circle-ir 4.15.0 → 4.16.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (135) hide show
  1. package/configs/sinks/golang.json +4 -4
  2. package/dist/analysis/config-loader.d.ts.map +1 -1
  3. package/dist/analysis/config-loader.js +11 -7
  4. package/dist/analysis/config-loader.js.map +1 -1
  5. package/dist/analysis/constant-caller-gate.d.ts +39 -0
  6. package/dist/analysis/constant-caller-gate.d.ts.map +1 -0
  7. package/dist/analysis/constant-caller-gate.js +298 -0
  8. package/dist/analysis/constant-caller-gate.js.map +1 -0
  9. package/dist/analysis/constant-propagation/evaluator.d.ts.map +1 -1
  10. package/dist/analysis/constant-propagation/evaluator.js +3 -1
  11. package/dist/analysis/constant-propagation/evaluator.js.map +1 -1
  12. package/dist/analysis/constant-propagation/patterns.d.ts +19 -0
  13. package/dist/analysis/constant-propagation/patterns.d.ts.map +1 -1
  14. package/dist/analysis/constant-propagation/patterns.js +33 -0
  15. package/dist/analysis/constant-propagation/patterns.js.map +1 -1
  16. package/dist/analysis/constant-propagation/propagator.d.ts +59 -0
  17. package/dist/analysis/constant-propagation/propagator.d.ts.map +1 -1
  18. package/dist/analysis/constant-propagation/propagator.js +396 -49
  19. package/dist/analysis/constant-propagation/propagator.js.map +1 -1
  20. package/dist/analysis/constant-propagation/types.d.ts +9 -0
  21. package/dist/analysis/constant-propagation/types.d.ts.map +1 -1
  22. package/dist/analysis/findings.d.ts.map +1 -1
  23. package/dist/analysis/findings.js +134 -0
  24. package/dist/analysis/findings.js.map +1 -1
  25. package/dist/analysis/interprocedural.d.ts.map +1 -1
  26. package/dist/analysis/interprocedural.js +127 -3
  27. package/dist/analysis/interprocedural.js.map +1 -1
  28. package/dist/analysis/java-array-element-keys.d.ts +14 -0
  29. package/dist/analysis/java-array-element-keys.d.ts.map +1 -0
  30. package/dist/analysis/java-array-element-keys.js +36 -0
  31. package/dist/analysis/java-array-element-keys.js.map +1 -0
  32. package/dist/analysis/js-crlf-strip.d.ts +26 -0
  33. package/dist/analysis/js-crlf-strip.d.ts.map +1 -0
  34. package/dist/analysis/js-crlf-strip.js +244 -0
  35. package/dist/analysis/js-crlf-strip.js.map +1 -0
  36. package/dist/analysis/passes/csharp-sanitizer-credit.d.ts +25 -0
  37. package/dist/analysis/passes/csharp-sanitizer-credit.d.ts.map +1 -1
  38. package/dist/analysis/passes/csharp-sanitizer-credit.js +157 -12
  39. package/dist/analysis/passes/csharp-sanitizer-credit.js.map +1 -1
  40. package/dist/analysis/passes/language-sources-pass.d.ts +38 -3
  41. package/dist/analysis/passes/language-sources-pass.d.ts.map +1 -1
  42. package/dist/analysis/passes/language-sources-pass.js +1234 -127
  43. package/dist/analysis/passes/language-sources-pass.js.map +1 -1
  44. package/dist/analysis/passes/param-source-flow-gate-pass.d.ts.map +1 -1
  45. package/dist/analysis/passes/param-source-flow-gate-pass.js +53 -1
  46. package/dist/analysis/passes/param-source-flow-gate-pass.js.map +1 -1
  47. package/dist/analysis/passes/plaintext-password-storage-pass.d.ts +6 -0
  48. package/dist/analysis/passes/plaintext-password-storage-pass.d.ts.map +1 -1
  49. package/dist/analysis/passes/plaintext-password-storage-pass.js +160 -0
  50. package/dist/analysis/passes/plaintext-password-storage-pass.js.map +1 -1
  51. package/dist/analysis/passes/scan-secrets-pass.d.ts +1 -1
  52. package/dist/analysis/passes/scan-secrets-pass.d.ts.map +1 -1
  53. package/dist/analysis/passes/scan-secrets-pass.js +7 -1
  54. package/dist/analysis/passes/scan-secrets-pass.js.map +1 -1
  55. package/dist/analysis/passes/sink-filter-pass.d.ts +1 -1
  56. package/dist/analysis/passes/sink-filter-pass.d.ts.map +1 -1
  57. package/dist/analysis/passes/sink-filter-pass.js +326 -34
  58. package/dist/analysis/passes/sink-filter-pass.js.map +1 -1
  59. package/dist/analysis/passes/taint-matcher-pass.d.ts.map +1 -1
  60. package/dist/analysis/passes/taint-matcher-pass.js +2 -2
  61. package/dist/analysis/passes/taint-matcher-pass.js.map +1 -1
  62. package/dist/analysis/passes/taint-propagation-pass.d.ts.map +1 -1
  63. package/dist/analysis/passes/taint-propagation-pass.js +1376 -58
  64. package/dist/analysis/passes/taint-propagation-pass.js.map +1 -1
  65. package/dist/analysis/passes/weak-hash-pass.d.ts +6 -0
  66. package/dist/analysis/passes/weak-hash-pass.d.ts.map +1 -1
  67. package/dist/analysis/passes/weak-hash-pass.js +191 -4
  68. package/dist/analysis/passes/weak-hash-pass.js.map +1 -1
  69. package/dist/analysis/passes/weak-random-pass.d.ts +5 -0
  70. package/dist/analysis/passes/weak-random-pass.d.ts.map +1 -1
  71. package/dist/analysis/passes/weak-random-pass.js +30 -0
  72. package/dist/analysis/passes/weak-random-pass.js.map +1 -1
  73. package/dist/analysis/project-callee-escape-gate.d.ts +40 -0
  74. package/dist/analysis/project-callee-escape-gate.d.ts.map +1 -0
  75. package/dist/analysis/project-callee-escape-gate.js +243 -0
  76. package/dist/analysis/project-callee-escape-gate.js.map +1 -0
  77. package/dist/analysis/rust-literal-guard-rebind.d.ts +34 -0
  78. package/dist/analysis/rust-literal-guard-rebind.d.ts.map +1 -0
  79. package/dist/analysis/rust-literal-guard-rebind.js +197 -0
  80. package/dist/analysis/rust-literal-guard-rebind.js.map +1 -0
  81. package/dist/analysis/rust-literal-guard.d.ts +33 -0
  82. package/dist/analysis/rust-literal-guard.d.ts.map +1 -0
  83. package/dist/analysis/rust-literal-guard.js +422 -0
  84. package/dist/analysis/rust-literal-guard.js.map +1 -0
  85. package/dist/analysis/scalar-derivation.d.ts +8 -0
  86. package/dist/analysis/scalar-derivation.d.ts.map +1 -1
  87. package/dist/analysis/scalar-derivation.js +172 -0
  88. package/dist/analysis/scalar-derivation.js.map +1 -1
  89. package/dist/analysis/taint-matcher.d.ts +2 -2
  90. package/dist/analysis/taint-matcher.d.ts.map +1 -1
  91. package/dist/analysis/taint-matcher.js +531 -31
  92. package/dist/analysis/taint-matcher.js.map +1 -1
  93. package/dist/analysis/taint-propagation.d.ts +26 -1
  94. package/dist/analysis/taint-propagation.d.ts.map +1 -1
  95. package/dist/analysis/taint-propagation.js +234 -13
  96. package/dist/analysis/taint-propagation.js.map +1 -1
  97. package/dist/analyzer.d.ts +16 -2
  98. package/dist/analyzer.d.ts.map +1 -1
  99. package/dist/analyzer.js +13 -3
  100. package/dist/analyzer.js.map +1 -1
  101. package/dist/browser/circle-ir.js +24432 -19928
  102. package/dist/core/circle-ir-core.cjs +4936 -3575
  103. package/dist/core/circle-ir-core.js +4936 -3575
  104. package/dist/core/extractors/calls.d.ts +12 -1
  105. package/dist/core/extractors/calls.d.ts.map +1 -1
  106. package/dist/core/extractors/calls.js +332 -22
  107. package/dist/core/extractors/calls.js.map +1 -1
  108. package/dist/core/extractors/dfg.js +102 -10
  109. package/dist/core/extractors/dfg.js.map +1 -1
  110. package/dist/core/extractors/types.d.ts +5 -1
  111. package/dist/core/extractors/types.d.ts.map +1 -1
  112. package/dist/core/extractors/types.js +62 -0
  113. package/dist/core/extractors/types.js.map +1 -1
  114. package/dist/graph/analysis-pass.d.ts +6 -1
  115. package/dist/graph/analysis-pass.d.ts.map +1 -1
  116. package/dist/graph/analysis-pass.js +2 -1
  117. package/dist/graph/analysis-pass.js.map +1 -1
  118. package/dist/navigation/navigation-index.d.ts +134 -14
  119. package/dist/navigation/navigation-index.d.ts.map +1 -1
  120. package/dist/navigation/navigation-index.js +1158 -104
  121. package/dist/navigation/navigation-index.js.map +1 -1
  122. package/dist/navigation/types.d.ts +59 -2
  123. package/dist/navigation/types.d.ts.map +1 -1
  124. package/dist/resolution/cross-file.d.ts +9 -0
  125. package/dist/resolution/cross-file.d.ts.map +1 -1
  126. package/dist/resolution/cross-file.js +102 -43
  127. package/dist/resolution/cross-file.js.map +1 -1
  128. package/dist/resolution/type-hierarchy.d.ts +3 -0
  129. package/dist/resolution/type-hierarchy.d.ts.map +1 -1
  130. package/dist/resolution/type-hierarchy.js +45 -1
  131. package/dist/resolution/type-hierarchy.js.map +1 -1
  132. package/dist/types/index.d.ts +33 -0
  133. package/dist/types/index.d.ts.map +1 -1
  134. package/dist/types/index.js.map +1 -1
  135. package/package.json +1 -1
@@ -18,6 +18,7 @@ import { attachSourceLineCode } from '../taint-matcher.js';
18
18
  import { javaReplaceCrLfCoverage } from '../java-crlf-strip.js';
19
19
  import { javaCallArgument, javaCanonicalBaseLacksSeparator } from '../java-canonical-prefix.js';
20
20
  import { findGoStringsReplacerSanitizers } from '../go-strings-replacer.js';
21
+ import { findRustLiteralEqualityGuardSanitizers } from '../rust-literal-guard.js';
21
22
  import { constLookupTables, isConstTableLookup } from '../const-table-lookup.js';
22
23
  // ---------------------------------------------------------------------------
23
24
  // Pattern tables (moved from analyzer.ts)
@@ -222,6 +223,22 @@ export class LanguageSourcesPass {
222
223
  });
223
224
  }
224
225
  }
226
+ // -- Java (#624): JSPWiki plugin `execute` return is HTML output ------------
227
+ if (language === 'java') {
228
+ const tm = ctx.getResult('taint-matcher');
229
+ const xssSan = tm.sanitizers.filter(sa => sa.sanitizes.includes('xss')).map(sa => ({ line: sa.line, method: sa.method }));
230
+ for (const r of findJavaWikiPluginReturnXSSSinks(code, types, graph.ir.imports, tm.sources, xssSan)) {
231
+ if (additionalSinks.some(s => s.line === r.sinkLine && s.type === 'xss'))
232
+ continue;
233
+ additionalSinks.push({
234
+ type: 'xss',
235
+ cwe: 'CWE-79',
236
+ line: r.sinkLine,
237
+ location: `WikiPlugin.execute returns HTML with user input at line ${r.sinkLine}`,
238
+ confidence: 0.9,
239
+ });
240
+ }
241
+ }
225
242
  // -- Python: assignment sources, trust-boundary sinks, return-XSS sinks --
226
243
  additionalSources.push(...findPythonAssignmentSources(code, language));
227
244
  const pyTaintedVars = language === 'python' ? buildPythonTaintedVars(code) : new Map();
@@ -342,6 +359,7 @@ export class LanguageSourcesPass {
342
359
  // recognition.
