Update some last docs, bnf grammars

18 files changed, 226 insertions, 113 deletions

diff --git a/readme.md b/readme.md
index e7fceab..bad92da 100644
--- a/readme.md
+++ b/readme.md
@@ -7,8 +7,6 @@ Crate docs are currently at
 
 ### Docs
 
-- [ ] (1) Go through all docs
-- [ ] (1) Go through all bnf docs
 - [ ] (1) Add overview docs on how everything works
 - [ ] (1) Add more examples
 
diff --git a/src/construct/autolink.rs b/src/construct/autolink.rs
index 9890aaf..b25dc32 100644
--- a/src/construct/autolink.rs
+++ b/src/construct/autolink.rs
@@ -1,8 +1,8 @@
-//! Autolinks occur in the [text][] content type.
+//! Autolink occurs in the [text][] content type.
 //!
 //! ## Grammar
 //!
-//! Autolinks form with the following BNF
+//! Autolink forms with the following BNF
 //! (<small>see [construct][crate::construct] for character groups</small>):
 //!
 //! ```bnf
diff --git a/src/construct/label_end.rs b/src/construct/label_end.rs
index 4752639..5e31444 100644
--- a/src/construct/label_end.rs
+++ b/src/construct/label_end.rs
@@ -1,11 +1,14 @@
-//! Label end is a construct that occurs in the [text][] content type.
+//! Label end occurs in the [text][] content type.
 //!
-//! It forms with the following BNF:
+//! ## Grammar
+//!
+//! Label end forms with the following BNF
+//! (<small>see [construct][crate::construct] for character groups</small>):
 //!
 //! ```bnf
-//! label_end ::= ']' [ resource | reference_full | reference_collapsed ]
+//! label_end ::= ']' [resource | reference_full | reference_collapsed]
 //!
-//! resource ::= '(' [ whitespace ] destination [ whitespace title ] [ whitespace ] ')'
+//! resource ::= '(' [space_or_tab_eol] destination [space_or_tab_eol title] [space_or_tab_eol] ')'
 //! reference_full ::= '[' label ']'
 //! reference_collapsed ::= '[' ']'
 //!
@@ -14,15 +17,7 @@
 //! ```
 //!
 //! See [`destination`][destination], [`label`][label], and [`title`][title]
-//! for grammar, notes, and recommendations.
-//!
-//! Label end does not, on its own, relate to anything in HTML.
-//! When matched with a [label start (link)][label_start_link], they together
-//! relate to the `<a>` element in HTML.
-//! See [*§ 4.5.1 The `a` element*][html-a] in the HTML spec for more info.
-//! It can also match with [label start (image)][label_start_image], in which
-//! case they form an `<img>` element.
-//! See [*§ 4.8.3 The `img` element*][html-img] in the HTML spec for more info.
+//! for grammar, notes, and recommendations on each part.
 //!
 //! In the case of a resource, the destination and title are given directly
 //! with the label end.
@@ -49,7 +44,7 @@
 //!
 //! When the resource or reference matches, the destination forms the `href`
 //! attribute in case of a [label start (link)][label_start_link], and an
-//! `src` attribute otherwise.
+//! `src` attribute in case of a [label start (image)][label_start_image].
 //! The title is formed, optionally, on either `<a>` or `<img>`.
 //!
 //! For info on how to encode characters in URLs, see
@@ -94,12 +89,29 @@
 //! <p>a [b <a href="#">c</a> d](#) e</p>
 //! ```
 //!
-//! This limiation is imposed because links in links is invalid according to
+//! This limitation is imposed because links in links is invalid according to
 //! HTML.
 //! Technically though, in markdown it is still possible to construct them by
 //! using an [autolink][] in a link.
 //! You definitely should not do that.
 //!
+//! ## HTML
+//!
+//! Label end does not, on its own, relate to anything in HTML.
+//! When matched with a [label start (link)][label_start_link], they together
+//! relate to the `<a>` element in HTML.
+//! See [*§ 4.5.1 The `a` element*][html_a] in the HTML spec for more info.
+//! It can also match with [label start (image)][label_start_image], in which
+//! case they form an `<img>` element.
+//! See [*§ 4.8.3 The `img` element*][html_img] in the HTML spec for more info.
+//!
+//! ## Recommendation
+//!
+//! It is recommended to use labels instead of [autolinks][autolink].
