@cesarandreslopez/occ 0.5.1 → 0.6.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (67) hide show
  1. package/README.md +7 -0
  2. package/dist/src/cli.d.ts +21 -0
  3. package/dist/src/cli.js +76 -2
  4. package/dist/src/cli.js.map +1 -1
  5. package/dist/src/code/build.d.ts +2 -1
  6. package/dist/src/code/build.js +6 -2
  7. package/dist/src/code/build.js.map +1 -1
  8. package/dist/src/code/chunk.d.ts +2 -1
  9. package/dist/src/code/chunk.js +5 -3
  10. package/dist/src/code/chunk.js.map +1 -1
  11. package/dist/src/code/command.js +48 -2
  12. package/dist/src/code/command.js.map +1 -1
  13. package/dist/src/code/output.d.ts +2 -0
  14. package/dist/src/code/output.js +20 -0
  15. package/dist/src/code/output.js.map +1 -1
  16. package/dist/src/code/query.d.ts +23 -0
  17. package/dist/src/code/query.js +64 -0
  18. package/dist/src/code/query.js.map +1 -1
  19. package/dist/src/doc/command.js +57 -0
  20. package/dist/src/doc/command.js.map +1 -1
  21. package/dist/src/doc/entities.d.ts +27 -0
  22. package/dist/src/doc/entities.js +102 -0
  23. package/dist/src/doc/entities.js.map +1 -0
  24. package/dist/src/doc/references.d.ts +20 -0
  25. package/dist/src/doc/references.js +112 -0
  26. package/dist/src/doc/references.js.map +1 -0
  27. package/dist/src/inspect/shared.js +2 -2
  28. package/dist/src/inspect/shared.js.map +1 -1
  29. package/dist/src/output/json.js +9 -0
  30. package/dist/src/output/json.js.map +1 -1
  31. package/dist/src/output/tabular.js +18 -10
  32. package/dist/src/output/tabular.js.map +1 -1
  33. package/dist/src/parsers/docx.js +5 -0
  34. package/dist/src/parsers/docx.js.map +1 -1
  35. package/dist/src/parsers/index.d.ts +1 -0
  36. package/dist/src/parsers/index.js +28 -3
  37. package/dist/src/parsers/index.js.map +1 -1
  38. package/dist/src/parsers/odf.js +15 -1
  39. package/dist/src/parsers/odf.js.map +1 -1
  40. package/dist/src/parsers/pdf.js +4 -0
  41. package/dist/src/parsers/pdf.js.map +1 -1
  42. package/dist/src/parsers/pptx.js +4 -0
  43. package/dist/src/parsers/pptx.js.map +1 -1
  44. package/dist/src/parsers/xlsx.js +5 -0
  45. package/dist/src/parsers/xlsx.js.map +1 -1
  46. package/dist/src/progress-event.d.ts +7 -0
  47. package/dist/src/progress-event.js +2 -0
  48. package/dist/src/progress-event.js.map +1 -0
  49. package/dist/src/stats.d.ts +4 -1
  50. package/dist/src/stats.js +18 -8
  51. package/dist/src/stats.js.map +1 -1
  52. package/dist/src/structure/extract.d.ts +2 -0
  53. package/dist/src/structure/extract.js +16 -0
  54. package/dist/src/structure/extract.js.map +1 -1
  55. package/dist/src/structure/index.d.ts +1 -1
  56. package/dist/src/structure/index.js +1 -1
  57. package/dist/src/structure/index.js.map +1 -1
  58. package/dist/src/structure/types.d.ts +2 -0
  59. package/dist/src/structure/types.js +4 -0
  60. package/dist/src/structure/types.js.map +1 -1
  61. package/dist/src/tokens.d.ts +16 -0
  62. package/dist/src/tokens.js +107 -0
  63. package/dist/src/tokens.js.map +1 -0
  64. package/dist/src/types.d.ts +10 -0
  65. package/dist/src/types.js +3 -0
  66. package/dist/src/types.js.map +1 -1
  67. package/package.json +27 -2
@@ -1,3 +1,3 @@
1
- export { extractFromMarkdown, findChunkSection, getSectionContent } from './extract.js';
1
+ export { extractFromMarkdown, findChunkSection, getSectionContent, enrichStructureWithMetrics } from './extract.js';