343
360
  additionalSanitizers.push(...findPythonRegexAllowlistWrapperSanitizers(code));
344
361
  additionalSanitizers.push(...findPythonSetMembershipXssGuardSanitizers(code));
362
+ additionalSanitizers.push(...findPythonLiteralMembershipGuardSanitizers(code));
345
363
  additionalSanitizers.push(...findPythonTraversalRejectGuardSanitizers(code));
346
364
  additionalSanitizers.push(...findPythonStringLiteralGuardSanitizers(code));
347
365
  additionalSanitizers.push(...findPythonDefaultSaxParserXxeSanitizers(code));
@@ -401,6 +419,8 @@ export class LanguageSourcesPass {
401
419
  if (language === 'rust') {
402
420
  additionalSanitizers.push(...findRustSetAllowlistGuardSanitizers(code));
403
421
  additionalSanitizers.push(...findRustCanonicalizeGuardSanitizers(code));
422
+ // cognium-dev#293: `if x != "a" && x != "b" { return Err(..) }` literal-equality guard.
423
+ additionalSanitizers.push(...findRustLiteralEqualityGuardSanitizers(code));
404
424
  // Sprint 77a (#216 Pattern X): argv-form Command::new(literal).arg(...)
405
425
  // sanitizer.
406
426
  additionalSanitizers.push(...findRustArgvCommandSanitizers(code));
@@ -544,8 +564,6 @@ export class LanguageSourcesPass {
544
564
  const tm = ctx.getResult('taint-matcher');
545
565
  const xssSan = tm.sanitizers.filter(sa => sa.sanitizes.includes('xss')).map(sa => sa.line);
546
566
  for (const r of findJavaSpringReturnXSSSinks(code, types, tm.sources, xssSan)) {
547
- if (additionalSinks.some(s => s.line === r.sinkLine && s.type === 'xss'))
548
- continue;
549
567
  additionalSinks.push({
550
568
  type: 'xss',
551
569
  cwe: 'CWE-79',
@@ -1610,6 +1628,50 @@ function findGoRequestBodySources(sourceCode, language) {
1610
1628
  }
1611
1629
  return sources;
1612
1630
  }
1631
+ /** cognium-dev#490 — split `a; b; c` on one line (no `{`), respecting string literals. */
1632
+ function splitCSharpDepth0Statements(line) {
1633
+ if (line.includes('{'))
1634
+ return [line];
1635
+ const statements = [];
1636
+ let start = 0;
1637
+ let depth = 0;
1638
+ let inStr = null;
1639
+ let esc = false;
1640
+ for (let i = 0; i < line.length; i++) {
1641
+ const c = line[i];
1642
+ if (inStr) {
1643
+ if (esc) {
1644
+ esc = false;
1645
+ continue;
1646
+ }
1647
+ if (c === '\\') {
1648
+ esc = true;
1649
+ continue;
1650
+ }
1651
+ if (c === inStr)
1652
+ inStr = null;
1653
+ continue;
1654
+ }
1655
+ if (c === '"' || c === "'") {
1656
+ inStr = c;
1657
+ continue;
1658
+ }
1659
+ if (c === '(' || c === '[')
1660
+ depth++;
1661
+ else if (c === ')' || c === ']')
1662
+ depth = Math.max(0, depth - 1);
1663
+ else if (c === ';' && depth === 0) {
1664
+ const part = line.slice(start, i + 1).trim();
1665
+ if (part)
1666
+ statements.push(part);
1667
+ start = i + 1;
1668
+ }
1669
+ }
1670
+ const tail = line.slice(start).trim();
1671
+ if (tail)
1672
+ statements.push(tail);
1673
+ return statements.length > 1 ? statements : [line];
1674
+ }
1613
1675
  function findCSharpRequestSources(sourceCode, language) {
1614
1676
  if (language !== 'csharp')
1615
1677
  return [];
@@ -1632,6 +1694,13 @@ function findCSharpRequestSources(sourceCode, language) {
1632
1694
  // of defect as #350 (Java try-with-resources): a resource-declaration syntax
1633
1695
  // the scanner never looked at.
1634
1696
  const assignRe = /^\s*(?:await\s+)?(?:using\s*\(?\s*)?(?:var\s+|[A-Za-z_][\w.<>\[\]]*\s+)?([A-Za-z_]\w*)\s*=\s*(.+?);?\s*$/;
1697
+ // cognium-dev#638 — `arr[i] = req.QueryString["k"]` does not match `assignRe`
1698
+ // (LHS is an indexer). Seed the array base so cross-file calls that pass
1699
+ // `arr` carry taint into `string[]` / `object[]` callee parameters.
1700
+ const arrayElementAssignRe = /^\s*([A-Za-z_]\w*)\s*\[[^\]]+\]\s*=\s*(.+?);?\s*$/;
1701
+ // cognium-dev#638 step 3 — `box.value = req.QueryString["k"]` seeds the
1702
+ // container local so cross-file calls that pass `box` carry taint.
1703
+ const memberFieldAssignRe = /^\s*([A-Za-z_]\w*)\.([A-Za-z_]\w*)\s*=\s*(.+?);?\s*$/;
1635
1704
  // `Request.*` is the inherited controller member. A parameter typed
1636
1705
  // `HttpRequest` / `HttpRequestBase` is the same surface under whatever
1637
1706
  // name the method uses (`req` in Juliet) — cognium-dev#501.
@@ -1761,97 +1830,136 @@ function findCSharpRequestSources(sourceCode, language) {
1761
1830
  // when it is not the line's code.
1762
1831
  const embeddedAssignRe = /(?:^|[{;])\s*(?:await\s+)?(?:using\s*\(?\s*)?(?:var\s+|[A-Za-z_][\w.<>\[\]]*\s+)?([A-Za-z_]\w*)\s*=(?!=)\s*([^;{}]+)/;
1763
1832
  for (let i = 0; i < lines.length; i++) {
1764
- const m = assignRe.exec(lines[i]);
1765
- // cognium-dev#502 — `assignRe` is anchored and greedy, so on a one-line
1766
- // block header that also contains the real assignment
1767
- //
1768
- // using (StreamReader sr = new StreamReader(tcp.GetStream())) { data = sr.ReadLine(); }
1769
- //
1770
- // it captures `sr` with EVERYTHING after `=` as the right-hand side. That
1771
- // RHS contains `sr.ReadLine()`, so the line was classified as a read and
1772
- // bound to `sr` — the reader, not `data`. The embedded-assignment path
1773
- // handles exactly this shape and finds both `sr` and `data`, but it only
1774
- // ran when `assignRe` failed to match at all. Prefer it whenever the
1775
- // captured RHS crosses a block boundary. (#504 was the mirror of this:
1776
- // there `assignRe` did not match and the inner assignment was lost.)
1777
- const rhsCrossesBlock = m !== null && m[2].includes('{');
1778
- if (!m || rhsCrossesBlock) {
1779
- const stmt = lines[i];
1780
- if (stmt.includes('{') && !/^\s*(?:\/\/|\/\*|\*)/.test(stmt)) {
1781
- const emb = new RegExp(embeddedAssignRe.source, 'g');
1782
- let em;
1783
- let seeded = false;
1784
- while ((em = emb.exec(stmt)) !== null) {
1785
- const varName = em[1];
1786
- const rhs = em[2];
1787
- const type = classify(rhs);
1788
- if (!type)
1789
- continue;
1833
+ for (const stmt of splitCSharpDepth0Statements(lines[i])) {
1834
+ const arrEl = arrayElementAssignRe.exec(stmt);
1835
+ if (arrEl) {
1836
+ const [, baseName, rhs] = arrEl;
1837
+ const type = classify(rhs);
1838
+ if (type && baseName) {
1790
1839
  const lineNumber = i + 1;
1791
- if (sources.some(s => s.line === lineNumber && s.variable === varName))
1792
- continue;
1793
- sources.push({
1794
- type,
1795
- location: `${varName} = ${rhs.trim().substring(0, 50)}${rhs.length > 50 ? '...' : ''}`,
1796
- // #502 — Java parity: a database row read is medium, I/O is high.
1797
- severity: type === 'db_input' ? 'medium' : 'high',
1798
- line: lineNumber,
1799
- confidence: 1.0,
1800
- variable: varName,
1801
- });
1802
- seeded = true;
1840
+ if (!sources.some(s => s.line === lineNumber && s.variable === baseName)) {
1841
+ sources.push({
1842
+ type,
1843
+ location: `${baseName}[…] = ${rhs.trim().substring(0, 50)}${rhs.length > 50 ? '...' : ''}`,
1844
+ severity: type === 'db_input' ? 'medium' : 'high',
1845
+ line: lineNumber,
1846
+ confidence: 1.0,
1847
+ variable: baseName,
1848
+ });
1849
+ }
1850
+ continue;
1803
1851
  }
1804
- if (seeded)
1852
+ }
1853
+ const memEl = memberFieldAssignRe.exec(stmt);
1854
+ if (memEl) {
1855
+ const [, baseName, fieldName, rhs] = memEl;
1856
+ const type = classify(rhs);
1857
+ if (type && baseName) {
1858
+ const lineNumber = i + 1;
1859
+ if (!sources.some(s => s.line === lineNumber && s.variable === baseName)) {
1860
+ sources.push({
1861
+ type,
1862
+ location: `${baseName}.${fieldName} = ${rhs.trim().substring(0, 50)}${rhs.length > 50 ? '...' : ''}`,
1863
+ severity: type === 'db_input' ? 'medium' : 'high',
1864
+ line: lineNumber,
1865
+ confidence: 1.0,
1866
+ variable: baseName,
1867
+ });
1868
+ }
1805
1869
  continue;
1870
+ }
1806
1871
  }
1807
- if (!exprStmtCallRe.test(stmt) || nonCallKeywordRe.test(stmt))
1872
+ const m = assignRe.exec(stmt);
1873
+ // cognium-dev#502 — `assignRe` is anchored and greedy, so on a one-line
1874
+ // block header that also contains the real assignment
1875
+ //
1876
+ // using (StreamReader sr = new StreamReader(tcp.GetStream())) { data = sr.ReadLine(); }
1877
+ //
1878
+ // it captures `sr` with EVERYTHING after `=` as the right-hand side. That
1879
+ // RHS contains `sr.ReadLine()`, so the line was classified as a read and
1880
+ // bound to `sr` — the reader, not `data`. The embedded-assignment path
1881
+ // handles exactly this shape and finds both `sr` and `data`, but it only
1882
+ // ran when `assignRe` failed to match at all. Prefer it whenever the
1883
+ // captured RHS crosses a block boundary. (#504 was the mirror of this:
1884
+ // there `assignRe` did not match and the inner assignment was lost.)
1885
+ const rhsCrossesBlock = m !== null && m[2].includes('{');
1886
+ if (!m || rhsCrossesBlock) {
1887
+ if (stmt.includes('{') && !/^\s*(?:\/\/|\/\*|\*)/.test(stmt)) {
1888
+ const emb = new RegExp(embeddedAssignRe.source, 'g');
1889
+ let em;
1890
+ let seeded = false;
1891
+ while ((em = emb.exec(stmt)) !== null) {
1892
+ const varName = em[1];
1893
+ const rhs = em[2];
1894
+ const type = classify(rhs);
1895
+ if (!type)
1896
+ continue;
1897
+ const lineNumber = i + 1;
1898
+ if (sources.some(s => s.line === lineNumber && s.variable === varName))
1899
+ continue;
1900
+ sources.push({
1901
+ type,
1902
+ location: `${varName} = ${rhs.trim().substring(0, 50)}${rhs.length > 50 ? '...' : ''}`,
1903
+ // #502 — Java parity: a database row read is medium, I/O is high.
