@json-to-office/jto-ops 1.4.0 → 1.11.0

package/dist/index.js

@@ -892,6 +892,36 @@ function toGenerationWarnings(raw) {
     }
   }));
 }
+function preparedThemeLabel(prepared) {
+  const label = prepared.metadata?.themeLabel;
+  return typeof label === "string" ? label : void 0;
+}
+var PREPARED_SOURCE = Symbol("jto.prepared.source");
+var PREPARED_WARNINGS = Symbol("jto.prepared.warnings");
+function stampPrepared(prepared, source, warnings) {
+  return Object.assign(prepared, {
+    [PREPARED_SOURCE]: source,
+    [PREPARED_WARNINGS]: warnings
+  });
+}
+function preparedFor(prepared, document) {
+  if (!prepared) return void 0;
+  const source = prepared[PREPARED_SOURCE];
+  return source !== void 0 && source === document ? prepared : void 0;
+}
+function warningKey(warning) {
+  return `${warning.component}|${warning.context?.code ?? ""}|${warning.message}`;
+}
+function withoutPreparedWarnings(warnings, prepared) {
+  const emitted = prepared?.[PREPARED_WARNINGS];
+  if (!emitted || emitted.length === 0) return warnings;
+  const seen = new Set(emitted.map(warningKey));
+  return warnings.filter((warning) => !seen.has(warningKey(warning)));
+}
+function documentRenderer(document) {
+  const renderer = document?.renderer;
+  return typeof renderer === "string" ? renderer : void 0;
+}
 var UNSAFE_KEYS = /* @__PURE__ */ new Set(["__proto__", "constructor", "prototype"]);
 function safeThemeKey(name) {
   return name && !UNSAFE_KEYS.has(name) ? name : "custom";
@@ -967,12 +997,24 @@ var DocxFormatAdapter = class {
   async generateBuffer(json, options) {
     const core = await import("@json-to-office/core-docx");
     const parsed = typeof json === "string" ? JSON.parse(json) : json;
-    const resolved = await this.resolveThemes(options);
-    const { document: docDefinition, customThemes } = withRequestedTheme(
-      parsed,
-      resolved.requested,
-      resolved.customThemes
+    const prepared = preparedFor(
+      options.prepared?.format === "docx" ? options.prepared : void 0,
+      parsed
     );
+    let docDefinition;
+    let customThemes;
+    if (prepared) {
+      docDefinition = prepared.model.authored;
+    } else {
+      const resolved = await this.resolveThemes(options);
+      const normalized = withRequestedTheme(
+        parsed,
+        resolved.requested,
+        resolved.customThemes
+      );
+      docDefinition = normalized.document;
+      customThemes = normalized.customThemes;
+    }
     const services = buildDocxServices();
     const warnings = [];
     const buffer = await core.generateBufferFromJson(docDefinition, {
@@ -986,10 +1028,13 @@ var DocxFormatAdapter = class {
       generatedAt: options.generatedAt,
       baseDir: options.baseDir,
       renderer: options.renderer,
+      svgRasterFallback: options.svgRasterFallback,
+      prepared,
       warnings
     });
-    emitGenerationWarnings(warnings);
-    options.warnings?.push(...toGenerationWarnings(warnings));
+    const emitted = withoutPreparedWarnings(warnings, prepared);
+    emitGenerationWarnings(emitted);
+    options.warnings?.push(...toGenerationWarnings(emitted));
     return buffer;
   }
   async createGenerator(plugins, options) {
@@ -997,32 +1042,46 @@ var DocxFormatAdapter = class {
     const hasPlugins = plugins.length > 0;
     const pluginNames = plugins.map((p) => p.name);
     const services = buildDocxServices();
-    const {
-      requested: requestedTheme,
-      customThemes,
-      label: themeLabel
-    } = await this.resolveThemes(options);
+    const prepared = !hasPlugins && options.prepared?.format === "docx" ? options.prepared : void 0;
+    const resolved = prepared ? void 0 : await this.resolveThemes(options);
+    const requestedTheme = resolved?.requested;
+    const customThemes = resolved?.customThemes;
+    const themeLabel = prepared ? preparedThemeLabel(prepared) : resolved?.label;
     if (!hasPlugins) {
+      let fallbackThemes;
+      const themesFor = () => fallbackThemes ??= resolved ? Promise.resolve(resolved) : this.resolveThemes(options);
       return {
         generateBuffer: async (document) => {