+//! Labels allow more characters in URLs, and allow relative URLs and `www.`
+//! URLs.
+//! They also allow for descriptive text to explain the URL in prose.
+//!
 //! ## Tokens
 //!
 //! *   [`Data`][Name::Data]
@@ -143,8 +155,8 @@
 //! [autolink]: crate::construct::autolink
 //! [sanitize_uri]: crate::util::sanitize_uri::sanitize_uri
 //! [normalize_identifier]: crate::util::normalize_identifier::normalize_identifier
-//! [html-a]: https://html.spec.whatwg.org/multipage/text-level-semantics.html#the-a-element
-//! [html-img]: https://html.spec.whatwg.org/multipage/embedded-content.html#the-img-element
+//! [html_a]: https://html.spec.whatwg.org/multipage/text-level-semantics.html#the-a-element
+//! [html_img]: https://html.spec.whatwg.org/multipage/embedded-content.html#the-img-element
 
 use crate::constant::RESOURCE_DESTINATION_BALANCE_MAX;
 use crate::construct::partial_space_or_tab_eol::space_or_tab_eol;
diff --git a/src/construct/label_start_image.rs b/src/construct/label_start_image.rs
index ffc1aee..8d35df2 100644
--- a/src/construct/label_start_image.rs
+++ b/src/construct/label_start_image.rs
@@ -1,16 +1,20 @@
-//! Label start (image) is a construct that occurs in the [text][] content
-//! type.
+//! Label start (image) occurs in the [text][] content type.
 //!
-//! It forms with the following BNF:
+//! ## Grammar
+//!
+//! Label start (image) forms with the following BNF
+//! (<small>see [construct][crate::construct] for character groups</small>):
 //!
 //! ```bnf
 //! label_start_image ::= '!' '['
 //! ```
 //!
+//! ## HTML
+//!
 //! Label start (image) does not, on its own, relate to anything in HTML.
 //! When matched with a [label end][label_end], they together relate to the
 //! `<img>` element in HTML.
-//! See [*§ 4.8.3 The `img` element*][html-img] in the HTML spec for more info.
+//! See [*§ 4.8.3 The `img` element*][html_img] in the HTML spec for more info.
 //! Without an end, the characters (`![`) are output.
 //!
 //! ## Tokens
@@ -26,7 +30,7 @@
 //!
 //! [text]: crate::construct::text
 //! [label_end]: crate::construct::label_end
-//! [html-img]: https://html.spec.whatwg.org/multipage/embedded-content.html#the-img-element
+//! [html_img]: https://html.spec.whatwg.org/multipage/embedded-content.html#the-img-element
 
 use crate::event::Name;
 use crate::resolve::Name as ResolveName;
diff --git a/src/construct/label_start_link.rs b/src/construct/label_start_link.rs
index dad6884..e079b2d 100644
--- a/src/construct/label_start_link.rs
+++ b/src/construct/label_start_link.rs
@@ -1,17 +1,21 @@
-//! Label start (link) is a construct that occurs in the [text][] content
-//! type.
+//! Label start (link) occurs in the [text][] content type.
 //!
-//! It forms with the following BNF:
+//! ## Grammar
+//!
+//! Label start (link) forms with the following BNF
+//! (<small>see [construct][crate::construct] for character groups</small>):
 //!
 //! ```bnf
 //! label_start_link ::= '['
 //! ```
 //!
+//! ## HTML
+//!
 //! Label start (link) does not, on its own, relate to anything in HTML.
 //! When matched with a [label end][label_end], they together relate to the
 //! `<a>` element in HTML.
-//! See [*§ 4.5.1 The `a` element*][html-a] in the HTML spec for more info.
-//! Without an end, the characters (`[`) are output.
+//! See [*§ 4.5.1 The `a` element*][html_a] in the HTML spec for more info.
+//! Without an end, the character (`[`) is output.
 //!
 //! ## Tokens
 //!
@@ -25,7 +29,7 @@
 //!
 //! [text]: crate::construct::text
 //! [label_end]: crate::construct::label_end
-//! [html-a]: https://html.spec.whatwg.org/multipage/text-level-semantics.html#the-a-element
+//! [html_a]: https://html.spec.whatwg.org/multipage/text-level-semantics.html#the-a-element
 
 use crate::event::Name;
 use crate::resolve::Name as ResolveName;
diff --git a/src/construct/list_item.rs b/src/construct/list_item.rs
index 09678dd..3e632fb 100644
--- a/src/construct/list_item.rs
+++ b/src/construct/list_item.rs
@@ -1,27 +1,41 @@
-//! List item is a construct that occurs in the [document][] content type.
