DarwinPatch Report

Baseline	solve@budget	95% CI	Avg attempts	Regressions blocked	Invalid rejected	Evidence packets	Unsolved cost	Route metadata matches
archive_no_routing	0.6	0.446-0.7365	1.8	12	15	0	32	0
clean_context_review	0.6	0.446-0.7365	1.8	12	15	0	32	0
evidence_aware_review	0.9	0.7695-0.9604	1.8	12	9	36	8	28
full_darwinpatch	0.9	0.7695-0.9604	1.8	12	9	36	8	28
linear_retry	0.6	0.446-0.7365	1.8	12	15	0	32	0
single_shot	0.2	0.105-0.3476	1.0	12	8	0	32	0

1. candidate_001 — promoted

Route: initial_repair
Selected by: initial_order
Intent: complete_spec_repair
Parent: none
Gen: 0
Failure: none
Fingerprint: none
Score: 1.0

scope_guard: PASS
patch_applies: PASS
python_ast: PASS
secret_scan: PASS
visible_tests: PASS
release_gate_regressions: PASS

No failure evidence; candidate promoted.

Patch diff: /Users/taneemishere/Developer/darwin-patch/darwinpatch/tasks/smoke_markdown_parser/candidate.patch

diff --git a/parser.py b/parser.py
index e6f2d6f..7db8a9f 100644
--- a/parser.py
+++ b/parser.py
@@ -1,13 +1,44 @@
+def _split_row(line):
+    cells = []
+    current = []
+    escaped = False
+    inner = line.strip()
+    if inner.startswith("|"):
+        inner = inner[1:]
+    if inner.endswith("|"):
+        inner = inner[:-1]
+
+    for char in inner:
+        if escaped:
+            current.append(char)
+            escaped = False
+            continue
+        if char == "\\":
+            escaped = True
+            continue
+        if char == "|":
+            cells.append("".join(current).strip())
+            current = []
+            continue
+        current.append(char)
+
+    if escaped:
+        current.append("\\")
+    cells.append("".join(current).strip())
+    return cells
+
+
 def parse_table(markdown):
     """Parse a simple Markdown table into row dictionaries."""
-    lines = [line.strip() for line in markdown.splitlines() if line.strip()]
-    table_lines = [line for line in lines if line.startswith("|")]
+    table_lines = [line.strip() for line in markdown.splitlines() if line.strip().startswith("|")]
     if len(table_lines) < 2:
         return []
 
-    headers = [cell.strip() for cell in table_lines[0].strip("|").split("|")]
+    headers = _split_row(table_lines[0])
     rows = []
     for line in table_lines[2:]:
-        values = [cell.strip() for cell in line.strip("|").split("|")]
+        values = _split_row(line)
+        if len(values) != len(headers):
+            raise ValueError("row has a different number of cells than the header")
         rows.append(dict(zip(headers, values)))
     return rows

1. candidate_001 — promoted

Route: initial_repair
Selected by: initial_order
Intent: complete_spec_repair
Parent: none
Gen: 0
Failure: none
Fingerprint: none
Score: 1.0

scope_guard: PASS
patch_applies: PASS
python_ast: PASS
secret_scan: PASS
visible_tests: PASS
release_gate_regressions: PASS

No failure evidence; candidate promoted.

Patch diff: /Users/taneemishere/Developer/darwin-patch/darwinpatch/tasks/smoke_markdown_parser/candidate.patch

diff --git a/parser.py b/parser.py
index e6f2d6f..7db8a9f 100644
--- a/parser.py
+++ b/parser.py
@@ -1,13 +1,44 @@
+def _split_row(line):
+    cells = []
+    current = []
+    escaped = False
+    inner = line.strip()
+    if inner.startswith("|"):
+        inner = inner[1:]
+    if inner.endswith("|"):
+        inner = inner[:-1]
+
+    for char in inner:
+        if escaped:
+            current.append(char)
+            escaped = False
+            continue
+        if char == "\\":
+            escaped = True
+            continue
+        if char == "|":
+            cells.append("".join(current).strip())
+            current = []
+            continue
+        current.append(char)
+
+    if escaped:
+        current.append("\\")
+    cells.append("".join(current).strip())
+    return cells
+
+
 def parse_table(markdown):
     """Parse a simple Markdown table into row dictionaries."""
-    lines = [line.strip() for line in markdown.splitlines() if line.strip()]
-    table_lines = [line for line in lines if line.startswith("|")]
+    table_lines = [line.strip() for line in markdown.splitlines() if line.strip().startswith("|")]
     if len(table_lines) < 2:
         return []
 
-    headers = [cell.strip() for cell in table_lines[0].strip("|").split("|")]
+    headers = _split_row(table_lines[0])
     rows = []
     for line in table_lines[2:]:
-        values = [cell.strip() for cell in line.strip("|").split("|")]
+        values = _split_row(line)
+        if len(values) != len(headers):
+            raise ValueError("row has a different number of cells than the header")
         rows.append(dict(zip(headers, values)))
     return rows

