Permalink
Commits on Feb 12, 2014
  1. iframeParser was sometimes not completing on Android 2.3. Waiting for…

    … the iframe window's onload addressed this.
    committed Feb 12, 2014
Commits on Feb 7, 2014
  1. If capturing then use `window.stop()` where available to override res…

    …ource downloading by the prescanner.
    committed Feb 7, 2014
  2. Allow STAGING_DOCUMENT_IS_INERT to be `true` on IE10 although the sta…

    …ging document URL is same as `document.URL`.
    committed Feb 7, 2014
Commits on Jan 13, 2014
  1. <title> of content document wasn't being merged into the view when ca…

    …pturing and decor document also had <title>.
    committed Jan 13, 2014
Commits on Dec 20, 2013
Commits on Dec 17, 2013
  1. Oops. Was only overriding resolveURL() for img@srcset, but source@src…

    …set needs the same override.
    committed Dec 17, 2013
  2. panner.options.loadFromString() makes capturing a bit more logical.

    Now normalize() and deneutralizeAll() are called as part of Capture.getDocument() instead of in decor.start(). This is the same as what happens for panned pages.
    committed Dec 17, 2013
Commits on Dec 16, 2013
  1. AttrDesc class makes code branching in resolveAll() more manageable:

    - script@src must always resolve
    - img@srcset and a@ping need special resolveURL code
    committed Dec 16, 2013
Commits on Dec 14, 2013
  1. On some pages with capturing enabled and being viewed on IE8 with an …

    …empty cache script.onreadystatechange handlers weren't being called. Wrapping enableScript() in setTimeout() resolved this issue. Not sure why.
    committed Dec 14, 2013
Commits on Dec 13, 2013
  1. Rename a few consts / vars.

    attrDesc.deref is merged with attrDesc.neutralize.
    If `neutralize` is true then the attribute needs to be neutralized before parsing. If `neutralize` is `> 0` then the attribute stays neutralized during the normalize() call. This means that if `neutralize` is less-than-zero the attribute will be neutralized for parsing and deneutralized for normalizing.
    committed Dec 13, 2013
  2. The HTMLLoader used in capturing wasn't passing `details` and hence `…

    …isNeutralized` notification
    committed Dec 13, 2013
Commits on Dec 12, 2013
  1. Neutralize URLs by prepending the attribute *value* with "meeko-href"…

    … rather than prepending the attribute *name* with "meeko-". This allows picture polyfill to work in the normalize(doc) call as it copies `source@src` to `img@src`.
    
    Also, if STAGING_DOCUMENT_IS_NEUTRAL then don't even need to keep neutralized URLs around for the normalize() call. See resolveAll() and deneutralizeAll().
    committed Dec 12, 2013
Commits on Dec 10, 2013
Commits on Dec 6, 2013
  1. Fix for IE9 swallowing <source> elements when not within <video> or <…

    …audio>.
    
    Needed if <picture> becomes best-practice.
    committed Dec 6, 2013
Commits on Dec 4, 2013
Commits on Nov 29, 2013
  1. some() is useful, like every().

    panner.navigate() now calls location.assign() (or replace() ) if history.pushState() not available.
    committed Nov 29, 2013
Commits on Nov 27, 2013
  1. HTML parsing performance on IE8 can be significantly improved by writ…

    …ing the <head> into an iframe and using element.innerHTML for the <body>. Can't hurt perf on other browsers.
    
    Also use DOMParser.parseFromString(html, 'text/html') on browsers that support it - that is, Firefox.
    committed Nov 27, 2013
  2. Undeclared variables

    committed Nov 27, 2013
Commits on Nov 25, 2013
  1. mergeElement() now removes @style.

    HTMLParser adds `style="display: none"` to <body>. This usually makes parsing a bit faster on IE8. The mergeElement() mod means this doesn't affect final view.
    committed Nov 25, 2013
  2. Add `mustResolve` field to resolveAll().

    Add `mustResolve` field to `details` object that is passed to HTMLLoader and HTMLParser
    Resolving URLs in the landing-page now runs over everything in <head>, not just elements after the self-marker
    committed Nov 25, 2013
Commits on Nov 22, 2013
  1. decor.placeHolders is probably a bad design choice and the javadoc-pa…

    …nner bookmarklet highlighted this bug.
    committed Nov 22, 2013
  2. Don't want to insert boot-script before html5shiv & visibility <style…

    …> when boot-script isn't in the head (as occurs in meeko-panner bookmarklet)
    committed Nov 22, 2013
  3. decor.placeHolders is probably a bad design choice,

    and capturing + long.html highlighted this bug.
    committed Nov 22, 2013
Commits on Nov 21, 2013
  1. <title> in test pages

    committed Nov 21, 2013
Commits on Nov 20, 2013
  1. Fully replace srcAttrs, hrefAttrs with urlAttrs.

    Implement resolveAll(doc, baseURL, isNeutralized) to be called from HTMLParser and HTMLLoader where appropriate.
    Not all URL attrs in the landing-page need to be resolved,
    because most elements will be paged out anyway when panning.
    For now just resolve script@src and link@href in <head>
    committed Nov 20, 2013
Commits on Nov 18, 2013