Reworked parser to be macro based.

docs/experiments/parsers/example.js

@@ -1,83 +1,80 @@
-import { parse } from './core.js'
+import { SYMBOLS, parse, isScope, isMacro, getTail, getText, getUnmatched, isCollection } from './core.js'
 
 const schemaTCE = `
-types
- section object, text as content, tail as pos
-  content string
-  pos number
+macros
+ primitive match number string
+ section
+  pos number tail
+  content string unmatched
   position number
- options object
+ options
   parameter1 number
   parameter2 string
+  literal unmatched string
+  unmatched primitive
 
-root object
+root
  title string
- options options
- options2 options
+ options options collection
  subsection section
-  collate collection
- list array
-  - string
- collection collection
-  section section
+ list
+  - string array
+ collection
+  section collection
+ collection2
+  unmatched collection
 `
 
 const schema = {
-  types: {
-    section: {
-      type: 'object',
-      text: 'content',
-      tail: 'pos',
-      values: {
-        content: 'string',
-        pos: 'number',
-        position: {
-          type: 'number'
-        }
-      }
+  macros: {
+    string({ tail }) {
+      return tail.toString()
     },
-    options: {
-      type: 'object',
-      values: {
-        parameter1: 'number',
-        parameter2: 'string'
-      }
+    number({ tail }) {
+      const num = +tail
+      if (isNaN(num) || tail === '') return
+      return num
+    },
+    primitive({ macros }) {
+      const num = macros.number(args)
+      return num !== undefined ? num : macros.string(args)
+    },
+    scope({ addScope, head, tail, line }, definition) {
+      return addScope({ definition, head, tail, line })
+    },
+    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.line
+      return args.macros.string(args)
     },
-    advList: {
-      type: 'array',
-      values: {
-        section: 'section'
-      }
-    }
   },
   root: {
-    type: 'object',
-    values: {
-      title: 'string',
-      options: 'options',
-      options2: 'options',
-      subsection: {
-        type: 'section',
-        // Allows a particular repeated key to collect itself under a multi-value root as an array.
-        collate: 'collection'
-      },
-      list: {
-        // Defines an array where all entries must conform to specific types. No other keys are permitted.
-        // Ex: [ value, value ]
-        type: 'array',
-        values: {
-          '-': 'string'
-        }
-      },
-      collection: {
-        // Defines an array where all entries must conform to specific types. They will be segregated by key. No other keys are permitted.
-        // Ex: [ {key: value }, { key: value }]
-        type: 'collection',
-        values: {
-          section: 'section'
-        }
-      }
-    }
+    title: isMacro('string'),
+    options: isMacro('options'),
+    subsection: isMacro('section'),
+    list: isScope({
+      '-': isMacro('string')
+    }),
+    collection: isScope({
+      section: isCollection(isMacro('section'))
+    }),
+    collection2: isScope({
+      [SYMBOLS.UNMATCHED]: getUnmatched(isMacro('any'))
+    })
   }
 }
 
@@ -87,7 +84,7 @@ const lines = [
   `options`,
   ` parameter1 30`,
   ` parameter2 Enim eu id anim minim reprehenderit nostrud eu amet deserunt ea ut do cupidatat ea.`,
-  `options2`,
+  `options`,
   ` parameter1 0`,
   ` parameter2 Esse incididunt et est adipisicing eiusmod aliqua enim ea aliqua id enim.`,
   `subsection`,
@@ -104,11 +101,18 @@ const lines = [
   ` section`,
   `  lorem ipsum 1`,
   ` section`,
-  `  lorem ipsum 2`
+  `  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.log(await parse(lines, schema))
+  console.dir(await parse(lines, schema), { depth: null })
 }
 
-main()
+main()