1904
+ severity: type === 'db_input' ? 'medium' : 'high',
1905
+ line: lineNumber,
1906
+ confidence: 1.0,
1907
+ variable: varName,
1908
+ });
1909
+ seeded = true;
1910
+ }
1911
+ if (seeded)
1912
+ continue;
1913
+ }
1914
+ if (!exprStmtCallRe.test(stmt) || nonCallKeywordRe.test(stmt))
1915
+ continue;
1916
+ const hit = inlineRead(stmt);
1917
+ if (!hit)
1918
+ continue;
1919
+ const { type, read } = hit;
1920
+ const lineNumber = i + 1;
1921
+ if (sources.some(s => s.line === lineNumber))
1922
+ continue;
1923
+ // `variable` is the `<inline>` marker (the same one the inline
1924
+ // colocation flow puts on its path steps), never a real name. It
1925
+ // must be set, and must match nothing:
1926
+ // - with NO variable, `findInitialTaint` seeds every def on the
1927
+ // FOLLOWING line by adjacency, so `Log(file.FileName);` followed by
1928
+ // `var x = "/tmp/a";` tainted the constant `x`;
1929
+ // - with the read text itself (`file.FileName`) as the variable, the
1930
+ // colocation path treats it as an LHS binding and hands the pair
1931
+ // to the variable scan, which only looks at strictly later sinks —
1932
+ // so the same-line flow this exists for was lost.
1933
+ // The marker reaches only a sink on this very statement.
1934
+ const code = stmt.trim();
1935
+ sources.push({
1936
+ type,
1937
+ location: `${read} in ${code.substring(0, 50)}${code.length > 50 ? '...' : ''}`,
1938
+ severity: type === 'db_input' ? 'medium' : 'high',
1939
+ line: lineNumber,
1940
+ confidence: 1.0,
1941
+ variable: '<inline>',
1942
+ code,
1943
+ });
1808
1944
  continue;
1809
- const hit = inlineRead(stmt);
1810
- if (!hit)
1945
+ }
1946
+ const [, varName, rhs] = m;
1947
+ const type = classify(rhs);
1948
+ if (!type)
1811
1949
  continue;
1812
- const { type, read } = hit;
1813
1950
  const lineNumber = i + 1;
1814
- if (sources.some(s => s.line === lineNumber))
1815
- continue;
1816
- // `variable` is the `<inline>` marker (the same one the inline
1817
- // colocation flow puts on its path steps), never a real name. It
1818
- // must be set, and must match nothing:
1819
- // - with NO variable, `findInitialTaint` seeds every def on the
1820
- // FOLLOWING line by adjacency, so `Log(file.FileName);` followed by
1821
- // `var x = "/tmp/a";` tainted the constant `x`;
1822
- // - with the read text itself (`file.FileName`) as the variable, the
1823
- // colocation path treats it as an LHS binding and hands the pair
1824
- // to the variable scan, which only looks at strictly later sinks —
1825
- // so the same-line flow this exists for was lost.
1826
- // The marker reaches only a sink on this very statement.
1827
- const code = stmt.trim();
1951
+ if (sources.some(s => s.line === lineNumber && s.variable === varName))
1952
+ continue;
1828
1953
  sources.push({
1829
1954
  type,
1830
- location: `${read} in ${code.substring(0, 50)}${code.length > 50 ? '...' : ''}`,
1955
+ location: `${varName} = ${rhs.trim().substring(0, 50)}${rhs.length > 50 ? '...' : ''}`,
1956
+ // #502 — Java parity: a database row read is medium, I/O is high.
1831
1957
  severity: type === 'db_input' ? 'medium' : 'high',
1832
1958
  line: lineNumber,
1833
1959
  confidence: 1.0,
1834
- variable: '<inline>',
1835
- code,
1960
+ variable: varName,
1836
1961
  });
1837
- continue;
1838
1962
  }
1839
- const [, varName, rhs] = m;
1840
- const type = classify(rhs);
1841
- if (!type)
1842
- continue;
1843
- const lineNumber = i + 1;
1844
- if (sources.some(s => s.line === lineNumber && s.variable === varName))
1845
- continue;
1846
- sources.push({
1847
- type,
1848
- location: `${varName} = ${rhs.trim().substring(0, 50)}${rhs.length > 50 ? '...' : ''}`,
1849
- // #502 — Java parity: a database row read is medium, I/O is high.
1850
- severity: type === 'db_input' ? 'medium' : 'high',
1851
- line: lineNumber,
1852
- confidence: 1.0,
1853
- variable: varName,
1854
- });
1855
1963
  }
1856
1964
  return sources;
1857
1965
  }
@@ -2003,29 +2111,52 @@ function findJavaScriptAssignmentSources(sourceCode, language) {
2003
2111
  // #667: `REGISTRY[req.query.mod]` on a constant table yields a value the
2004
2112
  // program wrote, not request data.
2005
2113
  const lookupTables = constLookupTables(sourceCode, language);
2114
+ // cognium-dev#582: scan every assignment on the line. A one-line
2115
+ // `if (c) t = "b"; else t = req.query.t` (or braced `var` arms) has the
2116
+ // source on a later assignment; matching only the first and truncating its
2117
+ // RHS dropped that source. After a control-flow header before an
2118
+ // assignment, that assignment owns only its own statement so a condition's
2119
+ // read is not bound to a body variable (`if (req.query.m) { t = "a"; } else
2120
+ // if (req.query.n) { … }`).
2121
+ // cognium-dev#490: a TypeScript declaration may carry a type annotation
2122
+ // (`const q: string = req.query.q`); the binding is the name before the colon,
2123
+ // not the annotation, so the source's own definition is found on the line.
2124
+ const assignmentRe = /(?:(?:var|let|const)\s+(\w+)\s*:\s*[^=;]+?\s*=(?![>=])|(?:(?:var|let|const)\s+)?(\w+)\s*=)\s*/g;
2125
+ const rhsOwnStmt = /^(?:"(?:[^"\\]|\\.)*"|'(?:[^'\\]|\\.)*'|`(?:[^`\\]|\\.)*`|[^;}"'`])*/;
2006
2126
  for (let lineNum = 0; lineNum < lines.length; lineNum++) {
2007
2127
  const line = lines[lineNum];
2008
2128
  const lineNumber = lineNum + 1;
2009
- const assignmentMatch = line.match(/(?:(?:var|let|const)\s+)?(\w+)\s*=\s*(.+)/);
2010
- if (!assignmentMatch)
2011
- continue;
2012
- const [, varName, rhs] = assignmentMatch;
2013
- if (isConstTableLookup(rhs, lookupTables))
2014
- continue;
2015
- for (const { pattern, type } of patterns) {
2016
- if (pattern.test(rhs)) {
2017
- const alreadyExists = sources.some(s => s.line === lineNumber && s.type === type);
2018
- if (!alreadyExists) {
2019
- sources.push({
2020
- type,
2021
- location: `${varName} = ${rhs.trim().substring(0, 50)}${rhs.length > 50 ? '...' : ''}`,
2022
- severity: 'high',
2023
- line: lineNumber,
2024
- confidence: 1.0,
2025
- variable: varName,
2026
- });
2129
+ assignmentRe.lastIndex = 0;
2130
+ for (const assignmentMatch of line.matchAll(assignmentRe)) {
2131
+ const varName = assignmentMatch[1] ?? assignmentMatch[2];
2132
+ if (!varName)
2133
+ continue;
2134
+ let rhs = line.slice((assignmentMatch.index ?? 0) + assignmentMatch[0].length);
2135
+ if (/\b(?:if|while|switch|else)\b/.test(line.slice(0, assignmentMatch.index ?? 0))) {
2136
+ rhs = rhsOwnStmt.exec(rhs)?.[0] ?? rhs;
2137
+ }
2138
+ else if (line.includes(';', (assignmentMatch.index ?? 0) + assignmentMatch[0].length)) {
2139
+ // Same-line later statements: do not let a greedy RHS attribute their
2140
+ // reads to this variable (`a = 1; b = req.query.t`).
2141
+ rhs = rhsOwnStmt.exec(rhs)?.[0] ?? rhs;
2142
+ }
2143
+ if (isConstTableLookup(rhs, lookupTables))
2144
+ continue;
2145
+ for (const { pattern, type } of patterns) {
2146
+ if (pattern.test(rhs)) {
2147
+ const alreadyExists = sources.some(s => s.line === lineNumber && s.type === type && s.variable === varName);
2148
+ if (!alreadyExists) {
2149
+ sources.push({
2150
+ type,
2151
+ location: `${varName} = ${rhs.trim().substring(0, 50)}${rhs.length > 50 ? '...' : ''}`,
2152
+ severity: 'high',
2153
+ line: lineNumber,
2154
+ confidence: 1.0,
2155
+ variable: varName,
2156
+ });
2157
+ }
2158
+ break;
2027
2159
  }
2028
- break;
2029
2160
  }
2030
2161
  }
2031
2162
  }
@@ -2041,8 +2172,8 @@ function findJavaScriptAssignmentSources(sourceCode, language) {
2041
2172
  // event-name allowlist to avoid firing on unrelated `.on('drain', cb)`
2042
2173
  // / `.on('close', cb)` etc. that carry non-user-input.
2043
2174
  sources.push(...findJavaScriptCallbackParamSources(sourceCode));
2044
- if (language === 'javascript') {
2045
- sources.push(...findExpressDestructuredRequestSources(sourceCode));
2175
+ sources.push(...findExpressDestructuredRequestSources(sourceCode));
2176
+ if (language === 'javascript' || language === 'typescript') {
2046
2177
  sources.push(...findGraphqlResolverArgSources(sourceCode));
2047
2178
  }
2048
2179
  return sources;
@@ -2062,6 +2193,36 @@ function matchBrace(text, open) {
2062
2193
  }
2063
2194
  return -1;
2064
2195
  }
2196
+ /**
2197
+ * Index just past an optional TypeScript `: <type>` annotation starting at
2198
+ * `from` (`}: Request<A, B>`), i.e. at the next top-level `,` or `)`. Returns
2199
+ * `from` unchanged when no annotation follows.
2200
+ */
2201
+ function skipTypeAnnotation(text, from) {
2202
+ const m = /^\s*:/.exec(text.slice(from));
2203
+ if (!m)
2204
+ return from;
2205
+ let depth = 0;
2206
+ for (let i = from + m[0].length; i < text.length; i++) {
2207
+ const ch = text[i];
2208
+ if (ch === '<' || ch === '(' || ch === '[' || ch === '{')
2209
+ depth++;
2210
+ else if (ch === '>' || ch === ']' || ch === '}')
2211
+ depth--;
2212
+ else if (ch === ')') {
2213
+ if (depth === 0)
2214
+ return i;
2215
+ depth--;
2216
+ }
2217
+ else if (ch === ',' && depth === 0)
2218
+ return i;
2219
+ else if (ch === '=' && text[i + 1] === '>' && depth === 0)
2220
+ return from;
2221
+ if (depth < 0)
2222
+ return from;
2223
+ }
2224
+ return from;
2225
+ }
2065
2226
  const EXPRESS_REQUEST_KEYS = {
2066
2227
  query: 'http_param',
2067
2228
  params: 'http_param',
@@ -2080,7 +2241,7 @@ const EXPRESS_REQUEST_KEYS = {
2080
2241
  function findExpressDestructuredRequestSources(sourceCode) {
2081
2242
  const sources = [];
2082
2243
  const receivers = new Set();
2083
- for (const m of sourceCode.matchAll(/\b(?:const|let|var)\s+(\w+)\s*=\s*(?:express\s*\(\s*\)|express\s*\.\s*Router\s*\(\s*\)|Router\s*\(\s*\))/g)) {
2244
+ for (const m of sourceCode.matchAll(/\b(?:const|let|var)\s+(\w+)\s*(?::\s*[\w.]+\s*)?=\s*(?:express\s*\(\s*\)|express\s*\.\s*Router\s*\(\s*\)|Router\s*\(\s*\))/g)) {
2084
2245
  receivers.add(m[1]);
2085
2246
  }
2086
2247
  if (receivers.size === 0)
@@ -2095,7 +2256,8 @@ function findExpressDestructuredRequestSources(sourceCode) {
2095
2256
  continue;
2096
2257
  // The pattern must be followed by another parameter or `)`, i.e. it is
2097
2258
  // the handler's first parameter.
2098
- if (!/^\s*[,)]/.test(sourceCode.slice(close)))
2259
+ // A TypeScript `: Request` annotation after the pattern is skipped first.
