dedup_inline_images filter is broken

[jmarantz]

A user reports that it's broken on his site here: https://groups.google.com/forum/?utm_medium=email&utm_source=footer#!msg/mod-pagespeed-discuss/ZlpKH_1O3A4/A57MfpGrAAAJ

And I think it's also broken in the same way on modpagespeed.com here: http://modpagespeed.com/dedup_inlined_images.html?ModPagespeed=on&ModPagespeedFilters=inline_images,dedup_inlined_images

[jeffkaufman]

From a scan of the email I thought at first this might be #1002, but this actually looks like the filter is seriously broken now?

[morlovich]

Seems easier to reproduce with local debug install than mps.com due to lack
of critical image stuff getting involved. Taking a look.

On Mon, Aug 1, 2016 at 6:43 PM, Jeff Kaufman notifications@github.com
wrote:

From a scan of the email I thought at first this might be #1002
#1002, but this
actually looks like the filter is seriously broken now?

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#1369 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/ADl1RLV96ohc6-p7VEvoYY8X66Eb5WcNks5qbnZ5gaJpZM4JZqal
.

[morlovich]

Well, this is obviously broken (irrelevant stuff removed) --- why are we getting all the duplicated (including the one that has the data) the same ID?:
<script id="pagespeed_script_1" data-pagespeed-no-defer>
pagespeed.dedupInlinedImages.inlineImg('pagespeed_img_98KheFixjv','pagespeed_img_98KheFixjv','pagespeed_script_1');
</script>

<script type="text/javascript" id="pagespeed_script_2" data-pagespeed-no-defer>
pagespeed.dedupInlinedImages.inlineImg('pagespeed_img_98KheFixjv','pagespeed_img_98KheFixjv','pagespeed_script_2');

[morlovich]

This might be enough (but I haven't run any actual tests):
--- net/instaweb/rewriter/dedup_inlined_images_filter.cc 2015-08-28 14:25:48.000000000 -0400
+++ net/instaweb/rewriter/dedup_inlined_images_filter.cc 2016-08-04 11:29:01.000000000 -0400
@@ -117,7 +117,8 @@
GoogleString element_id;
const char* id = element->AttributeValue(HtmlName::kId);
if (id == NULL || id[0] == '\0') {

•  element_id = StrCat("pagespeed_img_", hash);
  

•  element_id = StrCat("pagespeed_img_", hash,
  

•                      IntegerToString(++snippet_id_));
  

  driver()->AddAttribute(element, HtmlName::kId, element_id);
  } else {
  element_id = id;

[morlovich]

Let's try this again:

--- net/instaweb/rewriter/dedup_inlined_images_filter.cc      2015-08-28 14:25:48.000000000 -0400
+++ net/instaweb/rewriter/dedup_inlined_images_filter.cc       2016-08-04 11:29:01.000000000 -0400
@@ -117,7 +117,8 @@
     GoogleString element_id;
     const char* id = element->AttributeValue(HtmlName::kId);
     if (id == NULL || id[0] == '\0') {
-      element_id = StrCat("pagespeed_img_", hash);
+      element_id = StrCat("pagespeed_img_", hash,
+                          IntegerToString(++snippet_id_));
       driver()->AddAttribute(element, HtmlName::kId, element_id);
     } else {
       element_id = id;

[jeffkaufman]

Any idea how this could have gotten broken? Did it just never work?

[jmarantz]

I haven't dug in deeply enough to be sure but I assume it worked at one
point. One possible explanation is that maybe browsers were sloppier then
about allowing multiple elements with the same ID.

On Fri, Aug 5, 2016 at 1:48 AM, Jeff Kaufman notifications@github.com
wrote:

Any idea how this could have gotten broken? Did it just never work?

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
#1369 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AB2kPXjO0ZPkBi1ee8Lt_FmyNffXSeScks5qcs61gaJpZM4JZqal
.

[morlovich]

That would be very weird, since it wass basically doing
document.getElementById(to_id).src = document.getElementById(from_id).src, where to_id was equal to from_id.

[jeffkaufman]

This went out in 1.11.33.3

[jeffkaufman]

confirmed fixed