@gmb/bitmark-parser 6.11.0 → 7.0.0

package/dist/index.cjs

@@ -39,14 +39,16 @@ __export(src_exports, {
   canonicalize: () => canonicalize,
   convert: () => convert,
   countBits: () => countBits,
+  diff: () => diff,
+  diffBits: () => diffBits,
   info: () => info,
   init: () => init,
   initSync: () => initSync,
   legacy: () => legacy_exports,
-  lex: () => lex,
   objectsToBitmark: () => objectsToBitmark,
   patchEntry: () => patchEntry,
   register: () => register,
+  semanticTokens: () => semanticTokens,
   splitBits: () => splitBits,
   transform: () => transform,
   transformParallel: () => transformParallel,
@@ -66,7 +68,7 @@ var import_node_module = require("module");
 // src/generated/package_info.ts
 var PACKAGE_INFO = {
   "name": "@gmb/bitmark-parser",
-  "version": "6.11.0",
+  "version": "7.0.0",
   "author": "Get More Brain Ltd <info@getmorebrain.com>",
   "license": "ISC",
   "description": "A parser for bitmark text, powered by WebAssembly."
@@ -105,8 +107,11 @@ function getActive() {
 }
 function setActive(feature, module2) {
   for (const [type, data] of registrations) {
-    const error = module2.register(type, data);
-    if (error) throw new RegisterError(error.replace(/^error:\s*/, ""));
+    try {
+      module2.register(type, data);
+    } catch (e) {
+      throw new RegisterError(e instanceof Error ? e.message : String(e));
+    }
   }
   active = module2;
   activeFeature = feature;
@@ -122,38 +127,24 @@ function variant() {
 }
 
 // src/core/hooked-transform.ts
-function isError(s) {
-  return s.startsWith("error:");
-}
-function errorString(e) {
-  const m = e instanceof Error ? e.message : String(e);
-  return m.startsWith("error:") ? m : `error: ${m}`;
-}
 function stripArray(doc) {
   const t = doc.trim();
   return t.startsWith("[") && t.endsWith("]") ? t.slice(1, -1) : doc;
 }
-function normalisePatch(patch) {
-  if (patch === void 0) return [];
-  if (typeof patch === "string") {
-    const trimmed = patch.trim();
-    return trimmed === "" ? [] : JSON.parse(trimmed);
-  }
-  return patch;
+function patchJsonOf(patch) {
+  if (patch === void 0) return "";
+  if (typeof patch === "string") return patch.trim();
+  return JSON.stringify(patch);
 }
 function buildUnits(wasm, input, inputIsBitmark) {
   if (inputIsBitmark) {
-    const out2 = wasm.split_bits(input);
-    if (isError(out2)) throw new Error(out2);
-    const slices = JSON.parse(out2);
+    const slices = JSON.parse(wasm.split_bits(input));
     return slices.map((s) => ({
       view: { format: "bitmark", index: s.index, total: slices.length, source: s.source },
       source: s.source
     }));
   }
-  const out = wasm.split_json_bits(input);
-  if (isError(out)) throw new Error(out);
-  const envelopes = JSON.parse(out);
+  const envelopes = JSON.parse(wasm.split_json_bits(input));
   return envelopes.map((env, index) => ({
     view: { format: "json", index, total: envelopes.length, bit: env.bit ?? {} },
     // Per-bit transform consumes a one-bit JSON document.