+//! List item occurs in the [document][] content type.
 //!
-//! It forms with, roughly, the following BNF:
+//! ## Grammar
+//!
+//! List item forms with the following BNF
+//! (<small>see [construct][crate::construct] for character groups</small>):
 //!
 //! ```bnf
-//! ; Restriction: there must be `eol | space_or_tab` after the start.
+//! ; Restriction: if there is no space after the marker, the start must be followed by an `eol`.
 //! ; Restriction: if the first line after the marker is not blank and starts with `5( space_or_tab )`,
 //! ; only the first `space_or_tab` is part of the start.
-//! list_item_start ::= '*' | '+' | '-' | 1*9( ascii_decimal ) ( '.' | ')' ) [ 1*4 space_or_tab ]
+//! list_item_start ::= '*' | '+' | '-' | 1*9(ascii_decimal) ('.' | ')') [1*4 space_or_tab]
+//!
 //! ; Restriction: blank line allowed, except when this is the first continuation after a blank start.
 //! ; Restriction: if not blank, the line must be indented, exactly `n` times.
-//! list_item_cont ::= [ n( space_or_tab ) ]
+//! list_item_cont ::= [n(space_or_tab)]
 //! ```
 //!
-//! Further lines that are not prefixed with `list_item_cont` cause the item
-//! to be exited, except when those lines are lazy continuation.
-//! Like so many things in markdown, list (items) too, are very complex.
-//! See [*§ Phase 1: block structure*][commonmark-block] for more on parsing
-//! details.
+//! Further lines that are not prefixed with `list_item_cont` cause the list
+//! item to be exited, except when those lines are lazy continuation.
+//! Like so many things in markdown, list items too, are complex.
+//! See [*§ Phase 1: block structure* in `CommonMark`][commonmark_block] for
+//! more on parsing details.
+//!
+//! As list item is a container, it takes several bytes from the start of the
+//! line, while the rest of the line includes more containers or flow.
+//!
+//! ## HTML
+//!
+//! List item relates to the `<li>`, `<ol>`, and `<ul>` elements in HTML.
+//! See [*§ 4.4.8 The `li` element*][html_li],
+//! [*§ 4.4.5 The `ol` element*][html_ol], and
+//! [*§ 4.4.7 The `ul` element*][html_ul] in the HTML spec for more info.
+//!
+//! ## Recommendation
 //!
-//! Lists relates to the `<li>`, `<ol>`, and `<ul>` elements in HTML.
-//! See [*§ 4.4.8 The `li` element*][html-li],
-//! [*§ 4.4.5 The `ol` element*][html-ol], and
-//! [*§ 4.4.7 The `ul` element*][html-ul] in the HTML spec for more info.
+//! Use a single space after a marker.
+//! Never use lazy continuation.
 //!
 //! ## Tokens
 //!
@@ -39,10 +53,10 @@
 //! *   [*§ 5.3 Lists* in `CommonMark`](https://spec.commonmark.org/0.30/#lists)
 //!
 //! [document]: crate::construct::document
-//! [html-li]: https://html.spec.whatwg.org/multipage/grouping-content.html#the-li-element
-//! [html-ol]: https://html.spec.whatwg.org/multipage/grouping-content.html#the-ol-element
-//! [html-ul]: https://html.spec.whatwg.org/multipage/grouping-content.html#the-ul-element
-//! [commonmark-block]: https://spec.commonmark.org/0.30/#phase-1-block-structure
+//! [html_li]: https://html.spec.whatwg.org/multipage/grouping-content.html#the-li-element
+//! [html_ol]: https://html.spec.whatwg.org/multipage/grouping-content.html#the-ol-element
+//! [html_ul]: https://html.spec.whatwg.org/multipage/grouping-content.html#the-ul-element
+//! [commonmark_block]: https://spec.commonmark.org/0.30/#phase-1-block-structure
 
 use crate::constant::{LIST_ITEM_VALUE_SIZE_MAX, TAB_SIZE};
 use crate::construct::partial_space_or_tab::space_or_tab_min_max;
diff --git a/src/construct/mod.rs b/src/construct/mod.rs
index da2f5e8..5630143 100644
--- a/src/construct/mod.rs
+++ b/src/construct/mod.rs
@@ -75,12 +75,15 @@
 //! That is to say, they form illustrative, imperfect, but useful, examples.
