Here are some notes on the new item refactoring:

I'm writing some general-purpose routines to support managing content from multiple locations, in multiple formats.

location [edit section]

A tbwiki "item" is a logical unit of the wiki, and can be one of the following:

• page
• block
• section
• row
• field

A page is a whole file. A block is an item inside a file, defined by the triple-brace syntax. A section is an item inside a file, defined by the header syntax. A row is something that can be inside a page, block, or section. This is used to express moin table rows. A field can be a single-line entry appearing on a page, and defined by the definition list syntax.

content_format [edit section]

Each item is expressed in the raw text in a format:

• tbmarkup - this is page data, and may hold sub-items, but does not itself express a record
• moin_table - this is a set of rows of moin-formatted table lines (starting with ||)
• tbattr - this is a single record in my 'config' format (supporting single and multi-line entries, and comments
• tbwikidb - this represents data attributes with definition entries and sections (in tbmarkup)
• json - usually whole-page, supports single, multi-line and list entries
• yaml - usually whole-page, supports single, multi-line and list entries, as well as comments and ordering
• moin_table_row - this defines a single record in a moin_table

item_class [edit section]

A new set of routines is supposed to replace all previous access and update routines for content, and allow different systems to access data items in the wiki in a uniform manner. The different systems that need to access the items are:

• processors
• tables
• data handler (for tbclient)
• other external engines (Fuego)

The idea is that items are addressable with a uniform resource string.

item_class routines [edit section]

Here are the functions supported by the new item_class:

• __init__(name, location, source_file, parent)
• get_content() - load the content from it's source
• evalute_content() - execute processor, if any, for the content
• to_html() - convert item to html (for output)
• content_to_data() - convert content into data (parse it)
• data_to_content() - convert data into content (format it)
• update() - update one element of the data with a new value
• set_content() - push this content to parent (container item)
• write_content() - put content back on disk

internal helper routines:

• data_to_html() - convert data representation to html

example uses [edit section]

• in main page wiki:
  • show:
    • page = item_class(req.page_name, "page", req.page_filename)
    • html = page.to_html()
  • in section edit_form:
    • page = item_class(req.page_name, "page", req.page_filename)
    • section = item_class(section_name, "section", req.page_filename, page)
    • put section.content into edit form
  • in section_edit action:
    • page = item_class(req.page_name, "page", req.page_filename)
    • section = item_class(section_name, "section", req.page_filename, page)
    • section.update(value_from_form)
    • section.set_content()
    • page.write_content()
  • in Processor:
    • parse_conf(block)
  • in data_handler:
    • get_page_item
    • put_page_item
  • in table:
    • for source in source_list:
      • get_record(source)
        • page = item_class(req.page_name, "page", req.page_filename)
        • record = make_record(page)
          • data = page.content_to_data()
          • add field stuff
      • format for table