code wiki / (root) / nx_markdown_inline.nx
`nx_markdown_inline.nx` source

↩ module page · 235 lines · 8232 B
1// markdown_inline.nx -- render inline Markdown to HTML.
2//
3// Inline-level only -- headings / paragraphs / lists / code
4// blocks belong in a future markdown_block.nx that walks line-
5// by-line and feeds each line through THIS module.  Splitting
6// inline from block keeps each module small + testable.
7//
8// Supported inline constructs (CommonMark §6, practical subset):
9//   *em*             -> <em>em</em>
10//   _em_             -> <em>em</em>
11//   **strong**       -> <strong>strong</strong>
12//   __strong__       -> <strong>strong</strong>
13//   `code`           -> <code>code</code>
14//   [label](href)    -> <a href=\"href\">label</a>
15//
16// HTML-unsafe characters in surrounding text (<, >, &, \", ')
17// pass through html_escape.nx so we don't re-introduce XSS.
18//
19// Composes html_escape.nx.
20//
21// Invariants:
22//   MD1  No nested emphasis parsing today -- *outer *inner*
23//        outer* renders naively (matches emphasis greedily).
24//   MD2  Code spans (`...`) disable markup inside; their content
25//        IS html-escaped so `<script>` renders as code.
26//   MD3  Link href is NOT validated or escaped against
27//        javascript: / data: schemes.  For untrusted content
28//        callers should pre-filter href before feeding us.
29
30// nx_safety_envelope:
31//   intended_use:       AUTO_APPLIED -- primitive-specific tuning queued
32//   sil_target:         SIL1
33//   evidence:           [bulk_applied_2026-05-16, see-file-comment-for-detail]
34//   verdict:            NOT_YET_EVALUATED
35
36import "nx_syscalls.nx"
37import "nx_html_escape.nx"
38
39const MD_ERR_SHORT: i64 = -1
40
41func md_put(out: *u8, cap: i64, off: i64, src: *u8, n: i64) -> i64 {
  if off + n > cap { return MD_ERR_SHORT }
  var i: i64 = 0
  while i < n {
      out[off + i] = src[i]
      i = i + 1
  }
  return off + n
49}
50
51// Emit html_escape(src[a..b]) at out[off..].
52func md_put_esc(out: *u8, cap: i64, off: i64,
              src: *u8, a: i64, b: i64) -> i64 {
  let w: i64 = html_escape(out + off, cap - off, src + a, b - a)
  if w < 0 { return MD_ERR_SHORT }
  return off + w
57}
58
59// Render inline markdown src[0..n] into out[0..].  Returns bytes
60// written.  Parser is a single forward pass with peek-and-match
61// for delimiter pairs.
62func markdown_inline(src: *u8, n: i64, out: *u8, cap: i64) -> i64 {
  var i: i64 = 0
  var o: i64 = 0
  var plain_start: i64 = 0   // start of current uninterrupted run
66
  while i < n {
      let c: i64 = src[i]
69
      // Code span: backtick opens; find closing backtick.
      if c == 0x60 {
          // Flush pending plain text.
          o = md_put_esc(out, cap, o, src, plain_start, i)
          if o < 0 { return o }
          // Find the matching backtick.
          var j: i64 = i + 1
          while j < n {
              if src[j] == 0x60 { break }
              j = j + 1
          }
          if j < n {
              o = md_put(out, cap, o, "<code>", 6)
              if o < 0 { return o }
              o = md_put_esc(out, cap, o, src, i + 1, j)
              if o < 0 { return o }
              o = md_put(out, cap, o, "</code>", 7)
              if o < 0 { return o }
              i = j + 1
              plain_start = i
              continue
          }
          // Unterminated backtick: emit literal.
          i = i + 1
          continue
      }
96
      // Strong emphasis: ** or __ (same char doubled).