           const parsed = typeof document === "string" ? JSON.parse(document) : document;
-          const { document: docDefinition, customThemes: themes } = withRequestedTheme(parsed, requestedTheme, customThemes);
+          const usable = preparedFor(prepared, parsed);
+          const themes = usable ? void 0 : await themesFor();
+          const normalized = usable ? { document: usable.model.authored, customThemes: void 0 } : withRequestedTheme(
+            parsed,
+            themes?.requested,
+            themes?.customThemes
+          );
           const warnings = [];
-          const buffer = await core.generateBufferFromJson(docDefinition, {
-            customThemes: themes,
-            services,
-            fonts: options.fonts,
-            validation: {
-              allowUnknownFields: options.validation?.allowUnknownFields
-            },
-            deterministic: options.deterministic,
-            generatedAt: options.generatedAt,
-            baseDir: options.baseDir,
-            renderer: options.renderer,
-            warnings
-          });
-          emitGenerationWarnings(warnings);
-          options.warnings?.push(...toGenerationWarnings(warnings));
+          const buffer = await core.generateBufferFromJson(
+            normalized.document,
+            {
+              customThemes: normalized.customThemes,
+              services,
+              fonts: options.fonts,
+              validation: {
+                allowUnknownFields: options.validation?.allowUnknownFields
+              },
+              deterministic: options.deterministic,
+              generatedAt: options.generatedAt,
+              baseDir: options.baseDir,
+              renderer: options.renderer,
+              svgRasterFallback: options.svgRasterFallback,
+              prepared: usable,
+              warnings
+            }
+          );
+          const emitted = withoutPreparedWarnings(warnings, usable);
+          emitGenerationWarnings(emitted);
+          options.warnings?.push(...toGenerationWarnings(emitted));
           return buffer;
         },
         hasPlugins: false,
@@ -1045,7 +1104,8 @@ var DocxFormatAdapter = class {
       deterministic: options.deterministic,
       generatedAt: options.generatedAt,
       baseDir: options.baseDir,
-      renderer: options.renderer
+      renderer: options.renderer,
+      svgRasterFallback: options.svgRasterFallback
     });
     for (const plugin of plugins) {
       generator = generator.addComponent(plugin);
@@ -1065,7 +1125,8 @@ var DocxFormatAdapter = class {
           deterministic: options.deterministic,
           generatedAt: options.generatedAt,
           baseDir: options.baseDir,
-          renderer: options.renderer
+          renderer: options.renderer,
+          svgRasterFallback: options.svgRasterFallback
         });
         emitGenerationWarnings(result.warnings ?? []);
         options.warnings?.push(...toGenerationWarnings(result.warnings));
@@ -1104,6 +1165,64 @@ var DocxFormatAdapter = class {
       ...result.errors.length > 0 && { errors: result.errors }
     };
   }
+  async analyzeQuality(doc, options = {}) {
+    const core = await import("@json-to-office/core-docx");
+    const parsed = typeof doc === "string" ? JSON.parse(doc) : doc;
+    let prepared = preparedFor(
+      options.prepared?.format === "docx" ? options.prepared : void 0,
+      parsed
+    );
+    if (!prepared) {
+      try {
+        prepared = await this.prepareModel(parsed, options, []);
+      } catch {
+      }
+    }
+    return core.analyzeDocxQuality(prepared?.model.authored ?? parsed, {
+      ...prepared && { prepared },
+      renderer: options.renderer ?? documentRenderer(parsed),
+      profile: options.quality?.profile,
+      policy: options.quality?.policy
+    });
+  }
+  async prepareDocument(doc, options = {}) {
+    const warnings = [];
+    const prepared = await this.prepareModel(doc, options, warnings);
+    emitGenerationWarnings(warnings);
+    options.warnings?.push(...toGenerationWarnings(warnings));
+    return prepared;
+  }
