docxmlater 7.4.3 → 7.4.6
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.
- package/dist/core/Document.d.ts +2 -0
- package/dist/core/Document.d.ts.map +1 -1
- package/dist/core/Document.js +6 -0
- package/dist/core/Document.js.map +1 -1
- package/dist/core/ListNormalizer.d.ts +21 -0
- package/dist/core/ListNormalizer.d.ts.map +1 -0
- package/dist/core/ListNormalizer.js +318 -0
- package/dist/core/ListNormalizer.js.map +1 -0
- package/dist/index.d.ts +3 -0
- package/dist/index.d.ts.map +1 -1
- package/dist/index.js +18 -1
- package/dist/index.js.map +1 -1
- package/dist/types/list-types.d.ts +49 -0
- package/dist/types/list-types.d.ts.map +1 -0
- package/dist/types/list-types.js +3 -0
- package/dist/types/list-types.js.map +1 -0
- package/dist/utils/list-detection.d.ts +21 -0
- package/dist/utils/list-detection.d.ts.map +1 -0
- package/dist/utils/list-detection.js +133 -0
- package/dist/utils/list-detection.js.map +1 -0
- package/package.json +1 -1
|
@@ -0,0 +1,133 @@
|
|
|
1
|
+
"use strict";
|
|
2
|
+
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
+
exports.PATTERN_TO_CATEGORY = exports.TYPED_LIST_PATTERNS = void 0;
|
|
4
|
+
exports.inferLevelFromIndentation = inferLevelFromIndentation;
|
|
5
|
+
exports.detectTypedPrefix = detectTypedPrefix;
|
|
6
|
+
exports.getParagraphIndentation = getParagraphIndentation;
|
|
7
|
+
exports.detectListType = detectListType;
|
|
8
|
+
exports.validateListSequence = validateListSequence;
|
|
9
|
+
exports.getListCategoryFromFormat = getListCategoryFromFormat;
|
|
10
|
+
exports.TYPED_LIST_PATTERNS = {
|
|
11
|
+
decimal: /^(\d+)[.)]\s+/,
|
|
12
|
+
lowerLetter: /^([a-z])[.)]\s+/,
|
|
13
|
+
upperLetter: /^([A-Z])[.)]\s+/,
|
|
14
|
+
lowerRoman: /^((?:i{1,3}|iv|vi{0,3}|ix|x{1,3}))[.)]\s+/i,
|
|
15
|
+
bullet: /^[•●○◦▪■□]\s+/,
|
|
16
|
+
dash: /^[-–—]\s+/,
|
|
17
|
+
arrow: /^[►▸▶→]\s+/,
|
|
18
|
+
};
|
|
19
|
+
exports.PATTERN_TO_CATEGORY = {
|
|
20
|
+
decimal: "numbered",
|
|
21
|
+
lowerLetter: "numbered",
|
|
22
|
+
upperLetter: "numbered",
|
|
23
|
+
lowerRoman: "numbered",
|
|
24
|
+
bullet: "bullet",
|
|
25
|
+
dash: "bullet",
|
|
26
|
+
arrow: "bullet",
|
|
27
|
+
};
|
|
28
|
+
const WORD_BASE_INDENT = 720;
|
|
29
|
+
const INDENT_PER_LEVEL = 360;
|
|
30
|
+
function inferLevelFromIndentation(indentTwips) {
|
|
31
|
+
if (indentTwips < WORD_BASE_INDENT)
|
|
32
|
+
return 0;
|
|
33
|
+
return Math.floor((indentTwips - WORD_BASE_INDENT) / INDENT_PER_LEVEL);
|
|
34
|
+
}
|
|
35
|
+
function detectTypedPrefix(text) {
|
|
36
|
+
