use crate::lexer::*; use crate::*; struct Context<'a> { cur: [Option>;2], icur: usize, ll: Lexer<'a>, source: usize, } impl<'a> Context<'a> { pub fn new( mut ll: Lexer<'a>, source: usize ) -> Self { let cur = [ll.next(), ll.next()]; //println!(" - cur: {:?}, peek: {:?}", cur[0], cur[1] ); Self { cur, icur: 0, ll, source, } } pub fn next(&mut self) -> Option> { self.cur[self.icur] = self.ll.next(); self.icur = (self.icur+1)%2; //println!(" - cur: {:?}, peek: {:?}", self.cur(), self.peek() ); self.cur[self.icur] } pub fn cur(&self) -> Option> { self.cur[self.icur] } pub fn peek(&self) -> Option> { self.cur[(self.icur+1)%2] } pub fn check_unwrapped bool>(&self, context: &str, f: T) -> CrimResult { if self.cur().is_none() || self.peek().is_none() { Err(CrimError::eos( context )) } else { Ok(f( self.cur().unwrap().symbol(), self.peek().unwrap().symbol())) } } pub fn source(&self) -> usize { self.source } } trait SymbolHelper { fn is bool>(&self, context: &str, f: T ) -> CrimResult; fn is_valid_tag_name(&self) -> CrimResult; } impl<'a> SymbolHelper for Option> { fn is bool>(&self, context: &str, f: T ) -> CrimResult { if let Some(st) = self { Ok(f( &st.symbol() )) } else { Err(CrimError::eos( context )) } } fn is_valid_tag_name(&self) -> CrimResult { if let Some(st) = self { Ok(match st.symbol() { SymbolType::Token(_) | SymbolType::View | SymbolType::Output | SymbolType::Loop | SymbolType::If | SymbolType::ElIf | SymbolType::Else | SymbolType::Show => true, _ => false }) } else { Err(CrimError::parse( Position::none(), "Unexpeceted end of stream.".to_string() )) } } } pub struct Parser { } /** * input: input complete_tag * | input text * | * ; * * tag: unary_tag * | multinary_open_tag * | multinary_mid_tag * | multinary_close_tag * ; * * unary_tag: '[|' tag_guts '|]' * ; * * | '[|' tag_guts '|>' * | '<|' tag_guts '|>' * | '<|' tag_guts '|]' * ; * * tag_pair: open_tag tag_body close_tag * ; * * tag_body: tag_body tag * | tag_body tag_pair * | tag_body text * | * ; * * open_tag: '[|' tag_guts '|>' * ; * * close_tag: '<|' token '|]' * ; * * tag_guts: 'view' literal props * | 'section' literal * | 'output' literal * ; * * literal: '"' [^"]* '"' * ; * * props: props token '=' literal * | * ; * * disambiguate (tm): * * input: input tag * | input text * | * ; * * tag: open_tag_base unary_tag * | open_tag_base binary_tag * ; * * open_tag_base: '[|' tag_guts * ; * * unary_tag: '|]' * ; * * binary_tag: '|>' tag_body close_tag * ; * * tag_body: tag_body tag * | tag_body text * | * ; * * close_tag: '<|' token '|]' * ; * * tag_guts: 'view' literal props * | 'section' literal * | 'output' literal * ; * * literal: '"' [^"]* '"' * ; * * props: props token '=' literal * | * ; */ impl Parser { pub fn new() -> Self { Self { } } fn lex_error(&self, error: &Symbol) -> CrimResult { if let SymbolType::Error{what} = error.symbol() { Err(CrimError::parse( *error.start(), format!("What: {:?}", *what) )) } else { Err(CrimError::parse( Position::none(), "Not an error?".into())) } } pub fn parse(&self, src: &str, source: usize ) -> CrimResult> { let ll = Lexer::new( src ); let mut ctx = Context::new( ll, source ); // Parse the root of the file, the input context self.p_input( &mut ctx ) } fn p_input(&self, ctx: &mut Context ) -> CrimResult> { let mut children = Vec::new(); loop { if ctx.cur().is_none() { break; } match ctx.cur().unwrap().symbol() { SymbolType::Text(_) => { /* Skip top level text */ } SymbolType::StartFlat => { children.push( self.parse_tag_view( ctx )? ); } SymbolType::Error{..