2
2
  export type { StructureNode, DocumentStructure, PageMapping } from './types.js';
3
3
  export { flatten, getNodeById, getNodeByPath, toDict, fromDict } from './types.js';
@@ -1,3 +1,3 @@
1
- export { extractFromMarkdown, findChunkSection, getSectionContent } from './extract.js';
1
+ export { extractFromMarkdown, findChunkSection, getSectionContent, enrichStructureWithMetrics } from './extract.js';
2
2
  export { flatten, getNodeById, getNodeByPath, toDict, fromDict } from './types.js';
3
3
  //# sourceMappingURL=index.js.map
@@ -1 +1 @@
1
- {"version":3,"file":"index.js","sourceRoot":"","sources":["../../../src/structure/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,mBAAmB,EAAE,gBAAgB,EAAE,iBAAiB,EAAE,MAAM,cAAc,CAAC;AAExF,OAAO,EAAE,OAAO,EAAE,WAAW,EAAE,aAAa,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,YAAY,CAAC"}
1
+ {"version":3,"file":"index.js","sourceRoot":"","sources":["../../../src/structure/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,mBAAmB,EAAE,gBAAgB,EAAE,iBAAiB,EAAE,0BAA0B,EAAE,MAAM,cAAc,CAAC;AAEpH,OAAO,EAAE,OAAO,EAAE,WAAW,EAAE,aAAa,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,YAAY,CAAC"}
@@ -17,6 +17,8 @@ export interface StructureNode {
17
17
  endPage?: number;
18
18
  parentNodeId?: string;
19
19
  structureCode?: string;
20
+ wordCount?: number;
21
+ tokenEstimate?: number;
20
22
  children: StructureNode[];
21
23
  }
22
24
  export declare const DocumentStructureSchema: z.ZodObject<{
@@ -15,6 +15,8 @@ export const StructureNodeSchema = z.object({
15
15
  endPage: z.number().optional(),
16
16
  parentNodeId: z.string().optional(),
17
17
  structureCode: z.string().optional(),
18
+ wordCount: z.number().optional(),
19
+ tokenEstimate: z.number().optional(),
18
20
  children: z.lazy(() => z.array(StructureNodeSchema)),
19
21
  });
20
22
  export const DocumentStructureSchema = z.object({
@@ -61,6 +63,8 @@ export function toDict(structure) {
61
63
  endPage: node.endPage,
62
64
  parentNodeId: node.parentNodeId,
63
65
  structureCode: node.structureCode,
66
+ wordCount: node.wordCount,
67
+ tokenEstimate: node.tokenEstimate,
64
68
  children: node.children.map(nodeToDict),
65
69
  };
66
70
  }
@@ -1 +1 @@
1
- {"version":3,"file":"types.js","sourceRoot":"","sources":["../../../src/structure/types.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAExB,MAAM,CAAC,MAAM,iBAAiB,GAAG,CAAC,CAAC,MAAM,CAAC;IACxC,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE;IACtB,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE;IACrB,OAAO,EAAE,CAAC,CAAC,MAAM,EAAE;CACpB,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,mBAAmB,GAA6B,CAAC,CAAC,MAAM,CAAC;IACpE,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE;IAClB,KAAK,EAAE,CAAC,CAAC,MAAM,EAAE;IACjB,KAAK,EAAE,CAAC,CAAC,MAAM,EAAE;IACjB,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE;IACrB,OAAO,EAAE,CAAC,CAAC,MAAM,EAAE;IACnB,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE;IACrB,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAChC,OAAO,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAC9B,YAAY,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IACnC,aAAa,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IACpC,QAAQ,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,mBAAmB,CAAC,CAAC;CACrD,CAAC,CAAC;AAeH,MAAM,CAAC,MAAM,uBAAuB,GAAG,CAAC,CAAC,MAAM,CAAC;IAC9C,SAAS,EAAE,CAAC,CAAC,KAAK,CAAC,mBAAmB,CAAC;IACvC,YAAY,EAAE,CAAC,CAAC,KAAK,CAAC,iBAAiB,CAAC;IACxC,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE;IACtB,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE;CACrB,CAAC,CAAC;AAGH,MAAM,UAAU,OAAO,CAAC,KAAsB;IAC5C,MAAM,MAAM,GAAoB,EAAE,CAAC;IACnC,SAAS,IAAI,CAAC,IAAqB;QACjC,KAAK,MAAM,IAAI,IAAI,IAAI,EAAE,CAAC;YACxB,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YAClB,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QACtB,CAAC;IACH,CAAC;IACD,IAAI,CAAC,KAAK,CAAC,CAAC;IACZ,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,MAAM,UAAU,WAAW,CAAC,KAAsB,EAAE,MAAc;IAChE,KAAK,MAAM,IAAI,IAAI,OAAO,CAAC,KAAK,CAAC,EAAE,CAAC;QAClC,IAAI,IAAI,CAAC,MAAM,KAAK,MAAM;YAAE,OAAO,IAAI,CAAC;IAC1C,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,MAAM,UAAU,aAAa,CAAC,KAAsB,EAAE,aAAqB;IACzE,KAAK,MAAM,IAAI,IAAI,OAAO,CAAC,KAAK,CAAC,EAAE,CAAC;QAClC,IAAI,IAAI,CAAC,aAAa,KAAK,aAAa;YAAE,OAAO,IAAI,CAAC;IACxD,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,MAAM,UAAU,MAAM,CAAC,SAA4B;IACjD,SAAS,UAAU,CAAC,IAAmB;QACrC,OAAO;YACL,MAAM,EAAE,IAAI,CAAC,MAAM;YACnB,KAAK,EAAE,IAAI,CAAC,KAAK;YACjB,KAAK,EAAE,IAAI,CAAC,KAAK;YACjB,SAAS,EAAE,IAAI,CAAC,SAAS;YACzB,OAAO,EAAE,IAAI,CAAC,OAAO;YACrB,SAAS,EAAE,IAAI,CAAC,SAAS;YACzB,SAAS,EAAE,IAAI,CAAC,SAAS;YACzB,OAAO,EAAE,IAAI,CAAC,OAAO;YACrB,YAAY,EAAE,IAAI,CAAC,YAAY;YAC/B,aAAa,EAAE,IAAI,CAAC,aAAa;YACjC,QAAQ,EAAE,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,UAAU,CAAC;SACxC,CAAC;IACJ,CAAC;IACD,OAAO;QACL,SAAS,EAAE,SAAS,CAAC,SAAS,CAAC,GAAG,CAAC,UAAU,CAAC;QAC9C,YAAY,EAAE,SAAS,CAAC,YAAY;QACpC,UAAU,EAAE,SAAS,CAAC,UAAU;QAChC,QAAQ,EAAE,SAAS,CAAC,QAAQ;KAC7B,CAAC;AACJ,CAAC;AAED,MAAM,UAAU,QAAQ,CAAC,IAA6B;IACpD,OAAO,uBAAuB,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;AAC7C,CAAC"}
1
+ {"version":3,"file":"types.js","sourceRoot":"","sources":["../../../src/structure/types.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAExB,MAAM,CAAC,MAAM,iBAAiB,GAAG,CAAC,CAAC,MAAM,CAAC;IACxC,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE;IACtB,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE;IACrB,OAAO,EAAE,CAAC,CAAC,MAAM,EAAE;CACpB,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,mBAAmB,GAA6B,CAAC,CAAC,MAAM,CAAC;IACpE,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE;IAClB,KAAK,EAAE,CAAC,CAAC,MAAM,EAAE;IACjB,KAAK,EAAE,CAAC,CAAC,MAAM,EAAE;IACjB,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE;IACrB,OAAO,EAAE,CAAC,CAAC,MAAM,EAAE;IACnB,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE;IACrB,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAChC,OAAO,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAC9B,YAAY,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IACnC,aAAa,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IACpC,SAAS,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IAChC,aAAa,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;IACpC,QAAQ,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,mBAAmB,CAAC,CAAC;CACrD,CAAC,CAAC;AAiBH,MAAM,CAAC,MAAM,uBAAuB,GAAG,CAAC,CAAC,MAAM,CAAC;IAC9C,SAAS,EAAE,CAAC,CAAC,KAAK,CAAC,mBAAmB,CAAC;IACvC,YAAY,EAAE,CAAC,CAAC,KAAK,CAAC,iBAAiB,CAAC;IACxC,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE;IACtB,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE;CACrB,CAAC,CAAC;AAGH,MAAM,UAAU,OAAO,CAAC,KAAsB;IAC5C,MAAM,MAAM,GAAoB,EAAE,CAAC;IACnC,SAAS,IAAI,CAAC,IAAqB;QACjC,KAAK,MAAM,IAAI,IAAI,IAAI,EAAE,CAAC;YACxB,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