2260
+ if (!/^\s*[,)]/.test(sourceCode.slice(skipTypeAnnotation(sourceCode, close))))
2099
2261
  continue;
2100
2262
  const line = sourceCode.slice(0, m.index + m[0].length).split('\n').length;
2101
2263
  const inner = sourceCode.slice(open + 1, close - 1);
@@ -2140,11 +2302,127 @@ function findExpressDestructuredRequestSources(sourceCode) {
2140
2302
  }
2141
2303
  return sources;
2142
2304
  }
2305
+ /**
2306
+ * The parameters of the call / function whose `(` is at `open`: each top-level,
2307
+ * comma-separated entry with its offset into `text`. A comma inside `()`, `[]`,
2308
+ * `{}` or a type's `<>` (the `>` of `=>` excluded) does not split. Returns null
2309
+ * when the list does not close.
2310
+ */
2311
+ function readParameterList(text, open) {
2312
+ const params = [];
2313
+ let depth = 0;
2314
+ let start = open + 1;
2315
+ for (let i = open; i < text.length; i++) {
2316
+ const ch = text[i];
2317
+ if (ch === '"' || ch === "'" || ch === '`') {
2318
+ for (i++; i < text.length && text[i] !== ch; i++)
2319
+ if (text[i] === '\\')
2320
+ i++;
2321
+ continue;
2322
+ }
2323
+ if (ch === '(' || ch === '[' || ch === '{' || ch === '<')
2324
+ depth++;
2325
+ else if (ch === '>' && text[i - 1] !== '=')
2326
+ depth--;
2327
+ else if (ch === ')' || ch === ']' || ch === '}') {
2328
+ depth--;
2329
+ if (depth === 0) {
2330
+ if (text.slice(start, i).trim() !== '')
2331
+ params.push({ text: text.slice(start, i), start });
2332
+ return { params };
2333
+ }
2334
+ }
2335
+ else if (ch === ',' && depth === 1) {
2336
+ params.push({ text: text.slice(start, i), start });
2337
+ start = i + 1;
2338
+ }
2339
+ }
2340
+ return null;
2341
+ }
2342
+ /**
2343
+ * The names a resolver's `args` parameter binds: the identifier itself
2344
+ * (`args`, `args: Args`, `args?: Args`), or every name bound by an object
2345
+ * pattern (`{ sku }`, `{ sku: s }`, `{ filter: { id } }`, `{ id = 1 }`,
2346
+ * `{ ...rest }`), each with its offset into `param`. An array pattern or a
2347
+ * computed key binds nothing here.
2348
+ */
2349
+ function graphqlArgBindings(param) {
2350
+ const trimmed = param.replace(/^\s+/, '');
2351
+ const lead = param.length - trimmed.length;
2352
+ const ident = /^(?:\.\.\.)?([A-Za-z_$][\w$]*)\s*\??\s*(?::[\s\S]*|=[\s\S]*)?$/.exec(trimmed);
2353
+ if (ident)
2354
+ return [{ name: ident[1], offset: lead + trimmed.indexOf(ident[1]) }];
2355
+ if (!trimmed.startsWith('{'))
2356
+ return [];
2357
+ const close = matchBrace(trimmed, 0);
2358
+ if (close < 0)
2359
+ return [];
2360
+ return objectPatternBindings(trimmed, 1, close - 1).map(b => ({ name: b.name, offset: lead + b.offset }));
2361
+ }
2362
+ /** Every name a destructured (non-`args`) parameter binds, for the ambiguity check above. */
2363
+ function graphqlPatternNames(param) {
2364
+ const t = param.trim();
2365
+ if (!t.startsWith('{'))
2366
+ return [];
2367
+ const close = matchBrace(t, 0);
2368
+ if (close < 0)
2369
+ return [];
2370
+ return objectPatternBindings(t, 1, close - 1).map(b => b.name);
2371
+ }
2372
+ function objectPatternBindings(text, from, to) {
2373
+ const out = [];
2374
+ let depth = 0;
2375
+ let start = from;
2376
+ const entries = [];
2377
+ for (let i = from; i < to; i++) {
2378
+ const ch = text[i];
2379
+ if (ch === '{' || ch === '[' || ch === '(')
2380
+ depth++;
2381
+ else if (ch === '}' || ch === ']' || ch === ')')
2382
+ depth--;
2383
+ else if (ch === ',' && depth === 0) {
2384
+ entries.push([start, i]);
2385
+ start = i + 1;
2386
+ }
2387
+ }
2388
+ entries.push([start, to]);
2389
+ for (const [a, b] of entries) {
2390
+ const raw = text.slice(a, b);
2391
+ const body = raw.trimStart();
2392
+ const at = a + (raw.length - body.length);
2393
+ if (body.trim() === '')
2394
+ continue;
2395
+ const rest = /^\.\.\.\s*([A-Za-z_$][\w$]*)\s*$/.exec(body.trimEnd());
2396
+ if (rest) {
2397
+ out.push({ name: rest[1], offset: at + body.indexOf(rest[1]) });
2398
+ continue;
2399
+ }
2400
+ const keyed = /^(?:[A-Za-z_$][\w$]*|'[^']*'|"[^"]*")\s*:\s*/.exec(body);
2401
+ if (keyed) {
2402
+ const targetAt = at + keyed[0].length;
2403
+ const target = body.slice(keyed[0].length);
2404
+ if (target.startsWith('{')) {
2405
+ const end = matchBrace(text, targetAt);
2406
+ if (end > 0)
2407
+ out.push(...objectPatternBindings(text, targetAt + 1, end - 1));
2408
+ continue;
2409
+ }
2410
+ const id = /^([A-Za-z_$][\w$]*)\s*(?:=[\s\S]*)?$/.exec(target.trimEnd());
2411
+ if (id)
2412
+ out.push({ name: id[1], offset: targetAt });
2413
+ continue;
2414
+ }
2415
+ const short = /^([A-Za-z_$][\w$]*)\s*(?:=[\s\S]*)?$/.exec(body.trimEnd());
2416
+ if (short)
2417
+ out.push({ name: short[1], offset: at });
2418
+ }
2419
+ return out;
2420
+ }
2143
2421
  /**
2144
2422
  * cognium-dev#586 — GraphQL resolver-map arguments. The second parameter of a
2145
2423
  * function under `Query` / `Mutation` / `Subscription` in the object passed as
2146
2424
  * `resolvers` to `new ApolloServer({…})` / `makeExecutableSchema({…})` (inline,
2147
- * or a same-file `const resolvers = {…}`) is client-supplied. Parameter 0
2425
+ * or a same-file `const resolvers = {…}`, optionally type-annotated) is client-supplied. Parameter 0
2148
2426
  * (`parent`) and 2 (`context`) are never seeded.
2149
2427
  */
2150
2428
  function findGraphqlResolverArgSources(sourceCode) {
@@ -2166,11 +2444,17 @@ function findGraphqlResolverArgSources(sourceCode) {
2166
2444
  if (!named)
2167
2445
  continue;
2168
2446
  const ident = named[1] ?? 'resolvers';
2169
- const decl = new RegExp(`\\b(?:const|let|var)\\s+${ident}\\s*=\\s*\\{`).exec(sourceCode);
2447
+ // cognium-dev#752: an annotation between the name and `=` (`: Resolvers`,
2448
+ // `: Resolvers<Context>`, `: gql.Resolvers<Ctx, Parent>`) does not hide the map.
2449
+ const decl = new RegExp(`\\b(?:const|let|var)\\s+${ident}\\s*(?::\\s*[A-Za-z_$][\\w$]*(?:\\s*\\.\\s*[A-Za-z_$][\\w$]*)*` +
2450
+ `(?:\\s*<[^<>=]*(?:<[^<>=]*>[^<>=]*)*>)?\\s*)?=\\s*\\{`).exec(sourceCode);
2170
2451
  if (decl)
2171
2452
  maps.push(decl.index + decl[0].length - 1);
2172
2453
  }
2173
- const FN_RE = /(?:^|[\s,{])(\w+)\s*(?::\s*(?:async\s+)?(?:function\s*\w*\s*)?\(|\()\s*\w+\s*,\s*(\w+)\s*[,)]/g;
2454
+ // `name: (…) =>`, `name: async (…) =>`, `name: function (…)`, `name(…) {` and `async name(…)`.
2455
+ // The parameter list is read whole, so a type annotation or a destructured
2456
+ // pattern on any parameter does not hide the second one.
2457
+ const FN_RE = /(?:^|[\s,{])(\w+)\s*(?::\s*(?:async\s+)?(?:function\s*\w*\s*)?\(|\()/g;
2174
2458
  for (const open of maps) {
2175
2459
  const close = matchBrace(sourceCode, open);
2176
2460
  if (close < 0)
@@ -2193,18 +2477,46 @@ function findGraphqlResolverArgSources(sourceCode) {
2193
2477
  else if (ch === '}' || ch === ')')
2194
2478
  depth--;
2195
2479
  }
2480
+ const found = [];
2481
+ // Names the resolvers in this type object bind from `parent` or `context`.
2482
+ // A name shared with an `args` binding is ambiguous across functions, so
2483
+ // it is not seeded.
2484
+ const otherNames = new Set();
2196
2485
  for (const f of typeBody.matchAll(FN_RE)) {
2197
2486
  const at = f.index + f[0].indexOf(f[1]);
2198
2487
  if (depthAt[at] !== 0)
2199
2488
  continue;
2200
- const line = sourceCode.slice(0, tOpen + 1 + at).split('\n').length;
2489
+ const parenAt = f.index + f[0].length - 1;
2490
+ const list = readParameterList(typeBody, parenAt);
2491
+ if (!list)
2492
+ continue;
2493
+ list.params.forEach((param, idx) => {
2494
+ if (idx === 1)
2495
+ return;
2496
+ for (const b of graphqlPatternNames(param.text))
2497
+ otherNames.add(b);
2498
+ });
2499
+ // Only the second parameter (`args`) is client-supplied. `parent` and
2500
+ // `context` are never seeded, destructured or not.
2501
+ const second = list.params[1];
2502
+ if (!second)
2503
+ continue;
2504
+ const absStart = tOpen + 1 + second.start;
2505
+ for (const b of graphqlArgBindings(second.text)) {
2506
+ const line = sourceCode.slice(0, absStart + b.offset).split('\n').length;
2507
+ found.push({ fn: f[1], b: { name: b.name, line } });
2508
+ }
2509
+ }
2510
+ for (const { fn, b } of found) {
2511
+ if (otherNames.has(b.name))
2512
+ continue;
2201
2513
  sources.push({
2202
2514
  type: 'http_param',
2203
- location: `GraphQL ${t[1]}.${f[1]} resolver args '${f[2]}' at line ${line}`,
2515
+ location: `GraphQL ${t[1]}.${fn} resolver args '${b.name}' at line ${b.line}`,
2204
2516
  severity: 'high',
2205
- line,
2517
+ line: b.line,
2206
2518
  confidence: 0.9,
2207
- variable: f[2],
2519
+ variable: b.name,
2208
2520
  });
2209
2521
  }
2210
2522
  }
@@ -2832,6 +3144,480 @@ function isSafeJinjaReturnExpr(expr) {
2832
3144
  const trimmed = expr.trim();
2833
3145
  return JINJA_SAFE_RTS_RE.test(trimmed) || JINJA_SAFE_TEMPLATE_RE.test(trimmed);
2834
3146
  }
3147
+ /**
3148
+ * The statements on one Java source line, split at each `;` that is outside a
3149
+ * string / char literal and outside a comment. Comment text (`//` to the end of
3150
+ * the line, inline or unterminated `/* ... *\/`) is dropped; literals are kept.
3151
+ * `terminated` is false for a trailing piece with no `;` on this line.