} => { self.lex_error( &ctx.cur().unwrap() )?; } _ => { } } ctx.next(); } Ok(children) } fn parse_tag(&self, ctx: &mut Context) -> CrimResult { let start_sym = ctx.cur().unwrap(); let mut tb = TagBuilder::new(&start_sym); if ctx.next().is_some() { match ctx.cur().unwrap().symbol() { SymbolType::View | SymbolType::Show | SymbolType::Loop | SymbolType::If | SymbolType::ElIf | SymbolType::Else | SymbolType::Output => { let name_sym = ctx.cur().unwrap(); ctx.next(); tb.set_name( name_sym ); } _ => { return Err(CrimError::parse( *ctx.cur().unwrap().start(), "Unexpected symbol".to_string() )); } } } else { return Err(CrimError::eos("tag type")); } if tb.can_have_params() || tb.can_have_expr() { self.parse_tag_params( ctx, &mut tb )?; } if tb.can_have_props() { self.parse_tag_props( ctx, &mut tb )?; } if let Some(end_sym) = ctx.cur() { match end_sym.symbol() { SymbolType::EndPoint | SymbolType::EndFlat => { tb.set_end( end_sym )?; } _ => { return Err(CrimError::parse( *end_sym.start(), format!("Unexpected symbol {:?} looking for end of tag.", end_sym.symbol()) )); } } } ctx.next(); Ok(tb) } fn parse_tag_set<'a>(&self, ctx: &mut Context; base: &mut TagBuilder) -> CrimResult { if base.is_unary() { return Ok(base.build()); } if !base.is_multinary_open() { return Err(CrimError::broken("Found close/mid tag instead of opening")); } loop { if ctx.cur().is_none() { return Err(CrimError::eos(format!("close tag for {:?}", base.name()).as_str())); } match ctx.cur().unwrap().symbol() { SymbolType::Text(s) => { base.add_child(Token::Text(s.to_string())); ctx.next(); } SymbolType::StartFlat | SymbolType::StartPoint => { let mut tb = self.parse_tag( ctx )?; match tb.tag_type() { TagType::MultinaryOpen => { base.add_child( parse_tag_set( ctx, tb )? ); } TagType::MultinaryMid => { base.add_chain( parse_tag_set( ctx, tb )? ); } TagType::MultinaryClose => { if (tb.is_name(SymbolToken::If) && (base.is_name(SymbolToken::Elif) || base.is_name(SymbolToken::Else))) || tb.is_name(base.name().symbol()) { } else { return Err(CrimError::parse( *tb.name().start(), format!("Found {:?} looking for close of {:?}", tb.name().symbol(), base.name().symbol()) )); } } } } SymbolType::Error{..} => { return self.lex_error( &ctx.cur().unwrap() ); } _ => { return Err(CrimError::parse( *ctx.cur().unwrap().start(), format!("Unexpected token: {:?}", ctx.cur().unwrap().symbol()), )); } } } } fn parse_tag_view(&self, ctx: &mut Context) -> CrimResult { let tb = self.parse_open_tag( ctx )?; if !tb.is_name( &SymbolType::View ) { return Err(CrimError::parse(tb.start_pos(), "Unexpected tag type, only view allowed at root".into() )); } if tb.is_unary() { return tb.build_view(ctx, Vec::new()); } let mut children = Vec::new(); let mut outputs = Vec::new(); //println!("--parse-tag-view-- tag parsed, next token: {:?}", ctx.cur() ); loop { if ctx.cur().is_none() { return Err(CrimError::eos(format!("close tag for {:?}", tb.name()).as_str())) } match ctx.cur().unwrap().symbol() { SymbolType::Text(s) => { children.push( Token::Text(s.to_string()) ); ctx.next(); } SymbolType::StartFlat => { let subtb = self.parse_tag( ctx )?; if subtb.is_name(&SymbolType::Output) { outputs.push(subtb.build_output()?); } else { children.push(subtb.build()?); } } SymbolType::StartPoint => { let end = self.parse_end_tag( ctx )?; //println!("End tag: {:?}, open tag: {:?