 //! The code, in Rust, is considered to be normative.
 //!
-//! For example, the docs for [character escape][character_escape] contain:
+//! The actual syntax of markdown can be described in Backus–Naur form (BNF) as:
 //!
 //! ```bnf
-//! character_escape ::= '\\' ascii_punctuation
+//! markdown = .*
 //! ```
 //!
+//! No, that’s [not a typo][bnf]: markdown has no syntax errors; anything
+//! thrown at it renders *something*.
+//!
 //! These diagrams contain references to character group as defined by Rust on
 //! for example [char][], but also often on [u8][], which is what `micromark-rs`
 //! typically works on.
@@ -99,14 +102,14 @@
 //! ascii_alphabetic ::= ascii_lowercase | ascii_uppercase
 //! ; '0'..='9'
 //! ascii_digit ::= '0' | '1' | '2' | '3' | '4' | '5' | '6' | '7' | '8' | '9'
-//! ; '0'..='9'; 'A'..='F', 'a'..='f'
+//! ; '0'..='9', 'A'..='F', 'a'..='f'
 //! ascii_hexdigit ::= ascii_digit | 'a' | 'b' | 'c' | 'd' | 'e' | 'f' | 'A' | 'B' | 'C' | 'D' | 'E' | 'F'
-//! ; '0'..='9'; 'A'..='Z', 'a'..='z'
+//! ; '0'..='9', 'A'..='Z', 'a'..='z'
 //! ascii_alphanumeric ::= ascii_digit | ascii_alphabetic
-//! ; '!'..='/'; ':'..='@'; '['..='`'; '{'..='~'
+//! ; '!'..='/', ':'..='@', '['..='`', '{'..='~'
 //! ascii_punctuation ::= '!' | '"' | '#' | '$' | '%' | '&' | '\'' | '(' | ')' | '*' | '+' | ',' | '-' | '.' | '/' | ':' | ';' | '<' | '=' | '>' | '?' | '@' | '[' | '\' | ']' | '^' | '_' | '`' | '{' | '|' | '}' | '~'
-//! ; 0x00..=0x1F; 0x7F
-//! ascii_control ::= 0x00 | 0x01 | 0x02 | 0x03 | 0x04 | 0x05 | 0x06 | 0x07 | 0x08 | 0x09 | 0x0A | 0x0B | 0x0C | 0x0D | 0x0E | 0x0F | 0x10 | 0x12 | 0x13 | 0x14 | 0x15 | 0x16 | 0x17 | 0x18 | 0x19 | 0x1A | 0x1B | 0x1C | 0x1D | 0x1E | 0x1F | 0x7F
+//! ; 0x00..=0x1F, 0x7F
+//! ascii_control ::= 0x00 | 0x01 | 0x02 | 0x03 | 0x04 | 0x05 | 0x06 | 0x07 | 0x08 | 0x09 | 0x0A | 0x0B | 0x0C | 0x0D | 0x0E | 0x0F | 0x10 | 0x11 | 0x12 | 0x13 | 0x14 | 0x15 | 0x16 | 0x17 | 0x18 | 0x19 | 0x1A | 0x1B | 0x1C | 0x1D | 0x1E | 0x1F | 0x7F
 //!
 //! ; Markdown groups:
 //! ; Any byte (u8)
@@ -115,12 +118,14 @@
 //! eol ::= '\n' | '\r' | '\r\n'
 //! line ::= byte - eol
 //! text ::= line - space_or_tab
-//! space_or_tab_eol ::= 1*space_or_tab | 0*space_or_tab eol 0*space_or_tab
+//! space_or_tab_eol ::= 1*space_or_tab | *space_or_tab eol *space_or_tab
 //!
 //! ; Unicode groups:
 //! unicode_whitespace ::= ? ; See `char::is_whitespace`.
 //! unicode_punctuation ::= ? ; See `src/unicode.rs`.
 //! ```
+//!
+//! [bnf]: http://trevorjim.com/a-specification-for-markdown/
 
 pub mod attention;
 pub mod autolink;
diff --git a/src/construct/paragraph.rs b/src/construct/paragraph.rs
index 5f89ef9..0ddd0c4 100644
--- a/src/construct/paragraph.rs
+++ b/src/construct/paragraph.rs
@@ -1,15 +1,18 @@
-//! Paragraph is a construct that occurs in the [flow] content type.
