8dae91c2ca
Protect pending edits on window close, keep save-as paths and drafts consistent, copy relative image resources, reject oversized table imports without truncation, and preserve independent preview scrolling during refresh. Archive rendered MHTML resources with portable references and embedded local fonts. Preserve existing outputs on export cancellation or failure. Convert editor match positions and filter directory results in linear passes. Include the previously completed editor image/table/history improvements, undoable replacement, export snapshots, shared page settings, and cancellable PDF/Word export workflow. Update documentation and version to 0.3.1. Validation: 142 pytest cases passed; GUI smoke, improvements integration, and offline MHTML/scroll reliability smoke checks passed.
242 lines
11 KiB
Python
242 lines
11 KiB
Python
"""Save a rendered, self-contained browser archive without a visible browser."""
|
|
from __future__ import annotations
|
|
|
|
import os
|
|
import base64
|
|
import hashlib
|
|
import mimetypes
|
|
import re
|
|
import tempfile
|
|
from email import encoders, policy
|
|
from email.parser import BytesParser
|
|
from html import escape, unescape
|
|
from html.parser import HTMLParser
|
|
from pathlib import Path
|
|
from urllib.parse import urljoin
|
|
|
|
from PySide6.QtCore import QTimer, QUrl
|
|
from PySide6.QtWebEngineCore import QWebEngineDownloadRequest, QWebEnginePage, QWebEngineProfile
|
|
from PySide6.QtWidgets import QMessageBox, QProgressDialog
|
|
|
|
|
|
def portable_archive(data: bytes) -> bytes:
|
|
"""Use archive IDs instead of file URLs, and embed CSS font dependencies.
|
|
|
|
Chromium can block file: subresources even when present in an MHTML archive.
|
|
Native page saving also omits local fonts that were not painted on the page.
|
|
"""
|
|
import tinycss2
|
|
message = BytesParser(policy=policy.default).parsebytes(data)
|
|
parts = [p for p in message.walk() if not p.is_multipart()]
|
|
aliases = {}
|
|
for part in parts:
|
|
location = part.get("Content-Location", "")
|
|
if location:
|
|
identifier = "mdhelper-" + hashlib.sha256(location.encode()).hexdigest()[:24]
|
|
if "Content-ID" in part:
|
|
identifier = part["Content-ID"].strip("<>")
|
|
else:
|
|
part["Content-ID"] = f"<{identifier}>"
|
|
aliases[location] = "cid:" + identifier
|
|
|
|
def css_resource(url, base):
|
|
if url.startswith(("#", "data:", "cid:")):
|
|
return url
|
|
absolute = urljoin(base, url)
|
|
if absolute in aliases:
|
|
return aliases[absolute]
|
|
qurl = QUrl(absolute)
|
|
if qurl.isLocalFile():
|
|
path = Path(qurl.toLocalFile())
|
|
# Bundled styles list legacy font fallbacks not shipped with the app.
|
|
if not path.is_file():
|
|
return "data:,"
|
|
mime = mimetypes.guess_type(path.name)[0] or "application/octet-stream"
|
|
return f"data:{mime};base64," + base64.b64encode(path.read_bytes()).decode("ascii")
|
|
if qurl.scheme() in {"http", "https"}:
|
|
raise ValueError("以下样式资源未能包含在离线归档中,请改用本地资源后重试:" + absolute)
|
|
return absolute
|
|
|
|
def css_text(text, base):
|
|
tokens = tinycss2.parse_stylesheet(text)
|
|
def visit(items):
|
|
for token in items:
|
|
if token.type == "url":
|
|
value = css_resource(token.value, base)
|
|
token.value = value
|
|
token.representation = 'url("' + value.replace('"', '%22') + '")'
|
|
elif token.type == "function" and token.lower_name == "url":
