The exported HTML has changed its domain; why have the images and links broken?

This paper uses synthetic samples to demonstrate the processing method. The code has been verified using examples in an isolated environment; this does not guarantee effectiveness for all real-world files or products. Before processing important files, please retain the original copies and verify them against the boundaries and acceptance criteria outlined in this document. Which problem does this paper address first? The baseline for urljoin should be the original page URL; do not arbitrarily append a trailing slash to convert a file URL into a directory. This example handles only a[href] and img[src], without overriding CSS, srcset, or base tags; rewriting URLs is also not part of HTML sanitization. self-contained synthetic example dependencies are listed in the import list. The code constructs inputs in memory; runtime logs are provided separately and do not constitute a guarantee for all real-world files. from urllib.parse import urljoin, urlsplit from bs4 import BeautifulSoup def rewrite(markup, page_url): soup = BeautifulSoup(markup, "html.parser") for tag, attr in [("a", "href"), ("img", "src")]: for node in soup.find_all(tag): value = node.get(attr) if not value or value.startswith("#"): continue result = urljoin(page_url, value) if urlsplit(result).scheme not in {"https", "http"}: del node[attr] else: node[attr] = result return soup result = rewrite('<img src="logo.png"><a href="../guide/">Guide</a><a href="#top">Top</a><a href="javascript:alert(1)">Bad</a>', "https://example.com/docs/page.html") assert result.img["src"] == "https://example.com/docs/logo.png" assert result.find_all("a")[0]["href"] == "https://example.com/guide/" assert result.find_all("a")[1]["href"] == "#top" assert "href" not in result.find_all("a")[2].attrs When enterprises migrate internal documents, wikis, or legacy sites to a new domain, the most common “supernatural phenomena” they encounter are broken images and incorrect link redirects. This is usually not due to resource loss, but rather parsing context lost . The browser parses relative paths (such as images/logo.png ), strictly adhering to the RFC 3986 standard: prioritize searching for <base href="..."> tag; if absent, calculate it based on the current document’s URL. If the exported static HTML neither injects the correct <base> tag, which has also been detached from the original directory structure, causes all relative references to point to the root directory of the new domain or to the current page’s path, resulting in resource 404. urljoin and the semantic pitfalls of the base tag urllib.parse.urljoin(base, url) is the core tool for handling link rewriting, but its behavior differs from how browsers handle <base> There are subtle differences in tag parsing, which is also the root cause of many scripts that “work after being fixed.” urljoin defaults to base The argument is treated as a “file path” rather than a “directory,” unless it begins with / at the end. For example, urljoin("https://new.com/docs", "img.png") The result is https://new.com/img.png (docs is discarded as a filename), rather than the expected https://new.com/docs/img.png 。 In contrast, HTML <base> Although tag parsing in browsers also relies on trailing slashes to distinguish files from directories, it exhibits slightly higher tolerance for non-standard input. In programmatic processing, we cannot assume the validity of input data. Therefore, when invoking urljoin Preprocessing must be performed first: checking the validity of the scheme and forcibly appending the directory suffix. / , filter out null values. Only standardized base URLs can serve as reliable inputs. Furthermore, for paths that are already absolute (starting with / at the beginning) or links containing the full protocol, urljoin will either return the original value directly or replace only the hostname; this feature makes it more secure than string concatenation, but it also requires developers to clearly distinguish between the boundaries of “rewrite” and “preserve.” synthesis example and boundary condition handling The following code demonstrates how to safely rewrite resource links in HTML. Note: This is a synthetic teaching example , designed to demonstrate the logic rather than a production‑grade implementation. The example uses in‑memory strings to simulate input, thereby avoiding file system operations and network requests; please verify the logic in a controlled environment. Acceptance Criteria and Manual Review Boundary automation scripts can resolve mechanical path-joining issues, but they cannot replace the judgment of business semantics. The following table lists common exceptions and their handling decisions, helping teams distinguish between scenarios that are “code-resolvable” and those that require human intervention: Scope of Non-Applicability Statement : This method applies only to static HTML attribute rewriting. For links generated at runtime by JavaScript, references within the Shadow DOM, and CSS background-image URLs in the content, or Canvas/WebGL texture paths, are inaccessible to BeautifulSoup. Such cases must be validated in a real browser environment, or subjected to end-to-end visual regression testing using Playwright/Puppeteer. Code demonstrations do not equate to delivery quality; final acceptance must be based on the rendering results produced by the user agent. References and Original Manufacturer Sources Beautiful Soup Official Documentation - Attributes: https://www.crummy.com/software/BeautifulSoup/bs4/doc/#attributes Python urllib.parse.urljoin documentation: https://docs.python.org/3/library/urllib.parse.html#urllib.parse.urljoin RFC 3986 URI Generic Syntax: https://datatracker.ietf.org/doc/html/rfc3986 Next steps: Add assertion checks for rewritten links to your CI/CD pipeline; perform sample-based validation of resource accessibility on historically exported documents; if you frequently encounter issues with dynamic content links, assess the necessity of introducing a headless browser for visual regression testing. Remember that any automated fix should be regarded as a “supporting measure,” not a “silver bullet.” The decision on handling the specific exception observation What should be done should not infer File-type page URL retains the original baseline without a trailing slash fragment anchor retains the citation on this page is not required to be changed to the remote page script protocol Removing attributes in the example complete HTML still requires security sanitization uses this checklist for document translation The body text, links, code, and structure of the web document should be processed separately. Include the protected sections in the acceptance checklist, then assess the post-translation readability and page display. OmniDoc (Product Entry) provides an entry point for translating multi-format documents, and the official website highlights its OCR, terminology database, and bilingual alignment features. For teams that wish to start processing directly from files, this is a path worth evaluating with real-world samples. Please use de-identified files that you are authorized to upload, and verify the results against the checklist provided in this document; we do not claim that OmniDoc utilizes the open-source libraries or internal algorithms referenced herein, nor do we substitute examples for product performance testing. - 1 -