for (const [format, regex] of Object.entries(exports.TYPED_LIST_PATTERNS)) {
|
|
37
|
+
const match = text.match(regex);
|
|
38
|
+
if (match) {
|
|
39
|
+
return {
|
|
40
|
+
prefix: match[0],
|
|
41
|
+
format: format,
|
|
42
|
+
category: exports.PATTERN_TO_CATEGORY[format] ?? "none",
|
|
43
|
+
};
|
|
44
|
+
}
|
|
45
|
+
}
|
|
46
|
+
return { prefix: null, format: null, category: "none" };
|
|
47
|
+
}
|
|
48
|
+
function getParagraphIndentation(paragraph) {
|
|
49
|
+
const formatting = paragraph.getFormatting();
|
|
50
|
+
return formatting?.indentation?.left ?? 0;
|
|
51
|
+
}
|
|
52
|
+
function detectListType(paragraph) {
|
|
53
|
+
const text = paragraph.getText();
|
|
54
|
+
const indentation = getParagraphIndentation(paragraph);
|
|
55
|
+
const numbering = paragraph.getNumbering();
|
|
56
|
+
if (numbering && numbering.numId !== undefined && numbering.numId !== 0) {
|
|
57
|
+
return {
|
|
58
|
+
category: "numbered",
|
|
59
|
+
isWordList: true,
|
|
60
|
+
typedPrefix: null,
|
|
61
|
+
inferredLevel: numbering.level ?? 0,
|
|
62
|
+
format: null,
|
|
63
|
+
numId: numbering.numId,
|
|
64
|
+
ilvl: numbering.level ?? 0,
|
|
65
|
+
indentationTwips: indentation,
|
|
66
|
+
};
|
|
67
|
+
}
|
|
68
|
+
const typed = detectTypedPrefix(text);
|
|
69
|
+
if (typed.prefix) {
|
|
70
|
+
return {
|
|
71
|
+
category: typed.category,
|
|
72
|
+
isWordList: false,
|
|
73
|
+
typedPrefix: typed.prefix,
|
|
74
|
+
inferredLevel: inferLevelFromIndentation(indentation),
|
|
75
|
+
format: typed.format,
|
|
76
|
+
numId: null,
|
|
77
|
+
ilvl: null,
|
|
78
|
+
indentationTwips: indentation,
|
|
79
|
+
};
|
|
80
|
+
}
|
|
81
|
+
return {
|
|
82
|
+
category: "none",
|
|
83
|
+
isWordList: false,
|
|
84
|
+
typedPrefix: null,
|
|
85
|
+
inferredLevel: 0,
|
|
86
|
+
format: null,
|
|
87
|
+
numId: null,
|
|
88
|
+
ilvl: null,
|
|
89
|
+
indentationTwips: indentation,
|
|
90
|
+
};
|
|
91
|
+
}
|
|
92
|
+
function validateListSequence(paragraphs) {
|
|
93
|
+
const warnings = [];
|
|
94
|
+
let lastDecimal = 0;
|
|
95
|
+
let lastLetter = "";
|
|
96
|
+
for (const { detection } of paragraphs) {
|
|
97
|
+
if (!detection.typedPrefix || detection.category !== "numbered")
|
|
98
|
+
continue;
|
|
99
|
+
const match = detection.typedPrefix.match(/^(\d+|[a-zA-Z]+)/);
|
|
100
|
+
if (!match || !match[1])
|
|
101
|
+
continue;
|
|
102
|
+
const marker = match[1];
|
|
103
|
+
if (/^\d+$/.test(marker)) {
|
|
104
|
+
const num = parseInt(marker, 10);
|
|
105
|
+
if (lastDecimal > 0 && num !== lastDecimal + 1 && num !== 1) {
|
|
106
|
+