YAClB,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QACtB,CAAC;IACH,CAAC;IACD,IAAI,CAAC,KAAK,CAAC,CAAC;IACZ,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,MAAM,UAAU,WAAW,CAAC,KAAsB,EAAE,MAAc;IAChE,KAAK,MAAM,IAAI,IAAI,OAAO,CAAC,KAAK,CAAC,EAAE,CAAC;QAClC,IAAI,IAAI,CAAC,MAAM,KAAK,MAAM;YAAE,OAAO,IAAI,CAAC;IAC1C,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,MAAM,UAAU,aAAa,CAAC,KAAsB,EAAE,aAAqB;IACzE,KAAK,MAAM,IAAI,IAAI,OAAO,CAAC,KAAK,CAAC,EAAE,CAAC;QAClC,IAAI,IAAI,CAAC,aAAa,KAAK,aAAa;YAAE,OAAO,IAAI,CAAC;IACxD,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,MAAM,UAAU,MAAM,CAAC,SAA4B;IACjD,SAAS,UAAU,CAAC,IAAmB;QACrC,OAAO;YACL,MAAM,EAAE,IAAI,CAAC,MAAM;YACnB,KAAK,EAAE,IAAI,CAAC,KAAK;YACjB,KAAK,EAAE,IAAI,CAAC,KAAK;YACjB,SAAS,EAAE,IAAI,CAAC,SAAS;YACzB,OAAO,EAAE,IAAI,CAAC,OAAO;YACrB,SAAS,EAAE,IAAI,CAAC,SAAS;YACzB,SAAS,EAAE,IAAI,CAAC,SAAS;YACzB,OAAO,EAAE,IAAI,CAAC,OAAO;YACrB,YAAY,EAAE,IAAI,CAAC,YAAY;YAC/B,aAAa,EAAE,IAAI,CAAC,aAAa;YACjC,SAAS,EAAE,IAAI,CAAC,SAAS;YACzB,aAAa,EAAE,IAAI,CAAC,aAAa;YACjC,QAAQ,EAAE,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,UAAU,CAAC;SACxC,CAAC;IACJ,CAAC;IACD,OAAO;QACL,SAAS,EAAE,SAAS,CAAC,SAAS,CAAC,GAAG,CAAC,UAAU,CAAC;QAC9C,YAAY,EAAE,SAAS,CAAC,YAAY;QACpC,UAAU,EAAE,SAAS,CAAC,UAAU;QAChC,QAAQ,EAAE,SAAS,CAAC,QAAQ;KAC7B,CAAC;AACJ,CAAC;AAED,MAAM,UAAU,QAAQ,CAAC,IAA6B;IACpD,OAAO,uBAAuB,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;AAC7C,CAAC"}
@@ -0,0 +1,16 @@
1
+ /**
2
+ * Language-aware token estimation — tokenx-inspired approach.
3
+ * Replaces the naive `chars / 4` heuristic with segment-level analysis.
4
+ */
5
+ export interface LanguageConfig {
6
+ pattern: RegExp;
7
+ averageCharsPerToken: number;
8
+ }
9
+ export interface TokenEstimationOptions {
10
+ defaultCharsPerToken?: number;
11
+ languageConfigs?: LanguageConfig[];
12
+ }
13
+ /** Estimate token count using language-aware heuristics */
14
+ export declare function estimateTokenCount(text?: string | number, options?: TokenEstimationOptions): number;
15
+ /** Slice text by token positions (like Array.slice but for tokens) */
16
+ export declare function sliceByTokens(text: string, start?: number, end?: number, options?: TokenEstimationOptions): string;
@@ -0,0 +1,107 @@
1
+ /**
2
+ * Language-aware token estimation — tokenx-inspired approach.
3
+ * Replaces the naive `chars / 4` heuristic with segment-level analysis.
4
+ */
5
+ const PATTERNS = {
6
+ whitespace: /^\s+$/,
7
+ cjk: /[\u4E00-\u9FFF\u3400-\u4DBF\u3000-\u303F\uFF00-\uFFEF\u30A0-\u30FF\u2E80-\u2EFF\u31C0-\u31EF\u3200-\u32FF\u3300-\u33FF\uAC00-\uD7AF\u1100-\u11FF\u3130-\u318F\uA960-\uA97F\uD7B0-\uD7FF]/,
8
+ numeric: /^\d+(?:[.,]\d+)*$/,
9
+ punctuation: /[.,!?;(){}[\]<>:/\\|@#$%^&*+=`~_-]/,
10
+ alphanumeric: /^[a-zA-Z0-9\u00C0-\u00D6\u00D8-\u00F6\u00F8-\u00FF]+$/,
11
+ };
12
+ const TOKEN_SPLIT_PATTERN = new RegExp(`(\\s+|${PATTERNS.punctuation.source}+)`);
13
+ const DEFAULT_CHARS_PER_TOKEN = 6;
14
+ const SHORT_TOKEN_THRESHOLD = 3;
15
+ const DEFAULT_LANGUAGE_CONFIGS = [
16
+ { pattern: /[äöüßẞ]/i, averageCharsPerToken: 3 },
17
+ { pattern: /[éèêëàâîïôûùüÿçœæáíóúñ]/i, averageCharsPerToken: 3 },
18
+ { pattern: /[ąćęłńóśźżěščřžýůúďťň]/i, averageCharsPerToken: 3.5 },
19
+ ];
20
+ function getLanguageSpecificCharsPerToken(segment, configs) {
21
+ for (const config of configs) {
22
+ if (config.pattern.test(segment)) {
23
+ return config.averageCharsPerToken;