@@ -169,28 +160,54 @@ function planTransform(wasm, input, options) {
   const sel = options.inputFormat ?? "auto";
   const inputIsBitmark = sel === "bitmark" || sel !== "json" && wasm.is_bitmark(input);
   const inputFormat = inputIsBitmark ? "bitmark" : "json";
-  const staticPatch = normalisePatch(options.patch);
   const units = buildUnits(wasm, input, inputIsBitmark);
   const total = units.length;
-  const jobs = [];
-  const views = [];
+  const hooked = [];
   for (let index = 0; index < total; index++) {
-    const { view, source } = units[index];
+    const { view } = units[index];
     view.index = index;
     view.total = total;
     const result = options.preHook ? options.preHook(view) || {} : {};
-    if (result.drop) continue;
-    const patches = result.patches ? [...staticPatch, ...result.patches] : staticPatch;
-    jobs.push({
-      index,
-      source,
-      inputFormat,
-      outputFormat: result.outputFormat ?? output,
-      mode,
-      patchJson: JSON.stringify(patches),
-      spacesAroundValues
-    });
-    views.push(view);
+    hooked.push(result.drop ? null : { patches: result.patches ?? [], outputFormat: result.outputFormat ?? output });
+  }
+  const patchJson = patchJsonOf(options.patch);
+  let resolved;
+  if (patchJson === "") {
+    resolved = units.map((_, index) => ({ source: index, bit: null, patch: [] }));
+  } else {
+    resolved = JSON.parse(wasm.resolve_patch_document(input, patchJson));
+  }
+  const jobs = [];
+  const views = [];
+  for (const item of resolved) {
+    if (item.source !== null) {
+      const hook = hooked[item.source];
+      if (!hook) continue;
+      jobs.push({
+        index: item.source,
+        source: units[item.source].source,
+        inputFormat,
+        outputFormat: hook.outputFormat,
+        mode,
+        // The document's patches first, the hook's after (the hook contract).
+        patchJson: JSON.stringify([...item.patch, ...hook.patches]),
+        spacesAroundValues
+      });
+      views.push(units[item.source].view);
+    } else {
+      const bit = item.bit ?? {};
+      const index = jobs.length;
+      jobs.push({
+        index,
+        source: JSON.stringify([{ bit }]),
+        inputFormat: "json",
+        outputFormat: output,
+        mode,
+        patchJson: JSON.stringify(item.patch),
+        spacesAroundValues
+      });
+      views.push({ format: "json", index, total, bit });
+    }
   }
   return { jobs, views, output, pretty, indent };
 }
@@ -205,7 +222,6 @@ function computeBit(wasm, job) {
     2,
     job.spacesAroundValues
   );
-  if (isError(out)) return out;
   return job.outputFormat === "json" ? stripArray(out) : out;
 }
 function recombine(wasm, elements, output, pretty, indent) {
@@ -214,16 +230,10 @@ function recombine(wasm, elements, output, pretty, indent) {
   return pretty ? wasm.prettify(arr, indent) : arr;
 }
 function runHookedTransform(wasm, input, options) {
-  let plan;
-  try {
-    plan = planTransform(wasm, input, options);
-  } catch (e) {
-    return errorString(e);
-  }
+  const plan = planTransform(wasm, input, options);
   const rendered = [];
   for (let i = 0; i < plan.jobs.length; i++) {
     const element = computeBit(wasm, plan.jobs[i]);
-    if (isError(element)) return element;
     options.postHook?.(plan.views[i], element);
     rendered.push(element);
   }
@@ -232,20 +242,9 @@ function runHookedTransform(wasm, input, options) {
 
 // src/core/parallel.ts
 async function orchestrateParallel(wasm, input, options, runJobs) {
-  let plan;
-  try {
-    plan = planTransform(wasm, input, options);
-  } catch (e) {
-    return errorString(e);
-  }
-  let elements;
-  try {
-    elements = await runJobs(plan.jobs, options.workers ?? 4);
-  } catch (e) {
-    return errorString(e);
-  }
+  const plan = planTransform(wasm, input, options);
