 0 - 1 }
  let n: i64 = sys_read(fd, buf, cap - 1)
  sys_close(fd)
  if n < 0 { return 0 - 1 }
  buf[n] = 0 as u8
  return n
219}
220
221func pf_zero(c: *i64) -> i64 { var i: i64 = 0; while i < 16 { c[i] = 0; i = i + 1 } return 0 }
222
223func main(argc: i64, argv: *i64) -> i64 {
  if argc >= 3 {
      let v: *u8 = argv[1] as *u8
      var isc: i64 = 1
      let want: *u8 = "check" as *u8
      var q: i64 = 0
      while want[q] != (0 as u8) { if v[q] != want[q] { isc = 0 } q = q + 1 }
      if isc == 1 {
          let buf: *u8 = sys_mmap(PF_CAP)
          let n: i64 = pf_readfile(argv[2] as *u8, buf, PF_CAP)
          if n < 0 { pf_puts("PROOF-READ-FAIL\n" as *u8); return 2 }
          let counts: *i64 = sys_mmap(128) as *i64
          pf_zero(counts)
          let t: i64 = pf_check(buf, n, counts, 1)
          pf_puts("NX-PROOF findings=" as *u8); pf_num(t); pf_puts("\n" as *u8)
          return 0
      }
  }
241
  // ---- selftest: every rule proven able to FIRE, and a clean text proven to stay SILENT ----
  var pass: i64 = 0
  var tot: i64 = 0
  let c: *i64 = sys_mmap(128) as *i64
246
  pf_zero(c)
  pf_check("The the cat sat.\n" as *u8, 16, c, 0)
  tot = tot + 1
  if c[PF_RULE_DBL] == 1 { pass = pass + 1; pf_puts("  PASS R1 doubled word fires\n" as *u8) } else { pf_puts("  FAIL R1\n" as *u8) }
251
  pf_zero(c)
  pf_check("A  b.\n" as *u8, 6, c, 0)
  tot = tot + 1
  if c[PF_RULE_SP2] == 1 { pass = pass + 1; pf_puts("  PASS R2 double space fires\n" as *u8) } else { pf_puts("  FAIL R2\n" as *u8) }
256
  pf_zero(c)
  pf_check("One. two.\n" as *u8, 10, c, 0)
  tot = tot + 1
  if c[PF_RULE_LOW] == 1 { pass = pass + 1; pf_puts("  PASS R3 lowercase sentence start fires\n" as *u8) } else { pf_puts("  FAIL R3\n" as *u8) }
261
  pf_zero(c)
  pf_check("No terminator here\n" as *u8, 19, c, 0)
  tot = tot + 1
  if c[PF_RULE_END] == 1 { pass = pass + 1; pf_puts("  PASS R4 missing terminal punctuation fires\n" as *u8) } else { pf_puts("  FAIL R4\n" as *u8) }
266
  pf_zero(c)
  pf_check("A (b.\n" as *u8, 6, c, 0)
  tot = tot + 1
  if c[PF_RULE_BRK] == 1 { pass = pass + 1; pf_puts("  PASS R5 unbalanced bracket fires\n" as *u8) } else { pf_puts("  FAIL R5\n" as *u8) }
271
  pf_zero(c)
  pf_check("A \"b.\n" as *u8, 6, c, 0)
  tot = tot + 1
  if c[PF_RULE_QUO] == 1 { pass = pass + 1; pf_puts("  PASS R6 unbalanced quote fires\n" as *u8) } else { pf_puts("  FAIL R6\n" as *u8) }
276
  pf_zero(c)
  pf_check("Trailing space. \n" as *u8, 17, c, 0)
  tot = tot + 1
  if c[PF_RULE_TWS] == 1 { pass = pass + 1; pf_puts("  PASS R7 trailing whitespace fires\n" as *u8) } else { pf_puts("  FAIL R7\n" as *u8) }
281
  // NEGATIVE CONTROL: clean prose must produce ZERO findings, else every rule above is meaningless
  pf_zero(c)
  let clean: i64 = pf_check("The cat sat. It slept.\n" as *u8, 23, c, 0)
  tot = tot + 1
  if clean == 0 { pass = pass + 1; pf_puts("  PASS NEG clean prose -> 0 findings\n" as *u8) } else { pf_puts("  FAIL NEG clean prose flagged " as *u8); pf_num(clean); pf_puts("\n" as *u8) }
287
  // MARKUP CONTROL: the same defects INSIDE a tag must NOT fire
  pf_zero(c)
  let mk: i64 = pf_check("<a href=\"x  y\">Ok.</a>\n" as *u8, 23, c, 0)
  tot = tot + 1
  if mk == 0 { pass = pass + 1; pf_puts("  PASS MARKUP defects inside a tag are skipped\n" as *u8) } else { pf_puts("  FAIL MARKUP flagged " as *u8); pf_num(mk); pf_puts("\n" as *u8) }
293
  pf_puts("NX-PROOF-GATE " as *u8); pf_num(pass); pf_puts("/" as *u8); pf_num(tot)
  if pass == tot { pf_puts(" GREEN\n" as *u8); return 0 }
  pf_puts(" RED\n" as *u8)
  return 1
298}
299
nx_proof.nx source

`nx_proof.nx` source
