`432985d`

Reject placeholder HTML writes

Authored by mfwolffe <wolffemf@dukes.jmu.edu> 1 week ago

SHA: 432985dd455836297f8b8ed7d43f17dc1510cc78
Parents: 75beb49
Tree: ec386d1

4 changed files

Status	File	+	-
M	`src/loader/runtime/repair.py`	7	3
M	`src/loader/runtime/safeguard_services.py`	117	0
M	`tests/test_repair.py`	7	0
M	`tests/test_safeguard_services.py`	64	0

src/loader/runtime/repair.pymodified

                  '<!DOCTYPE html><html lang="en"><head><meta charset="UTF-8">'
                  '<meta name="viewport" content="width=device-width, initial-scale=1.0">'
                  f"<title>{label}</title></head><body><div class=\"container\"><h1>{label}</h1>"
 -                "<p>Starter overview for this guide.</p><nav><ul>"
 +                f"<p>{label} introduces the guide topic, the practical workflow, and the "
 +                "chapter sequence readers can follow from setup through verification.</p><nav><ul>"
                  '<li><a href="chapters/01-introduction.html">Chapter 1: Introduction</a></li>'
                  '<li><a href="chapters/02-installation.html">Chapter 2: Installation</a></li>'
                  "</ul></nav></div></body></html>"
              '<!DOCTYPE html><html lang="en"><head><meta charset="UTF-8">'
              '<meta name="viewport" content="width=device-width, initial-scale=1.0">'
              f"<title>{label}</title></head><body><div class=\"container\"><h1>{label}</h1>"
 -            "<p>Starter content for this chapter.</p><h2>Overview</h2><p>Key concepts go here.</p>"
 -            "<h2>Key Steps</h2><p>Practical steps go here.</p>"
 +            f"<p>{label} frames the chapter topic and connects it to the guide workflow.</p>"
 +            "<h2>Core Concepts</h2><p>This section introduces the essential ideas, "
 +            "configuration choices, and operational tradeoffs.</p>"
 +            "<h2>Practical Workflow</h2><p>This section walks through the actions, checks, "
 +            "and expected outcomes for a real environment.</p>"
              '<p><a href="../index.html">Back to Main Guide Index</a></p>'
              "</div></body></html>"
+         )

src/loader/runtime/safeguard_services.pymodified

  class PreActionValidator:
      """Validates tool calls before execution to catch problematic actions."""
 +    HTML_PLACEHOLDER_PATTERNS = [
 +        (
 +            re.compile(r"\bstarter\s+(?:content|overview)\b", re.IGNORECASE),
 +            "starter content",
 +        ),
 +        (
 +            re.compile(r"\bkey\s+concepts\s+go\s+here\b", re.IGNORECASE),
 +            "key concepts go here",
 +        ),
 +        (
 +            re.compile(r"\bpractical\s+steps\s+go\s+here\b", re.IGNORECASE),
 +            "practical steps go here",
 +        ),
 +        (
 +            re.compile(r"\blorem\s+ipsum\b", re.IGNORECASE),
 +            "lorem ipsum",
 +        ),
 +        (
 +            re.compile(r"\bcoming\s+soon\b", re.IGNORECASE),
 +            "coming soon",
 +        ),
 +        (
 +            re.compile(r"\bto\s+be\s+(?:added|written|completed|filled\s+in)\b", re.IGNORECASE),
 +            "to be added/written",
 +        ),
 +    ]
++
      DANGEROUS_PATTERNS = [
          (r'rm\s+(-[rf]+\s+)?/', "Dangerous: removing from root directory"),
          (r'rm\s+-rf\s+~', "Dangerous: removing home directory"),
                  severity="warning",
+             )
 +        html_placeholder_result = self._validate_html_placeholder_content(
 +            str(file_path),
 +            str(content),
 +        )
 +        if not html_placeholder_result.valid:
 +            return html_placeholder_result
++
          sensitive_paths = ['/etc/', '/usr/', '/bin/', '/sbin/', '/boot/', '/sys/', '/proc/']
          for sensitive in sensitive_paths:
              if file_path.startswith(sensitive):
              str(new_string),
+         )
 +        html_placeholder_result = self._validate_html_placeholder_content(
 +            str(file_path),
 +            prospective_content,
 +        )
 +        if not html_placeholder_result.valid:
 +            return html_placeholder_result
++
          html_index_result = self._validate_html_index_links(
              str(file_path),
              prospective_content,
                  severity="error",
+             )
 +        html_placeholder_result = self._validate_html_placeholder_patch(
 +            str(file_path),
 +            hunks,
 +            raw_patch,
 +        )
 +        if not html_placeholder_result.valid:
 +            return html_placeholder_result
++
          return ValidationResult(valid=True)
 +    def _validate_html_placeholder_content(
 +        self,
 +        file_path: str,
 +        content: str,
 +    ) -> ValidationResult:
 +        normalized = Path(file_path).expanduser()
 +        if normalized.suffix.lower() not in {".html", ".htm"}:
 +            return ValidationResult(valid=True)
++
 +        matched_labels = [
 +            label
 +            for pattern, label in self.HTML_PLACEHOLDER_PATTERNS
 +            if pattern.search(content)
 +        ]
 +        if not matched_labels:
 +            return ValidationResult(valid=True)
++
 +        preview = ", ".join(matched_labels[:3])
 +        if len(matched_labels) > 3:
 +            preview += ", ..."
 +        return ValidationResult(
 +            valid=False,
 +            reason="HTML content contains placeholder or stub text",
 +            suggestion=(
 +                "Replace placeholder phrases with concrete user-facing content before "
 +                f"writing the HTML artifact. Placeholder phrase(s): {preview}. Include "
 +                "specific explanations, examples, commands, or structured prose instead."
 +            ),
 +            severity="error",
 +        )
++
 +    def _validate_html_placeholder_patch(
 +        self,
 +        file_path: str,
 +        hunks: object,
 +        raw_patch: object,
 +    ) -> ValidationResult:
 +        normalized = Path(file_path).expanduser()
 +        if normalized.suffix.lower() not in {".html", ".htm"}:
 +            return ValidationResult(valid=True)
++
 +        added_fragments: list[str] = []
 +        if isinstance(raw_patch, str):
 +            for line in raw_patch.splitlines():
 +                if line.startswith("+") and not line.startswith("+++"):
 +                    added_fragments.append(line[1:])
++
 +        if isinstance(hunks, list):
 +            for hunk in hunks:
 +                if not isinstance(hunk, dict):
 +                    continue
 +                new_lines = hunk.get("new_lines")
 +                if isinstance(new_lines, list):
 +                    added_fragments.extend(str(line) for line in new_lines)
 +                lines = hunk.get("lines")
 +                if isinstance(lines, list):
 +                    for line in lines:
 +                        text = str(line)
 +                        if text.startswith("+") and not text.startswith("+++"):
 +                            added_fragments.append(text[1:])
++
 +        if not added_fragments:
 +            return ValidationResult(valid=True)
 +        return self._validate_html_placeholder_content(
 +            str(file_path),
 +            "\n".join(added_fragments),
 +        )
++
      def _validate_numbered_sibling_conflict(self, file_path: str) -> ValidationResult:
          path = Path(file_path).expanduser()
          if path.exists() or not path.suffix or not path.parent.exists():