3152
+ */
3153
+ function javaLineStatements(line) {
3154
+ const out = [];
3155
+ let cur = '';
3156
+ let inStr = null;
3157
+ let escaped = false;
3158
+ for (let i = 0; i < line.length; i++) {
3159
+ const ch = line[i];
3160
+ if (inStr) {
3161
+ cur += ch;
3162
+ if (escaped) {
3163
+ escaped = false;
3164
+ continue;
3165
+ }
3166
+ if (ch === '\\') {
3167
+ escaped = true;
3168
+ continue;
3169
+ }
3170
+ if (ch === inStr)
3171
+ inStr = null;
3172
+ continue;
3173
+ }
3174
+ if (ch === '"' || ch === "'") {
3175
+ inStr = ch;
3176
+ cur += ch;
3177
+ continue;
3178
+ }
3179
+ if (ch === '/' && line[i + 1] === '/')
3180
+ break;
3181
+ if (ch === '/' && line[i + 1] === '*') {
3182
+ const end = line.indexOf('*/', i + 2);
3183
+ if (end < 0)
3184
+ break;
3185
+ cur += ' ';
3186
+ i = end + 1;
3187
+ continue;
3188
+ }
3189
+ if (ch === ';') {
3190
+ out.push({ text: cur, terminated: true });
3191
+ cur = '';
3192
+ continue;
3193
+ }
3194
+ cur += ch;
3195
+ }
3196
+ if (cur.trim())
3197
+ out.push({ text: cur, terminated: false });
3198
+ return out;
3199
+ }
3200
+ /**
3201
+ * `[modifiers] [Type] name = rhs` as ONE statement (no `;`; see
3202
+ * `javaLineStatements`). `==` is not an assignment; `+=` and field targets
3203
+ * (`a.b = x`) are not matched. Group 1 is the name, group 2 the right-hand side.
3204
+ */
3205
+ const JAVA_STMT_ASSIGN = /^\s*(?:(?:final|public|private|protected|static|volatile|transient)\s+)*(?:[\p{L}_][\p{L}\p{N}_.]*(?:\s*<[^=]*>)?(?:\s*\[\s*\])*\s+)?([\p{L}_][\p{L}\p{N}_]*)\s*=(?!=)\s*([\s\S]*)$/u;
3206
+ /**
3207
+ * Arguments of `receiver.append(...)` on one statement, including a chain of
3208
+ * further `.append(...)` calls on the same expression. Stops at any other
3209
+ * method (e.g. `.indexOf(...)`) or at the end of the statement. Nested calls
3210
+ * inside an argument (e.g. `append(escape(who))`) stay inside that argument.
3211
+ * String / char literal bodies are walked so a `)` inside a string does not
3212
+ * close the argument.
3213
+ */
3214
+ function javaAppendArgExprs(line) {
3215
+ const start = /^\s*([\p{L}_][\p{L}\p{N}_]*)\s*\.\s*append\s*\(/u.exec(line);
3216
+ if (!start)
3217
+ return null;
3218
+ const receiver = start[1];
3219
+ const args = [];
3220
+ let i = start[0].length;
3221
+ const takeArg = () => {
3222
+ let depth = 1;
3223
+ let inStr = null;
3224
+ let escaped = false;
3225
+ const from = i;
3226
+ for (; i < line.length; i++) {
3227
+ const ch = line[i];
3228
+ if (inStr) {
3229
+ if (escaped) {
3230
+ escaped = false;
3231
+ continue;
3232
+ }
3233
+ if (ch === '\\') {
3234
+ escaped = true;
3235
+ continue;
3236
+ }
3237
+ if (ch === inStr)
3238
+ inStr = null;
3239
+ continue;
3240
+ }
3241
+ if (ch === '"' || ch === "'") {
3242
+ inStr = ch;
3243
+ continue;
3244
+ }
3245
+ if (ch === '(') {
3246
+ depth++;
3247
+ continue;
3248
+ }
3249
+ if (ch === ')') {
3250
+ depth--;
3251
+ if (depth === 0) {
3252
+ const arg = line.slice(from, i);
3253
+ i++; // past ')'
3254
+ return arg;
3255
+ }
3256
+ }
3257
+ }
3258
+ return null;
3259
+ };
3260
+ while (true) {
3261
+ const arg = takeArg();
3262
+ if (arg === null)
3263
+ return null;
3264
+ args.push(arg);
3265
+ // More chained .append(…)? Otherwise stop (other method, ';', end).
3266
+ const rest = line.slice(i);
3267
+ const next = /^\s*\.\s*append\s*\(/.exec(rest);
3268
+ if (!next)
3269
+ break;
3270
+ i += next[0].length;
3271
+ }
3272
+ return { receiver, args };
3273
+ }
3274
+ /**
3275
+ * The source call named by a source location (`req.getParameter() in execute`,
3276
+ * `getParameter() in execute`) as a regex over `codeOnly` text: the receiver
3277
+ * chain must match too (any spacing between its tokens), and a call with no
3278
+ * receiver must not follow a `.`. `byName` is the method name alone, for a
3279
+ * receiver whose text is not on the line as written. `method` / `receiver`
3280
+ * are the unescaped names from the location. Null when the location names no
3281
+ * call.
3282
+ */
3283
+ function javaSourceCallRe(location) {
3284
+ const callee = /^([\s\S]*?)\(\)(?:\s+in\s+[^()]*)?$/u.exec((location ?? '').trim())?.[1]?.trim();
3285
+ if (!callee)
3286
+ return null;
3287
+ const cm = /^(?:([\s\S]*?)\s*\.\s*)?([\p{L}_$][\p{L}\p{N}_$]*)$/u.exec(callee);
3288
+ if (!cm)
3289
+ return null;
3290
+ const esc = (t) => t.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
3291
+ const methodName = cm[2];
3292
+ const method = esc(methodName);
3293
+ const byName = new RegExp(`(?<![\\p{L}\\p{N}_$])${method}\\s*\\(`, 'u');
3294
+ if (!cm[1]) {
3295
+ return {
3296
+ exact: new RegExp(`(?<![\\p{L}\\p{N}_$.])(?<!\\.\\s+)${method}\\s*\\(`, 'u'),
3297
+ byName, method: methodName, receiver: null,
3298
+ };
3299
+ }
3300
+ // Receiver text as written in the location, with literal bodies blanked like the
3301
+ // statement text it is matched against.
3302
+ const recv = cm[1]
3303
+ .replace(/"(?:[^"\\]|\\.)*"/g, ' ')
3304
+ .replace(/'(?:[^'\\]|\\.)'/g, ' ');
3305
+ const toks = recv.match(/[\p{L}\p{N}_$]+|\S/gu);
3306
+ if (!toks)
3307
+ return null;
3308
+ return {
3309
+ exact: new RegExp(`(?<![\\p{L}\\p{N}_$.])(?<!\\.\\s+)${toks.map(esc).join('\\s*')}\\s*\\.\\s*${method}\\s*\\(`, 'u'),
3310
+ byName, method: methodName, receiver: toks.join(''),
3311
+ };
3312
+ }
3313
+ /** Keywords that must not be treated as a Java type name in a declaration. */
3314
+ const JAVA_NON_TYPE_NAMES = new Set([
3315
+ 'return', 'if', 'else', 'while', 'for', 'switch', 'case', 'new', 'throw', 'try',
3316
+ 'catch', 'class', 'interface', 'package', 'import', 'assert', 'synchronized',
3317
+ 'do', 'break', 'continue', 'finally', 'default', 'this', 'super', 'true',
3318
+ 'false', 'null', 'instanceof', 'extends', 'implements', 'throws', 'enum',
3319
+ ]);
3320
+ /**
3321
+ * Simple type name → last segment of an FQN, generics stripped
3322
+ * (`javax.servlet.http.HttpServletRequest` → `HttpServletRequest`).
3323
+ */
3324
+ function javaSimpleTypeName(type) {
3325
+ return type.replace(/\s+/g, '').replace(/<.*$/, '').replace(/^.*\./, '');
3326
+ }
3327
+ /**
3328
+ * Locals and parameters declared in `[startLine, endLine]` (1-based), mapped to
3329
+ * their simple type name. Used so a line that lost a second request-source to
3330
+ * per-line dedup can still seed every receiver of that source class.
3331
+ */
3332
+ function javaMethodVarTypes(lines, startLine, endLine) {
3333
+ const out = new Map();
3334
+ // Optional modifiers, then Type (FQN / simple, optional generics), then name,
3335
+ // then `=` `,` `;` or `)` (parameter list). Strings / comments blanked first.
3336
+ const decl = /(?:(?:public|private|protected|static|final|volatile|transient)\s+)*(?<![\p{L}\p{N}_$.])((?:[\p{L}_$][\p{L}\p{N}_$]*\s*\.\s*)*[\p{L}_$][\p{L}\p{N}_$]*)\s*(?:<[^;{}()=]*>)?\s+([\p{L}_$][\p{L}\p{N}_$]*)\s*(?:=|,|;|\))/gu;
3337
+ for (let i = startLine - 1; i < endLine && i < lines.length; i++) {
3338
+ const text = (lines[i] ?? '')
3339
+ .replace(/"(?:[^"\\]|\\.)*"/g, ' ')
3340
+ .replace(/'(?:[^'\\]|\\.)'/g, ' ')
3341
+ .replace(/\/\*.*?\*\//g, ' ')
3342
+ .replace(/\/\/.*$/, '');
3343
+ decl.lastIndex = 0;
3344
+ for (let m = decl.exec(text); m; m = decl.exec(text)) {
3345
+ const simple = javaSimpleTypeName(m[1]);
3346
+ if (!simple || JAVA_NON_TYPE_NAMES.has(simple))
3347
+ continue;
3348
+ out.set(m[2], simple);
3349
+ }
3350
+ }
3351
+ return out;
3352
+ }
3353
+ /**
3354
+ * Call matcher for a source location that accepts every receiver declared with
3355
+ * the same simple type as the location's stored receiver (not only that
3356
+ * receiver). Falls back to `exact` when the stored receiver's type is unknown.
3357
+ * When the location names no receiver, returns `exact` (method-only form).
3358
+ */
3359
+ function javaSameClassCallRe(srcRes, varTypes) {
3360
+ if (!srcRes.receiver)
3361
+ return srcRes.exact;
3362
+ const esc = (t) => t.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
3363
+ // Location receiver may be a chain (`this.req`); the declared name is the
3364
+ // last identifier.
3365
+ const recvId = /[\p{L}_$][\p{L}\p{N}_$]*$/u.exec(srcRes.receiver)?.[0] ?? srcRes.receiver;
3366
+ const names = new Set([recvId]);
3367
+ const ty = varTypes.get(recvId);
3368
+ if (ty) {
3369
+ for (const [n, t] of varTypes)
3370
+ if (t === ty)
3371
+ names.add(n);
3372
+ }
3373
+ const alts = [...names].map(esc).join('|');
3374
+ return new RegExp(`(?<![\\p{L}\\p{N}_$.])(?<!\\.\\s+)(?:${alts})\\s*\\.\\s*${esc(srcRes.method)}\\s*\\(`, 'u');
3375
+ }
3376
+ /**
3377
+ * `expr` (already `codeOnly`) with every call to the named method blanked out,
3378
+ * from its name through its closing `)` (or the end of `expr` when the call
3379
+ * does not close there). `found` is true when at least one call was blanked.
3380
+ */
3381
+ function blankJavaCalls(expr, method) {
3382
+ const re = new RegExp(`(?<![\\p{L}\\p{N}_$])${method.replace(/[$]/g, '\\$&')}\\s*\\(`, 'gu');
3383
+ let text = expr;
3384
+ let found = false;
3385
+ for (let m = re.exec(text); m; m = re.exec(text)) {
3386
+ let depth = 1;
3387
+ let k = m.index + m[0].length;
3388
+ for (; k < text.length && depth > 0; k++) {
3389
+ if (text[k] === '(')
3390
+ depth++;
3391
+ else if (text[k] === ')')
3392
+ depth--;
3393
+ }
3394
+ text = `${text.slice(0, m.index)} 0 ${text.slice(k)}`;
3395
+ re.lastIndex = m.index + 3;
3396
+ found = true;
3397
+ }
3398
+ return { text, found };
3399
+ }
3400
+ /**
3401
+ * cognium-dev#624 (minimal slice) — JSPWiki plugin output is rendered as HTML.
3402
+ *
3403
+ * The `String` returned from `execute(...)` of a class that implements
3404
+ * `org.apache.wiki.api.plugin.WikiPlugin` (or the older `...plugin.Plugin`) is
3405
+ * written into the page unescaped, so a request- or markup-derived value that
3406
+ * reaches the return is reflected XSS. The rule keys on the interface (from the
3407
+ * import or a fully-qualified `implements`), never on the method name: another
3408
+ * class's `execute(...)` is untouched.