+  /** Prepare into a caller-owned sink; `prepareDocument` owns the reporting. */
+  async prepareModel(doc, options, warnings) {
+    const core = await import("@json-to-office/core-docx");
+    const parsed = typeof doc === "string" ? JSON.parse(doc) : doc;
+    const resolved = await this.resolveThemes(options);
+    const normalized = withRequestedTheme(
+      parsed,
+      resolved.requested,
+      resolved.customThemes
+    );
+    const prepared = core.prepareDocxQualityDocument(
+      normalized.document,
+      {
+        customThemes: normalized.customThemes,
+        fonts: options.fonts,
+        renderer: options.renderer ?? documentRenderer(parsed),
+        warnings
+      }
+    );
+    return stampPrepared(
+      {
+        ...prepared,
+        metadata: {
+          ...prepared.metadata,
+          ...resolved.label && { themeLabel: resolved.label }
+        }
+      },
+      parsed,
+      toGenerationWarnings(warnings)
+    );
+  }
   generateSchema(_options) {
     return null;
   }
@@ -1115,6 +1234,10 @@ var DocxFormatAdapter = class {
       return {};
     }
   }
+  async getBuiltinThemeValues() {
+    const core = await import("@json-to-office/core-docx");
+    return core.themes || {};
+  }
   async resolveTheme(options) {
     const core = await import("@json-to-office/core-docx");
     const { requested } = await this.resolveThemes(options);
@@ -1227,12 +1350,24 @@ var PptxFormatAdapter = class {
   async generateBuffer(json, options) {
     const core = await import("@json-to-office/core-pptx");
     const parsed = typeof json === "string" ? JSON.parse(json) : json;
-    const resolved = await this.resolveThemes(options);
-    const { document: docDefinition, customThemes } = withRequestedTheme(
-      parsed,
-      resolved.requested,
-      resolved.customThemes
+    const prepared = preparedFor(
+      options.prepared?.format === "pptx" ? options.prepared : void 0,
+      parsed
     );
+    let docDefinition;
+    let customThemes;
+    if (prepared) {
+      docDefinition = prepared.model.authored;
+    } else {
+      const resolved = await this.resolveThemes(options);
+      const normalized = withRequestedTheme(
+        parsed,
+        resolved.requested,
+        resolved.customThemes
+      );
+      docDefinition = normalized.document;
+      customThemes = normalized.customThemes;
+    }
     const services = buildServicesFromEnv();
     const result = await core.generateBufferWithWarnings(docDefinition, {
       customThemes,
@@ -1244,11 +1379,15 @@ var PptxFormatAdapter = class {
       deterministic: options.deterministic,
       generatedAt: options.generatedAt,
       baseDir: options.baseDir,
-      renderer: options.renderer
+      renderer: options.renderer,
+      prepared
     });
-    const normalized = toGenerationWarnings(result.warnings);
-    emitGenerationWarnings(normalized);
-    options.warnings?.push(...normalized);
+    const emitted = withoutPreparedWarnings(
+      toGenerationWarnings(result.warnings),
+      prepared
+    );
+    emitGenerationWarnings(emitted);
+    options.warnings?.push(...emitted);
     return result.buffer;
   }
   async createGenerator(plugins, options) {
@@ -1256,31 +1395,46 @@ var PptxFormatAdapter = class {
     const hasPlugins = plugins.length > 0;
     const pluginNames = plugins.map((p) => p.name);
     const services = buildServicesFromEnv();
-    const {
-      requested: requestedTheme,
-      customThemes,
-      label: themeLabel
-    } = await this.resolveThemes(options);
+    const prepared = !hasPlugins && options.prepared?.format === "pptx" ? options.prepared : void 0;
+    const resolved = prepared ? void 0 : await this.resolveThemes(options);
+    const requestedTheme = resolved?.requested;
+    const customThemes = resolved?.customThemes;
+    const themeLabel = prepared ? preparedThemeLabel(prepared) : resolved?.label;
     if (!hasPlugins) {
+      let fallbackThemes;
+      const themesFor = () => fallbackThemes ??= resolved ? Promise.resolve(resolved) : this.resolveThemes(options);
       return {
         generateBuffer: async (document) => {
