Add support for collection keys in toObject.

docs/experiments/parsers/v4/example.js

@@ -40,7 +40,11 @@ const lines = [
   ` Further comments below. As I will now demonstrate, there is no simple`,
   `  even if embedded`,
   ` way of dealing with this problem.`,
-  ``
+  `author`,
+  ` name Second Person`,
+  ` email second@secondperson.com`,
+  ` More text,`,
+  ` across two lines.`,
 ]
 
 // Schema
@@ -76,7 +80,10 @@ async function main() {
     }),
     'scripts': () => toObject({ '#any': true }),
     'devDependencies': () => toObject(),
-    'author': () => toObject({ name: true, email: true, '#text': true })
+    'author': {
+      type: 'collection',
+      handle: () => toObject({ name: true, email: true, '#text': true })
+    }
   })
 
   console.dir(structure, { depth: null })

packages/js/core/dist/document.cjs

@@ -1 +1 @@
-"use strict";Object.defineProperties(exports,{__esModule:{value:!0},[Symbol.toStringTag]:{value:"Module"}});const d=require("./parser.cjs");function D(v,b=" "){let a=0;const n=d.createLineData("",b);async function u(){switch(a){case 0:a=1;break;case 2:return a=1,!1}const e=await v.next();return e===null?(a=3,!0):(n.line=e,d.parseLine(n),!1)}const y=()=>a=2,r=()=>n.level,s=()=>n.line.slice(n.offsetHead),c=()=>n.line.slice(n.offsetHead,n.offsetTail),l=()=>n.line.slice(n.offsetTail),j=e=>e===c();async function w(e){const t=a===0?-1:r();for(;;){if(await u())return;if(r()<=t)return y();if(await e())return}}async function f(e=-1,t=[s()]){var i;return e===-1&&(e=r()+1),await u()?t:r()<e?(y(),t):(t.push(((i=n.line)==null?void 0:i.slice(e))||""),f(e,t))}async function x(e){const t={};return await w(async()=>{const i=c();if(!i)return;const o=e?e[i]||e["#any"]:null;if(e?o===!0?t[i]=l().trim():o?t[i]=await o():e!=null&&e["#text"]&&(t["#text"]=await f(r())):t[i]=l().trim(),e&&Object.keys(e).every(p=>t[p]!==void 0))return!0}),t}async function T(){const e=[["root",[]]];for(;!await u();){const t=r(),i=t+1,o=e[t];if(!o)continue;e.length=i;const p=e[i]=[s(),[]];o[1].push(p)}return e[0]}return{next:u,line:s,head:c,tail:l,level:r,match:j,each:w,blockAsText:f,toObject:x,toLineArray:T}}exports.useDocument=D;
+"use strict";Object.defineProperties(exports,{__esModule:{value:!0},[Symbol.toStringTag]:{value:"Module"}});const h=require("./parser.cjs");function O(m,b=" "){let o=0;const i=h.createLineData("",b);async function l(){switch(o){case 0:o=1;break;case 2:return o=1,!1}const e=await m.next();return e===null?(o=3,!0):(i.line=e,h.parseLine(i),!1)}const p=()=>o=2,a=()=>i.level,s=()=>i.line.slice(i.offsetHead),f=()=>i.line.slice(i.offsetHead,i.offsetTail),u=()=>i.line.slice(i.offsetTail),j=e=>e===f();async function y(e){const r=o===0?-1:a();for(;;){if(await l())return;if(a()<=r)return p();if(await e())return}}async function d(e=-1,r=[s()]){var t;return e===-1&&(e=a()+1),await l()?r:a()<e?(p(),r):(r.push(((t=i.line)==null?void 0:t.slice(e))||""),d(e,r))}async function v(e={}){const r={};let t={};return Object.keys(e).length?Object.keys(e).forEach(n=>{e[n]===!0&&(t[n]={type:"normal",handle:()=>u().trim()}),typeof e[n]=="function"&&(t[n]={type:"normal",handle:e[n]}),typeof e[n]=="object"&&(t[n]=e[n])}):t={"#any":{type:"normal",handle:()=>u().trim()}},await y(async()=>{const n=f();if(!n)return;const c=t[n]||t["#any"];if(!!c&&(c.type==="normal"?r[n]=await c.handle():c.type==="collection"?(r[n]||(r[n]=[]),r[n].push(await c.handle())):t!=null&&t["#text"]&&(r["#text"]=await d(a())),t&&Object.keys(t).every(w=>["collection"].includes(t[w].type)?!1:r[w]!==void 0)))return!0}),r}async function x(){const e=[["root",[]]];for(;!await l();){const r=a(),t=r+1,n=e[r];if(!n)continue;e.length=t;const c=e[t]=[s(),[]];n[1].push(c)}return e[0]}return{next:l,line:s,head:f,tail:u,level:a,match:j,each:y,blockAsText:d,toObject:v,toLineArray:x}}exports.useDocument=O;