tests/test_repair.pymodified

      assert "<title>Nginx Guide</title>" in decision.retry_message
      assert 'href="chapters/01-introduction.html"' in decision.retry_message
      assert "../index.html" not in decision.retry_message
 +    assert "Starter overview" not in decision.retry_message
 +    assert "go here" not in decision.retry_message
  def test_empty_response_retry_prefers_pending_index_over_broad_directory_headline(
      assert "If blanking continues, use this minimal starter payload shape" in decision.retry_message
      assert "<title>Chapter 1: Introduction to Nginx</title>" in decision.retry_message
      assert "../index.html" in decision.retry_message
 +    assert "Starter content" not in decision.retry_message
 +    assert "Key concepts go here" not in decision.retry_message
 +    assert "Practical steps go here" not in decision.retry_message
  def test_empty_response_retry_surfaces_minimal_child_html_payload_earlier_after_progress(
      assert decision.retry_message is not None
      assert "If blanking continues, use this minimal starter payload shape" in decision.retry_message
      assert "<title>Chapter 1: Introduction to Nginx</title>" in decision.retry_message
 +    assert "Starter content" not in decision.retry_message
 +    assert "go here" not in decision.retry_message
  def test_final_empty_response_retry_uses_short_exact_write_call(

tests/test_safeguard_services.pymodified

      assert result == ValidationResult(valid=True)
 +def test_pre_action_validator_blocks_placeholder_html_write(tmp_path: Path) -> None:
 +    validator = PreActionValidator()
++
 +    result = validator.validate(
 +        "write",
 +        {
 +            "file_path": str(tmp_path / "guide" / "chapters" / "01-introduction.html"),
 +            "content": (
 +                "<html><body><h1>Introduction</h1>"
 +                "<p>Starter content for this chapter.</p>"
 +                "<h2>Overview</h2><p>Key concepts go here.</p>"
 +                "</body></html>"
 +            ),
 +        },
 +    )
++
 +    assert result.valid is False
 +    assert result.reason == "HTML content contains placeholder or stub text"
 +    assert "concrete user-facing content" in result.suggestion
 +    assert "starter content" in result.suggestion
++
++
 +def test_pre_action_validator_blocks_placeholder_html_edit(tmp_path: Path) -> None:
 +    validator = PreActionValidator()
 +    page = tmp_path / "guide" / "chapters" / "01-introduction.html"
 +    page.parent.mkdir(parents=True)
 +    page.write_text("<html><body><h1>Introduction</h1></body></html>")
++
 +    result = validator.validate(
 +        "edit",
 +        {
 +            "file_path": str(page),
 +            "old_string": "</body>",
 +            "new_string": "<p>Practical steps go here.</p></body>",
 +        },
 +    )
++
 +    assert result.valid is False
 +    assert result.reason == "HTML content contains placeholder or stub text"
 +    assert "practical steps go here" in result.suggestion
++
++
 +def test_pre_action_validator_blocks_placeholder_html_patch(tmp_path: Path) -> None:
 +    validator = PreActionValidator()
++
 +    result = validator.validate(
 +        "patch",
 +        {
 +            "file_path": str(tmp_path / "guide" / "chapters" / "02-installation.html"),
 +            "patch": (
 +                "--- a/guide/chapters/02-installation.html\n"
 +                "+++ b/guide/chapters/02-installation.html\n"
 +                "@@ -1,1 +1,2 @@\n"
 +                " <h1>Installation</h1>\n"
 +                "+<p>Coming soon.</p>\n"
 +            ),
 +        },
 +    )
++
 +    assert result.valid is False
 +    assert result.reason == "HTML content contains placeholder or stub text"
 +    assert "coming soon" in result.suggestion
++
++
  def test_pre_action_validator_blocks_shell_text_rewrite_for_html_target() -> None:
      validator = PreActionValidator()