24
+ }
25
+ }
26
+ return undefined;
27
+ }
28
+ function getCharacterCount(text) {
29
+ return Array.from(text).length;
30
+ }
31
+ function estimateSegmentTokens(segment, languageConfigs, defaultCharsPerToken) {
32
+ if (PATTERNS.whitespace.test(segment))
33
+ return 0;
34
+ if (PATTERNS.cjk.test(segment))
35
+ return getCharacterCount(segment);
36
+ if (PATTERNS.numeric.test(segment))
37
+ return 1;
38
+ if (segment.length <= SHORT_TOKEN_THRESHOLD)
39
+ return 1;
40
+ if (PATTERNS.punctuation.test(segment))
41
+ return segment.length > 1 ? Math.ceil(segment.length / 2) : 1;
42
+ const charsPerToken = getLanguageSpecificCharsPerToken(segment, languageConfigs) ?? defaultCharsPerToken;
43
+ return Math.ceil(segment.length / charsPerToken);
44
+ }
45
+ /** Estimate token count using language-aware heuristics */
46
+ export function estimateTokenCount(text, options = {}) {
47
+ if (text == null)
48
+ return 0;
49
+ if (typeof text === 'number') {
50
+ // Fallback for numeric input (character count) — use default ratio
51
+ return Math.max(0, Math.ceil(text / (options.defaultCharsPerToken ?? DEFAULT_CHARS_PER_TOKEN)));
52
+ }
53
+ if (!text)
54
+ return 0;
55
+ const { defaultCharsPerToken = DEFAULT_CHARS_PER_TOKEN, languageConfigs = DEFAULT_LANGUAGE_CONFIGS, } = options;
56
+ const segments = text.split(TOKEN_SPLIT_PATTERN).filter(Boolean);
57
+ let tokenCount = 0;
58
+ for (const segment of segments) {
59
+ tokenCount += estimateSegmentTokens(segment, languageConfigs, defaultCharsPerToken);
60
+ }
61
+ return tokenCount;
62
+ }
63
+ /** Slice text by token positions (like Array.slice but for tokens) */
64
+ export function sliceByTokens(text, start = 0, end, options = {}) {
65
+ if (!text)
66
+ return '';
67
+ const { defaultCharsPerToken = DEFAULT_CHARS_PER_TOKEN, languageConfigs = DEFAULT_LANGUAGE_CONFIGS } = options;
68
+ let totalTokens = 0;
69
+ if (start < 0 || (end !== undefined && end < 0)) {
70
+ totalTokens = estimateTokenCount(text, options);
71
+ }
72
+ const normalizedStart = start < 0 ? Math.max(0, totalTokens + start) : Math.max(0, start);
73
+ const normalizedEnd = end === undefined ? Infinity : end < 0 ? Math.max(0, totalTokens + end) : end;
74
+ if (normalizedStart >= normalizedEnd)
75
+ return '';
76
+ const segments = text.split(TOKEN_SPLIT_PATTERN).filter(Boolean);
77
+ const parts = [];
78
+ let currentTokenPos = 0;
79
+ for (const segment of segments) {
80
+ if (currentTokenPos >= normalizedEnd)
81
+ break;
82
+ const tokenCount = estimateSegmentTokens(segment, languageConfigs, defaultCharsPerToken);
83
+ const segmentEnd = currentTokenPos + tokenCount;
84
+ if (segmentEnd > normalizedStart && currentTokenPos < normalizedEnd) {
85
+ if (tokenCount === 0) {
86
+ if (currentTokenPos >= normalizedStart && currentTokenPos < normalizedEnd) {
87
+ parts.push(segment);
88
+ }
89
+ }
90
+ else {
91
+ const overlapStart = Math.max(0, normalizedStart - currentTokenPos);
92
+ const overlapEnd = Math.min(tokenCount, normalizedEnd - currentTokenPos);
93
+ if (overlapStart === 0 && overlapEnd === tokenCount) {
94
+ parts.push(segment);
95
+ }
96
+ else {
97
+ const charStart = Math.floor((overlapStart / tokenCount) * segment.length);
98
+ const charEnd = Math.ceil((overlapEnd / tokenCount) * segment.length);
99
+ parts.push(segment.slice(charStart, charEnd));
100
+ }
101
+ }
102
+ }
103
+ currentTokenPos += tokenCount;
104
+ }
105
