Commit 8: Add location tests exposing term-location off-by-one (pre-fix)

Add location test cases to test.cpp covering various C++ string literal forms:
- r_plain:     "a+b"          — plain, offset 1 (already correct)
- r_raw:       R"(a+b)"       — raw, offset 3 (R"( = 3); currently uses 1 (WRONG)
- r_raw2:      R"(\s+$)"      — raw with metacharacters; currently wrong
- r_raw3:      R"(\(([,\w]+)+\)$)" — complex raw; currently wrong
- r_raw4:      R"x(a+b)x"    — custom-delimiter raw, offset 4; currently uses 1 (WRONG)
- r_wide:      L"a+b"         — L" prefix, offset 2; currently uses 1 (WRONG)
- r_wide_raw:  LR"(a+b)"      — combined LR"( prefix, offset 4; currently uses 1 (WRONG)
- r_esc1:      "\\s+"         — escape-containing plain; offset 1 (correct)
- r_esc2:      "a\\.b"        — escaped dot; offset 1 (correct)

Added std::wregex typedef to stubs (for L"..." and LR"(...)" wide-char literals).

New locations.ql query reports per-term: litStartCol, valueStart, valueEnd,
termStartCol, termEndCol — restricted to test_locations() function.

Generated locations.expected via codeql test run --learn (CodeQL 2.26.1).
All 3 tests pass. The expected output captures the CURRENT (pre-fix) columns:
raw/prefixed rows show termStartCol = litStartCol + 1 (wrong);
plain rows show litStartCol + 1 (correct).
Commit 9 changes the raw/prefixed rows to their correct values.

Author: Copilot

cpp/ql/test/library-tests/regex/locations.expected

@@ -0,0 +1,40 @@
+| test.cpp:144:23:144:23 | a | 22 | 0 | 1 | 23 | 23 |
+| test.cpp:144:23:144:24 | a+ | 22 | 0 | 2 | 23 | 24 |
+| test.cpp:144:23:144:25 | a+b | 22 | 0 | 3 | 23 | 25 |
+| test.cpp:144:25:144:25 | b | 22 | 2 | 3 | 25 | 25 |
+| test.cpp:147:21:147:21 | a | 20 | 0 | 1 | 21 | 21 |
+| test.cpp:147:21:147:22 | a+ | 20 | 0 | 2 | 21 | 22 |
+| test.cpp:147:21:147:23 | a+b | 20 | 0 | 3 | 21 | 23 |
+| test.cpp:147:23:147:23 | b | 20 | 2 | 3 | 23 | 23 |
+| test.cpp:150:22:150:23 | \\s | 21 | 0 | 2 | 22 | 23 |
+| test.cpp:150:22:150:24 | \\s+ | 21 | 0 | 3 | 22 | 24 |
+| test.cpp:150:22:150:25 | \\s+$ | 21 | 0 | 4 | 22 | 25 |
+| test.cpp:150:25:150:25 | $ | 21 | 3 | 4 | 25 | 25 |
+| test.cpp:153:22:153:23 | \\( | 21 | 0 | 2 | 22 | 23 |
+| test.cpp:153:22:153:35 | \\(([,\\w]+)+\\)$ | 21 | 0 | 14 | 22 | 35 |
+| test.cpp:153:24:153:31 | ([,\\w]+) | 21 | 2 | 10 | 24 | 31 |
+| test.cpp:153:24:153:32 | ([,\\w]+)+ | 21 | 2 | 11 | 24 | 32 |
+| test.cpp:153:25:153:29 | [,\\w] | 21 | 3 | 8 | 25 | 29 |
+| test.cpp:153:25:153:30 | [,\\w]+ | 21 | 3 | 9 | 25 | 30 |
+| test.cpp:153:26:153:26 | , | 21 | 4 | 5 | 26 | 26 |
+| test.cpp:153:27:153:28 | \\w | 21 | 5 | 7 | 27 | 28 |
+| test.cpp:153:33:153:34 | \\) | 21 | 11 | 13 | 33 | 34 |
+| test.cpp:153:35:153:35 | $ | 21 | 13 | 14 | 35 | 35 |
+| test.cpp:156:22:156:22 | a | 21 | 0 | 1 | 22 | 22 |
+| test.cpp:156:22:156:23 | a+ | 21 | 0 | 2 | 22 | 23 |
+| test.cpp:156:22:156:24 | a+b | 21 | 0 | 3 | 22 | 24 |
+| test.cpp:156:24:156:24 | b | 21 | 2 | 3 | 24 | 24 |
+| test.cpp:159:23:159:23 | a | 22 | 0 | 1 | 23 | 23 |
+| test.cpp:159:23:159:24 | a+ | 22 | 0 | 2 | 23 | 24 |
+| test.cpp:159:23:159:25 | a+b | 22 | 0 | 3 | 23 | 25 |
+| test.cpp:159:25:159:25 | b | 22 | 2 | 3 | 25 | 25 |
+| test.cpp:162:27:162:27 | a | 26 | 0 | 1 | 27 | 27 |
+| test.cpp:162:27:162:28 | a+ | 26 | 0 | 2 | 27 | 28 |
+| test.cpp:162:27:162:29 | a+b | 26 | 0 | 3 | 27 | 29 |
+| test.cpp:162:29:162:29 | b | 26 | 2 | 3 | 29 | 29 |
+| test.cpp:166:22:166:23 | \\s | 21 | 0 | 2 | 22 | 23 |
+| test.cpp:166:22:166:24 | \\s+ | 21 | 0 | 3 | 22 | 24 |
+| test.cpp:169:22:169:22 | a | 21 | 0 | 1 | 22 | 22 |
+| test.cpp:169:22:169:25 | a\\.b | 21 | 0 | 4 | 22 | 25 |
+| test.cpp:169:23:169:24 | \\. | 21 | 1 | 3 | 23 | 24 |
+| test.cpp:169:25:169:25 | b | 21 | 3 | 4 | 25 | 25 |

cpp/ql/test/library-tests/regex/locations.ql

@@ -0,0 +1,29 @@
+/**
+ * Reports per-term location info for each regex term, showing the literal's
+ * start column, the term's value offsets, and the computed start/end columns.
+ * Used to verify that `hasLocationInfo` produces correct columns for plain,
+ * raw, and encoding-prefixed C++ string literals.
+ *
+ * Commit 8 captures pre-fix (wrong) columns for raw/prefixed literals.
+ * Commit 9 fixes them; plain "..." rows must remain unchanged.
+ */
+
+import cpp
+import semmle.code.cpp.regex.RegexTreeView as RE
+
+query predicate locations(
+  RE::RegExpTerm t,
+  int litStartCol,
+  int valueStart,
+  int valueEnd,
+  int termStartCol,
+  int termEndCol
+) {
+  // Only report terms from the location test function (test_locations)
+  // to keep the output focused.
+  t.getRegExp().getLocation().getStartLine() >= 142 and // test_locations starts here
+  t.getRegExp().getLocation().hasLocationInfo(_, _, litStartCol, _, _) and
+  valueStart = t.getStart() and
+  valueEnd = t.getEnd() and
+  t.hasLocationInfo(_, _, termStartCol, _, termEndCol)
+}