Skip to content
Commits on Jan 2, 2016
  1. @jvshahid
  2. @jvshahid
  3. @mkristian @jvshahid

    parsing numeric character references needs an update of nekohtml

    mkristian committed with jvshahid
    the new version of nekohtml brought a few regressions. this commit fixes
    but two error warning ones.
    
    it avoids to autocomplete the tbody tag around tr tags of a table. the check
    of unknown html did change upstream and got adjusted.
    
    fixes #1113
    
    Sponsored by Lookout Inc.
  4. @mkristian @jvshahid

    fix encoding when writing data

    mkristian committed with jvshahid
    this is partial fix for #1113 to NOT use character entities when the encoding
    of the document can encode the data.
    
    Sponsored by Lookout Inc.
Commits on Jan 1, 2016
  1. @larskanis
Commits on Dec 31, 2015
  1. @flavorjones
Commits on Dec 30, 2015
  1. @jvshahid

    Merge pull request #1400 from mehdi-farsi/fix-typo-in-changelog

    jvshahid committed
    Fix typo in CHANGELOG.rdoc
  2. @mehdi-farsi
Commits on Dec 26, 2015
  1. @nurse

    Explicitly specify file encoding to avoid test failure

    nurse committed
    The encoding of strings read from files are default_external by default.
    On Windows, it is locale by default though many current Unix uses UTF-8.
    https://ci.appveyor.com/project/flavorjones/nokogiri/branch/master/job/pnwl8bft96oeyidt
Commits on Dec 18, 2015
  1. @larskanis

    Fix test failure when Encoding.default_external is not UTF-8.

    larskanis committed
    This fixes a test failure on Windows, that was caused
    by loading a HTML document per File.read(). This function
    returns a string in some cpXXX encoding on Windows, which
    was used as the document encoding later on, so that the test
    failed.
  2. @larskanis
  3. @flavorjones
  4. @jvshahid
Commits on Dec 17, 2015
  1. @tenderlove

    remove Japanese translation

    tenderlove committed
    we don't have time to translate this, and nobody seems to be interested
    in doing the work.  Lets rm it now.  We can bring it back if we find
    translation volunteers.
  2. @flavorjones

    update changelogs

    flavorjones committed
  3. @flavorjones

    version bump to 1.6.8.rc1

    flavorjones committed
  4. @flavorjones
  5. @flavorjones

    Ensure SyntaxError on empty XML in strict mode

    flavorjones committed
    Previously raised RuntimeError in MRI for zero-length docs.
    
    Fixes #1349. Related to #1005 and 2c9b7f1
  6. @flavorjones

    update changelogs

    flavorjones committed
  7. @flavorjones
  8. @twalpole @flavorjones
  9. @twalpole @flavorjones
  10. @flavorjones

    update changelogs

    flavorjones committed
  11. @flavorjones
  12. @larskanis @flavorjones

    Revert capturing of errors while xmlCopyDoc()/xmlDocCopyNode() and sy…

    larskanis committed with flavorjones
    …nc behavior with JRuby.
    
    This was introduced due to #1196
    and #1208 .
    
    However it turned out, that the change in libxml-2.9.2 was a regression, that was
    fixed in: https://bugzilla.gnome.org/show_bug.cgi?id=737840 and libxml-2.9.3.
    
    If I read the libxml sources right, it seems, that xmlDocCopyNode() is not
    intended to emit any such warnings at all. Only errors leading to a failure
    of the function are emitted. However these errors should be reported to ruby
    space in the form of exceptions (this is not yet implemented - currently either
    nil is returned or a generic error text is raised).
    
    This patch also synchronizes the behavior on MRI to that of JRuby, so that
    the error list is filled from the parser only and that it is shared after
    Document#dup .
  13. @larskanis @flavorjones
  14. @larskanis @flavorjones

    Revert silencing of dup errors.

    larskanis committed with flavorjones
    This reverts part of commit e4d587e
  15. @larskanis @flavorjones

    [MRI] Update to libxml-2.9.3.

    larskanis committed with flavorjones
    The new release fixes no less than 10 CVE assigned security issues,
    so the recommended libxml version is increased as well.
    
    The patch to config.guess is removed, because libxml-2.9.3 ships
    a reasonably new version of this file (2015-01-01).
    
    See https://mail.gnome.org/archives/xml/2015-November/msg00012.html
  16. @flavorjones
  17. @flavorjones
  18. @flavorjones

    Fix memory management of xmlNs xpath results

    flavorjones committed
    this commit introduces a bifurcation in how Namespace objects are
    managed; Namespace objects wrapped around xmlNs structs from an xpath
    query now have their own Ruby object lifecycle and are GCed
    independently from their original document.
    
    See comments in xml_node_set.c and xml_namespace.c for more details.
    
    Related to #1155
  19. @flavorjones

    normalize how we remove nodes from a node set

    flavorjones committed
    previously we had one code path (in #minus) that might try to free xmlNs
    structs in the xmlNodeset.
  20. @flavorjones

    DRY up type checks in xml_node_set.c

    flavorjones committed
    introduce Check_Node_Set_Node_Type to replace repeated logic.
  21. @flavorjones

    Nokogiri_wrap_xml_node_set handles NULL

    flavorjones committed
    This allows us to move repeated NULL checks out of callers.
  22. @flavorjones

    Reverting a portion of 2675a75

    flavorjones committed
    to remove nokogiriNodeSetTuple which is no longer necessary
    
    Related to #1155
Something went wrong with that request. Please try again.