+ return parts.join('');
106
+ }
107
+ //# sourceMappingURL=tokens.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"tokens.js","sourceRoot":"","sources":["../../src/tokens.ts"],"names":[],"mappings":"AAAA;;;GAGG;AAYH,MAAM,QAAQ,GAAG;IACf,UAAU,EAAE,OAAO;IACnB,GAAG,EAAE,0LAA0L;IAC/L,OAAO,EAAE,mBAAmB;IAC5B,WAAW,EAAE,oCAAoC;IACjD,YAAY,EAAE,uDAAuD;CAC7D,CAAC;AAEX,MAAM,mBAAmB,GAAG,IAAI,MAAM,CAAC,SAAS,QAAQ,CAAC,WAAW,CAAC,MAAM,IAAI,CAAC,CAAC;AACjF,MAAM,uBAAuB,GAAG,CAAC,CAAC;AAClC,MAAM,qBAAqB,GAAG,CAAC,CAAC;AAEhC,MAAM,wBAAwB,GAAqB;IACjD,EAAE,OAAO,EAAE,UAAU,EAAE,oBAAoB,EAAE,CAAC,EAAE;IAChD,EAAE,OAAO,EAAE,0BAA0B,EAAE,oBAAoB,EAAE,CAAC,EAAE;IAChE,EAAE,OAAO,EAAE,yBAAyB,EAAE,oBAAoB,EAAE,GAAG,EAAE;CAClE,CAAC;AAEF,SAAS,gCAAgC,CAAC,OAAe,EAAE,OAAyB;IAClF,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;QAC7B,IAAI,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;YACjC,OAAO,MAAM,CAAC,oBAAoB,CAAC;QACrC,CAAC;IACH,CAAC;IACD,OAAO,SAAS,CAAC;AACnB,CAAC;AAED,SAAS,iBAAiB,CAAC,IAAY;IACrC,OAAO,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC;AACjC,CAAC;AAED,SAAS,qBAAqB,CAC5B,OAAe,EACf,eAAiC,EACjC,oBAA4B;IAE5B,IAAI,QAAQ,CAAC,UAAU,CAAC,IAAI,CAAC,OAAO,CAAC;QAAE,OAAO,CAAC,CAAC;IAChD,IAAI,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,OAAO,CAAC;QAAE,OAAO,iBAAiB,CAAC,OAAO,CAAC,CAAC;IAClE,IAAI,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC;QAAE,OAAO,CAAC,CAAC;IAC7C,IAAI,OAAO,CAAC,MAAM,IAAI,qBAAqB;QAAE,OAAO,CAAC,CAAC;IACtD,IAAI,QAAQ,CAAC,WAAW,CAAC,IAAI,CAAC,OAAO,CAAC;QAAE,OAAO,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAEtG,MAAM,aAAa,GAAG,gCAAgC,CAAC,OAAO,EAAE,eAAe,CAAC,IAAI,oBAAoB,CAAC;IACzG,OAAO,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,GAAG,aAAa,CAAC,CAAC;AACnD,CAAC;AAED,2DAA2D;AAC3D,MAAM,UAAU,kBAAkB,CAAC,IAAsB,EAAE,UAAkC,EAAE;IAC7F,IAAI,IAAI,IAAI,IAAI;QAAE,OAAO,CAAC,CAAC;IAC3B,IAAI,OAAO,IAAI,KAAK,QAAQ,EAAE,CAAC;QAC7B,mEAAmE;QACnE,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,IAAI,CAAC,IAAI,GAAG,CAAC,OAAO,CAAC,oBAAoB,IAAI,uBAAuB,CAAC,CAAC,CAAC,CAAC;IAClG,CAAC;IACD,IAAI,CAAC,IAAI;QAAE,OAAO,CAAC,CAAC;IAEpB,MAAM,EACJ,oBAAoB,GAAG,uBAAuB,EAC9C,eAAe,GAAG,wBAAwB,GAC3C,GAAG,OAAO,CAAC;IAEZ,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,mBAAmB,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC;IACjE,IAAI,UAAU,GAAG,CAAC,CAAC;IACnB,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;QAC/B,UAAU,IAAI,qBAAqB,CAAC,OAAO,EAAE,eAAe,EAAE,oBAAoB,CAAC,CAAC;IACtF,CAAC;IACD,OAAO,UAAU,CAAC;AACpB,CAAC;AAED,sEAAsE;AACtE,MAAM,UAAU,aAAa,CAC3B,IAAY,EACZ,KAAK,GAAG,CAAC,EACT,GAAY,EACZ,UAAkC,EAAE;IAEpC,IAAI,CAAC,IAAI;QAAE,OAAO,EAAE,CAAC;IAErB,MAAM,EAAE,oBAAoB,GAAG,uBAAuB,EAAE,eAAe,GAAG,wBAAwB,EAAE,GAAG,OAAO,CAAC;IAE/G,IAAI,WAAW,GAAG,CAAC,CAAC;IACpB,IAAI,KAAK,GAAG,CAAC,IAAI,CAAC,GAAG,KAAK,SAAS,IAAI,GAAG,GAAG,CAAC,CAAC,EAAE,CAAC;QAChD,WAAW,GAAG,kBAAkB,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IAClD,CAAC;IAED,MAAM,eAAe,GAAG,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,WAAW,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC;IAC1F,MAAM,aAAa,GAAG,GAAG,KAAK,SAAS,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,WAAW,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC;IAEpG,IAAI,eAAe,IAAI,aAAa;QAAE,OAAO,EAAE,CAAC;IAEhD,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,mBAAmB,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC;IACjE,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,IAAI,eAAe,GAAG,CAAC