+//! Paragraph occurs in the [flow][] content type.
 //!
-//! They’re formed with the following BNF:
+//! ## Grammar
+//!
+//! Paragraph forms with the following BNF
+//! (<small>see [construct][crate::construct] for character groups</small>):
 //!
 //! ```bnf
 //! ; Restriction: lines cannot start other flow constructs.
 //! ; Restriction: lines cannot be blank.
-//! paragraph ::= 1*line *( eol 1*line )
+//! paragraph ::= 1*line *(eol 1*line)
 //! ```
 //!
-//! Paragraphs in markdown relate to the `<p>` element in HTML.
-//! See [*§ 4.4.1 The `p` element* in the HTML spec][html] for more info.
+//! As this construct occurs in flow, like all flow constructs, it must be
+//! followed by an eol (line ending) or eof (end of file).
 //!
 //! Paragraphs can contain line endings and whitespace, but they are not
 //! allowed to contain blank lines, or to be blank themselves.
@@ -17,6 +20,11 @@
 //! The paragraph is interpreted as the [text][] content type.
 //! That means that [autolinks][autolink], [code (text)][code_text], etc are allowed.
 //!
+//! ## HTML
+//!
+//! Paragraphs in markdown relate to the `<p>` element in HTML.
+//! See [*§ 4.4.1 The `p` element* in the HTML spec][html] for more info.
+//!
 //! ## Tokens
 //!
 //! *   [`Paragraph`][Name::Paragraph]
diff --git a/src/construct/partial_bom.rs b/src/construct/partial_bom.rs
index 1818ef4..3f772e9 100644
--- a/src/construct/partial_bom.rs
+++ b/src/construct/partial_bom.rs
@@ -1,6 +1,17 @@
 //! Byte order mark occurs at the start of the document.
 //!
-//! It’s the three bytes 0xEF, 0xBB, and 0xBF.
+//! ## Grammar
+//!
+//! Byte order mark forms with the following BNF
+//! (<small>see [construct][crate::construct] for character groups</small>):
+//!
+//! ```bnf
+//! byte_order_mark ::= 0xEF 0xBB 0xBF
+//! ```
+//!
+//! ## Recommendation
+//!
+//! Don’t use BOMs.
 //!
 //! ## Tokens
 //!
diff --git a/src/construct/partial_data.rs b/src/construct/partial_data.rs
index adbfae1..19f5f3b 100644
--- a/src/construct/partial_data.rs
+++ b/src/construct/partial_data.rs
@@ -1,4 +1,4 @@
-//! Data occurs in [text][] and [string][].
+//! Data occurs in the [string][] and [text][] content types.
 //!
 //! It can include anything (except for line endings) and stops at certain
 //! characters.
diff --git a/src/construct/partial_destination.rs b/src/construct/partial_destination.rs
index b5254d3..99b5c5a 100644
--- a/src/construct/partial_destination.rs
+++ b/src/construct/partial_destination.rs
@@ -1,17 +1,21 @@
 //! Destination occurs in [definition][] and [label end][label_end].
 //!
-//! They’re formed with the following BNF:
+//! ## Grammar
+//!
+//! Destination forms with the following BNF
+//! (<small>see [construct][crate::construct] for character groups</small>):
 //!
 //! ```bnf
 //! destination ::= destination_enclosed | destination_raw
 //!
-//! destination_enclosed ::= '<' *( destination_enclosed_text | destination_enclosed_escape ) '>'
-//! destination_enclosed_text ::= code - '<' - '\\' - '>' - eol
-//! destination_enclosed_escape ::= '\\' [ '<' | '\\' | '>' ]
-//! destination_raw ::= 1*( destination_raw_text | destination_raw_escape )
+//! destination_enclosed ::= '<' *(destination_enclosed_byte | destination_enclosed_escape) '>'
+//! destination_enclosed_byte ::= line - '<' - '\\' - '>'
+//! destination_enclosed_escape ::= '\\' ['<' | '\\' | '>']
+//!
+//! destination_raw ::= 1*(destination_raw_byte | destination_raw_escape)
 //! ; Restriction: unbalanced `)` characters are not allowed.