|
|
arguments = [item for item in token.arguments if item.type not in {"whitespace", "comment"}]
|
|
value = arguments[0].value if len(arguments) == 1 and arguments[0].type == "string" else tinycss2.serialize(token.arguments).strip()
|
|
token.arguments = tinycss2.parse_component_value_list('"' + css_resource(value, base).replace('"', '%22') + '"')
|
|
else:
|
|
if token.type == "at-rule" and token.lower_at_keyword == "import":
|
|
for value in token.prelude:
|
|
if value.type == "string":
|
|
value.value = css_resource(value.value, base)
|
|
value.representation = '"' + value.value + '"'
|
|
for name in ("content", "prelude", "arguments"):
|
|
if getattr(token, name, None):
|
|
visit(getattr(token, name))
|
|
visit(tokens)
|
|
return tinycss2.serialize(tokens)
|
|
|
|
class HtmlResources(HTMLParser):
|
|
def __init__(self, text):
|
|
super().__init__(convert_charrefs=False)
|
|
self.changes = []
|
|
self.offsets = [0] + [m.end() for m in re.finditer("\n", text)]
|
|
self.feed(text)
|
|
|
|
def handle_starttag(self, tag, attrs):
|
|
raw = self.get_starttag_text()
|
|
line, column = self.getpos()
|
|
offset = self.offsets[line - 1] + column
|
|
for match in re.finditer(r'''\b(?:src|href|poster)\s*=\s*(["'])(.*?)\1''', raw, re.S):
|
|
value = unescape(match[2])
|
|
if value in aliases:
|
|
self.changes.append((offset + match.start(2), offset + match.end(2), escape(aliases[value], quote=True)))
|
|
|
|
for part in parts:
|
|
kind = part.get_content_type()
|
|
if kind not in {"text/html", "text/css"}:
|
|
continue
|
|
text = part.get_payload(decode=True).decode(part.get_content_charset() or "utf-8")
|
|
if kind == "text/css":
|
|
text = css_text(text, part.get("Content-Location", ""))
|
|
else:
|
|
chunks, offset = [], 0
|
|
for start, end, replacement in HtmlResources(text).changes:
|
|
chunks.extend((text[offset:start], replacement)); offset = end
|
|
chunks.append(text[offset:])
|
|
text = "".join(chunks)
|
|
if "Content-Transfer-Encoding" in part:
|
|
del part["Content-Transfer-Encoding"]
|
|
part.set_param("charset", "utf-8")
|
|
part.set_payload(text.encode("utf-8"))
|
|
encoders.encode_base64(part)
|
|
# Chromium expects literal Content-Location values, not RFC 2047 folded URLs.
|
|
return message.as_bytes(policy=policy.SMTP.clone(max_line_length=0, refold_source="none", utf8=True))
|
|
|
|
|
|
class MhtmlExportController:
|
|
def __init__(self, owner, output: Path, content: str, source_path: Path):
|
|
self.owner, self.output, self.content = owner, output.resolve(), content
|
|
self.base_url = QUrl.fromLocalFile(str(source_path.parent.resolve()) + os.sep)
|
|
self.finished = False
|
|
self.request = self.page = self.profile = self.temporary = None
|
|
self.progress = QProgressDialog("正在准备离线文档…", "取消", 0, 0, owner)
|
|
self.progress.setWindowTitle("导出 MHTML")
|
|
self.progress.setMinimumDuration(0)
|
|
self.progress.setAutoClose(False)
|
|
self.progress.canceled.connect(self.cancel)
|
|
self.timeout = QTimer(self.progress)
|
|
self.timeout.setSingleShot(True)
|
|
self.timeout.timeout.connect(lambda: self.fail("离线文档导出超时,请检查图片和公式资源。"))
|
|
|
|
def start(self):
|
|
try:
|
|
self.output.parent.mkdir(parents=True, exist_ok=True)
|
|
self.temporary = tempfile.TemporaryDirectory(prefix=".mdhelper-mhtml-", dir=self.output.parent, ignore_cleanup_errors=True)
|
|
self.staged = Path(self.temporary.name) / "document.mhtml"
