Throw a validation report as a single message

Collapse a variable number of validation reports into a single message that can be formatted for the CLI or GitHub.

Usage

throw_heading_warnings(VAL)

throw_div_warnings(VAL)

throw_link_warnings(VAL)

Arguments

VAL: [data.frame] a validation report derived from one of the validate functions.

Value

NULL, invisibly. This is used for it's side-effect of formatting and issuing messages via issue_warning().

Details

One of the key features of pegboard is the ability to parse and validate markdown elements. These functions provide a standard way of creating the reports that are for the user based on whether or not they are on the CLI or on GitHub. The prerequisites of these functions are the input data frame (generated from the actual validation function) and an internal set of known templating vectors that contain templates for each test to show the actual error along with general information that can help correct the error (see below).

Input Data Frame

The validations are initially reported in a data frame that has the following properties:

one row per element
columns that indicate the parsed attributes of the element, source file, source position, and the actual element XML node object.
boolean columns that indicate the tests for each element, used with collect_labels() to add a "labels" column to the data.

Templating vectors

These vectors come in two forms [thing]_tests and [thing]_info (e.g. for validate_links(), we have link_tests and link_info). These are named vectors that match the boolean columns of the data frame produced by the validation function. The [thing]_tests vector contains templates that describes the error and shows the text that caused the error. The [thing]_info contains general information about how to address that particular error. For example, one common link error is that a link is not descriptive (e.g. the link text says "click here"). The column in the VAL data frame that contains the result of this test is called "descriptive", so if we look at the values from the link info and tests vectors:

link_info["descriptive"]
#>                                                                                          descriptive 
#> "Avoid uninformative link phrases <https://webaim.org/techniques/hypertext/link_text#uninformative>"
link_tests["descriptive"]
#>                                   descriptive 
#> "[uninformative link text]: [{text}]({orig})"

If the throw_*_warnings() functions detect any errors, they will use the info and tests vectors to construct a composite message.

Process

The throw_*_warnings() functions all do the same basic procedure (and indeed could be consolidated into a single function in the future)

pass data to collect_labels(), which will parse the [thing]_tests templating vector and label each failing element in VAL with the appropriate failure message
gather the source information for each failure
pass failures with the [thing]_info elements that matched the unique failures to issue_warning()