packages/js/core/dist/document.js

@@ -1,70 +1,73 @@
-import { parseLine as j, createLineData as D } from "./parser.js";
-function k(x, d = " ") {
-  let o = 0;
-  const n = D("", d);
-  async function c() {
-    switch (o) {
+import { parseLine as v, createLineData as O } from "./parser.js";
+function D(h, m = " ") {
+  let c = 0;
+  const r = O("", m);
+  async function l() {
+    switch (c) {
       case 0:
-        o = 1;
+        c = 1;
         break;
       case 2:
-        return o = 1, !1;
+        return c = 1, !1;
     }
-    const e = await x.next();
-    return e === null ? (o = 3, !0) : (n.line = e, j(n), !1);
+    const e = await h.next();
+    return e === null ? (c = 3, !0) : (r.line = e, v(r), !1);
   }
-  const w = () => o = 2, r = () => n.level, u = () => n.line.slice(n.offsetHead), s = () => n.line.slice(n.offsetHead, n.offsetTail), f = () => n.line.slice(n.offsetTail), v = (e) => e === s();
-  async function y(e) {
-    const t = o === 0 ? -1 : r();
+  const y = () => c = 2, a = () => r.level, f = () => r.line.slice(r.offsetHead), s = () => r.line.slice(r.offsetHead, r.offsetTail), u = () => r.line.slice(r.offsetTail), b = (e) => e === s();
+  async function w(e) {
+    const i = c === 0 ? -1 : a();
     for (; ; ) {
-      if (await c())
+      if (await l())
         return;
-      if (r() <= t)
-        return w();
+      if (a() <= i)
+        return y();
       if (await e())
         return;
     }
   }
-  async function l(e = -1, t = [u()]) {
-    var i;
-    return e === -1 && (e = r() + 1), await c() ? t : r() < e ? (w(), t) : (t.push(((i = n.line) == null ? void 0 : i.slice(e)) || ""), l(e, t));
+  async function p(e = -1, i = [f()]) {
+    var t;
+    return e === -1 && (e = a() + 1), await l() ? i : a() < e ? (y(), i) : (i.push(((t = r.line) == null ? void 0 : t.slice(e)) || ""), p(e, i));
   }
-  async function T(e) {
-    const t = {};
-    return await y(async () => {
-      const i = s();
-      if (!i)
+  async function j(e = {}) {
+    const i = {};
+    let t = {};
+    return Object.keys(e).length ? Object.keys(e).forEach((n) => {
+      e[n] === !0 && (t[n] = { type: "normal", handle: () => u().trim() }), typeof e[n] == "function" && (t[n] = { type: "normal", handle: e[n] }), typeof e[n] == "object" && (t[n] = e[n]);
+    }) : t = { "#any": { type: "normal", handle: () => u().trim() } }, await w(async () => {
+      const n = s();
+      if (!n)
         return;
-      const a = e ? e[i] || e["#any"] : null;
-      if (e ? a === !0 ? t[i] = f().trim() : a ? t[i] = await a() : e != null && e["#text"] && (t["#text"] = await l(r())) : t[i] = f().trim(), e && Object.keys(e).every((p) => t[p] !== void 0))
+      const o = t[n] || t["#any"];
+      if (!!o && (o.type === "normal" ? i[n] = await o.handle() : o.type === "collection" ? (i[n] || (i[n] = []), i[n].push(await o.handle())) : t != null && t["#text"] && (i["#text"] = await p(a())), t && Object.keys(t).every((d) => ["collection"].includes(t[d].type) ? !1 : i[d] !== void 0)))
         return !0;
-    }), t;
+    }), i;
   }
-  async function b() {
+  async function x() {
     const e = [["root", []]];
-    for (; !await c(); ) {
-      const t = r(), i = t + 1, a = e[t];
-      if (!a)
+    for (; !await l(); ) {
+      const i = a(), t = i + 1, n = e[i];
+      if (!n)
         continue;
-      e.length = i;
-      const p = e[i] = [u(), []];
-      a[1].push(p);
+      e.length = t;
+      const o = e[t] = [f(), []];
+      n[1].push(o);
     }
     return e[0];
   }
   return {
-    next: c,
-    line: u,
+    next: l,
+    line: f,
     head: s,
-    tail: f,
-    level: r,
-    match: v,
-    each: y,
-    blockAsText: l,
-    toObject: T,
-    toLineArray: b
+    tail: u,
+    level: a,
+    match: b,
+    each: w,
+    blockAsText: p,
+    toObject: j,
+    toLineArray: x
   };
 }
 export {
-  k as useDocument
+  D as useDocument
 };