}", end, tb.name() ); if tb.is_name(end.symbol()) { // They match, time to decide what we're doing. let any_outputs = outputs.len() > 0; let all_text = children.iter().all( |t| matches!(t, Token::Text(..)) ); if !any_outputs { // Create an implicit output called "content" outputs.push(Output{ name: "content".to_string(), code: children, }); } else { if !all_text { return Err(CrimError::parse( *end.start(), "You cannot mix non-output and output tags in a view.".to_string() )); } } return tb.build_view(ctx,outputs); } else { // They don't match, complain. return Err(CrimError::parse( *end.start(), "Mismatched open and closing tags.".to_string() )); } } SymbolType::Error{..} => { return self.lex_error( &ctx.cur().unwrap() ); } _ => { return Err(CrimError::parse( *ctx.cur().unwrap().start(), format!("Unexpected tag in view root: {:?}", ctx.cur().unwrap().symbol()) )); } } } } fn parse_open_tag<'a>(&self, ctx: &mut Context<'a>) -> CrimResult> { let start_sym = ctx.cur().unwrap(); let mut tb = TagBuilder::new(&start_sym); if ctx.next().is_some() { match ctx.cur().unwrap().symbol() { SymbolType::View | SymbolType::Show | SymbolType::Loop | SymbolType::If | SymbolType::ElIf | SymbolType::Else | SymbolType::Output => { let name_sym = ctx.cur().unwrap(); ctx.next(); tb.set_name( name_sym ); } _ => { return Err(CrimError::parse( *ctx.cur().unwrap().start(), "Unexpected symbol".to_string() )); } } } else { return Err(CrimError::eos("tag type")); } self.parse_tag_params( ctx, &mut tb )?; self.parse_tag_props( ctx, &mut tb )?; if let Some(end_sym) = ctx.cur() { match end_sym.symbol() { SymbolType::EndPoint => { tb.set_type( TagType::MultinaryOpen ); } SymbolType::EndFlat => { tb.set_type( TagType::Unary ); } _ => { return Err(CrimError::parse( *end_sym.start(), format!("Unexpected symbol {:?} looking for end of tag.", end_sym.symbol()) )); } } } ctx.next(); Ok(tb) } fn parse_end_tag<'a>(&self, ctx: &mut Context<'a> ) -> CrimResult> { if !ctx.cur().is("end tag",|s| *s == SymbolType::StartPoint )? { return Err(CrimError::parse( *ctx.cur().unwrap().start(), "Invalid end tag?".to_string() )); } if !ctx.next().is_valid_tag_name()? { return Err(CrimError::parse( *ctx.cur().unwrap().start(), "Invalid tag name".to_string() )); } let name = ctx.cur().unwrap(); if !ctx.next().is("end of end tag", |s| *s == SymbolType::EndFlat )? { return Err(CrimError::parse( *ctx.cur().unwrap().start(), "Tag should be <| |] style end tag.".to_string(), )); } ctx.next(); Ok(name) } fn parse_tag_body<'a>(&self, ctx: &mut Context<'a> ) -> CrimResult> { let mut tb = self.parse_open_tag( ctx )?; if tb.is_unary() { return Ok(tb); } loop { if ctx.cur().is_none() { return Err(CrimError::eos(format!("close tag for {:?}", tb.name()).as_str())); } match ctx.cur().unwrap().symbol() { SymbolType::Text(s) => { tb.add_child(Token:: Text(s.to_string())); ctx.next(); } SymbolType::StartFlat => { tb.add_child( self.parse_tag_body( ctx )?.build()? ); } SymbolType::StartPoint => { let end = self.parse_end_tag( ctx )?; if tb.is_name(end.symbol()) { // They match, end. return Ok(tb); } else { // They don't match, complain. return Err(CrimError::parse( *end.start(), "Mismatched open and closing tags.".to_string() )); } } SymbolType::Error{..} => { return self.lex_error( &ctx.cur().unwrap() ); } _ => { return Err(CrimError::parse( *ctx.cur().unwrap().start(), format!("Unexpected token: {:?