+  const elements = await runJobs(plan.jobs, options.workers ?? 4);
   for (let i = 0; i < elements.length; i++) {
-    if (isError(elements[i])) return elements[i];
     options.postHook?.(plan.views[i], elements[i]);
   }
   return recombine(wasm, elements, plan.output, plan.pretty, plan.indent);
@@ -600,10 +599,7 @@ var Builder = class {
       }
     }
     this.parseStringTexts(bit);
-    const out = wasm.convert(JSON.stringify([{ bit }]), "json", "json", "optimized", false, 2, 1, false, false);
-    if (out.startsWith("error:")) {
-      throw new Error(out.slice("error:".length).trim());
-    }
+    const out = wasm.convert(JSON.stringify([{ bit }]), "json", "json", "optimized", false, 2, 1, false, false, "", "");
     const entries = JSON.parse(out);
     const built = entries[0]?.bit ?? bit;
     for (const [key, value] of Object.entries(bit)) {
@@ -740,8 +736,7 @@ var BitmarkParserGenerator = class {
     const emit = coreFormat === "bitmark" ? prettyIndent(bitmarkOptions.prettifyJson) : { pretty, indent };
     const spaces = spacesAround(bitmarkOptions.spacesAroundValues);
     const includeUnknown = !jsonOptions.excludeUnknownProperties;
-    const out = wasm.convert(source, isBitmark ? "bitmark" : "json", coreFormat, "optimized", emit.pretty, emit.indent, spaces, includeUnknown, false);
-    this.throwOnError(out);
+    const out = wasm.convert(source, isBitmark ? "bitmark" : "json", coreFormat, "optimized", emit.pretty, emit.indent, spaces, includeUnknown, false, "", "");
     if (outputFormat === "bitmark" || outputFormat === "text") {
       return this.deliver(out, opts.outputFile, opts.fileOptions);
     }
@@ -775,7 +770,6 @@ var BitmarkParserGenerator = class {
     );
     const spaces = spacesAround(bitmarkOptions.spacesAroundValues);
     const out = wasm.canonicalize(source, isBitmark ? "bitmark" : "json", "optimized", pretty, indent, spaces);
-    this.throwOnError(out);
     if (isBitmark) return this.deliver(out, opts.outputFile, opts.fileOptions);
     if (opts.outputFile) return this.deliver(out, opts.outputFile, opts.fileOptions);
     if (pretty || jsonOptions.stringify) return out;
@@ -792,7 +786,6 @@ var BitmarkParserGenerator = class {
     const { pretty, indent } = prettyIndent(opts.prettify);
     const coreFormat = format === "pojo" ? "json" : format;
     const out = wasm.info(type, coreFormat, opts.bit ?? "", pretty, indent, false, false, true, "");
-    this.throwOnError(out);
     if (format === "pojo") return JSON.parse(out);
     return out;
   }
@@ -824,8 +817,7 @@ var BitmarkParserGenerator = class {
     const isHtmlInput = opts.inputFormat === "html" || opts.inputFormat === void 0 && typeof input === "string" && /<table[\s>]/iu.test(inStr);
     if (!isHtmlInput) {
       const { source, isBitmark } = this.resolveInput(input, void 0);
-      const out2 = wasm.convert(source, isBitmark ? "bitmark" : "json", "html", "optimized", false, 2, 1, false, false);
-      this.throwOnError(out2);
+      const out2 = wasm.convert(source, isBitmark ? "bitmark" : "json", "html", "optimized", false, 2, 1, false, false, "", "");
       return this.deliver(extractHtmlTables(out2).join("\n"), opts.outputFile, opts.fileOptions);
     }
     const bitType = opts.tableFormat === "table" ? "table" : "table-extended";
@@ -834,13 +826,11 @@ var BitmarkParserGenerator = class {
     const { pretty, indent } = prettyIndent(jsonOptions.prettify);
     const spaces = spacesAround(bitmarkOptions.spacesAroundValues);
     if (outputFormat === "html") {
-      const out2 = wasm.convert(wrapped, "html", "html", "optimized", false, 2, 1, false, false);
-      this.throwOnError(out2);
+      const out2 = wasm.convert(wrapped, "html", "html", "optimized", false, 2, 1, false, false, "", "");
       return this.deliver(extractHtmlTables(out2).join("\n"), opts.outputFile, opts.fileOptions);
     }
