; Left-Factored ABNF Grammar for Markdown with Embedded HTML
; Based on RFC 5234 (ABNF) syntax

; Top-level document structure
document = *( block-element / CRLF )

; Block-level elements (left-factored by common prefixes)
block-element = heading / code-block / blockquote / list / horizontal-rule / 
                html-block / paragraph

; Headings (ATX-style: # ## ### etc.)
heading = heading-prefix heading-content
heading-prefix = 1*6"#" *WSP
heading-content = *( inline-element ) line-ending

; Code blocks (fenced with ``` or indented)
code-block = fenced-code-block / indented-code-block
fenced-code-block = code-fence [ language-info ] line-ending
                   *( code-line )
                   code-fence line-ending
code-fence = "```"
language-info = 1*( ALPHA / DIGIT / "-" / "+" / "." )
indented-code-block = 1*( indented-code-line )
indented-code-line = 4*WSP 1*VCHAR line-ending
code-line = *VCHAR line-ending

; Blockquotes
blockquote = 1*( blockquote-line )
blockquote-line = ">" [ WSP ] *( inline-element ) line-ending

; Lists (left-factored by list marker)
list = unordered-list / ordered-list
unordered-list = 1*( unordered-list-item )
ordered-list = 1*( ordered-list-item )
unordered-list-item = unordered-marker list-item-content
ordered-list-item = ordered-marker list-item-content
unordered-marker = ( "*" / "-" / "+" ) WSP
ordered-marker = 1*DIGIT "." WSP
list-item-content = *( inline-element ) line-ending
                   *( list-continuation )
list-continuation = 2*WSP *( inline-element ) line-ending

; Horizontal rule
horizontal-rule = hr-dashes / hr-asterisks / hr-underscores
hr-dashes = 3*( "-" ) *( WSP / "-" ) line-ending
hr-asterisks = 3*( "*" ) *( WSP / "*" ) line-ending  
hr-underscores = 3*( "_" ) *( WSP / "_" ) line-ending

; HTML blocks
html-block = html-block-tag / html-comment-block / html-declaration
html-block-tag = "<" tag-name *( WSP attribute ) [ WSP ] [ "/" ] ">"
                *( html-content )
                [ "</" tag-name ">" ]
                line-ending
html-comment-block = "<!--" *( CHAR ) "-->" line-ending
html-declaration = "<!" 1*ALPHA *( WSP / VCHAR ) ">" line-ending

; Paragraphs (catch-all for regular text)
paragraph = paragraph-line 1*( paragraph-continuation )
paragraph-line = *( inline-element ) line-ending
paragraph-continuation = !block-element-start *( inline-element ) line-ending
block-element-start = "#" / ">" / ( 1*DIGIT "." WSP ) / 
                     ( ( "*" / "-" / "+" ) WSP ) / 
                     "```" / 4*WSP / "<"

; Inline elements (left-factored by opening characters)
inline-element = emphasis / strong / code-span / link / image / 
                html-inline / line-break / escaped-char / plain-text

; Emphasis and strong (left-factored by delimiter)
emphasis = emphasis-asterisk / emphasis-underscore
strong = strong-asterisk / strong-underscore
emphasis-asterisk = "*" emphasis-content "*"
emphasis-underscore = "_" emphasis-content "_"
strong-asterisk = "**" strong-content "**"
strong-underscore = "__" strong-content "__"
emphasis-content = 1*( !( "*" / line-ending ) inline-element )
strong-content = 1*( !( "**" / line-ending ) inline-element )

; Code spans
code-span = code-delimiter code-span-content code-delimiter
code-delimiter = 1*"`"
code-span-content = 1*( !"`" CHAR )

; Links and images (left-factored by opening bracket)
link = link-reference / link-inline
image = image-reference / image-inline
link-reference = "[" link-text "]" "[" reference-id "]"
link-inline = "[" link-text "]" "(" url [ WSP title ] ")"
image-reference = "!" "[" alt-text "]" "[" reference-id "]"
image-inline = "!" "[" alt-text "]" "(" url [ WSP title ] ")"
link-text = 1*( !( "]" / line-ending ) inline-element )
alt-text = 1*( !( "]" / line-ending ) CHAR )
reference-id = 1*( !( "]" / line-ending ) CHAR )
url = 1*( !( WSP / ")" / line-ending ) CHAR )
title = ( DQUOTE title-content DQUOTE ) / 
        ( "'" title-content "'" ) /
        ( "(" title-content ")" )
title-content = *( !( DQUOTE / "'" / ")" ) CHAR )

; Inline HTML
html-inline = html-tag-inline / html-entity
html-tag-inline = "<" tag-name *( WSP attribute ) [ WSP ] [ "/" ] ">"
html-entity = "&" entity-name ";"
entity-name = 1*( ALPHA / DIGIT )

; HTML attributes and content
tag-name = ALPHA *( ALPHA / DIGIT / "-" / ":" )
attribute = attribute-name [ "=" attribute-value ]
attribute-name = ( ALPHA / "_" / ":" ) *( ALPHA / DIGIT / "-" / "_" / ":" / "." )
attribute-value = ( DQUOTE *( !DQUOTE CHAR ) DQUOTE ) /
                  ( "'" *( !"'" CHAR ) "'" ) /
                  ( 1*( !( WSP / ">" ) VCHAR ) )
html-content = *( !( "</" ) CHAR )

; Line breaks and escaped characters
line-break = hard-line-break / soft-line-break
hard-line-break = 2*WSP line-ending
soft-line-break = line-ending
escaped-char = "\" ( VCHAR / WSP )

; Plain text (everything else)
plain-text = 1*( !special-char CHAR )
special-char = "*" / "_" / "`" / "[" / "]" / "(" / ")" / 
               "<" / ">" / "#" / "+" / "-" / "." / "!" /
               "&" / "\" / line-ending

; Basic definitions
line-ending = CRLF / LF / CR
WSP = SP / HTAB
CHAR = %x00-10FFFF  ; Any Unicode character
VCHAR = %x21-7E     ; Visible ASCII characters
ALPHA = %x41-5A / %x61-7A  ; A-Z / a-z  
DIGIT = %x30-39     ; 0-9
SP = %x20           ; Space
HTAB = %x09         ; Horizontal tab
CR = %x0D           ; Carriage return
LF = %x0A           ; Line feed
CRLF = CR LF        ; Internet standard newline
DQUOTE = %x22       ; Double quote