fix(coverage): stop reporting a duplicate range and a line past EOF (#963)

scripts/setup-windows.ps1 has 326 lines and its parse-coverage report read
"113-113,113-113,245-327" — the same line named twice, and an end line that
does not exist. Two separate faults, both in cbm_error_regions_push.

Past-EOF end line. A tree-sitter node that ends at column 0 stopped right
after the previous line's newline, so it holds no text on the row it points
at. Adding 1 to that row named a line past the end of the file whenever the
region ran to EOF. The node here is start=(244,2) end=(326,0). Clamp the end
to the row above when the end column is 0 and the node spans more than one
row.

Duplicate range. Line 113 carries two separate ERROR nodes, at columns 25-29
and 31-32, and each pushed its own range. A line range says nothing new the
second time. Drop a range that exactly repeats the one already open. The
drop runs BEFORE the cap check, so a repeat is never miscounted as a range
the cap threw away.

Only an EXACT repeat is dropped, never a range that merely overlaps. Each
range is judged separately afterwards by cbm_region_is_recovered, which asks
whether definitions starting inside that range cover it. Two ranges holding
the same numbers always get the same verdict, so dropping the repeat changes
nothing. Two different ranges do not. Merging 3-3 into 2-3 hands the wider
range's covering definition to an error that definition does not explain, and
a real parse failure then vanishes from the report.

That is not hypothetical. An earlier version of this commit merged on overlap
and broke perl_malformed_source_remains_partial_issue1838, the test added
with the Perl grammar refresh in 17b5a43. The malformed fixture produces two
ERROR nodes, at lines 2-3 and 3-3. Merged, the 2-3 range looks fully covered
by before_error and is removed, so parse_incomplete comes back false on a file
that plainly does not parse. That test now pins this boundary.

The real file reports "113-113,245-326".

Two tests, both proved RED first with the exact expected text:
  coverage_repeated_error_line_reports_one_range_issue963  "2-2,2-2" != "2-2"
  coverage_range_never_ends_past_the_last_line_issue963    "1-5"    != "1-4"

Suites run on this change: parse_coverage 34, extraction 325, pipeline 264,
mcp 246, index_resilience 7 — all passing.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Joshua Richter <jrichter5781@gmail.com>

Author: CaptainMittens

internal/cbm/cbm.c

@@ -790,12 +790,47 @@ typedef struct {
 } cbm_error_regions_t;
 
 static void cbm_error_regions_push(cbm_error_regions_t *acc, TSNode n) {
+    TSPoint start = ts_node_start_point(n);
+    TSPoint end = ts_node_end_point(n);
+    uint32_t start_line = start.row + 1;
+    uint32_t end_line = end.row + 1;
+
+    /* A node that ends at column 0 stopped right after the previous line's
+     * newline, so it holds no text on the row it points at. Counting that row
+     * named a line past the end of the file whenever the region ran to EOF:
+     * scripts/setup-windows.ps1 has 326 lines and reported "245-327". */
+    if (end.column == 0 && end.row > start.row) {
+        end_line = end.row;
+    }
+
+    /* One line can carry several error nodes, and repeating the same line range
+     * says nothing new. Line 113 of scripts/setup-windows.ps1 has two error
+     * nodes, at columns 25-29 and 31-32, and the report read "113-113,113-113".
+     * Drop the repeat.
+     *
+     * Only an EXACT repeat of the range already open is dropped. Do not merge
+     * ranges that merely overlap. Each range is judged separately later by
+     * cbm_region_is_recovered, which asks whether definitions starting inside
+     * that range cover it. Two ranges with the same numbers always get the same
+     * verdict, so collapsing them changes nothing. Two DIFFERENT ranges do not:
+     * merging 3-3 into 2-3 hands the wider range's covering definition to an
+     * error the definition does not explain, and a real parse failure then
+     * disappears from the report. tests/test_parse_coverage.c pins that case in
+     * perl_malformed_source_remains_partial_issue1838.
+     *
+     * This runs BEFORE the cap check, so a dropped repeat never counts as a
+     * range the cap threw away. */
+    if (acc->count > 0 && start_line == acc->starts[acc->count - 1] &&
+        end_line == acc->ends[acc->count - 1]) {
+        return;
+    }
+
     if (acc->count >= CBM_MAX_ERROR_REGIONS) {
         acc->dropped++;
         return;
     }
-    acc->starts[acc->count] = ts_node_start_point(n).row + 1;
-    acc->ends[acc->count] = ts_node_end_point(n).row + 1;
+    acc->starts[acc->count] = start_line;
+    acc->ends[acc->count] = end_line;
     acc->count++;
 }
 