-//! destination_raw_text ::= code - '\\' - ascii_control - space_or_tab - eol
-//! destination_raw_escape ::= '\\' [ '(' | ')' | '\\' ]
+//! destination_raw_byte ::= text - '\\' - ascii_control
+//! destination_raw_escape ::= '\\' ['(' | ')' | '\\']
 //! ```
 //!
 //! Balanced parens allowed in raw destinations.
@@ -28,18 +32,19 @@
 //! The grammar for enclosed destinations (`<x>`) prohibits the use of `<`,
 //! `>`, and line endings to form URLs.
 //! The angle brackets can be encoded as a character reference, character
-//! escape, or percent encoding: for `<` as `&lt;`, `\<`, or `%3c` and for
-//! `>` as `&gt;`, `\>`, or `%3e`.
+//! escape, or percent encoding:
+//!
+//! *   `<` as `&lt;`, `\<`, or `%3c`
+//! *   `>` as `&gt;`, `\>`, or `%3e`
 //!
 //! The grammar for raw destinations (`x`) prohibits space (` `) and all
-//! [ASCII control][char::is_ascii_control] characters, which thus must be
+//! [ASCII control][u8::is_ascii_control] characters, which thus must be
 //! encoded.
-//! Unbalanced arens can be encoded as a character reference, character escape,
-//! or percent encoding: for `(` as `&lpar;`, `\(`, or `%28` and for `)` as
-//! `&rpar;`, `\)`, or `%29`.
+//! Unbalanced parens can be encoded as a character reference, character escape,
+//! or percent encoding:
 //!
-//! It is recommended to use the enclosed variant of destinations, as it allows
-//! the most characters, including arbitrary parens, in URLs.
+//! *   `(` as `&lpar;`, `\(`, or `%28`
+//! *   `)` as `&rpar;`, `\)`, or `%29`
 //!
 //! There are several cases where incorrect encoding of URLs would, in other
 //! languages, result in a parse error.
@@ -60,6 +65,11 @@
 //! <p><a href="https://a%F0%9F%91%8Db%25">x</a></p>
 //! ```
 //!
+//! ## Recommendation
+//!
+//! It is recommended to use the enclosed variant of destinations, as it allows
+//! the most characters, including arbitrary parens, in URLs.
+//!
 //! ## References
 //!
 //! *   [`micromark-factory-destination/index.js` in `micromark`](https://github.com/micromark/micromark/blob/main/packages/micromark-factory-destination/dev/index.js)
diff --git a/src/construct/partial_label.rs b/src/construct/partial_label.rs
index 00d613a..1c1dcec 100644
--- a/src/construct/partial_label.rs
+++ b/src/construct/partial_label.rs
@@ -1,14 +1,17 @@
-//! Label occurs in [definition][] and label end.
+//! Label occurs in [definition][] and [label end][label_end].
 //!
-//! They’re formed with the following BNF:
+//! ## Grammar
+//!
+//! Label forms with the following BNF
+//! (<small>see [construct][crate::construct] for character groups</small>):
 //!
 //! ```bnf
 //! ; Restriction: maximum `999` codes allowed between brackets.
 //! ; Restriction: no blank lines.
-//! ; Restriction: at least 1 non-space and non-eol code must exist.
-//! label ::= '[' *( label_text | label_escape ) ']'
-//! label_text ::= code - '[' - '\\' - ']'
-//! label_escape ::= '\\' [ '[' | '\\' | ']' ]
+//! ; Restriction: at least 1 `text` byte must exist.
+//! label ::= '[' *(label_byte | label_escape) ']'
+//! label_byte ::= code - '[' - '\\' - ']'
+//! label_escape ::= '\\' ['[' | '\\' | ']']
 //! ```
 //!
 //! The maximum allowed size of the label, without the brackets, is `999`
diff --git a/src/construct/partial_non_lazy_continuation.rs b/src/construct/partial_non_lazy_continuation.rs
index 5eec71f..35641ee 100644
--- a/src/construct/partial_non_lazy_continuation.rs
+++ b/src/construct/partial_non_lazy_continuation.rs
@@ -3,7 +3,7 @@
 //! This is a tiny helper that [flow][] constructs can use to make sure that
 //! the following line is not lazy.
 //! For example, [html (flow)][html_flow] and code ([fenced][code_fenced],
-//! [indented][code_indented]), stop when next line is lazy.
+//! [indented][code_indented]), stop when the next line is lazy.
 //!