3409
+ *
3410
+ * Mirrors `findPythonReturnXSSSinks` (#368): one sink per tainted `return`, and
3411
+ * `TaintPropagationPass` links it to a source. `StringBuilder.append` is
3412
+ * deliberately not an xss sink on its own (#268) — a builder only matters here
3413
+ * when it reaches the return. An xss-covering sanitizer between the source and
3414
+ * the return suppresses the site.
3415
+ */
3416
+ export function findJavaWikiPluginReturnXSSSinks(code, types, imports, sources, xssSanitizers) {
3417
+ if (sources.length === 0)
3418
+ return [];
3419
+ const PKG = 'org.apache.wiki.api.plugin';
3420
+ // Each simple name counts only when that name was imported (a wildcard of the
3421
+ // plugin package accepts both): importing `WikiPlugin` does not make an
3422
+ // `implements Plugin` (some other interface) a wiki plugin, nor the reverse.
3423
+ const importedSimple = new Set();
3424
+ let wildcardImported = false;
3425
+ for (const i of imports) {
3426
+ if (i.from_package !== PKG)
3427
+ continue;
3428
+ if (i.is_wildcard)
3429
+ wildcardImported = true;
3430
+ else if (i.imported_name === 'WikiPlugin' || i.imported_name === 'Plugin')
3431
+ importedSimple.add(i.imported_name);
3432
+ }
3433
+ const isPluginIface = (name) => {
3434
+ const bare = name.replace(/<.*$/, '').trim();
3435
+ if (bare === `${PKG}.WikiPlugin` || bare === `${PKG}.Plugin`)
3436
+ return true;
3437
+ return (bare === 'WikiPlugin' || bare === 'Plugin') && (wildcardImported || importedSimple.has(bare));
3438
+ };
3439
+ const lines = code.split('\n');
3440
+ // Code positions only: string / char literal bodies blanked, then inline `/* */` and
3441
+ // trailing `//` comments dropped (after the literals, so a `"http://…"` is not cut).
3442
+ const codeOnly = (expr) => expr
3443
+ .replace(/"(?:[^"\\]|\\.)*"/g, ' ')
3444
+ .replace(/'(?:[^'\\]|\\.)'/g, ' ')
3445
+ .replace(/\/\*.*?\*\//g, ' ')
3446
+ .replace(/\/\/.*$/, '');
3447
+ const refs = (expr, name) => new RegExp(`(?<![\\p{L}\\p{N}_.])${name.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')}(?![\\p{L}\\p{N}_])`, 'u').test(expr);
3448
+ const out = [];
3449
+ for (const t of types) {
3450
+ if (!t.implements.some(isPluginIface))
3451
+ continue;
3452
+ for (const m of t.methods) {
3453
+ if (m.name !== 'execute' || m.return_type !== 'String')
3454
+ continue;
3455
+ const inMethod = sources.filter(sc => sc.line >= m.start_line && sc.line <= m.end_line);
3456
+ if (inMethod.length === 0)
3457
+ continue;
3458
+ // Sources that name a variable (the carriers are walked per return below).
3459
+ const perSource = inMethod
3460
+ .filter((sc) => !!sc.variable)
3461
+ .map(sc => ({ sc }));
3462
+ if (perSource.length === 0)
3463
+ continue;
3464
+ for (let i = m.start_line - 1; i < m.end_line; i++) {
3465
+ const rm = /^\s*return\s+(.+?);\s*$/.exec(lines[i] ?? '');
3466
+ if (!rm)
3467
+ continue;
3468
+ const sinkLine = i + 1;
3469
+ let best = null;
3470
+ for (const { sc } of perSource) {
3471
+ if (sc.line > sinkLine)
3472
+ continue;
3473
+ // Carriers: the names the source's value reaches before the return, with
3474
+ // the line each became tainted. Walked one statement at a time, in order,
3475
+ // from the source's own statement: a name is live only after the statement
3476
+ // that gives it a tainted value (the source call, or an already-live
3477
+ // carrier), so a statement earlier on the same line sees nothing. Only
3478
+ // names inside a statement's append(...) arguments (chained .append and
3479
+ // nested calls included) taint its receiver, only its `+=` right-hand
3480
+ // side taints its target, and only its `=` right-hand side taints the
3481
+ // assigned name. A name in another method or statement does not.
3482
+ const carriers = new Map();
3483
+ const refsCarrier = (expr) => {
3484
+ const c = codeOnly(expr);
3485
+ return [...carriers.keys()].some(v => refs(c, v));
3486
+ };
3487
+ // Where the source enters on its line. Call sources are deduped to one
3488
+ // per line, so the surviving location names only one receiver
3489
+ // ("req.getParameter() in execute"). Accept every call whose receiver
3490
+ // shares that source class (another HttpServletRequest on the line),
3491
+ // not only the stored receiver — otherwise a second request call never
3492
+ // seeds its value. `prefs.getParameter` stays out: different class.
3493
+ // `byName` (method alone) must not run when the line has another
3494
+ // same-named call: that reintroduces the method-name FP. A parameter /
3495
+ // annotation source (no call) is live from its own assignment on the
3496
+ // line, or from the start of the line.
3497
+ const srcStmts = javaLineStatements(lines[sc.line - 1] ?? '');
3498
+ const srcRes = javaSourceCallRe(sc.location);
3499
+ const varTypes = javaMethodVarTypes(lines, m.start_line, m.end_line);
3500
+ let callRe = null;
3501
+ if (srcRes) {
3502
+ const hasExact = srcStmts.some(st => srcRes.exact.test(codeOnly(st.text)));
3503
+ const byNameHits = srcStmts.filter(st => srcRes.byName.test(codeOnly(st.text))).length;
3504
+ if (hasExact)
3505
+ callRe = javaSameClassCallRe(srcRes, varTypes);
3506
+ else if (byNameHits === 1)
3507
+ callRe = srcRes.byName;
3508
+ else
3509
+ callRe = srcRes.receiver ? javaSameClassCallRe(srcRes, varTypes) : null;
3510
+ }
3511
+ const callsSource = (expr) => !!callRe && callRe.test(codeOnly(expr));
3512
+ const seeded = srcStmts.some(st => callsSource(st.text));
3513
+ // xss sanitizer calls on the source line: a value that reaches a
3514
+ // statement only inside one of them is not tainted by it. A sanitizer
3515
+ // that cannot be found in the line's text covers the whole line.
3516
+ const srcLineSan = sc.line < sinkLine
3517
+ ? xssSanitizers.filter(sa => sa.line === sc.line).map(sa => ({
3518
+ method: /([\p{L}_$][\p{L}\p{N}_$]*)\s*(?:\(\))?\s*$/u.exec(sa.method ?? '')?.[1],
3519
+ }))
3520
+ : [];
3521
+ const sanMethods = srcLineSan.map(sa => sa.method).filter((x) => !!x);
3522
+ const srcLineSanFound = srcLineSan.length > 0 && srcLineSan.every(sa => !!sa.method && srcStmts.some(st => blankJavaCalls(codeOnly(st.text), sa.method).found));
3523
+ const unsanitized = (expr) => {
3524
+ let text = codeOnly(expr);
3525
+ if (srcLineSanFound)
3526
+ for (const sm of sanMethods)
3527
+ text = blankJavaCalls(text, sm).text;
3528
+ return text;
3529
+ };
3530
+ // Unseeded: index of the statement after which `sc.variable` is live.
3531
+ let liveAfter = -1;
3532
+ if (!seeded)
3533
+ liveAfter = srcStmts.findIndex(st => JAVA_STMT_ASSIGN.exec(st.text)?.[1] === sc.variable);
3534
+ // A source on the return line itself is checked against the return as is.
3535
+ if ((!seeded && liveAfter < 0) || sc.line === sinkLine)
3536
+ carriers.set(sc.variable, sc.line);
3537
+ for (let j = sc.line - 1; j < i; j++) {
3538
+ const onSrcLine = j === sc.line - 1;
3539
+ const stmts = onSrcLine ? srcStmts : javaLineStatements(lines[j] ?? '');
3540
+ for (let k = 0; k < stmts.length; k++) {
3541
+ const st = stmts[k];
3542
+ const raw = (expr) => refsCarrier(expr) || (onSrcLine && seeded && callsSource(expr));
3543
+ const tainted = (expr) => (onSrcLine ? raw(unsanitized(expr)) : raw(expr));
3544
+ const as = JAVA_STMT_ASSIGN.exec(st.text);
3545
+ // A name reassigned on the source line from a sanitized value only
3546
+ // (`who = escapeHtml4(who)`) no longer carries the source.
3547
+ if (as && onSrcLine && srcLineSanFound && carriers.has(as[1]) && raw(as[2]) && !tainted(as[2]))
3548
+ carriers.delete(as[1]);
3549
+ else if (as && !carriers.has(as[1]) && tainted(as[2]))
3550
+ carriers.set(as[1], j + 1);
3551
+ const ap = javaAppendArgExprs(st.text);
3552
+ if (ap && !carriers.has(ap.receiver) && ap.args.some(a => tainted(a)))
3553
+ carriers.set(ap.receiver, j + 1);
3554
+ const cm = st.terminated ? /^\s*([\p{L}_][\p{L}\p{N}_]*)\s*\+=\s*(.+?)\s*$/u.exec(st.text) : null;
3555
+ if (cm && !carriers.has(cm[1]) && tainted(cm[2]))
3556
+ carriers.set(cm[1], j + 1);
3557
+ if (onSrcLine && !seeded && k === liveAfter && !carriers.has(sc.variable))
3558
+ carriers.set(sc.variable, sc.line);
3559
+ }
3560
+ }
3561
+ if (!refsCarrier(rm[1]))
3562
+ continue;
3563
+ // An xss-covering sanitizer only counts for this source when it is on a
3564
+ // later line that names one of its carriers. One on the source's own
3565
+ // line already took effect in the walk above (only the value inside
3566
+ // its argument is clean); one there that is not found in the line's
3567
+ // text, or a source on the return line, still covers the line.
3568
+ const sanitized = xssSanitizers.some(({ line: l }) => l >= sc.line && l <= sinkLine &&
3569
+ (l === sc.line ? !srcLineSanFound : refsCarrier(lines[l - 1] ?? '')));
3570
+ if (sanitized)
3571
+ continue;
3572
+ if (!best || sc.line > best.line)
3573
+ best = { line: sc.line };
3574
+ }
3575
+ if (best)
3576
+ out.push({ sinkLine, sourceLine: best.line });
3577
+ }
3578
+ }
3579
+ }
3580
+ return out;
3581
+ }
3582
+ const JSPWIKI_TAG_PKG = 'org.apache.wiki.tags';
3583
+ const JSPWIKI_TAG_SUPER_NAMES = ['WikiTagBase', 'WikiBodyTag', 'WikiLinkTag'];
3584
+ /**
3585
+ * cognium-dev#624 (WikiTag slice) — JSPWiki tag handlers inherit `protected
3586
+ * WikiContext m_wikiContext` from `WikiTagBase`. The configured
3587
+ * `WikiContext.getPage` source is class-scoped, so `m_wikiContext.getPage()`
3588
+ * misses unless the enclosing type extends a JSPWiki tag base (import-gated,
3589
+ * never from the field name alone).