     const coreFormat = outputFormat === "ast" ? "json" : outputFormat;
-    const out = wasm.convert(wrapped, "html", coreFormat, "optimized", pretty, indent, spaces, false, false);
-    this.throwOnError(out);
+    const out = wasm.convert(wrapped, "html", coreFormat, "optimized", pretty, indent, spaces, false, false, "", "");
     if (outputFormat === "bitmark" || outputFormat === "text") {
       return this.deliver(out, opts.outputFile, opts.fileOptions);
     }
@@ -864,8 +854,7 @@ var BitmarkParserGenerator = class {
     const opts = Object.assign({}, options);
     const wasm = getLegacyWasm();
     const { source, isBitmark } = this.resolveInput(input, opts.inputFormat);
-    const out = wasm.convert(source, isBitmark ? "bitmark" : "json", "json", "optimized", false, 2, 1, true, false);
-    this.throwOnError(out);
+    const out = wasm.convert(source, isBitmark ? "bitmark" : "json", "json", "optimized", false, 2, 1, true, false, "", "");
     return makePseudoAst(JSON.parse(out));
   }
   /**
@@ -917,11 +906,9 @@ var BitmarkParserGenerator = class {
     }
     if (ast !== void 0) {
       const out2 = isBitmarkText ? wasm.convert_text(JSON.stringify(ast), "astToText", "bitmark++", textLocation) : wasm.convert_text(JSON.stringify(ast), "astToPlainText", "plainText", textLocation);
-      this.throwOnError(out2);
       return this.deliver(out2, opts.outputFile, opts.fileOptions);
     }
     const out = wasm.convert_text(inStr, "textToAst", isBitmarkText ? "bitmark++" : "plainText", "body");
-    this.throwOnError(out);
     const { pretty, indent } = prettyIndent(jsonOptions.prettify);
     const outStr = pretty ? wasm.prettify(out, indent) : out;
     if (opts.outputFile) return this.deliver(outStr, opts.outputFile, opts.fileOptions);
@@ -939,7 +926,6 @@ var BitmarkParserGenerator = class {
       "bitmark++",
       "body"
     );
-    this.throwOnError(out);
     return out;
   }
   /**
@@ -954,7 +940,6 @@ var BitmarkParserGenerator = class {
     if (inputFormat === "bitmarkText") {
       const ast = wasm.convert_text(String(input), "textToAst", "bitmark++", "body");
       const out2 = wasm.convert_text(ast, "astToPlainText", "bitmark++", "body");
-      this.throwOnError(out2);
       return out2;
     }
     if (typeof input === "string" && this.parseTextAstString(input) === void 0 && !wasm.is_bitmark(input)) {
@@ -971,8 +956,7 @@ var BitmarkParserGenerator = class {
       }
     }
     const { source, isBitmark } = this.resolveInput(input, void 0);
-    const out = wasm.convert(source, isBitmark ? "bitmark" : "json", "text", "optimized", false, 2, 1, false, false);
-    this.throwOnError(out);
+    const out = wasm.convert(source, isBitmark ? "bitmark" : "json", "text", "optimized", false, 2, 1, false, false, "", "");
     return out;
   }
   /** If `s` parses as a TextAst JSON array, return it. */
@@ -1076,12 +1060,6 @@ var BitmarkParserGenerator = class {
       encoding: fileOptions?.encoding ?? "utf8"
     });
   }
-  /** The core reports failures as `error:`-prefixed strings; bpg throws. */
-  throwOnError(out) {
-    if (out.startsWith("error:")) {
-      throw new Error(out.slice("error:".length).trim());
-    }
-  }
 };
 function extractHtmlTables(html) {
   const tables = [];
@@ -1118,7 +1096,6 @@ function extractHtmlTables(html) {
 }
 function bitmarkTextParse(text) {
   const out = getLegacyWasm().convert_text(text, "textToAst", "bitmark++", "body");
-  if (out.startsWith("error:")) throw new Error(out.slice("error:".length).trim());
   return JSON.parse(out);
 }
 
@@ -1448,11 +1425,6 @@ function patchEntry(path2, op, value) {
 var ALL_LANGUAGES = "all";
 
 // src/core/api.ts
-function lex(input, options = {}) {
-  const wasm = getActive();
-  if (typeof wasm.lex !== "function") throw new UnsupportedFeatureError("lex", variant());
-  return wasm.lex(input, options.stage ?? "lex");
-}
