//
// HTML
//
// Fork from:
// https://github.com/mathiversen/html-parser/blob/v0.6.3/src/grammar/rules.pest
//
item = _{
SOI
~ comment*
~ doctype?
~ node*
~ EOI
}
// Other
other = { !(comment | node_element | server | text) ~ ANY }
// Code
code = ${ "" ~ (!"" ~ ANY)* ~ "" }
//
// DOCTYPE
//
doctype = { chevron_left_bang ~ doctype_name ~ attr* ~ chevron_right_normal }
doctype_name = @{ ^"doctype" }
//
// SERVER CODE
//
server = ${ chevron_left_server ~ server_code ~ chevron_right_server }
server_code = { (!chevron_right_server ~ ANY)* }
chevron_left_server = { "<%" }
chevron_right_server = { "%>" }
//
// NODES
//
node = _{ code | server | comment | node_element | text | other }
comment = { comment_if | comment_normal }
text = ${ (!(chevron_left_normal | comment_tag_start ) ~ ANY)+ } // NOTE: Should we be able to write < in text? ^^
node_element = { el_void | el_void_xml | el_process_instruct | javascript_text | style_text | el_raw_text | el_normal | el_dangling }
//
// COMMENTS
//
comment_normal = _{ comment_tag_start ~ (!comment_tag_end ~ ANY)* ~ comment_tag_end }
comment_tag_start = _{ chevron_left_bang ~ "--" }
comment_tag_end = _{ "--" ~ chevron_right_normal }
// Compatability with old IE browsers... This is not necessary for newer browsers
comment_if = _{ comment_if_start ~ (!comment_if_end ~ ANY)* ~ comment_if_end }
comment_if_start = _{ comment_tag_start ~ "[" ~ ^"if" }
comment_if_end = _{ chevron_left_bang ~ "[" ~ ^"endif" ~ "]" ~ comment_tag_end }
//
// ATTRIBUTES
//
attr = { attr_key ~ (equal ~ (attr_non_quoted | attr_quoted ))? }
attr_quoted = ${PUSH(quote) ~ attr_value ~ PUSH(quote) }
attr_non_quoted = @{ !quote ~ (!(WHITESPACE | chevron_right) ~ ANY)* }
attr_key = ${ !WHITESPACE ~ (":" | "@" | "#" | ".")? ~ ASCII_ALPHA ~ text_chars* }
attr_value = { WHITESPACE* ~ (!PEEK ~ ANY)* ~ WHITESPACE* }
//
// ELEMENTS
//
el_name = @{ ASCII_ALPHA ~ text_chars* }
// Void element aka self-closing element
// Ex:
el_void_name_html = @{
^"area"
| ^"base"
| ^"br"
| ^"col"
| ^"command"
| ^"embed"
| ^"hr"
| ^"img"
| ^"input"
| ^"keygen"
| ^"link"
| ^"meta"
| ^"param"
| ^"source"
| ^"track"
| ^"wbr"
| ^"meta"
}
// NOTE: This should not have to be a rule, but people doesn't know what void elements are...
el_void_name_svg = @{
^"path"
| ^"polygon"
| ^"rect"
| ^"circle"
}
el_void_name = @{ el_void_name_html | el_void_name_svg }
el_void = _{ chevron_left_normal ~ el_void_name ~ attr* ~ (chevron_right_normal | chevron_right_closed) }
el_void_xml = _{ chevron_left_normal ~ el_name ~ attr* ~ chevron_right_closed }
// Open elements are default element that can take children
// and have both a start tag and an end tag
// Ex:
el_normal = _{ el_normal_start ~ (!el_normal_end ~ node)* ~ el_normal_end }
el_normal_start = _{ chevron_left_normal ~ PUSH(el_name) ~ attr* ~ chevron_right_normal}
el_normal_end = _{ chevron_left_closed ~ PUSH(el_name) ~ chevron_right_normal}
// Raw text elements are elements with text/script content that
// might interfere with the normal html syntax
el_raw_text_name = {
^"title"
| ^"textarea"
}
el_raw_text_content = ${ (!el_raw_text_end ~ ANY)* }
el_raw_text = _{ el_raw_text_start ~ el_raw_text_content ~ el_raw_text_end }
el_raw_text_start = _{ chevron_left_normal ~ PUSH(el_raw_text_name) ~ attr* ~ chevron_right_normal}
el_raw_text_end = { chevron_left_closed ~ PUSH(el_raw_text_name) ~ chevron_right_normal}
inline_javascript = ${ (!javascript_end ~ ANY)* }
javascript_text = _{ javascript_start ~ inline_javascript ~ javascript_end }
el_javascript_name = @{ ^"script" }
javascript_start = _{ chevron_left_normal ~ PUSH(el_javascript_name) ~ attr* ~ chevron_right_normal}
javascript_end = { chevron_left_closed ~ PUSH(el_javascript_name) ~ chevron_right_normal}
inline_style = ${ (!style_end ~ ANY)* }
el_style_name = @{ ^"style" }
style_text = _{ style_start ~ inline_style ~ style_end }
style_start = _{ chevron_left_normal ~ PUSH(el_style_name) ~ attr* ~ chevron_right_normal}
style_end = { chevron_left_closed ~ PUSH(el_style_name) ~ chevron_right_normal}
// XML processing instruction
// Ex:
el_process_instruct = { chevron_left_question ~ el_name? ~ attr* ~ chevron_right_question }
// Catch dangling elements
// Ex:
el_dangling = { chevron_left_closed ~ el_name ~ chevron_right_normal}
//
// SYMBOLS / CHARACTERS
//
text_chars = _{'a'..'z' | 'A'..'Z' | "_" | "-" | ":" |'0'..'9'}
chevron_left_normal = { "<"}
chevron_left_closed = { "</" }
chevron_left_bang = { "<!" }
chevron_left_question = { "<?" }
chevron_right_normal = { ">" }
chevron_right_closed = { "/>" }
chevron_right_question = { "?>" }
chevron_right = _{
chevron_right_normal
| chevron_right_closed
| chevron_right_question
}
equal = { "=" }
quote_dubble = { "\"" }
quote_single = { "'" }
quote = _{ quote_dubble | quote_single }
WHITESPACE = { " " | "\t" | "\r" | "\n" }