 //! [flow]: crate::construct::flow
 //! [code_fenced]: crate::construct::code_fenced
diff --git a/src/construct/partial_space_or_tab.rs b/src/construct/partial_space_or_tab.rs
index 43cfd45..198dae2 100644
--- a/src/construct/partial_space_or_tab.rs
+++ b/src/construct/partial_space_or_tab.rs
@@ -1,4 +1,13 @@
-//! Several helpers to parse whitespace (`space_or_tab`, `space_or_tab_eol`).
+//! Space or tab occurs in tons of places.
+//!
+//! ## Grammar
+//!
+//! Space or tab forms with the following BNF
+//! (<small>see [construct][crate::construct] for character groups</small>):
+//!
+//! ```bnf
+//! space_or_tab ::= 1*('\t' | ' ')
+//! ```
 //!
 //! ## References
 //!
@@ -9,16 +18,16 @@ use crate::state::{Name as StateName, State};
 use crate::subtokenize::link;
 use crate::tokenizer::Tokenizer;
 
-/// Options to parse `space_or_tab`.
+/// Configuration.
 #[derive(Debug)]
 pub struct Options {
     /// Minimum allowed bytes (inclusive).
     pub min: usize,
     /// Maximum allowed bytes (inclusive).
     pub max: usize,
-    /// Token type to use for whitespace events.
+    /// Name to use for events.
     pub kind: Name,
-    /// Connect this whitespace to the previous.
+    /// Connect this event to the previous.
     pub connect: bool,
     /// Embedded content type to use.
     pub content: Option<Content>,
diff --git a/src/construct/partial_space_or_tab_eol.rs b/src/construct/partial_space_or_tab_eol.rs
index 427cb11..01f440e 100644
--- a/src/construct/partial_space_or_tab_eol.rs
+++ b/src/construct/partial_space_or_tab_eol.rs
@@ -1,8 +1,23 @@
-//! Several helpers to parse whitespace (`space_or_tab`, `space_or_tab_eol`).
+//! Space or tab (eol) occurs in [destination][], [label][], and [title][].
+//!
+//! ## Grammar
+//!
+//! Space or tab (eol) forms with the following BNF
+//! (<small>see [construct][crate::construct] for character groups</small>):
+//!
+//! ```bnf
+//! space_or_tab_eol ::= 1*space_or_tab | *space_or_tab eol *space_or_tab
+//! ```
+//!
+//! Importantly, this allows one line ending, but not blank lines.
 //!
 //! ## References
 //!
 //! *   [`micromark-factory-space/index.js` in `micromark`](https://github.com/micromark/micromark/blob/main/packages/micromark-factory-space/dev/index.js)
+//!
+//! [destination]: crate::construct::partial_destination
+//! [label]: crate::construct::partial_label
+//! [title]: crate::construct::partial_title
 
 use crate::construct::partial_space_or_tab::{
     space_or_tab_with_options, Options as SpaceOrTabOptions,
@@ -12,7 +27,7 @@ use crate::state::{Name as StateName, State};
 use crate::subtokenize::link;
 use crate::tokenizer::Tokenizer;
 
-/// Options to parse `space_or_tab` and one optional eol, but no blank line.
+/// Configuration.
 #[derive(Debug)]
 pub struct Options {
     /// Connect this whitespace to the previous.
@@ -21,12 +36,7 @@ pub struct Options {
     pub content: Option<Content>,
 }
 
-/// `space_or_tab`, or optionally `space_or_tab`, one `eol`, and
-/// optionally `space_or_tab`.
-///
-/// ```bnf
-/// space_or_tab_eol ::= 1*( ' ' '\t' ) | 0*( ' ' '\t' ) eol 0*( ' ' '\t' )
-/// ```
+/// `space_or_tab_eol`
 pub fn space_or_tab_eol(tokenizer: &mut Tokenizer) -> StateName {
     space_or_tab_eol_with_options(
         tokenizer,
diff --git a/src/construct/partial_title.rs b/src/construct/partial_title.rs
index 50d26be..a878be3 100644
--- a/src/construct/partial_title.rs
+++ b/src/construct/partial_title.rs
@@ -1,11 +1,16 @@
 //! Title occurs in [definition][] and [label end][label_end].
 //!
-//! They’re formed with the following BNF:
+//! ## Grammar
+//!
+//! Title forms with the following BNF
+//! (<small>see [construct][crate::construct] for character groups</small>):
 //!
 //! ```bnf
 //! ; Restriction: no blank lines.