 function bitmarkToObjects(input, options = {}) {
   const out = getActive().convert(
     input,
@@ -1463,9 +1435,10 @@ function bitmarkToObjects(input, options = {}) {
     2,
     1,
     options.includeUnknownProperties ?? false,
-    false
+    false,
+    "",
+    ""
   );
-  if (out.startsWith("error:")) throw new Error(out.slice("error:".length).trim());
   return JSON.parse(out);
 }
 function objectsToBitmark(bits, options = {}) {
@@ -1481,9 +1454,10 @@ function objectsToBitmark(bits, options = {}) {
     2,
     1,
     false,
-    false
+    false,
+    "",
+    ""
   );
-  if (out.startsWith("error:")) throw new Error(out.slice("error:".length).trim());
   return out;
 }
 function convert(input, options = {}) {
@@ -1496,7 +1470,19 @@ function convert(input, options = {}) {
     options.indent ?? 2,
     options.spacesAroundValues ?? 1,
     options.includeUnknownProperties ?? false,
-    options.mappingReport ?? false
+    options.mappingReport ?? false,
+    options.positionEncoding ?? "",
+    options.tokensLayout ?? ""
+  );
+}
+function semanticTokens(input, options = {}) {
+  return JSON.parse(
+    convert(input, {
+      inputFormat: "bitmark",
+      outputFormat: "semantic-tokens",
+      positionEncoding: options.positionEncoding,
+      tokensLayout: options.tokensLayout
+    })
   );
 }
 function canonicalize(input, options = {}) {
@@ -1510,9 +1496,7 @@ function canonicalize(input, options = {}) {
   );
 }
 function countBits(input) {
-  const out = getActive().count_bits(input);
-  if (out.startsWith("error:")) throw new Error(out);
-  return Number.parseInt(out, 10);
+  return Number.parseInt(getActive().count_bits(input), 10);
 }
 function splitBits(input) {
   return JSON.parse(getActive().split_bits(input));
@@ -1534,6 +1518,25 @@ function transform(input, options = {}) {
     options.spacesAroundValues ?? 1
   );
 }
+function diff(a, b, options = {}) {
+  const wasm = getActive();
+  if (typeof wasm.diff !== "function") throw new UnsupportedFeatureError("diff", variant());
+  const out = wasm.diff(
+    a,
+    b,
+    options.inputFormat ?? "auto",
+    options.outputFormat ?? "bitmark",
+    options.context ?? 3,
+    options.locate ?? false,
+    options.similarity ?? 0.6,
+    options.bboxTolerance ?? 8,
+    options.spacesAroundValues ?? 0
+  );
+  return out;
+}
+function diffBits(a, b, options = {}) {
+  return JSON.parse(diff(a, b, { ...options, outputFormat: "json" }));
+}
 function breakscapeText(input, options = {}) {
   const format = options.format ?? "bitmark++";
   const location = options.location ?? "body";
@@ -1554,7 +1557,11 @@ function info(options = {}) {
   const onlyDeprecated = options.onlyDeprecated ?? false;
   const full = options.full ?? false;
   const language = typeof options.language === "string" ? options.language : options.language?.join(",") ?? "";
-  return getActive().info(
+  const wasm = getActive();
+  if (format !== "json" && typeof wasm.info_text_supported !== "function") {
+    throw new UnsupportedFeatureError('info (format: "text")', variant());
+  }
+  return wasm.info(
     infoType,
     format,
     bit,
@@ -1567,8 +1574,11 @@ function info(options = {}) {
   );
 }
 function register(options) {
-  const result = getActive().register(options.type, options.data);
-  if (result) throw new RegisterError(result.replace(/^error:\s*/, ""));
+  try {
+    getActive().register(options.type, options.data);
+  } catch (e) {
+    throw new RegisterError(e instanceof Error ? e.message : String(e));
+  }
   rememberRegistration(options.type, options.data);
 }
 
@@ -1612,14 +1622,16 @@ function transformParallel(input, options = {}) {
   canonicalize,
   convert,
   countBits,
+  diff,
+  diffBits,
   info,
   init,
   initSync,
   legacy,
-  lex,
   objectsToBitmark,
   patchEntry,
   register,
+  semanticTokens,
   splitBits,
   transform,
   transformParallel,