Working on simpler array-based implementation.

docs/experiments/parsers/v3/core.js

@@ -0,0 +1,121 @@
+import { createLineData, useDocument } from '@terrace/core'
+import { createStringReader } from '@terrace/core/readers/js-string'
+
+export const SYMBOLS = {
+  TAIL: Symbol('TAIL'),
+  UNMATCHED: Symbol('UNMATCHED')
+}
+
+export const BASE_MACROS = {
+  // string ({ line }) {
+  //   return tail(line).toString()
+  // },
+  // number ({ line }) {
+  //   const num = +tail(line)
+  //   if (isNaN(num) || line === '') return
+  //   return num
+  // },
+  // primitive (args) {
+  //   const num = macros.number(args)
+  //   return num !== undefined ? num : macros.string(args)
+  // },
+  // any (args) {
+  //   const macro = args.macros[args.head]
+  //   if (macro) return macro(args)
+  //   const numResult = args.macros.number(args)
+  //   if (numResult !== undefined) return numResult
+  //   args.tail = args.line
+  //   return args.macros.string(args)
+  // },
+  // scope ({ addScope, head, tail, line }, definition) {
+  //   return addScope({ definition, head, tail, line })
+  // }
+}
+
+export function getTail(key, macro) {
+  return args => {
+    const scope = args.scope
+    const result = macro(args)
+    if (result === undefined) return
+    if (!scope[key]) scope[key] = []
+    scope[key].push(result)
+  }
+}
+
+export function getText(key, macro) {
+  return (args) => {
+    const scope = args.scope
+    const result = macro({...args, tail: args.line })
+    if (result === undefined) return
+    if (!scope[key]) scope[key] = []
+    scope[key].push(result)
+  }
+}
+
+export function getUnmatched(macro) {
+  return (args) => {
+    const key = args.head
+    const scope = args.scope
+    const result = macro(args)
+    if (result === undefined) return
+    if (!scope[key]) scope[key] = []
+    scope[key].push(result)
+  }
+}
+
+export function isMacro (macro) {
+  return (args) => args.macros[macro](args)
+}
+
+export function isCollection (macro) {
+  return (args) => ({ [args.head]: macro(args) })
+}
+
+export function isScope (definition) {
+  return (args) => args.macros.scope(args, definition)
+}
+
+export async function parse(lines, schema) {
+  const doc = useDocument(createStringReader(lines))
+
+  const macros = schema.macros
+  const scopes = [[]]
+
+  function addScope() {
+    const scope = []
+    scopes[scopes.length] = scope
+    return scope
+  }
+
+  let ended = false
+  let lastLevel = 0
+  while (!ended) {
+    ended = await doc.next()
+    if (ended) break;
+    const level = doc.level()
+    scopes.length = level + 1
+
+    if (lastLevel == level) {
+      console.log('||DECREASE: ', doc.line())
+    }
+
+    lastLevel = level
+
+    let entry = [doc.line()]
+
+    Object.keys(schema.root).forEach(key => {
+      if (entry[0] === key || entry[0].startsWith(`${key} `)) {
+        entry = schema.root[key]({ entry, macros })
+      }
+    })
+
+    const scope = scopes[level] || []
+    if (!scopes[level]) {
+      scopes[level] = scope
+      scopes[level - 1].at(-1)[1] = scopes[level]
+    }
+    scope.push(entry)
+  }
+
+  return scopes[0]
+}

docs/experiments/parsers/v3/example.js

@@ -0,0 +1,139 @@
+import { SYMBOLS, parse, isScope, isMacro, getTail, getText, getUnmatched, isCollection } from './core.js'
+
+const schemaTCE = `
+macros
+ primitive match number string
+ section
+  pos number tail
+  content string unmatched
+  position number
+ options
+  parameter1 number
+  parameter2 string
+  literal unmatched string
+  unmatched primitive
+
+root
+ title string
+ options options collection
+ subsection section
+ list
+  - string array
+ collection
+  section collection
+ collection2
+  unmatched collection
+`
+
+function head(line) {
+  return line.split(' ')[0]
+}
+
+function tail(line) {
+  return line.split(' ').slice(1).join(' ')
+}
+
+function chain(...macroNames) {
+  return (args) => {
+    for (const macroName of macroNames) {
+      if (args.macros[macroName]) args.entry = args.macros[macroName](args)
+    }
+    return args.entry
+  }
+}
+
+const schema = {
+  macros: {
+    tail({ entry }) {
+      return [head(entry[0]), tail(entry[0])]
+    },
+    string({ entry }) {
+      return [entry[0], entry[1].toString()]
+    },
+    number({ entry }) {
+      const num = +entry[1]
+      if (isNaN(num) || entry[1] === '') return entry
+      return [entry[0], num]
+    },
+    primitive(args) {
+      const num = macros.number(args)
+      return num !== undefined ? num : macros.string(args)
+    },
+    scope({ addScope, head, tail, entry }, definition) {
+      return addScope({ definition, head, tail, entry })
+    },
+    section: isScope({
+      [SYMBOLS.TAIL]: getTail('pos', isMacro('number')),
+      [SYMBOLS.UNMATCHED]: getText('content', isMacro('string')),
+      position: isMacro('number')
+    }),
+    options: isScope({
+      parameter1: isMacro('number'),
+      parameter2: isMacro('string'),
+      unmatched: isMacro('string'),
+      [SYMBOLS.UNMATCHED]: getUnmatched(isMacro('any')),
+    }),
+    any(args) {
+      const macro = args.macros[args.head]
+      if (macro) return macro(args)
+      const numResult = args.macros.number(args)
+      if (numResult !== undefined) return numResult
+      args.tail = args.entry
+      return args.macros.string(args)
+    },
+  },
+  root: {
+    // Move tail into value position and cast it to string.
+    title: chain('tail', 'string'),
+    // options: isMacro('options'),
+    // subsection: isMacro('section'),
+    // list: isScope({
+    //   '-': isMacro('string')
+    // }),
+    // collection: isScope({
+    //   section: isCollection(isMacro('section'))
+    // }),
+    // collection2: isScope({
+    //   [SYMBOLS.UNMATCHED]: getUnmatched(isMacro('any'))
+    // })
+  }
+}
+
+
+const lines = [
+  `title Example`,
+  `options`,
+  ` parameter1 30`,
+  ` parameter2 Enim eu id anim minim reprehenderit nostrud eu amet deserunt ea ut do cupidatat ea.`,
+  `options`,
+  ` parameter1 0`,
+  ` parameter2 Esse incididunt et est adipisicing eiusmod aliqua enim ea aliqua id enim.`,
+  `subsection`,
+  ` position 1`,
+  ` Ea dolore in aliquip fugiat anim adipisicing amet aute tempor et deserunt est duis sint.`,
+  `subsection 2`,
+  ` position 2`,
+  ` Aute deserunt incididunt ad in sint adipisicing est officia velit pariatur ipsum deserunt quis nulla.`,
+  ` Ea dolore in aliquip fugiat anim adipisicing amet aute tempor et deserunt est duis sint.`,
+  `list`,
+  ` - item 1`,
+  ` - item 2`,
+  `collection`,
+  ` section`,
+  `  lorem ipsum 1`,
+  ` section`,
+  `  lorem ipsum 2`,
+  `collection2`,
+  ` section`,
+  `  position 3`,
+  `  Laborum aute anim occaecat occaecat pariatur tempor proident magna sit magna non non.`,
+  ` list`,
+  `  1`,
+  `  2`
+]
+
+async function main() {
+  console.dir(await parse(lines, schema), { depth: null })
+}
+
+main()