packages/js/core/src/document.ts

@@ -84,25 +84,52 @@ export function useDocument (reader: Reader, indent: string = ' '): Document {
     return blockAsText(startLevel, blockLines)
   }
 
-  async function toObject (matchers?: { [key: string]: Function|boolean }) {
+  async function toObject (inputMatchers: { [key: string]: Function|boolean|{ type: string, handle: Function } } = {}) {
     const obj: { [key: string]: any } = {}
+
+    let matchers: { [key: string]: {type: string, handle: Function } } = {}
+
+    // Normalize the matchers to an object-based format despite allowing flexible input types for convenience.
+    // TODO: Decide whether to enforce verbose input once a DSL has been created.
+    if (!Object.keys(inputMatchers).length) {
+      // Default matcher
+      matchers = { '#any': { type: 'normal', handle: () => tail().trim() } }
+    } else {
+      Object.keys(inputMatchers).forEach(key => {
+        // If a matcher is specified as `true`, treat as a key-value pair where { [head]: tail }
+        if(inputMatchers[key] === true) matchers[key] = { type: 'normal', handle: () => tail().trim() }
+        // If a matcher is specified as a function, treat as a key-value pair where { [head]: handle() }
+        if (typeof inputMatchers[key] === 'function') matchers[key] = { type: 'normal', handle: inputMatchers[key] as Function }
+        // If a matcher is specified as an object, allow customization of the type and handle for various cases.
+        if (typeof inputMatchers[key] === 'object') matchers[key] = inputMatchers[key] as { type: string, handle: Function }
+      })
+    }
+
     await each(async () => {
       const currHead = head()
       if (!currHead) return
 
-      const propertyMatcher = matchers ? matchers[currHead] || matchers['#any'] : null
-      // Set the object key matching the current head to tail() if no matchers are specified.
-      if (!matchers) obj[currHead] = tail().trim()
-      // Or if matchers[currHead] is `true`, or if #any is true.
-      else if (propertyMatcher === true) obj[currHead] = tail().trim()
+      const currentMatcher = matchers[currHead] || matchers['#any']
+      if (!currentMatcher) return
+
+      if (currentMatcher.type === 'normal') obj[currHead] = await currentMatcher.handle()
+      // Allows matching the same key more than once.
+      else if (currentMatcher.type === 'collection') {
+        if (!obj[currHead]) obj[currHead] = []
+        obj[currHead].push(await currentMatcher.handle())
+      }
       // If matchers[currHead] or matchers[#any] is a function, set object key to its output.
-      else if (propertyMatcher) obj[currHead] = await propertyMatcher()
       // If we get to this point and matchers[#text] is set, parse all remaining block contents as text.
       // TODO: I still don't like this.
       else if (matchers?.['#text']) obj['#text'] = await blockAsText(level())
 
       // Bail early as soon as we know all keys have been matched.
-      if (matchers && Object.keys(matchers).every(k => obj[k] !== undefined)) return true
+      if (matchers && Object.keys(matchers).every(key => {
+        // If we have any collection keys, we have to continue searching all the way to the end of the current block
+        // as there may be more than one entry.
+        if (['collection'].includes(matchers[key].type)) return false
+        return obj[key] !== undefined
+      })) return true
     })
     return obj
   }