,CAAC;IAExB,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;QAC/B,IAAI,eAAe,IAAI,aAAa;YAAE,MAAM;QAE5C,MAAM,UAAU,GAAG,qBAAqB,CAAC,OAAO,EAAE,eAAe,EAAE,oBAAoB,CAAC,CAAC;QACzF,MAAM,UAAU,GAAG,eAAe,GAAG,UAAU,CAAC;QAEhD,IAAI,UAAU,GAAG,eAAe,IAAI,eAAe,GAAG,aAAa,EAAE,CAAC;YACpE,IAAI,UAAU,KAAK,CAAC,EAAE,CAAC;gBACrB,IAAI,eAAe,IAAI,eAAe,IAAI,eAAe,GAAG,aAAa,EAAE,CAAC;oBAC1E,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;gBACtB,CAAC;YACH,CAAC;iBAAM,CAAC;gBACN,MAAM,YAAY,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,eAAe,GAAG,eAAe,CAAC,CAAC;gBACpE,MAAM,UAAU,GAAG,IAAI,CAAC,GAAG,CAAC,UAAU,EAAE,aAAa,GAAG,eAAe,CAAC,CAAC;gBACzE,IAAI,YAAY,KAAK,CAAC,IAAI,UAAU,KAAK,UAAU,EAAE,CAAC;oBACpD,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;gBACtB,CAAC;qBAAM,CAAC;oBACN,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,YAAY,GAAG,UAAU,CAAC,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC;oBAC3E,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC,UAAU,GAAG,UAAU,CAAC,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC;oBACtE,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,SAAS,EAAE,OAAO,CAAC,CAAC,CAAC;gBAChD,CAAC;YACH,CAAC;QACH,CAAC;QAED,eAAe,IAAI,UAAU,CAAC;IAChC,CAAC;IAED,OAAO,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;AACxB,CAAC"}
@@ -13,13 +13,23 @@ export type SkippedEntry = z.infer<typeof SkippedEntrySchema>;
13
13
  export declare const ParserOutputSchema: z.ZodObject<{
14
14
  fileType: z.ZodString;
15
15
  metrics: z.ZodRecord<z.ZodString, z.ZodNumber>;
16
+ confidence: z.ZodOptional<z.ZodRecord<z.ZodString, z.ZodEnum<{
17
+ exact: "exact";
18
+ estimated: "estimated";
19
+ }>>>;
16
20
  }, z.core.$strip>;
17
21
  export type ParserOutput = z.infer<typeof ParserOutputSchema>;
22
+ export type ConfidenceLevel = 'exact' | 'estimated';
18
23
  export declare const ParseResultSchema: z.ZodObject<{
19
24
  filePath: z.ZodString;
20
25
  size: z.ZodNumber;
21
26
  success: z.ZodBoolean;
22
27
  fileType: z.ZodString;
23
28
  metrics: z.ZodNullable<z.ZodRecord<z.ZodString, z.ZodNumber>>;
29
+ confidence: z.ZodOptional<z.ZodRecord<z.ZodString, z.ZodEnum<{
30
+ exact: "exact";
31
+ estimated: "estimated";
32
+ }>>>;
33
+ suggestion: z.ZodOptional<z.ZodString>;
24
34
  }, z.core.$strip>;
25
35
  export type ParseResult = z.infer<typeof ParseResultSchema>;
package/dist/src/types.js CHANGED
@@ -11,6 +11,7 @@ export const SkippedEntrySchema = z.object({
11
11
  export const ParserOutputSchema = z.object({
12
12
  fileType: z.string(),
13
13
  metrics: z.record(z.string(), z.number()),
14
+ confidence: z.record(z.string(), z.enum(['exact', 'estimated'])).optional(),
14
15
  });
15
16
  export const ParseResultSchema = z.object({
16
17
  filePath: z.string(),
@@ -18,5 +19,7 @@ export const ParseResultSchema = z.object({
18
19
  success: z.boolean(),
19
20
  fileType: z.string(),
20
21
  metrics: z.record(z.string(), z.number()).nullable(),
22
+ confidence: z.record(z.string(), z.enum(['exact', 'estimated'])).optional(),
23
+ suggestion: z.string().optional(),
21
24
  });
22
25
  //# sourceMappingURL=types.js.map
@@ -1 +1 @@
1