      if c == 0x2A {
          if i + 1 < n {
              if src[i + 1] == 0x2A {
                  // find the closing **
                  var j: i64 = i + 2
                  while j + 1 < n {
                      if src[j] == 0x2A {
                          if src[j + 1] == 0x2A { break }
                      }
                      j = j + 1
                  }
                  if j + 1 < n {
                      o = md_put_esc(out, cap, o, src, plain_start, i)
                      if o < 0 { return o }
                      o = md_put(out, cap, o, "<strong>", 8)
                      if o < 0 { return o }
                      o = md_put_esc(out, cap, o, src, i + 2, j)
                      if o < 0 { return o }
                      o = md_put(out, cap, o, "</strong>", 9)
                      if o < 0 { return o }
                      i = j + 2
                      plain_start = i
                      continue
                  }
              }
          }
          // Single '*' emphasis.
          var j: i64 = i + 1
          while j < n {
              if src[j] == 0x2A { break }
              j = j + 1
          }
          if j < n {
              o = md_put_esc(out, cap, o, src, plain_start, i)
              if o < 0 { return o }
              o = md_put(out, cap, o, "<em>", 4)
              if o < 0 { return o }
              o = md_put_esc(out, cap, o, src, i + 1, j)
              if o < 0 { return o }
              o = md_put(out, cap, o, "</em>", 5)
              if o < 0 { return o }
              i = j + 1
              plain_start = i
              continue
          }
          i = i + 1
          continue
      }
146
      // Link: [label](href)
      if c == 0x5B {
          // Find closing ']'
          var close_label: i64 = i + 1
          while close_label < n {
              if src[close_label] == 0x5D { break }
              close_label = close_label + 1
          }
          if close_label < n {
              if close_label + 1 < n {
                  if src[close_label + 1] == 0x28 {
                      // Find closing ')'
                      var close_url: i64 = close_label + 2
                      while close_url < n {
                          if src[close_url] == 0x29 { break }
                          close_url = close_url + 1
                      }
                      if close_url < n {
                          o = md_put_esc(out, cap, o, src, plain_start, i)
                          if o < 0 { return o }
                          o = md_put(out, cap, o, "<a href=\"", 9)
                          if o < 0 { return o }
                          o = md_put_esc(out, cap, o, src,
                                         close_label + 2, close_url)
                          if o < 0 { return o }
                          o = md_put(out, cap, o, "\">", 2)
                          if o < 0 { return o }
                          o = md_put_esc(out, cap, o, src,
                                         i + 1, close_label)
                          if o < 0 { return o }
                          o = md_put(out, cap, o, "</a>", 4)
                          if o < 0 { return o }
                          i = close_url + 1
                          plain_start = i
                          continue
                      }
                  }
              }
          }
      }
187
      i = i + 1
  }
  // Flush trailing plain text.
  o = md_put_esc(out, cap, o, src, plain_start, n)
  return o
193}
194
195// Compile-only smoke.
196func main() -> i64 {
  let out: *u8 = sys_mmap(512)
198
  // Plain + escape.
  let n1: i64 = markdown_inline("a<b>c", 5, out, 512)
  // Expected: "a&lt;b&gt;c" = 11 bytes.
  if n1 != 11 { return 1 }
203
  // Strong + em.
  let n2: i64 = markdown_inline("**bold** and *em*", 17, out, 512)
  if n2 <= 0 { return 2 }
  // Must contain <strong>
  var has_strong: i64 = 0
  var i: i64 = 0
  while i < n2 - 8 {
      if out[i] == 0x3C {
          if out[i + 1] == 0x73 {       // 's'
              if out[i + 2] == 0x74 {   // 't'
                  has_strong = 1
                  break
              }
          }
      }
      i = i + 1
  }
  if has_strong != 1 { return 3 }
222
  // Code span.
  let n3: i64 = markdown_inline("use `printf(\"hi\")`", 18, out, 512)
  if n3 <= 0 { return 4 }
226
  // Link.
  let n4: i64 = markdown_inline("[Nishi](https://nishifamily.com)", 32,
                                out, 512)
  if n4 <= 0 { return 5 }
  if out[0] != 0x3C { return 6 }       // '<'
  if out[1] != 0x61 { return 7 }       // 'a'
233
  return 0
235}
nx_markdown_inline.nx source

`nx_markdown_inline.nx` source