1. candidate_001 — rejected

Route: initial_repair
Selected by: initial_order
Intent: first_pass_attempt_for_gate_coverage
Parent: none
Gen: 0
Failure: visible_tests
Fingerprint: visible_tests:tests/visible/test_parser_visible.py::test_escaped_pipe_stays_inside_cell
Score: 0.4

scope_guard: PASS
patch_applies: PASS
python_ast: PASS
secret_scan: PASS
visible_tests: FAIL

visible developer tests failed

Evidence ID: ev_3355097a5d9f

{
  "returncode": 1,
  "stderr_tail": "",
  "stdout_tail": ".F                                                                       [100%]\n=================================== FAILURES ===================================\n_____________________ test_escaped_pipe_stays_inside_cell ______________________\n\n    def test_escaped_pipe_stays_inside_cell():\n        markdown = r\"\"\"\n        | name | value |\n        | --- | --- |\n        | alpha | one \\| two |\n        \"\"\"\n    \n>       assert parse_table(markdown) == [{\"name\": \"alpha\", \"value\": \"one | two\"}]\nE       AssertionError: assert [{'name': 'al...e': 'one \\\\'}] == [{'name': 'al... 'one | two'}]\nE         \nE         At index 0 diff: {'name': 'alpha', 'value': 'one \\\\'} != {'name': 'alpha', 'value': 'one | two'}\nE         Use -v to get more diff\n\ntests/visible/test_parser_visible.py:22: AssertionError\n=========================== short test summary info ============================\nFAILED tests/visible/test_parser_visible.py::test_escaped_pipe_stays_inside_cell\n1 failed, 1 passed in 0.02s\n"
}

Patch diff: /Users/taneemishere/Developer/darwin-patch/darwinpatch/tasks/smoke_markdown_parser/candidate_visible_fail.patch

diff --git a/parser.py b/parser.py
index e6f2d6f..96d3ee1 100644
--- a/parser.py
+++ b/parser.py
@@ -1,13 +1,13 @@
 def parse_table(markdown):
     """Parse a simple Markdown table into row dictionaries."""
-    lines = [line.strip() for line in markdown.splitlines() if line.strip()]
-    table_lines = [line for line in lines if line.startswith("|")]
+    table_lines = [line.strip() for line in markdown.splitlines() if line.strip().startswith("|")]
     if len(table_lines) < 2:
         return []
 
     headers = [cell.strip() for cell in table_lines[0].strip("|").split("|")]
     rows = []
     for line in table_lines[2:]:
         values = [cell.strip() for cell in line.strip("|").split("|")]
+        values = values[: len(headers)]
         rows.append(dict(zip(headers, values)))
     return rows

2. candidate_002 — promoted

Route: behavior_repair
Selected by: metadata_route_match:behavior_repair
Intent: complete_spec_repair
Parent: candidate_001
Gen: 1
Failure: none
Fingerprint: none
Score: 1.0

scope_guard: PASS
patch_applies: PASS
python_ast: PASS
secret_scan: PASS
visible_tests: PASS
release_gate_regressions: PASS

No failure evidence; candidate promoted.

Patch diff: /Users/taneemishere/Developer/darwin-patch/darwinpatch/tasks/smoke_markdown_parser/candidate.patch

diff --git a/parser.py b/parser.py
index e6f2d6f..7db8a9f 100644
--- a/parser.py
+++ b/parser.py
@@ -1,13 +1,44 @@
+def _split_row(line):
+    cells = []
+    current = []
+    escaped = False
+    inner = line.strip()
+    if inner.startswith("|"):
+        inner = inner[1:]
+    if inner.endswith("|"):
+        inner = inner[:-1]
+
+    for char in inner:
+        if escaped:
+            current.append(char)
+            escaped = False
+            continue
+        if char == "\\":
+            escaped = True
+            continue
+        if char == "|":
+            cells.append("".join(current).strip())
+            current = []
+            continue
+        current.append(char)
+
+    if escaped:
+        current.append("\\")
+    cells.append("".join(current).strip())
+    return cells
+
+
 def parse_table(markdown):
     """Parse a simple Markdown table into row dictionaries."""
-    lines = [line.strip() for line in markdown.splitlines() if line.strip()]
-    table_lines = [line for line in lines if line.startswith("|")]
+    table_lines = [line.strip() for line in markdown.splitlines() if line.strip().startswith("|")]
     if len(table_lines) < 2:
         return []
 
-    headers = [cell.strip() for cell in table_lines[0].strip("|").split("|")]
+    headers = _split_row(table_lines[0])
     rows = []
     for line in table_lines[2:]:
-        values = [cell.strip() for cell in line.strip("|").split("|")]
+        values = _split_row(line)
+        if len(values) != len(headers):
+            raise ValueError("row has a different number of cells than the header")
         rows.append(dict(zip(headers, values)))
     return rows

DarwinPatch Report

Benchmark summary: phase3_smoke_suite

Claim-to-artifact links

parser_single_success_01 / full_darwinpatch

1. candidate_001 — promoted

parser_single_success_02 / full_darwinpatch

1. candidate_001 — promoted

parser_visible_then_success_01 / full_darwinpatch

1. candidate_001 — rejected

2. candidate_002 — promoted