|
|
# A separate profile keeps unrelated downloads out of this operation.
|
|
self.profile = QWebEngineProfile(self.owner)
|
|
self.page = QWebEnginePage(self.profile, self.owner)
|
|
self.owner._configure_web_settings(self.page.settings())
|
|
self.profile.downloadRequested.connect(self.download_requested)
|
|
self.page.loadFinished.connect(self.loaded)
|
|
self.timeout.start(120_000)
|
|
self.page.setHtml(self.content, self.base_url)
|
|
self.progress.show()
|
|
except OSError as exc:
|
|
self.fail(str(exc))
|
|
|
|
def loaded(self, ok):
|
|
if self.finished:
|
|
return
|
|
if not ok:
|
|
self.fail("离线文档页面加载失败。")
|
|
return
|
|
self.progress.setLabelText("正在载入图片、公式和字体…")
|
|
self.wait_ready()
|
|
|
|
def wait_ready(self):
|
|
if self.finished:
|
|
return
|
|
self.page.runJavaScript(r"""(() => {
|
|
const images = [...document.images].filter(i => i.getAttribute('src') || i.getAttribute('srcset'));
|
|
const lazy = images.filter(i => i.loading === 'lazy');
|
|
lazy.forEach(i => i.loading = 'eager');
|
|
if (lazy.length || images.some(i => !i.complete) ||
|
|
(document.fonts && document.fonts.status !== 'loaded') ||
|
|
window.mdhelperRenderReady === false) return 'waiting';
|
|
const missing = images.filter(i => !i.naturalWidth);
|
|
return missing.length ? 'missing:' + missing.map(i => i.src).join('\n') : 'ready';
|
|
})();""", self.ready)
|
|
|
|
def ready(self, state):
|
|
if self.finished:
|
|
return
|
|
if isinstance(state, str) and state.startswith("missing:"):
|
|
self.fail("部分图片未能加载,未生成离线文档。请修复图片路径后重试。\n" + state[8:])
|
|
elif state == "ready":
|
|
self.progress.setLabelText("正在封装离线资源…")
|
|
self.page.save(str(self.staged), QWebEngineDownloadRequest.SavePageFormat.MimeHtmlSaveFormat)
|
|
else:
|
|
QTimer.singleShot(100, self.wait_ready)
|
|
|
|
def download_requested(self, request):
|
|
if self.finished or request.page() != self.page:
|
|
return
|
|
self.request = request
|
|
request.isFinishedChanged.connect(self.download_finished)
|
|
if request.state() == QWebEngineDownloadRequest.DownloadState.DownloadRequested:
|
|
request.accept()
|
|
self.download_finished()
|
|
|
|
def download_finished(self):
|
|
if self.finished or self.request is None or not self.request.isFinished():
|
|
return
|
|
if self.request.state() != QWebEngineDownloadRequest.DownloadState.DownloadCompleted:
|
|
self.fail("离线资源保存失败:" + self.request.interruptReasonString())
|
|
return
|
|
try:
|
|
if not self.staged.exists() or not self.staged.stat().st_size:
|
|
raise OSError("浏览器未生成有效的离线文档。")
|
|
self.staged.write_bytes(portable_archive(self.staged.read_bytes()))
|
|
os.replace(self.staged, self.output)
|
|
except (OSError, ValueError, UnicodeError) as exc:
|
|
self.fail(str(exc))
|
|
return
|
|
self.owner.status_label.setText(f"已导出:{self.output}")
|
|
self.cleanup()
|
|
|
|
def cancel(self):
|
|
if not self.finished:
|
|
self.owner.status_label.setText("已取消 MHTML 导出,原有输出文件保持不变")
|
|
self.cleanup()
|
|
|
|
def fail(self, message):
|
|
if not self.finished:
|
|
self.cleanup()
|
|
QMessageBox.warning(self.owner, "导出失败", message)
|
|
|
|
def cleanup(self):
|
|
self.finished = True
|
|
self.timeout.stop()
|
|
if self.request is not None and not self.request.isFinished():
|
|
self.request.cancel()
|
|
if self.page is not None:
|
|
self.page.triggerAction(QWebEnginePage.WebAction.Stop)
|
|
# Destroy the page before its profile, as required by Qt WebEngine.
|
|
self.page.destroyed.connect(self.profile.deleteLater)
|
|
self.page.deleteLater()
|
|
self.page = None
|
|
if self.temporary is not None:
|
|
self.temporary.cleanup()
|
|
self.temporary = None
|
|
self.progress.blockSignals(True)
|
|
self.progress.close()
|
|
self.progress.deleteLater()
|
|
self.owner._mhtml_exports.discard(self)
|