From ce5c224211a94bfd4c898b51d15febdf2ed9d6f2 Mon Sep 17 00:00:00 2001
From: Charlie Stanton <charlie@shtanton.xyz>
Date: Fri, 26 Aug 2022 11:51:46 +0100
Subject: Refactors some stuff and adds lexing and parsing

---
 main/parse.go | 145 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 145 insertions(+)
 create mode 100644 main/parse.go

(limited to 'main/parse.go')

diff --git a/main/parse.go b/main/parse.go
new file mode 100644
index 0000000..e876010
--- /dev/null
+++ b/main/parse.go
@@ -0,0 +1,145 @@
+package main
+
+import (
+	"strings"
+	"strconv"
+	"fmt"
+)
+
+type parser struct {
+	tokenStream chan Token
+	rewinds []Token
+}
+func (p *parser) next() Token {
+	if len(p.rewinds) == 0 {
+		return <- p.tokenStream
+	}
+	token := p.rewinds[len(p.rewinds)-1]
+	p.rewinds = p.rewinds[:len(p.rewinds)-1]
+	return token
+}
+func (p *parser) rewind(token Token) {
+	p.rewinds = append(p.rewinds, token)
+}
+func (p *parser) peek() Token {
+	token := p.next()
+	p.rewind(token)
+	return token
+}
+
+// TODO: make a pratt parser
+func (p *parser) parsePathPatternFilter() PathFilterAST {
+	var segments []PathFilterAST
+	loop: for {
+		token := p.next()
+		switch token.typ {
+			case TokenHash:
+				stringIndex := p.next()
+				if stringIndex.typ != TokenPatternStringIndex {
+					panic("Expected string index after # in pattern")
+				}
+				segments = append(segments, StringSegmentPathFilterAST{stringIndex.val})
+			case TokenAt:
+				intIndex := p.next()
+				if intIndex.typ != TokenPatternIntegerIndex {
+					panic("Expected integer index after @ in pattern")
+				}
+				index, err := strconv.Atoi(intIndex.val)
+				if err != nil {
+					panic("Expected integer index after @ in pattern")
+				}
+				segments = append(segments, IntegerSegmentPathFilterAST{index})
+			case TokenDot:
+				segments = append(segments, AnySegmentPathFilterAST{})
+			case TokenAst:
+				if len(segments) == 0 {
+					panic("Invalid * in pattern, * must go after something")
+				}
+				segments[len(segments) - 1] = RepeatPathFilterAST {segments[len(segments)-1]}
+			default:
+				p.rewind(token)
+				break loop
+		}
+	}
+	return SequencePathFilterAST {segments}
+}
+
+// TODO: should only return a single filter
+func (p *parser) parseFilter() []Filter {
+	var filters []Filter
+	token := p.next()
+	switch token.typ {
+		case TokenHash, TokenAt, TokenDot:
+			p.rewind(token)
+			filterAst := p.parsePathPatternFilter()
+			filters = append(filters, compilePathFilterAST(filterAst))
+			token = p.next()
+	}
+	if len(filters) == 0 {
+		panic("Missing filter")
+	}
+	p.rewind(token)
+	return filters
+}
+
+func (p *parser) parseBasicCommand(commandChar rune) Command {
+	switch commandChar {
+		case 'p':
+			return PrintValueCommand{}
+		case 'd':
+			return DeleteAllCommand{}
+		default:
+			panic("Invalid command")
+	}
+}
+
+func (p *parser) parseCommand() Command {
+	token := p.next()
+	switch token.typ {
+		case TokenHash, TokenAt, TokenDot:
+			p.rewind(token)
+			filters := p.parseFilter()
+			command := p.parseCommand()
+			for _, filter := range filters {
+				command = FilteredCommand {
+					filter: filter,
+					command: command,
+				}
+			}
+			return command
+		case TokenCommand:
+			commandChar, _, err := strings.NewReader(token.val).ReadRune()
+			if err != nil {
+				panic("Error reading a command character!?")
+			}
+			return p.parseBasicCommand(commandChar)
+		default:
+			fmt.Println(token)
+			panic("Invalid token, expected command")
+	}
+}
+
+func (p *parser) parseCommands() []Command {
+	var commands []Command
+	for {
+		nextToken := p.peek()
+		if nextToken.typ == TokenEOF || nextToken.typ == TokenRBrace {
+			return commands
+		}
+		commands = append(commands, p.parseCommand())
+		semicolon := p.next()
+		if semicolon.typ == TokenEOF || semicolon.typ == TokenRBrace {
+			return commands
+		}
+		if semicolon.typ != TokenSemicolon {
+			panic("Expected ; after command")
+		}
+	}
+}
+
+func Parse(tokens chan Token) []Command {
+	p := parser {
+		tokenStream: tokens,
+	}
+	return p.parseCommands()
+}
-- 
cgit v1.2.3