 //! ; Restriction: markers must match (in case of `(` with `)`).
-//! title ::= marker [  *( code - '\\' | '\\' [ marker ] ) ] marker
+//! title ::= marker *(title_byte | title_escape) marker
+//! title_byte ::= code - '\\' - marker
+//! title_escape ::= '\\' ['\\' | marker]
 //! marker ::= '"' | '\'' | '('
 //! ```
 //!
diff --git a/src/construct/partial_whitespace.rs b/src/construct/partial_whitespace.rs
index 04016cb..faaff73 100644
--- a/src/construct/partial_whitespace.rs
+++ b/src/construct/partial_whitespace.rs
@@ -1,25 +1,25 @@
 //! Trailing whitespace occurs in [string][] and [text][].
 //!
-//! It occurs around line endings, and, in the case of text content it also
-//! occurs at the start or end of the whole.
+//! ## Grammar
 //!
-//! They’re formed with the following BNF:
+//! Trailing whitespace forms with the following BNF
+//! (<small>see [construct][crate::construct] for character groups</small>):
 //!
 //! ```bnf
 //! ; Restriction: the start and end here count as an eol in the case of `text`.
-//! whitespace ::= 0.*space_or_tab eol 0.*space_or_tab
+//! whitespace ::= *space_or_tab eol *space_or_tab
 //! ```
 //!
+//! It occurs around line endings and, in the case of text content, it also
+//! occurs at the start or end of the whole.
+//!
 //! Normally this whitespace is ignored.
 //! In the case of text content, whitespace before a line ending that
 //! consistents solely of spaces, at least 2, forms a hard break (trailing).
 //!
-//! The minimum number of the spaces is defined in
+//! The minimum number of those spaces is defined in
 //! [`HARD_BREAK_PREFIX_SIZE_MIN`][hard_break_prefix_size_min].
 //!
-//! Hard breaks in markdown relate to the HTML element `<br>`.
-//! See [*§ 4.5.27 The `br` element* in the HTML spec][html] for more info.
-//!
 //! It is also possible to create a hard break with a similar construct: a
 //! [hard break (escape)][hard_break_escape] is a backslash followed
 //! by a line ending.
@@ -28,6 +28,18 @@
 //! “escaped” in other languages.
 //! Trailing spaces are typically invisible in editors, or even automatically
 //! removed, making hard break (trailing) hard to use.
+//!
+//! ## HTML
+//!
+//! Hard breaks in markdown relate to the HTML element `<br>`.
+//! See [*§ 4.5.27 The `br` element* in the HTML spec][html] for more info.
+//!
+//! ## Recommendation
+//!
+//! Do not use trailing whitespace.
+//! It is never needed when using [hard break (escape)][hard_break_escape]
+//! to create hard breaks.
+//!
 //! ## Tokens
 //!
 //! *   [`HardBreakTrailing`][Name::HardBreakTrailing]
diff --git a/src/construct/thematic_break.rs b/src/construct/thematic_break.rs
index 9496d98..0a8ebe9 100644
--- a/src/construct/thematic_break.rs
+++ b/src/construct/thematic_break.rs
@@ -1,20 +1,28 @@
-//! Thematic breaks, sometimes called horizontal rules, are a construct that
-//! occurs in the [flow][] content type.
+//! Thematic break occurs in the [flow][] content type.
 //!
-//! They’re formed with the following BNF:
+//! ## Grammar
+//!
+//! Thematic break forms with the following BNF
+//! (<small>see [construct][crate::construct] for character groups</small>):
 //!
 //! ```bnf
 //! ; Restriction: all markers must be identical.
 //! ; Restriction: at least 3 markers must be used.
 //! thematic_break ::= *space_or_tab 1*(1*marker *space_or_tab)
 //!
-//! space_or_tab ::= ' ' | '\t'
 //! marker ::= '*' | '-' | '_'
 //! ```
 //!
+//! As this construct occurs in flow, like all flow constructs, it must be
+//! followed by an eol (line ending) or eof (end of file).
+//!
+//! ## HTML
+//!
 //! Thematic breaks in markdown typically relate to the HTML element `<hr>`.
 //! See [*§ 4.4.2 The `hr` element* in the HTML spec][html] for more info.
 //!
+//! ## Recommendation
+//!
 //! It is recommended to use exactly three asterisks without whitespace when
 //! writing markdown.
 //! As using more than three markers has no effect other than wasting space,