           const parsed = typeof document === "string" ? JSON.parse(document) : document;
-          const { document: docDefinition, customThemes: themes } = withRequestedTheme(parsed, requestedTheme, customThemes);
-          const result = await core.generateBufferWithWarnings(docDefinition, {
-            customThemes: themes,
-            services,
-            fonts: options.fonts,
-            validation: {
-              allowUnknownFields: options.validation?.allowUnknownFields
-            },
-            deterministic: options.deterministic,
-            generatedAt: options.generatedAt,
-            baseDir: options.baseDir,
-            renderer: options.renderer
-          });
-          const normalized = toGenerationWarnings(result.warnings);
-          emitGenerationWarnings(normalized);
-          options.warnings?.push(...normalized);
+          const usable = preparedFor(prepared, parsed);
+          const themes = usable ? void 0 : await themesFor();
+          const normalized = usable ? { document: usable.model.authored, customThemes: void 0 } : withRequestedTheme(
+            parsed,
+            themes?.requested,
+            themes?.customThemes
+          );
+          const result = await core.generateBufferWithWarnings(
+            normalized.document,
+            {
+              customThemes: normalized.customThemes,
+              services,
+              fonts: options.fonts,
+              validation: {
+                allowUnknownFields: options.validation?.allowUnknownFields
+              },
+              deterministic: options.deterministic,
+              generatedAt: options.generatedAt,
+              baseDir: options.baseDir,
+              renderer: options.renderer,
+              prepared: usable
+            }
+          );
+          const warnings = withoutPreparedWarnings(
+            toGenerationWarnings(result.warnings),
+            usable
+          );
+          emitGenerationWarnings(warnings);
+          options.warnings?.push(...warnings);
           return result.buffer;
         },
         hasPlugins: false,
@@ -1345,6 +1499,66 @@ var PptxFormatAdapter = class {
       ...result.errors.length > 0 && { errors: result.errors }
     };
   }
+  async analyzeQuality(doc, options = {}) {
+    const core = await import("@json-to-office/core-pptx");
+    const parsed = typeof doc === "string" ? JSON.parse(doc) : doc;
+    let prepared = preparedFor(
+      options.prepared?.format === "pptx" ? options.prepared : void 0,
+      parsed
+    );
+    if (!prepared) {
+      try {
+        prepared = await this.prepareModel(parsed, options, []);
+      } catch {
+      }
+    }
+    return core.analyzePptxQuality(prepared?.model.authored ?? parsed, {
+      ...prepared && { prepared },
+      renderer: options.renderer ?? documentRenderer(parsed),
+      profile: options.quality?.profile,
+      policy: options.quality?.policy
+    });
+  }
+  async prepareDocument(doc, options = {}) {
+    const warnings = [];
+    const prepared = await this.prepareModel(doc, options, warnings);
+    const normalized = toGenerationWarnings(warnings);
+    emitGenerationWarnings(normalized);
+    options.warnings?.push(...normalized);
+    return prepared;
+  }
+  /** Prepare into a caller-owned sink; `prepareDocument` owns the reporting. */
+  async prepareModel(doc, options, warnings) {
+    const core = await import("@json-to-office/core-pptx");
+    const parsed = typeof doc === "string" ? JSON.parse(doc) : doc;
+    const resolved = await this.resolveThemes(options);
+    const normalized = withRequestedTheme(
+      parsed,
+      resolved.requested,
+      resolved.customThemes
+    );
+    const prepared = core.preparePptxQualityDocument(
+      normalized.document,
+      {
+        customThemes: normalized.customThemes,
+        fonts: options.fonts,
+        services: buildServicesFromEnv(),
+        renderer: options.renderer ?? documentRenderer(parsed),
+        warnings
+      }
+    );
+    return stampPrepared(
+      {
+        ...prepared,
+        metadata: {
+          ...prepared.metadata,
+          ...resolved.label && { themeLabel: resolved.label }