}", ctx.cur().unwrap().symbol()), )); } } } } fn parse_tag_params(&self, ctx: &mut Context, tb: &mut TagBuilder ) -> CrimResult<()> { loop { if ctx.cur().is_none() { return Err(CrimError::eos("tag parameters")); } match ctx.cur().unwrap().symbol() { SymbolType::Literal(s) => { tb.add_param( ParamValue::Literal(s.to_string()) ); ctx.next(); } SymbolType::Token(_) => { if ctx.peek().is("tag data", |s| *s == SymbolType::Equals)? { break; } tb.add_param( self.parse_identifier( ctx )? ); } _ => { break; } } } Ok(()) } fn parse_identifier(&self, ctx: &mut Context ) -> CrimResult { let mut id = Identifier::new(); loop { if ctx.cur().is("identifier",|s| matches!( s, SymbolType::Token(_) ))? { if let SymbolType::Token(s) = ctx.cur().unwrap().symbol() { id.push( IdentifierValue::Name(s.to_string()) ); } } else { return Err(CrimError::parse( ctx.cur().unwrap().start().clone(), format!("Expeceted identifier, found {:?}", ctx.cur().unwrap().symbol() ) ) ); } if !ctx.next().is("identifier seperator",|s| *s == SymbolType::Period )? { break; } ctx.next(); } Ok(ParamValue::Identifier(id)) } fn parse_tag_props(&self, ctx: &mut Context, tb: &mut TagBuilder ) -> CrimResult<()> { loop { if ctx.cur().is_none() { return Err(CrimError::eos("tag properties")); } if let SymbolType::Token(s) = ctx.cur().unwrap().symbol() { if ctx.peek().is_some_and(|s|s.check_type(|t| matches!(t,SymbolType::Equals))) { ctx.next(); if let Some(sym) = ctx.next() && let SymbolType::Literal(lv) = sym.symbol() { tb.add_prop( s.to_string(), lv.to_string() ); } else { return Err(CrimError::parse(Position::none(),"Expected quoted literal string".to_string())); } } else { break; } } else { break; } ctx.next(); } Ok(()) } } #[derive(PartialEq,Copy,Clone,Debug)] enum TagType { Unknown, Unary, MultinaryOpen, MultinaryMid, MultinaryClose, } struct TagBuilder<'a> { name: Option>, params: Vec, props: Properties, children: Vec, tag_type: TagType, start: Symbol<'a>, chain: Vec, } #[derive(Debug)] enum ParamValue { Literal(String), Identifier(Identifier), } enum BuildResult { View(View), Output(Output), Token(Token), } impl<'a> TagBuilder<'a> { pub fn new(start: &Symbol<'a>) -> TagBuilder<'a> { TagBuilder { name: None, params: Vec::new(), props: Properties::new(), children: Vec::new(), tag_type: TagType::Unknown, start: start.clone(), chain: Vec::new(), } } pub fn set_end(&mut self, end: &Symbol) -> CrimResult<()> { if start.symbol() == SymbolType::StartFlat { if end.symbol() == SymbolType::EndFlat { self.tag_type = TagType::Unary; } else if end.symbol() == SymbolType::EndPoint { self.tag_type = TagType::MultinaryOpen; } else { return Err(CrimError::broken( *self.end.start(), "Invalid bracket token type.")); } } else if start.symbol() == SymbolType::StartPoint { if end.symbol() == SymbolType::EndFlat { self.tag_type = TagType::MultinaryClose; } else if end.symbol() == SymbolType::EndPoint { self.tag_type = TagType::MultinaryMid; } else { return Err(CrimError::broken( *self.end.start(), "Invalid bracket token type.")); } } else { return Err(CrimError::broken( *self.start.start(), "Invalid bracket token type.")); } } pub fn start_pos(&self) -> Position { self.start.start() } pub fn is_unary(&self) -> bool { self.tag_type == TagType::Unary } pub fn is_multinary_open(&self) -> bool { self.tag_type == TagType::MultinaryOpen } pub fn can_have_params(&self) -> bool { match self.name.symbol() { SymbolType::View | SymbolType::Output | SymbolType::Loop => true, SymbolType::Show | SymbolType::If | SymbolType::ElIf | SymbolType::Else => false, } } pub fn can_have_expr(&self) -> bool { match self.name.symbol() { SymbolType::View | SymbolType::Output | SymbolType::Loop => false, SymbolType::Show | SymbolType::If | SymbolType::ElIf | SymbolType::Else => true, } } pub fn can_have_props(&self) -> bool { true } pub fn can_have_children(&self) -> bool { if self.tag_type == TagType::MultinaryOpen || self.tag_type == TagType::MultinaryMid { true } else { false } } pub fn is_name(&self, name: &SymbolType<'a>) -> bool { if let Some(a) = self.name { a.symbol() == name } else { false } } pub fn name(&self) -> Option> { self.name } pub fn params(&self) -> &Vec { &self.params } pub fn set_name(&mut self, name: Symbol<'a>) { self.name = Some(name); } pub fn add_param(&mut self, param: ParamValue) { self.params.push( param ); } pub fn add_prop(&mut self, key: String, value: String) { self.props.insert( key, value ); } pub fn add_child(&mut self, token: Token) { self.children.push( token ); } pub fn append_children(&mut self, children: &mut Vec::) { self.children.append( children ); } pub fn set_type(&mut self, tag_type: TagType) { self.tag_type = tag_type; } pub fn tag_type(&self) -> TagType { self.tag_type } pub fn build(mut self) -> CrimResult { if let Some(sym) = self.name { match sym.symbol() { SymbolType::View => { let name = if let ParamValue::Literal(s) = self.params.swap_remove(0) { s.to_string() } else { return Err(CrimError::parse(Position::none(), "Expected string literal for view name.".to_string())); }; Ok(BuildResult::View(View { name: name, source: ctx.source(), outputs: outputs, //theme: Option layout: self.props.get("layout").cloned(), })) } SymbolType::Output => { let name = if let ParamValue::Literal(s) = self.params.swap_remove(0) { s.to_string() } else { return Err(CrimError::parse(Position::none(), "Expected string literal for output name.".into())); }; Ok(BuildResult::Output(Output { name: name, code: self.children, })) } SymbolType::Loop => { let id = if let ParamValue::Identifier(id) = self.params.swap_remove(0) { id } else { return Err(CrimError::parse( Position::none(), "Identifier for loop variable name.".into()) ); }; Ok(BuildResult::Token(Token::Loop(id, self.children))) } SymbolType::Show => { let id = if let ParamValue::Identifier(id) = self.params.swap_remove(0) { id } else { return Err(CrimError::parse( Position::none(), "Identifier for show variable name.".into()) ); }; Ok(BuildResult::Token(Token::Show(id, self.props))) } SymbolType::If => { let id = if let ParamValue::Identifier(id) = self.params.swap_remove(0) { id } else { return Err(CrimError::parse( Position::none(), "Identifier for if variable name.".into()) ); }; if self.chain.len() == 1 && let Token::If(id, mut children, chain) = self.chain[0] && id.len() == 0 { self.chain.clear(); self.chain.append(children); } Ok(BuildResult::Token(Token::If(id, self.children, self.chain))) } SymbolType::ElIf => { let id = if let ParamValue::Identifier(id) = self.params.swap_remove(0) { id } else { return Err(CrimError::parse( Position::none(), "Identifier for elif variable name.".into()) ); }; Ok(BuildResult::Token(Token::If(Identifier::new(), self.children, self.chain))) } SymbolType::Else => { Ok(BuildResult::Token(Token::If(Identifier::new(), self.children, self.chain))) } _ => { Err(CrimError::parse( self.start_pos, "Bad tag type".into())) } } } else { Err(CrimError::parse(self.start_pos, "Bad tag type".into())) } } }