@smart-cloud/publisher-exporter 1.2.0 → 1.2.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/README.md CHANGED
@@ -41,6 +41,76 @@ manifests. The declaration does not require longer render readiness waits.
41
41
 
42
42
  ## Release notes
43
43
 
44
+ ### 1.2.2
45
+
46
+ - Save the original main HTTP response HTML in both local and delegated
47
+ rendering, including generated 404 pages. Browser DOM mutations, runtime
48
+ mounts, tokens and transient UI are never serialized into published HTML.
49
+ Normal URL rewriting still runs on the saved source HTML afterward.
50
+ - Remove DOM serialization fallbacks, partial-DOM recovery saves and runtime
51
+ mount snapshotting from the publishing paths. If the response body cannot be
52
+ read, retry or fail the page instead of substituting different content. Read
53
+ the response belonging to that navigation; do not issue a second page GET.
54
+ - Keep browser execution for DOM/network asset discovery and dynamically created
55
+ page links. Saving source HTML does not disable discovery inside shadow roots,
56
+ declared dependencies or JavaScript-generated links and resources.
57
+ - Retain the main HTTP response independently of `page.goto()` completion.
58
+ A synchronous script in the document head can keep the browser parser from
59
+ creating the body even when the complete source HTML has arrived.
60
+ - Before browser discovery, require a parsed HTML document with a body within
61
+ the configured readiness timeout. A navigation timeout can recover if parsing
62
+ finishes within that budget; it does not cause a partial-DOM export.
63
+ A zero or invalid readiness timeout uses a bounded 1.5-second document check
64
+ instead of disabling this mandatory check.
65
+ - Retry failed navigation or document readiness up to three times, closing the
66
+ failed page and context before a one-second, then two-second backoff. Ordinary
67
+ HTTP 404 skipping, generated 404 capture and HTTP 5xx retries remain separate.
68
+ - Log the response status, document readiness and grouped pending-request origins
69
+ and resource types. Pending-request diagnostics omit paths, query strings,
70
+ credentials and headers. Persistent failures retain the incomplete-crawl
71
+ deployment barrier and never save an empty successful page.
72
+
73
+ Upgrade the coordinator and rebuild delegated worker images with 1.2.2 after
74
+ active jobs stop. Node.js, Playwright, browser launch flags, WordPress plugins
75
+ and cache adapters are unchanged; this patch alone does not require reinstalling
76
+ browser binaries or changing the worker protocol. Run crawl/publish afterward;
77
+ deploy alone cannot finish a failed render. Use a full publish once to replace
78
+ all previously saved DOM-based HTML: incremental mode can reuse existing
79
+ unchanged outputs and does not automatically convert their capture format.
80
+
81
+ ### 1.2.1
82
+
83
+ - Serialize a detached document clone instead of destructively cleaning the
84
+ live browser page. Preserve the HTML/head/body scaffold even when scripts
85
+ created it, and restore structural runtime mounts from their pristine
86
+ snapshots rather than deleting them.
87
+ - Continue removing transient runtime roots, overlays, initialization markers
88
+ and scroll locks from exported HTML without changing the live DOM or running
89
+ custom-element constructors during snapshot restoration. Repeated captures
90
+ produce the same result.
91
+ - Reject missing-body and non-HTML documents with explicit serialization
92
+ diagnostics. Do not hide an incomplete render with optional chaining or
93
+ manufacture an empty successful export; the incomplete-crawl deploy barrier
94
+ remains enforced.
95
+
96
+ Upgrade the coordinator after active jobs stop. Delegated render workers use
97
+ the same serializer and need an image rebuild to receive this correction.
98
+ Node 24.22.1 and Playwright 1.63.0 remain unchanged, so a browser reinstall is
99
+ not required solely for this patch. No WordPress plugin update is needed.
100
+
101
+ A failure at HTML capture is still a crawl failure, not a deploy failure.
102
+ Run a new crawl/publish after upgrading. An incremental crawl can reuse
103
+ successfully checkpointed outputs when the existing manifest and change
104
+ tokens permit it; a failed full crawl may not have such checkpoints. Do not
105
+ use `--resume-rewrite` or delete the incomplete marker to bypass unfinished
106
+ rendering, asset discovery/downloads or rewrite.
107
+
108
+ Release audit: the production high-severity gate passes. Two inherited low
109
+ findings refer to [deepmerge GHSA-gcqx-w48m-w2h4](https://github.com/advisories/GHSA-gcqx-w48m-w2h4)
110
+ through `@wordpress/data`; no patched deepmerge version is currently published.
111
+ This targeted hotfix does not apply npm's suggested breaking WordPress package
112
+ downgrade, and does not claim a zero-finding audit.
113
+
44
114
  ### 1.1.87
45
115
 
46
116
  - Require patched supported Node.js runtimes: Node 22.23.4 or newer within