+        }
+      },
+      parsed,
+      toGenerationWarnings(warnings)
+    );
+  }
   generateSchema(_options) {
     return null;
   }
@@ -1356,6 +1570,10 @@ var PptxFormatAdapter = class {
       return {};
     }
   }
+  async getBuiltinThemeValues() {
+    const core = await import("@json-to-office/core-pptx");
+    return core.pptxThemes || {};
+  }
   async resolveTheme(options) {
     const core = await import("@json-to-office/core-pptx");
     const themes = core.pptxThemes || {};
@@ -1448,6 +1666,116 @@ function createAdapter(format) {
       throw new Error(`Unknown format: ${format}`);
   }
 }
+
+// src/pdf-text-geometry.ts
+import { execFile as execFile2 } from "child_process";
+import * as path6 from "path";
+var ENTITIES = {
+  "&": "&",
+  "&lt;": "<",
+  "&gt;": ">",
+  "&quot;": '"',
+  "&apos;": "'",
+  "&#34;": '"',
+  "&#39;": "'"
+};
+function decodeEntities(value) {
+  return value.replace(
+    /&(?:amp|lt|gt|quot|apos|#34|#39);/g,
+    (entity) => ENTITIES[entity] ?? entity
+  );
+}
+var PAGE_PATTERN = /<page\s+width="([\d.]+)"\s+height="([\d.]+)">([\s\S]*?)<\/page>/g;
+var WORD_PATTERN = /<word\s+xMin="(-?[\d.]+)"\s+yMin="(-?[\d.]+)"\s+xMax="(-?[\d.]+)"\s+yMax="(-?[\d.]+)">([\s\S]*?)<\/word>/g;
+function parsePdfTextBbox(bboxXml) {
+  const pages = [];
+  for (const pageMatch of bboxXml.matchAll(PAGE_PATTERN)) {
+    const words = [];
+    for (const wordMatch of pageMatch[3].matchAll(WORD_PATTERN)) {
+      words.push({
+        xMin: Number(wordMatch[1]),
+        yMin: Number(wordMatch[2]),
+        xMax: Number(wordMatch[3]),
+        yMax: Number(wordMatch[4]),
+        text: decodeEntities(wordMatch[5])
+      });
+    }
+    pages.push({
+      widthPt: Number(pageMatch[1]),
+      heightPt: Number(pageMatch[2]),
+      words
+    });
+  }
+  return pages;
+}
+function pdftotextCandidates() {
+  const candidates = [];
+  const configured = process.env.PDFTOTEXT_PATH?.trim();
+  if (configured) candidates.push(configured);
+  candidates.push("pdftotext");
+  return [...new Set(candidates)];
+}
+async function run(binary, args, timeoutMs) {
+  return new Promise((resolve2, reject) => {
+    execFile2(
+      binary,
+      args,
+      { timeout: timeoutMs, maxBuffer: 64 * 1024 * 1024 },
+      (error, stdout) => {
+        if (error) reject(error);
+        else resolve2(stdout);
+      }
+    );
+  });
+}
+var pdftotextPromise;
+async function resolvePdftotext() {
+  if (!pdftotextPromise) {
+    pdftotextPromise = (async () => {
+      for (const candidate of pdftotextCandidates()) {
+        if (candidate.includes(path6.sep)) {
+          try {
+            await run(candidate, ["-v"], 1e4);
+            return candidate;
+          } catch {
+            continue;
+          }
+        }
+        try {
+          await run(candidate, ["-v"], 1e4);
+          return candidate;
+        } catch (error) {
+          const code = error.code;
+          if (code !== "ENOENT" && code !== "EACCES") return candidate;
+        }
+      }
+      throw new Error(
+        `Text geometry extraction needs pdftotext (poppler), which was not found. Install poppler-utils or set PDFTOTEXT_PATH (searched: ${pdftotextCandidates().join(", ")}).`
+      );
+    })().catch((error) => {
+      pdftotextPromise = void 0;
+      throw error;
+    });
+  }
+  return pdftotextPromise;
+}
+async function pdftotextAvailable() {
+  try {
+    await resolvePdftotext();
+    return true;
+  } catch {
+    return false;
+  }
+}
+async function extractPdfTextGeometry(pdfPath, options = {}) {
+  const binary = await resolvePdftotext();
+  const stdout = await run(
+    binary,
+    ["-bbox", pdfPath, "-"],
+    options.timeoutMs ?? 6e4
+  );
+  return parsePdfTextBbox(stdout);
+}
 export {
   DocxFormatAdapter,
   FontconfigStager,
@@ -1460,8 +1788,11 @@ export {
   createLibreOfficePptxBatchRasterizer,
   createLibreOfficePptxRasterizer,
   emitDiagnostic,
+  extractPdfTextGeometry,
   getFontStager,
   getRasterizerCacheStats,
+  parsePdfTextBbox,
+  pdftotextAvailable,
   runWithDiagnosticSink,
   stderrDiagnosticSink
 };