tests/test_parse_coverage.c

@@ -763,6 +763,51 @@ TEST(c_thread_local_grammar_limit_is_pinned_issue963) {
     PASS();
 }
 
+/* Two error nodes can sit on ONE line. Line 113 of scripts/setup-windows.ps1
+ * does exactly that, and the report used to read "113-113,113-113" — the same
+ * line named twice. A line range says nothing new the second time, so repeated
+ * or overlapping regions must collapse into one. */
+static const char *PS_TWO_ERRORS_ONE_LINE = "Write-Host \"start\"\n"             /* 1 */
+                                            "wsl.exe -- bash -c $Command 2>&1\n" /* 2 */
+                                            "Write-Host \"end\"\n";              /* 3 */
+
+/* An error region that runs to the end of the file stops just after the last
+ * newline. Tree-sitter calls that position row N, column 0 — a row that holds
+ * no text. Reading it as a line number named a line past the end of the file:
+ * scripts/setup-windows.ps1 has 326 lines and the report said "245-327". */
+static const char *PS_ERROR_TO_EOF = "} else {\n"             /* 1 */
+                                     "    if ($a) {\n"        /* 2 */
+                                     "        Write-Host x\n" /* 3 */
+                                     "}\n";                   /* 4 */
+
+TEST(coverage_repeated_error_line_reports_one_range_issue963) {
+    CBMFileResult *r =
+        cbm_extract_file(PS_TWO_ERRORS_ONE_LINE, (int)strlen(PS_TWO_ERRORS_ONE_LINE),
+                         CBM_LANG_POWERSHELL, "covproj", "two_errors.ps1", 0, NULL, NULL);
+    ASSERT_NOT_NULL(r);
+    ASSERT_TRUE(r->parse_incomplete);
+    ASSERT_NOT_NULL(r->error_ranges);
+    /* Line 2 carries two separate error nodes. It must be named once. */
+    ASSERT_STR_EQ(r->error_ranges, "2-2");
+    ASSERT_EQ(r->error_region_count, 1);
+    cbm_free_result(r);
+    PASS();
+}
+
+TEST(coverage_range_never_ends_past_the_last_line_issue963) {
+    int len = (int)strlen(PS_ERROR_TO_EOF);
+    CBMFileResult *r = cbm_extract_file(PS_ERROR_TO_EOF, len, CBM_LANG_POWERSHELL, "covproj",
+                                        "error_to_eof.ps1", 0, NULL, NULL);
+    ASSERT_NOT_NULL(r);
+    ASSERT_TRUE(r->parse_incomplete);
+    ASSERT_NOT_NULL(r->error_ranges);
+    /* The file has four lines and ends with a newline. Line 5 does not exist. */
+    ASSERT_STR_EQ(r->error_ranges, "1-4");
+    ASSERT_NULL(strstr(r->error_ranges, "5"));
+    cbm_free_result(r);
+    PASS();
+}
+
 SUITE(parse_coverage) {
     RUN_TEST(c_ifdef_split_brace_sets_parse_incomplete);
     RUN_TEST(c_ifdef_split_brace_neighbors_still_extracted);
@@ -796,4 +841,6 @@ SUITE(parse_coverage) {
     RUN_TEST(perl_format_followed_by_named_sub_is_complete_issue1838);
     RUN_TEST(perl_malformed_source_remains_partial_issue1838);
     RUN_TEST(c_thread_local_grammar_limit_is_pinned_issue963);
+    RUN_TEST(coverage_repeated_error_line_reports_one_range_issue963);
+    RUN_TEST(coverage_range_never_ends_past_the_last_line_issue963);
 }