3590
+ */
3591
+ export function javaWikiTagInheritedContextReceiver(receiver, className, call, types, imports) {
3592
+ if (className !== 'WikiContext' || receiver !== 'm_wikiContext')
3593
+ return false;
3594
+ const line = call.location.line;
3595
+ const enclosing = types.find(t => line >= t.start_line && line <= t.end_line &&
3596
+ (!call.in_method || t.methods.some(m => m.name === call.in_method && line >= m.start_line && line <= m.end_line)));
3597
+ if (!enclosing?.extends)
3598
+ return false;
3599
+ const importedSimple = new Set();
3600
+ let wildcardImported = false;
3601
+ for (const i of imports) {
3602
+ if (i.from_package !== JSPWIKI_TAG_PKG)
3603
+ continue;
3604
+ if (i.is_wildcard)
3605
+ wildcardImported = true;
3606
+ else if (JSPWIKI_TAG_SUPER_NAMES.includes(i.imported_name)) {
3607
+ importedSimple.add(i.imported_name);
3608
+ }
3609
+ }
3610
+ const isTagSuper = (name) => {
3611
+ const bare = name.replace(/<.*$/, '').trim();
3612
+ if (bare === `${JSPWIKI_TAG_PKG}.WikiTagBase` ||
3613
+ bare === `${JSPWIKI_TAG_PKG}.WikiBodyTag` ||
3614
+ bare === `${JSPWIKI_TAG_PKG}.WikiLinkTag`)
3615
+ return true;
3616
+ return JSPWIKI_TAG_SUPER_NAMES.includes(bare) &&
3617
+ (wildcardImported || importedSimple.has(bare));
3618
+ };
3619
+ return isTagSuper(enclosing.extends);
3620
+ }
2835
3621
  export function findPythonReturnXSSSinks(sourceCode, taintedVars) {
2836
3622
  if (taintedVars.size === 0)
2837
3623
  return [];
@@ -3099,6 +3885,43 @@ export function buildRustTaintedVars(sourceCode, seedVars) {
3099
3885
  }
3100
3886
  return derived;
3101
3887
  }
3888
+ /**
3889
+ * Join an assignment / declaration whose call argument list wraps over several
3890
+ * lines into one line (`label = String.format("%s.tar",\n picked);`), or null
3891
+ * when line `i` does not start such a statement. The statement must start with
3892
+ * `[Type] name =`, leave a parenthesis open on its first line, and close it and
3893
+ * end in `;` within a few lines, with no `{`, `}`, lambda arrow or second `;`
3894
+ * on the way (a lambda, anonymous class, switch expression or block is not a
3895
+ * plain wrapped value).
3896
+ */
3897
+ function joinWrappedJavaStatement(lines, i) {
3898
+ const strip = (l) => l.replace(/"(?:[^"\\]|\\.)*"|'(?:[^'\\]|\\.)'/g, '""').replace(/\/\/.*$/, '');
3899
+ const first = strip(lines[i]);
3900
+ if (!/^\s*(?:(?:final\s+)?[A-Za-z_][\w.]*(?:\s*<[^>]*>)?(?:\s*\[\s*\])*\s+)?[A-Za-z_]\w*\s*=(?!=)/.test(first))
3901
+ return null;
3902
+ if (/[;{}]|->/.test(first))
3903
+ return null;
3904
+ const depthOf = (t) => (t.match(/\(/g)?.length ?? 0) - (t.match(/\)/g)?.length ?? 0);
3905
+ let depth = depthOf(first);
3906
+ if (depth <= 0)
3907
+ return null;
3908
+ let text = lines[i].replace(/\/\/.*$/, '');
3909
+ for (let j = i + 1; j < lines.length && j <= i + 12; j++) {
3910
+ const next = strip(lines[j]);
3911
+ if (/[{}]|->/.test(next))
3912
+ return null;
3913
+ const semi = next.indexOf(';');
3914
+ if (semi >= 0 && next.slice(semi + 1).trim() !== '')
3915
+ return null;
3916
+ depth += depthOf(next);
3917
+ if (depth < 0)
3918
+ return null;
3919
+ text += ' ' + lines[j].replace(/\/\/.*$/, '').trim();
3920
+ if (semi >= 0)
3921
+ return depth === 0 && text.trimEnd().endsWith(';') ? text : null;
3922
+ }
3923
+ return null;
3924
+ }
3102
3925
  /**
3103
3926
  * cognium-dev #220 — Java derived-taint expansion.
3104
3927
  *
@@ -3163,8 +3986,46 @@ export function buildJavaTaintedVars(sourceCode, seedVars) {
3163
3986
  const trimmed = line.trimStart();
3164
3987
  if (trimmed.startsWith('//') || trimmed.startsWith('*'))
3165
3988
  continue;
3166
- const declMatch = declRe.exec(line);
3167
- const assignMatch = !declMatch ? assignRe.exec(line) : null;
3989
+ // cognium-dev#476: non-assignment statements (must run before the #753
3990
+ // wrap-join early-continue, which would skip them).
3991
+ const arrayCopyRe = /\bSystem\.arraycopy\s*\(\s*([A-Za-z_]\w*)\s*,\s*[^,]+\s*,\s*([A-Za-z_]\w*)\s*,/;
3992
+ const acMatch = arrayCopyRe.exec(line);
3993
+ if (acMatch) {
3994
+ const srcArr = acMatch[1];
3995
+ const dstArr = acMatch[2];
3996
+ if (knownTainted.has(srcArr) && !knownTainted.has(dstArr)) {
3997
+ derived.set(dstArr, i + 1);
3998
+ knownTainted.add(dstArr);
3999
+ changed = true;
4000
+ }
4001
+ continue;
4002
+ }
4003
+ const listAddRe = /^\s*([A-Za-z_]\w*)\.add\s*\(\s*([A-Za-z_]\w*)\s*\)\s*;?\s*$/;
4004
+ const addMatch = listAddRe.exec(line.trimEnd());
4005
+ if (addMatch) {
4006
+ const listVar = addMatch[1];
4007
+ const elem = addMatch[2];
4008
+ if (knownTainted.has(elem) && !knownTainted.has(listVar)) {
4009
+ derived.set(listVar, i + 1);
4010
+ knownTainted.add(listVar);
4011
+ changed = true;
4012
+ }
4013
+ continue;
4014
+ }
4015
+ let statement = line;
4016
+ let declMatch = declRe.exec(statement);
4017
+ let assignMatch = !declMatch ? assignRe.exec(statement) : null;
4018
+ if (!declMatch && !assignMatch) {
4019
+ // cognium-dev#753: `label = String.format("%s.tar",\n picked);` — a
4020
+ // statement whose call argument list wraps over several lines reads like
4021
+ // the same statement on one line.
4022
+ const joined = joinWrappedJavaStatement(lines, i);
4023
+ if (joined === null)
4024
+ continue;
4025
+ statement = joined;
4026
+ declMatch = declRe.exec(statement);
4027
+ assignMatch = !declMatch ? assignRe.exec(statement) : null;
4028
+ }
3168
4029
  const m = declMatch ?? assignMatch;
3169
4030
  if (!m)
3170
4031
  continue;
@@ -4517,6 +5378,196 @@ function findPythonRegexAllowlistWrapperSanitizers(code) {
4517
5378
  }
4518
5379
  return sanitizers;
4519
5380
  }
5381
+ function isPythonQuotedStringLiteral(token) {
5382
+ const t = token.trim();
5383
+ return (/^"(?:[^"\\]|\\.)*"$/.test(t) ||
5384
+ /^'(?:[^'\\]|\\.)*'$/.test(t));
5385
+ }
5386
+ function pythonRhsIsLiteralStringCollection(rhs) {
5387
+ let s = rhs.trim();
5388
+ if (/^frozenset\s*\(/.test(s) && s.endsWith(')')) {
5389
+ s = s.slice(s.indexOf('(') + 1, -1).trim();
5390
+ }
5391
+ const splitElems = (inner) => {
5392
+ const parts = [];
5393
+ let depth = 0;
5394
+ let quote = null;
5395
+ let start = 0;
5396
+ for (let i = 0; i < inner.length; i++) {
5397
+ const c = inner[i];
5398
+ if (quote) {
5399
+ if (c === '\\') {
5400
+ i++;
5401
+ continue;
5402
+ }
5403
+ if (c === quote)
5404
+ quote = null;
5405
+ continue;
5406
+ }
5407
+ if (c === '"' || c === "'") {
5408
+ quote = c;
5409
+ continue;
5410
+ }
5411
+ if (c === '(' || c === '[' || c === '{')
5412
+ depth++;
5413
+ else if (c === ')' || c === ']' || c === '}')
5414
+ depth--;
5415
+ else if (c === ',' && depth === 0) {
5416
+ parts.push(inner.slice(start, i).trim());
5417
+ start = i + 1;
5418
+ }
5419
+ }
5420
+ const last = inner.slice(start).trim();
5421
+ if (last || parts.length > 0)
5422
+ parts.push(last);
5423
+ return parts.filter(Boolean);
5424
+ };
5425
+ if (s.startsWith('{') && s.endsWith('}')) {
5426
+ const inner = s.slice(1, -1).trim();
5427
+ if (!inner)
5428
+ return true;
5429
+ return splitElems(inner).every(isPythonQuotedStringLiteral);
5430
+ }
5431
+ if (s.startsWith('[') && s.endsWith(']')) {
5432
+ const inner = s.slice(1, -1).trim();
5433
+ if (!inner)
5434
+ return false;
5435
+ return splitElems(inner).every(isPythonQuotedStringLiteral);
5436
+ }
5437
+ if (s.startsWith('(') && s.endsWith(')')) {
5438
+ const inner = s.slice(1, -1).trim();
5439
+ if (!inner || !inner.includes(','))
5440
+ return false;
5441
+ return splitElems(inner).every(isPythonQuotedStringLiteral);
5442
+ }
5443
+ return false;
5444
+ }
5445
+ function collectPythonLiteralMembershipCollections(code) {
5446
+ const names = new Set();
5447
+ const lines = code.split('\n');
5448
+ for (let i = 0; i < lines.length; i++) {
5449
+ const raw = lines[i];
5450
+ const trimmed = raw.trim();
5451
+ if (!trimmed || trimmed.startsWith('#'))
5452
+ continue;
5453
+ const indent = raw.length - raw.trimStart().length;
5454
+ if (indent !== 0)
5455
+ continue;
5456
+ const am = /^([A-Za-z_]\w*)\s*=\s*(.+)$/.exec(trimmed);
5457
+ if (!am)
5458
+ continue;
5459
+ const name = am[1];
5460
+ if (!pythonRhsIsLiteralStringCollection(am[2]))
5461
+ continue;
5462
+ if (new RegExp(String.raw `\b${name}\s*(?:\.add\s*\(|\.update\s*\(|\|=)|\b${name}\s*\[[^\]]+\]\s*=`).test(code)) {
5463
+ continue;
5464
+ }
5465
+ names.add(name);
5466
+ }
5467
+ return names;
5468
+ }
5469
+ function pythonEnclosingDefEndLine(lines, fromLine) {
5470
+ let defLine = -1;
5471
+ let defIndent = -1;
5472
+ for (let k = fromLine; k >= 0; k--) {
5473
+ const dm = /^(\s*)def\s+/.exec(lines[k]);
5474
+ if (dm) {
5475
+ defLine = k;
5476
+ defIndent = dm[1].length;
5477
+ break;
5478
+ }
5479
+ }
5480
+ if (defLine < 0)
5481
+ return lines.length;
5482
+ for (let j = defLine + 1; j < lines.length; j++) {
5483
+ const line = lines[j];
5484
+ if (line.trim() === '')
5485
+ continue;
5486
+ const indent = line.length - line.trimStart().length;
5487
+ if (indent <= defIndent && /^\s*(?:def\s+|@|class\s+)/.test(line)) {
5488
+ return j;
5489
+ }
5490
+ }
5491
+ return lines.length;
5492
+ }
5493
+ /**
5494
+ * Python literal set/list/tuple membership guard (cognium-dev #563).
5495
+ *
5496
+ * if theme not in THEMES:
5497
+ * abort(400)
5498
+ * session["theme"] = theme
5499
+ *
5500
+ * Past the guard, `theme` is one of the module-level string literals in
5501
+ * `THEMES`. The collection must be a literal of string constants that the
5502
+ * file never mutates.