- {"version":3,"file":"types.js","sourceRoot":"","sources":["../../src/types.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAExB,MAAM,CAAC,MAAM,eAAe,GAAG,CAAC,CAAC,MAAM,CAAC;IACtC,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;CACjB,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,kBAAkB,GAAG,CAAC,CAAC,MAAM,CAAC;IACzC,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE;IAClB,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;CACjB,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,kBAAkB,GAAG,CAAC,CAAC,MAAM,CAAC;IACzC,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE;IACpB,OAAO,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC;CAC1C,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,iBAAiB,GAAG,CAAC,CAAC,MAAM,CAAC;IACxC,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE;IACpB,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,OAAO,EAAE,CAAC,CAAC,OAAO,EAAE;IACpB,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE;IACpB,OAAO,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,CAAC,QAAQ,EAAE;CACrD,CAAC,CAAC"}
1
+ {"version":3,"file":"types.js","sourceRoot":"","sources":["../../src/types.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAExB,MAAM,CAAC,MAAM,eAAe,GAAG,CAAC,CAAC,MAAM,CAAC;IACtC,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;CACjB,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,kBAAkB,GAAG,CAAC,CAAC,MAAM,CAAC;IACzC,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE;IAClB,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;CACjB,CAAC,CAAC;AAGH,MAAM,CAAC,MAAM,kBAAkB,GAAG,CAAC,CAAC,MAAM,CAAC;IACzC,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE;IACpB,OAAO,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC;IACzC,UAAU,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,WAAW,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE;CAC5E,CAAC,CAAC;AAKH,MAAM,CAAC,MAAM,iBAAiB,GAAG,CAAC,CAAC,MAAM,CAAC;IACxC,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE;IACpB,IAAI,EAAE,CAAC,CAAC,MAAM,EAAE;IAChB,OAAO,EAAE,CAAC,CAAC,OAAO,EAAE;IACpB,QAAQ,EAAE,CAAC,CAAC,MAAM,EAAE;IACpB,OAAO,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,CAAC,QAAQ,EAAE;IACpD,UAAU,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,WAAW,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE;IAC3E,UAAU,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE;CAClC,CAAC,CAAC"}
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@cesarandreslopez/occ",
3
- "version": "0.5.1",
3
+ "version": "0.6.1",
4
4
  "description": "Office Cloc and Count — scc-style summary tables for office documents",
5
5
  "type": "module",
6
6
  "bin": {
@@ -30,7 +30,8 @@
30
30
  "pptx",
31
31
  "pdf",
32
32
  "scc",
33
- "cli"
33
+ "cli",
34
+ "confidence"
34
35
  ],
35
36
  "exports": {
36
37
  ".": {
@@ -104,6 +105,30 @@
104
105
  "./table/inspect": {
105
106
  "types": "./dist/src/table/inspect.d.ts",
106
107
  "default": "./dist/src/table/inspect.js"
108
+ },
109
+ "./tokens": {
110
+ "types": "./dist/src/tokens.d.ts",
111
+ "default": "./dist/src/tokens.js"
112
+ },
113
+ "./doc/entities": {
114
+ "types": "./dist/src/doc/entities.d.ts",
115
+ "default": "./dist/src/doc/entities.js"
116
+ },
117
+ "./doc/references": {
118
+ "types": "./dist/src/doc/references.d.ts",
119
+ "default": "./dist/src/doc/references.js"
120
+ },
121
+ "./progress-event": {
122
+ "types": "./dist/src/progress-event.d.ts",
123
+ "default": "./dist/src/progress-event.js"
124
+ },
125
+ "./types": {
126
+ "types": "./dist/src/types.d.ts",
127
+ "default": "./dist/src/types.js"
128
+ },
129
+ "./stats": {
130
+ "types": "./dist/src/stats.d.ts",
131
+ "default": "./dist/src/stats.js"
107
132
  }
108
133
  },
109
134
  "files": [