fixed current unit tests for importing individual pages. still needs more coverage, but at least all current test stuff is greenlight

fixed all tests for the UI. Automated testing can now import sample content unattended and successfully

documentation repair. A number of form UI text help blocks got lost in the conversion from #value to #markup elements. Made all text blobs inserted into forms to be #type=markup explicitly, not magically

Looks like that error must have been around for a while - it would DELETE menu items on the SECOND time being imported, because menu.module required it to be flagged as 'enabled' every time we save ... byt does not flag it as enabled when loaded, so I have to re-iterate that it's on every time. This was making items mysteriously go away even if they worked the first time. This was horrid. Testing was only ever on clean runs - not repeats

A pattern like

  //xhtml:div[@id = "block-menu_block-1"]//xhtml:li[contains(@class, "active active-trail")]//text()

Will mean that the text of the menu item found in a site menu will be used when creating menu items,

This is needed for cases when the page title is too long and cumbersome, and the url path is not useful enough.
Normally it falls bake to a version of the url path when the title is too damn long. However they are often ugly.

This allows you to now very clearly specify which title on the page to use when creating nav.
Note this can't be added retroactively, as menu building only happens in the first import, it doesn't update if you change the rules later.

Standardize the use of IMPORT_HTML_MAX_LABEL_LENGTH and increased it a little. Still should look at making it configurable.

system_check_directory() is NOT what I should use to ceate my subdir. It locks things down with .htaccess and just is the wrong choice. Make the dir myself the normal way.

when rebuilding/importing a feature, the new setting didn't turn up on its own. flush the menu cache

Radically improved the behavior of the demo form. Shifted the result output back into page body not the help zone where it was just messed up.

Error-protection. Don't ask for MIME on remote URLs, ensure dblog is enabled for the results logging page, don't die if write-access to .htaccess is prevented

Added none as an option for the mapper, so the UI does not imply that an undefined source will be mapped to a defined target. Just a UI improvement

term support needs revisiting. removong a redundant function taxonomy_get_vocabulary_by_name now available in core

Added taxonomy tag support - taxonomy_term_reference - was not ported to D7 before. The old auto-tagging needs deprecating or revisiting.

need to adjust the weight of the pre-scraper so it runs before field tries to serialize the data. Made the promiscuous grab-all-tagged elements process optional. stopped the IU from adding too many empty replacement patterns