5503
+ */
5504
+ function findPythonLiteralMembershipGuardSanitizers(code) {
5505
+ const sanitizers = [];
5506
+ const lines = code.split('\n');
5507
+ const literalCollections = collectPythonLiteralMembershipCollections(code);
5508
+ if (literalCollections.size === 0)
5509
+ return sanitizers;
5510
+ const guardOpen = /^(\s*)if\s+([A-Za-z_][A-Za-z0-9_]*)\s+not\s+in\s+([A-Za-z_][A-Za-z0-9_]*)\s*:\s*$/;
5511
+ const terminator = /\b(return|raise|abort\s*\(|sys\.exit\s*\()/;
5512
+ for (let i = 0; i < lines.length; i++) {
5513
+ const m = guardOpen.exec(lines[i]);
5514
+ if (!m)
5515
+ continue;
5516
+ const guardIndent = m[1].length;
5517
+ const guardedVar = m[2];
5518
+ const collectionName = m[3];
5519
+ if (!literalCollections.has(collectionName))
5520
+ continue;
5521
+ let bodyHasTerminator = false;
5522
+ let blockEnd = -1;
5523
+ const maxScan = Math.min(lines.length, i + 26);
5524
+ for (let j = i + 1; j < maxScan; j++) {
5525
+ const line = lines[j];
5526
+ if (line.trim() === '')
5527
+ continue;
5528
+ const indent = line.length - line.trimStart().length;
5529
+ if (indent <= guardIndent) {
5530
+ blockEnd = j - 1;
5531
+ break;
5532
+ }
5533
+ if (terminator.test(line))
5534
+ bodyHasTerminator = true;
5535
+ }
5536
+ if (blockEnd === -1)
5537
+ blockEnd = Math.min(lines.length - 1, i + 25);
5538
+ if (!bodyHasTerminator)
5539
+ continue;
5540
+ const varRe = new RegExp(String.raw `\b${guardedVar}\b`);
5541
+ const reassign = new RegExp(String.raw `^\s*${guardedVar}\s*=(?!=)`);
5542
+ const funcEnd = pythonEnclosingDefEndLine(lines, i);
5543
+ for (let l = blockEnd + 2; l <= funcEnd; l++) {
5544
+ const lineText = lines[l - 1];
5545
+ if (!lineText || !varRe.test(lineText))
5546
+ continue;
5547
+ if (reassign.test(lineText))
5548
+ break;
5549
+ sanitizers.push({
5550
+ type: 'python_literal_membership_guard',
5551
+ method: 'if',
5552
+ line: l,
5553
+ sanitizes: [
5554
+ 'trust_boundary',
5555
+ 'sql_injection',
5556
+ 'command_injection',
5557
+ 'path_traversal',
5558
+ 'ssrf',
5559
+ 'ldap_injection',
5560
+ 'xpath_injection',
5561
+ 'log_injection',
5562
+ 'xss',
5563
+ 'open_redirect',
5564
+ 'external_taint_escape',
5565
+ ],
5566
+ });
5567
+ }
5568
+ }
5569
+ return sanitizers;
5570
+ }
4520
5571
  /**
4521
5572
  * Detect Python set-membership allow-list guards that should sanitize
4522
5573
  * `xss` flows (cognium-dev #216 Sprint 74 Pattern B, SSTI fixture).
@@ -6395,10 +7446,14 @@ function findCSharpEarlyReturnGuardSanitizers(code, language) {
6395
7446
  const lines = code.split('\n');
6396
7447
  const terminatorRe = /\b(throw|return)\b/;
6397
7448
  // Allow-list membership, rejecting form:
6398
- // if (Array.IndexOf(Allowed, x) < 0) return;
7449
+ // if (Array.IndexOf(Allowed, x) < 0) return; (also System.Array, == -1)
6399
7450
  // if (!Allowed.Contains(x)) return;
7451
+ // cognium-dev#809: the IndexOf form credits only a `static readonly`
7452
+ // string-literal array field the file never writes (see
7453
+ // `collectCSharpLiteralArrayFields`); any other array may hold the value.
7454
+ const literalArrays = collectCSharpLiteralArrayFields(code);
7455
+ const indexOfRe = /\bif\s*\(\s*(?:System\s*\.\s*)?Array\s*\.\s*IndexOf\s*\(\s*([A-Za-z_]\w*)\s*,\s*([A-Za-z_][\w.]*)\s*\)\s*(?:<\s*0|==\s*-\s*1)\s*\)/;
6400
7456
  const allowlistRes = [
6401
- /\bif\s*\(\s*Array\s*\.\s*IndexOf\s*\([^,]+,\s*([A-Za-z_][\w.]*)\s*\)\s*<\s*0\s*\)/,
6402
7457
  /\bif\s*\(\s*!\s*[A-Za-z_][\w.]*\s*\.\s*Contains\s*\(\s*([A-Za-z_][\w.]*)\s*\)\s*\)/,
6403
7458
  ];
6404
7459
  // Character-class validation, rejecting form:
@@ -6406,13 +7461,17 @@ function findCSharpEarlyReturnGuardSanitizers(code, language) {
6406
7461
  const charClassRe = /\bif\s*\(\s*!\s*([A-Za-z_][\w.]*)\s*\.\s*All\s*\(\s*char\s*\.\s*Is[A-Za-z]+\s*\)\s*\)/;
6407
7462
  for (let i = 0; i < lines.length; i++) {
6408
7463
  let guardedExpr = null;
6409
- for (const re of allowlistRes) {
6410
- const m = re.exec(lines[i]);
6411
- if (m) {
6412
- guardedExpr = m[1];
6413
- break;
7464
+ const im = indexOfRe.exec(lines[i]);
7465
+ if (im && literalArrays.has(im[1]))
7466
+ guardedExpr = im[2];
7467
+ if (!guardedExpr)
7468
+ for (const re of allowlistRes) {
7469
+ const m = re.exec(lines[i]);
7470
+ if (m) {
7471
+ guardedExpr = m[1];
7472
+ break;
7473
+ }
6414
7474
  }
6415
- }
6416
7475
  if (!guardedExpr) {
6417
7476
  const m = charClassRe.exec(lines[i]);
6418
7477
  if (m)
@@ -6456,6 +7515,10 @@ function findCSharpEarlyReturnGuardSanitizers(code, language) {
6456
7515
  const a = assignRe.exec(lines[l]);
6457
7516
  if (a && mentions(a[2]))
6458
7517
  guarded.add(a[1]);
7518
+ // cognium-dev#809: a reassignment from anything unguarded ends the
7519
+ // credit (`target = Console.ReadLine();` after the guard).
7520
+ else if (a)
7521
+ guarded.delete(a[1]);
6459
7522
  if (mentions(lines[l])) {
6460
7523
  sanitizers.push({
6461
7524
  type: 'csharp_early_return_guard',
@@ -6469,6 +7532,44 @@ function findCSharpEarlyReturnGuardSanitizers(code, language) {
6469
7532
  }
6470
7533
  return sanitizers;
6471
7534
  }
7535
+ /**
7536
+ * cognium-dev#809 — names of `static readonly string[]` fields initialised
7537
+ * with string literals only, that the file never writes:
7538
+ *
7539
+ * private static readonly string[] KnownHosts = { "a.example", "b.example" };
7540
+ *
7541
+ * Past `if (Array.IndexOf(KnownHosts, x) < 0) throw …;`, `x` is one of those
7542
+ * literals. The C# counterpart of the Python literal-collection membership
7543
+ * guard (#563) and the Rust literal-equality guard (#293).
7544
+ *
7545
+ * `readonly` stops the field being rebound outside an initialiser, but array
7546
+ * elements stay writable, so any element store, `ref`/`out` pass,
7547
+ * `SetValue`, `Array.<mutator>(…)` naming the field, or plain assignment
7548
+ * after the declaration disqualifies it.
7549
+ */
7550
+ function collectCSharpLiteralArrayFields(code) {
7551
+ const names = new Set();
7552
+ const lit = String.raw `(?:@"(?:[^"]|"")*"|"(?:[^"\\\n]|\\.)*")`;
7553
+ const elems = new RegExp(String.raw `^\s*${lit}(?:\s*,\s*${lit})*\s*,?\s*$`);
7554
+ const declRe = /((?:\b(?:public|private|protected|internal|static|readonly)\s+)+)string\s*\[\s*\]\s+([A-Za-z_]\w*)\s*=\s*(?:new\s+(?:string\s*)?\[\s*\]\s*)?\{([^{}]*)\}\s*;/g;
7555
+ for (const m of code.matchAll(declRe)) {
7556
+ const mods = m[1];
7557
+ const name = m[2];
7558
+ if (!/\bstatic\b/.test(mods) || !/\breadonly\b/.test(mods))
7559
+ continue;
7560
+ if (!elems.test(m[3]))
7561
+ continue;
7562
+ const rest = code.slice(0, m.index) + code.slice(m.index + m[0].length);
7563
+ const written = new RegExp(String.raw `\b${name}\s*(?:\[[^\]]*\]\s*)?(?:[-+*/%&|^]|\?\?)?=(?!=)` +
7564
+ String.raw `|\b(?:ref|out)\s+${name}\b` +
7565
+ String.raw `|\b${name}\s*\.\s*SetValue\s*\(` +
7566
+ String.raw `|\bArray\s*\.\s*(?:Clear|Copy|ConstrainedCopy|Fill|Resize|Reverse|Sort)\s*\([^;]*\b${name}\b`);
7567
+ if (written.test(rest))
7568
+ continue;
7569
+ names.add(name);
7570
+ }
7571
+ return names;
7572
+ }
6472
7573
  function findCSharpFullPathStartsWithGuardSanitizers(code) {
6473
7574
  const sanitizers = [];
6474
7575
  const lines = code.split('\n');
@@ -11226,8 +12327,8 @@ function javaParensBalanced(expr) {
11226
12327
  * - the method carries a `@*Mapping` annotation;
11227
12328
  * - it returns `String` under a body-writing controller, or
11228
12329
  * `ResponseEntity<String>` via `ResponseEntity.ok(x)` /
11229
- * `new ResponseEntity<>(x, …)` (the builder `.body(x)` form is left to
11230
- * SinkFilterPass Stage 15c's text/html content-type rule);
12330
+ * `new ResponseEntity<>(x, …)`, or `ResponseEntity.ok().body(x)` /
12331
+ * `ResponseEntity.status(...).body(x)` on a body-writing controller);
11231
12332
  * - the returned expression references a request-bound parameter
11232
12333
  * (`@RequestParam`, `@PathVariable`, `@RequestHeader`, `@CookieValue`,
11233
12334
  * `@MatrixVariable`, a `String` `@RequestBody`) or an in-method source,
@@ -11312,14 +12413,20 @@ export function findJavaSpringReturnXSSSinks(code, types, sources, xssSanitizerL
11312
12413
  const sinkLine = i + 1;
11313
12414
  let expr = stmt;
11314
12415
  if (entity) {
11315
- // `ResponseEntity.ok(x)` / `new ResponseEntity<>(x, status)` only. The
11316
- // builder `.body(x)` form stays with SinkFilterPass Stage 15c, which
11317
- // keeps it only next to an explicit text/html signal.
11318
- const em = /^\s*ResponseEntity\s*\.\s*ok\s*\(([\s\S]+)\)\s*$/.exec(stmt)
12416
+ // `ResponseEntity.ok(x)` / `new ResponseEntity<>(x, status)` /
12417
+ // `ResponseEntity.ok().body(x)` / `ResponseEntity.status(...).body(x)`.
12418
+ const em = /^\s*ResponseEntity\s*\.\s*ok\s*\(\s*\)\s*\.\s*body\s*\(([\s\S]+)\)\s*$/.exec(stmt)
12419
+ ?? /^\s*ResponseEntity\s*\.\s*status\s*\(([\s\S]+)\)\s*\.\s*body\s*\(([\s\S]+)\)\s*$/.exec(stmt)
12420
+ ?? /^\s*ResponseEntity\s*\.\s*ok\s*\(([\s\S]+)\)\s*$/.exec(stmt)
11319
12421
  ?? /^\s*new\s+ResponseEntity\s*(?:<[^>]*>)?\s*\(([\s\S]+),[^,()]*\)\s*$/.exec(stmt);
11320
- if (!em || !javaParensBalanced(em[1]))
12422
+ if (!em)
12423
+ continue;
12424
+ const payload = em.length === 3 ? em[2] : em[1];
12425
+ if (!payload || !javaParensBalanced(payload))
12426
+ continue;
12427
+ if (em.length === 3 && !javaParensBalanced(em[1]))
11321
12428
  continue;
11322
- expr = em[1];
12429
+ expr = payload;
11323
12430
  }
11324
12431
  const hit = [...carriers].find(([c, info]) => info.assignLine <= sinkLine && refers(expr, c));
11325
12432
  if (!hit)