chore: commit local changes
This commit is contained in:
1 parent
c1336ad5b3
commit
e6e0832c8a
15 files changed
+291
-24
No files matched your search
+57
@@ -53,8 +53,18 @@ _MARKDOWN_MIN_CHARS = 100
|
||||
_MARKDOWN_INSTRUCTION = (
|
||||
"\n\n【输出要求】请将本次任务的最终结果生成为一个 Markdown(.md) 格式的文件,"
|
||||
"保存到 /mnt/user-data/outputs 目录,并使用 present_files 展示该文件。"
|
||||
"Markdown 内的目录/锚点跳转必须只跳转到当前这个 Markdown 文件内部:"
|
||||
"只使用 [标题](#section-id),不要使用 [标题](xxx.md#section-id)、"
|
||||
"[标题](xxx.html#section-id) 或任何跨文件锚点。"
|
||||
)
|
||||
|
||||
_LOCAL_DOC_MARKDOWN_LINK_RE = re.compile(
|
||||
r"^(?![a-z][a-z0-9+.-]*:)(?!//)(?P<path>[^#?]*\.(?:html?|md|markdown))(?P<query>\?[^#]*)?(?P<fragment>#.+)?$",
|
||||
re.IGNORECASE,
|
||||
)
|
||||
_MARKDOWN_INLINE_LINK_RE = re.compile(r"(?<!!)\[([^\]]+)\]\(([^)\s]+)(?:\s+\"[^\"]*\")?\)")
|
||||
_MARKDOWN_REFERENCE_LINK_RE = re.compile(r"^(\s*\[[^\]]+\]:\s*)(\S+)(.*)$", re.MULTILINE)
|
||||
|
||||
|
||||
class MarkdownRequirementError(Exception):
|
||||
"""Raised when a ``require_markdown`` task did not produce a valid .md file.
|
||||
@@ -262,6 +272,43 @@ def _has_valid_markdown_file(result_files: list[dict[str, Any]] | None) -> bool:
|
||||
return False
|
||||
|
||||
|
||||
def _normalize_same_page_markdown_target(target: str) -> str | None:
|
||||
value = (target or "").strip()
|
||||
if not value or value.startswith("#"):
|
||||
return value
|
||||
match = _LOCAL_DOC_MARKDOWN_LINK_RE.match(value)
|
||||
if not match:
|
||||
return value
|
||||
fragment = match.group("fragment")
|
||||
return fragment or None
|
||||
|
||||
|
||||
def _constrain_markdown_same_page_anchor_links(content: str) -> str:
|
||||
"""Keep generated Markdown document jumps inside the current Markdown file."""
|
||||
if not content or not isinstance(content, str):
|
||||
return content or ""
|
||||
|
||||
def replace_inline(match: re.Match[str]) -> str:
|
||||
label = match.group(1)
|
||||
target = match.group(2)
|
||||
normalized = _normalize_same_page_markdown_target(target)
|
||||
if normalized is None:
|
||||
return label
|
||||
return f"[{label}]({normalized})"
|
||||
|
||||
def replace_reference(match: re.Match[str]) -> str:
|
||||
prefix = match.group(1)
|
||||
target = match.group(2)
|
||||
suffix = match.group(3)
|
||||
normalized = _normalize_same_page_markdown_target(target)
|
||||
if normalized is None:
|
||||
return f"{prefix}#"
|
||||
return f"{prefix}{normalized}{suffix}"
|
||||
|
||||
content = _MARKDOWN_INLINE_LINK_RE.sub(replace_inline, content)
|
||||
return _MARKDOWN_REFERENCE_LINK_RE.sub(replace_reference, content)
|
||||
|
||||
|
||||
def _build_result(
|
||||
task: dict[str, Any],
|
||||
*,
|
||||
@@ -1169,6 +1216,7 @@ class ScheduledTaskService:
|
||||
|
||||
# 4) Sanitize + validate, with one repair pass on failure.
|
||||
html, _removed = html_page_utils.sanitize_html(html)
|
||||
html, _link_changes = html_page_utils.constrain_same_page_anchor_links(html)
|
||||
errors = html_page_utils.validate_html(html)
|
||||
if errors:
|
||||
logger.info("Generated HTML failed validation (%s); attempting repair", errors)
|
||||
@@ -1181,6 +1229,7 @@ class ScheduledTaskService:
|
||||
fixed = repaired.get("fixed_html")
|
||||
if isinstance(fixed, str) and "<" in fixed:
|
||||
fixed_clean, _ = html_page_utils.sanitize_html(fixed)
|
||||
fixed_clean, _ = html_page_utils.constrain_same_page_anchor_links(fixed_clean)
|
||||
if not html_page_utils.validate_html(fixed_clean):
|
||||
html = fixed_clean
|
||||
except Exception:
|
||||
@@ -1540,6 +1589,14 @@ class ScheduledTaskService:
|
||||
index += 1
|
||||
|
||||
shutil.copy2(path, candidate)
|
||||
if candidate.suffix.lower() in {".md", ".markdown"}:
|
||||
try:
|
||||
original = candidate.read_text(encoding="utf-8", errors="ignore")
|
||||
constrained = _constrain_markdown_same_page_anchor_links(original)
|
||||
if constrained != original:
|
||||
candidate.write_text(constrained, encoding="utf-8", newline="")
|
||||
except Exception:
|
||||
logger.debug("Failed to constrain scheduled markdown links in %s", candidate, exc_info=True)
|
||||
return candidate
|
||||
|
||||
def _build_file_result(self, *, virtual_path: str, source_path: Path, archive_path: Path) -> dict[str, Any]:
|
||||
|
||||
Reference in new issue
Block a user