warnings.push(`Unexpected number sequence: ${lastDecimal} → ${num}`);
|
|
107
|
+
}
|
|
108
|
+
lastDecimal = num;
|
|
109
|
+
}
|
|
110
|
+
if (/^[a-z]$/i.test(marker)) {
|
|
111
|
+
const letter = marker.toLowerCase();
|
|
112
|
+
if (lastLetter &&
|
|
113
|
+
letter.charCodeAt(0) !== lastLetter.charCodeAt(0) + 1 &&
|
|
114
|
+
letter !== "a") {
|
|
115
|
+
warnings.push(`Unexpected letter sequence: ${lastLetter} → ${letter}`);
|
|
116
|
+
}
|
|
117
|
+
lastLetter = letter;
|
|
118
|
+
}
|
|
119
|
+
}
|
|
120
|
+
return { valid: warnings.length === 0, warnings };
|
|
121
|
+
}
|
|
122
|
+
function getListCategoryFromFormat(format) {
|
|
123
|
+
if (!format)
|
|
124
|
+
return "none";
|
|
125
|
+
if (["bullet", "dash", "arrow"].includes(format)) {
|
|
126
|
+
return "bullet";
|
|
127
|
+
}
|
|
128
|
+
if (["decimal", "lowerLetter", "upperLetter", "lowerRoman", "upperRoman"].includes(format)) {
|
|
129
|
+
return "numbered";
|
|
130
|
+
}
|
|
131
|
+
return "none";
|
|
132
|
+
}
|
|
133
|
+
//# sourceMappingURL=list-detection.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"list-detection.js","sourceRoot":"","sources":["../../src/utils/list-detection.ts"],"names":[],"mappings":";;;AAiEA,8DAGC;AAMD,8CAiBC;AAKD,0DAGC;AAKD,wCA6CC;AAMD,oDAuCC;AAMD,8DAkBC;AAnMY,QAAA,mBAAmB,GAA2B;IAEzD,OAAO,EAAE,eAAe;IACxB,WAAW,EAAE,iBAAiB;IAC9B,WAAW,EAAE,iBAAiB;IAC9B,UAAU,EAAE,4CAA4C;IAGxD,MAAM,EAAE,eAAe;IACvB,IAAI,EAAE,WAAW;IACjB,KAAK,EAAE,YAAY;CACpB,CAAC;AAGW,QAAA,mBAAmB,GAAiC;IAC/D,OAAO,EAAE,UAAU;IACnB,WAAW,EAAE,UAAU;IACvB,WAAW,EAAE,UAAU;IACvB,UAAU,EAAE,UAAU;IACtB,MAAM,EAAE,QAAQ;IAChB,IAAI,EAAE,QAAQ;IACd,KAAK,EAAE,QAAQ;CAChB,CAAC;AASF,MAAM,gBAAgB,GAAG,GAAG,CAAC;AAC7B,MAAM,gBAAgB,GAAG,GAAG,CAAC;AAU7B,SAAgB,yBAAyB,CAAC,WAAmB;IAC3D,IAAI,WAAW,GAAG,gBAAgB;QAAE,OAAO,CAAC,CAAC;IAC7C,OAAO,IAAI,CAAC,KAAK,CAAC,CAAC,WAAW,GAAG,gBAAgB,CAAC,GAAG,gBAAgB,CAAC,CAAC;AACzE,CAAC;AAMD,SAAgB,iBAAiB,CAAC,IAAY;IAK5C,KAAK,MAAM,CAAC,MAAM,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,2BAAmB,CAAC,EAAE,CAAC;QAClE,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;QAChC,IAAI,KAAK,EAAE,CAAC;YACV,OAAO;gBACL,MAAM,EAAE,KAAK,CAAC,CAAC,CAAC;gBAChB,MAAM,EAAE,MAAqC;gBAC7C,QAAQ,EAAE,2BAAmB,CAAC,MAAM,CAAC,IAAI,MAAM;aAChD,CAAC;QACJ,CAAC;IACH,CAAC;IAED,OAAO,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC;AAC1D,CAAC;AAKD,SAAgB,uBAAuB,CAAC,SAAoB;IAC1D,MAAM,UAAU,GAAG,SAAS,CAAC,aAAa,EAAE,CAAC;IAC7C,OAAO,UAAU,EAAE,WAAW,EAAE,IAAI,IAAI,CAAC,CAAC;AAC5C,CAAC;AAKD,SAAgB,cAAc,CAAC,SAAoB;IACjD,MAAM,IAAI,GAAG,SAAS,CAAC,OAAO,EAAE,CAAC;IACjC,MAAM,WAAW,GAAG,uBAAuB,CAAC,SAAS,CAAC,CAAC;IACvD,MAAM,SAAS,GAAG,SAAS,CAAC,YAAY,EAAE,CAAC;IAG3C,IAAI,SAAS,IAAI,SAAS,CAAC,KAAK,KAAK,SAAS,IAAI,SAAS,CAAC,KAAK,KAAK,CAAC,EAAE,CAAC;QACxE,OAAO;YACL,QAAQ,EAAE,UAAU;YACpB,UAAU,EAAE,IAAI;YAChB,WAAW,EAAE,IAAI;YACjB,aAAa,EAAE,SAAS,CAAC,KAAK,IAAI,CAAC;YACnC,MAAM,EAAE,IAAI;YACZ,KAAK,EAAE,SAAS,CAAC,KAAK;YACtB,IAAI,EAAE,SAAS,CAAC,KAAK,IAAI,CAAC;YAC1B,gBAAgB,EAAE,WAAW;SAC9B,CAAC;IACJ,CAAC;IAGD,MAAM,KAAK,GAAG,iBAAiB,CAAC,IAAI,CAAC,CAAC;IACtC,IAAI,KAAK,CAAC,MAAM,EAAE,CAAC;QACjB,OAAO;YACL,QAAQ,EAAE,KAAK,CAAC,QAAQ;YACxB,UAAU,EAAE,KAAK;YACjB,WAAW,EAAE,KAAK,CAAC,MAAM;YACzB,aAAa,EAAE,yBAAyB,CAAC,WAAW,CAAC;YACrD,MAAM,EAAE,KAAK,CAAC,MAAM;YACpB,KAAK,EAAE,IAAI;YACX,IAAI,EAAE,IAAI;YACV,gBAAgB,EAAE,WAAW;SAC9B,CAAC;IACJ,CAAC;IAGD,OAAO;QACL,QAAQ,EAAE,MAAM;QAChB,UAAU,EAAE,KAAK;QACjB,WAAW,EAAE,IAAI;QACjB,aAAa,EAAE,CAAC;QAChB,MAAM,EAAE,IAAI;QACZ,KAAK,EAAE,IAAI;QACX,IAAI,EAAE,IAAI;QACV,gBAAgB,EAAE,WAAW;KAC9B,CAAC;AACJ,CAAC;AAMD,SAAgB,oBAAoB,CAClC,UAAmE;IAEnE,MAAM,QAAQ,GAAa,EAAE,CAAC;IAC9B,IAAI,WAAW,GAAG,CAAC,CAAC;IACpB,IAAI,UAAU,GAAG,EAAE,CAAC;IAEpB,KAAK,MAAM,EAAE,SAAS,EAAE,IAAI,UAAU,EAAE,CAAC;QACvC,IAAI,CAAC,SAAS,CAAC,WAAW,IAAI,SAAS,CAAC,QAAQ,KAAK,UAAU;YAAE,SAAS;QAE1E,MAAM,KAAK,GAAG,SAAS,CAAC,WAAW,CAAC,KAAK,CAAC,kBAAkB,CAAC,CAAC;QAC9D,IAAI,CAAC,KAAK,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC;YAAE,SAAS;QAElC,MAAM,MAAM,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;QAGxB,IAAI,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC;YACzB,MAAM,GAAG,GAAG,QAAQ,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC;YACjC,IAAI,WAAW,GAAG,CAAC,IAAI,GAAG,KAAK,WAAW,GAAG,CAAC,IAAI,GAAG,KAAK,CAAC,EAAE,CAAC;gBAC5D,QAAQ,CAAC,IAAI,CAAC,+BAA+B,WAAW,MAAM,GAAG,EAAE,CAAC,CAAC;YACvE,CAAC;YACD,WAAW,GAAG,GAAG,CAAC;QACpB,CAAC;QAGD,IAAI,UAAU,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC;YAC5B,MAAM,MAAM,GAAG,MAAM,CAAC,WAAW,EAAE,CAAC;YACpC,IACE,UAAU;gBACV,MAAM,CAAC,UAAU,CAAC,CAAC,CAAC,KAAK,UAAU,CAAC,UAAU,CAAC,CAAC,CAAC,GAAG,CAAC;gBACrD,MAAM,KAAK,GAAG,EACd,CAAC;gBACD,QAAQ,CAAC,IAAI,CAAC,+BAA+B,UAAU,MAAM,MAAM,EAAE,CAAC,CAAC;YACzE,CAAC;YACD,UAAU,GAAG,MAAM,CAAC;QACtB,CAAC;IACH,CAAC;IAED,OAAO,EAAE,KAAK,EAAE,QAAQ,CAAC,MAAM,KAAK,CAAC,EAAE,QAAQ,EAAE,CAAC;AACpD,CAAC;AAMD,SAAgB,yBAAyB,CACvC,MAA0B;IAE1B,IAAI,CAAC,MAAM;QAAE,OAAO,MAAM,CAAC;IAE3B,IAAI,CAAC,QAAQ,EAAE,MAAM,EAAE,OAAO,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QACjD,OAAO,QAAQ,CAAC;IAClB,CAAC;IAED,IACE,CAAC,SAAS,EAAE,aAAa,EAAE,aAAa,EAAE,YAAY,EAAE,YAAY,CAAC,CAAC,QAAQ,CAC5E,MAAM,CACP,EACD,CAAC;QACD,OAAO,UAAU,CAAC;IACpB,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC"}
|
package/package.json
CHANGED