This project is not being developed. Feel free to fork!
gem install tablerize
Tablerize is a format for writing tables using YAML/JSON-compatible data structures, and Ruby code to convert it to HTML.
You can use it in Ruby...
require 'tablerize'
puts Tablerize.load_file(path).to_html
# or
puts Tablerize.load(yaml_string).to_html
# or
puts Tablerize.make_table(object_from_yaml_or_json).to_html
...or from the command line
tablerize path/to/yaml-table.yml [...]
Markdown is easy on the eyes. It helps you write formatted documents in plain text in a way that is meaningful even without rendering. One thing that the original Markdown specification doesn't support is tables. But many authors writing in Markdown want to write tables, so Markdown libraries have come up with various but similar ways of representing tables in Markdown.
Tables exist to help you line things up. But these Markdown tables force you to either line things up yourself, or deal with the unreadable results. Here's some a table from the Markdown Cheatsheet:
| Tables | Are | Cool |
| ------------- |:-------------:| -----:|
| col 3 is | right-aligned | $1600 |
| col 2 is | centered | $12 |
| zebra stripes | are neat | $1 |
This is pretty readable, but unless you know exactly what the table looks like beforehand, lining everything up takes a lot of time! But that's okay, you don't actually need to do that, you can just have this mess instead:
Markdown | Less | Pretty
--- | --- | ---
*Still* | `renders` | **nicely**
1 | 2 | 3
Why is this less pretty? Because, while tables exist to organize data, the data looks pretty disorganized here. This goes against the concept of having raw Markdown be readable!
Depending what Markdown library you're using, the examples above might not even render. Many flavors of Markdown tables exist, and accept table syntax in varying degrees of leniency. Not only does this make writing tables even harder ("which Markdown library am I using and does it support table feature X?"), but it suggests that these these libraries are just making concessions with the constraints of representing a table in this way. Tables take data, something that should be computer-readable, and make them human-readable. Unrendered Markdown tables, sadly, usually are neither.
Oh, and with this traditional syntax, you can pretty much forget about nesting tables.
Tablerize attempts to solve these problems. In its purest form, it is a specification of a human-readable representation of tables in YAML/JSON- compatible data. Since YAML is human-readable, so is Tablerize. This project also includes a Ruby library and command-line tool to convert this YAML-based format into HTML tables. It can be run with:
tablerize path/to/yaml-table.yml [...]
A complementary project, kramdown-tablerize, allows embedding of YAML tables into kramdown Markdown documents.
Here's example: Searching for "statistics" on Google news, I come across a Forbes article with a neat little table I want to type up. Let's try it now (examples/example-1.yml):
class: [statistics-table, nsa-surveillance-details]
cols:
- name: authority
- name: num_orders
- name: num_targets
data:
- class: table-header
authority: Legal Authority
num_orders: Annual Number of Orders
num_targets: Estimated Number of Targets Affected
- authority: |
__FISA Orders__
Based on probable cause
(Title I and III of FISA, Sections 703 and 704 of FISA)
num_orders: "1,167 orders"
num_targets: "1,144"
- authority: |
__Section 702__
of FISA
num_orders: "1 order"
num_targets: "89,138"
- authority: |
__FISA Pen Register/Trap and Trace__
(Title IV of FISA)
num_orders: "131 orders"
num_targets: "319"
Here's what it looks like as HTML, using a common Markdown stylesheet ("GitHub" on Mou/Macdown):
Here's an example that illustrates some of the more advanced features of Tablerize (examples/example-2.yml):
class: [http-spec-exchange, another-class] # this line is optional
cols: # column specifications and ordering
- name: k # is used to identify the column below
class: http-key # is applied to each cell (td) in the column
- name: v
class: http-value
data: # data, by row
- v: GET # v corresponds to cols.1.name above
k: Method # k corresponds to cols.0.name above
- k: Parameters # the order of the columns in data doesn't matter
v:
# nest tables by nesting another YAML dictionary, in the same format
class: http-spec-params
cols:
- name: k
- name: v
data:
- k: '`client_id`' # backticks must be quoted!
v: |
A client ID for your service as set in your configuration.
a new line, wow! Let's see regular Markdown tables do that...
# <p>...</p> gets inserted only if there are multiple paragraphs
- k: '`type`'
v: '`code`'
- k: '`state`'
v: An anti-forgery token provided by the API.
- k: redirect_uri
v: '`https://example.com/api/{{your_service}}/authorize`'
With the right CSS, it becomes this:
-
YAML tip: Backticks
`
and some other characters need to be quoted because they have special meaing in YAML or are otherwise not allowed to be unquoted by YAML. Other suspicious characters include commas,
, ampersands&
, and asterisks*
and have been implicated in similar crimes. If readability isn't an issue and there's been a syntax error spree in your area, you can go ahead and quote every string just be safe. -
Don't use
class
as a column name, since it is used for classes. Unless you really want to. In which case you can. But it still will be used for classes. -
Auto column classes: As a convenience, if a table has a class
my-table
and a column is namedcolumn-1
, then all the cells in the column will have the classmy-table-column-1
. This only happens for the first table class listed. If you don't want this to happen, make the first table classnull
; it will be ignored and columns will not have automatically-generated classes.
-
Support using representing two-column tables as key and value. YAML doesn't support ordered dictionaries, so this will be done by looking at the only key-value pair inside each dictionary in a list:
data: - wake up: done - brush teeth: done - eat breakfast: not done
-
Allow HTML attributes to be placed anywhere classes currently can.
-
Allow arbitrary HTML and tables as siblings together inside cells. This will probably be implemented by recursively calling kramdown-tablerize, if installed.
-
Allow empty cells to be created by simply not including keys or making keys with no value (represented as
null
/nil
/None
). Possibly add a "strict mode" setting that causes these to error instead. -
Support
thead
andtfoot
. Perhaps add supporttbody
, which will be a synonym fordata
. -
For simple tables, allow outputting to Markdown, for GitHub and other sites that don't allow HTML in Markdown.
-
textmate/yaml.tmbundle, the YAML syntax highlighter used by TextMate, Sublime Text, and GitHub isn't perfect. Fix the plugin and use the examples in this README as test cases. Update: GitHub now highlights the YAML in this file almost correctly, but textmate/yaml.tmbundle doesn't seem to be updated. GitHub's probably using something else to do syntax highlighting.
Tablerize was originally designed and written by @szhu at @IFTTT.