@effortlessmotion/pdf-to-md 0.1.0
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/LICENSE +21 -0
- package/README.md +157 -0
- package/bin/cli.js +9 -0
- package/dist/cjs/analyze.d.cts +35 -0
- package/dist/cjs/analyze.d.ts +35 -0
- package/dist/cjs/analyze.js +620 -0
- package/dist/cjs/analyze.js.map +1 -0
- package/dist/cjs/audit.d.cts +38 -0
- package/dist/cjs/audit.d.ts +38 -0
- package/dist/cjs/audit.js +252 -0
- package/dist/cjs/audit.js.map +1 -0
- package/dist/cjs/cli/audit.d.cts +3 -0
- package/dist/cjs/cli/audit.d.ts +3 -0
- package/dist/cjs/cli/audit.js +269 -0
- package/dist/cjs/cli/audit.js.map +1 -0
- package/dist/cjs/cli/extract.d.cts +3 -0
- package/dist/cjs/cli/extract.d.ts +3 -0
- package/dist/cjs/cli/extract.js +704 -0
- package/dist/cjs/cli/extract.js.map +1 -0
- package/dist/cjs/cli/index.d.cts +5 -0
- package/dist/cjs/cli/index.d.ts +5 -0
- package/dist/cjs/cli/index.js +58 -0
- package/dist/cjs/cli/index.js.map +1 -0
- package/dist/cjs/cli/render.d.cts +3 -0
- package/dist/cjs/cli/render.d.ts +3 -0
- package/dist/cjs/cli/render.js +361 -0
- package/dist/cjs/cli/render.js.map +1 -0
- package/dist/cjs/cli/types.d.cts +10 -0
- package/dist/cjs/cli/types.d.ts +10 -0
- package/dist/cjs/cli/types.js +7 -0
- package/dist/cjs/cli/types.js.map +1 -0
- package/dist/cjs/cli.d.cts +2 -0
- package/dist/cjs/cli.d.ts +2 -0
- package/dist/cjs/cli.js +345 -0
- package/dist/cjs/cli.js.map +1 -0
- package/dist/cjs/collect.d.cts +6 -0
- package/dist/cjs/collect.d.ts +6 -0
- package/dist/cjs/collect.js +370 -0
- package/dist/cjs/collect.js.map +1 -0
- package/dist/cjs/dict.d.cts +8 -0
- package/dist/cjs/dict.d.ts +8 -0
- package/dist/cjs/dict.js +55 -0
- package/dist/cjs/dict.js.map +1 -0
- package/dist/cjs/emit.d.cts +27 -0
- package/dist/cjs/emit.d.ts +27 -0
- package/dist/cjs/emit.js +187 -0
- package/dist/cjs/emit.js.map +1 -0
- package/dist/cjs/engines/tesseract.d.cts +30 -0
- package/dist/cjs/engines/tesseract.d.ts +30 -0
- package/dist/cjs/engines/tesseract.js +650 -0
- package/dist/cjs/engines/tesseract.js.map +1 -0
- package/dist/cjs/errors.d.cts +21 -0
- package/dist/cjs/errors.d.ts +21 -0
- package/dist/cjs/errors.js +124 -0
- package/dist/cjs/errors.js.map +1 -0
- package/dist/cjs/extract.d.cts +88 -0
- package/dist/cjs/extract.d.ts +88 -0
- package/dist/cjs/extract.js +436 -0
- package/dist/cjs/extract.js.map +1 -0
- package/dist/cjs/geometry.d.cts +63 -0
- package/dist/cjs/geometry.d.ts +63 -0
- package/dist/cjs/geometry.js +521 -0
- package/dist/cjs/geometry.js.map +1 -0
- package/dist/cjs/index.d.cts +15 -0
- package/dist/cjs/index.d.ts +15 -0
- package/dist/cjs/index.js +56 -0
- package/dist/cjs/index.js.map +1 -0
- package/dist/cjs/lines.d.cts +3 -0
- package/dist/cjs/lines.d.ts +3 -0
- package/dist/cjs/lines.js +181 -0
- package/dist/cjs/lines.js.map +1 -0
- package/dist/cjs/markdown.d.cts +2 -0
- package/dist/cjs/markdown.d.ts +2 -0
- package/dist/cjs/markdown.js +222 -0
- package/dist/cjs/markdown.js.map +1 -0
- package/dist/cjs/package.json +1 -0
- package/dist/cjs/page-marker.d.cts +16 -0
- package/dist/cjs/page-marker.d.ts +16 -0
- package/dist/cjs/page-marker.js +85 -0
- package/dist/cjs/page-marker.js.map +1 -0
- package/dist/cjs/pdf-open.d.cts +12 -0
- package/dist/cjs/pdf-open.d.ts +12 -0
- package/dist/cjs/pdf-open.js +257 -0
- package/dist/cjs/pdf-open.js.map +1 -0
- package/dist/cjs/raster.d.cts +26 -0
- package/dist/cjs/raster.d.ts +26 -0
- package/dist/cjs/raster.js +262 -0
- package/dist/cjs/raster.js.map +1 -0
- package/dist/cjs/render-analysis.d.cts +25 -0
- package/dist/cjs/render-analysis.d.ts +25 -0
- package/dist/cjs/render-analysis.js +319 -0
- package/dist/cjs/render-analysis.js.map +1 -0
- package/dist/cjs/report.d.cts +12 -0
- package/dist/cjs/report.d.ts +12 -0
- package/dist/cjs/report.js +392 -0
- package/dist/cjs/report.js.map +1 -0
- package/dist/cjs/resolve.d.cts +1 -0
- package/dist/cjs/resolve.d.ts +1 -0
- package/dist/cjs/resolve.js +18 -0
- package/dist/cjs/resolve.js.map +1 -0
- package/dist/cjs/types.d.cts +171 -0
- package/dist/cjs/types.d.ts +171 -0
- package/dist/cjs/types.js +9 -0
- package/dist/cjs/types.js.map +1 -0
- package/dist/cjs/validate.d.cts +14 -0
- package/dist/cjs/validate.d.ts +14 -0
- package/dist/cjs/validate.js +96 -0
- package/dist/cjs/validate.js.map +1 -0
- package/dist/esm/analyze.d.ts +35 -0
- package/dist/esm/analyze.js +270 -0
- package/dist/esm/analyze.js.map +1 -0
- package/dist/esm/audit.d.ts +38 -0
- package/dist/esm/audit.js +113 -0
- package/dist/esm/audit.js.map +1 -0
- package/dist/esm/cli/audit.d.ts +3 -0
- package/dist/esm/cli/audit.js +20 -0
- package/dist/esm/cli/audit.js.map +1 -0
- package/dist/esm/cli/extract.d.ts +3 -0
- package/dist/esm/cli/extract.js +313 -0
- package/dist/esm/cli/extract.js.map +1 -0
- package/dist/esm/cli/index.d.ts +5 -0
- package/dist/esm/cli/index.js +7 -0
- package/dist/esm/cli/index.js.map +1 -0
- package/dist/esm/cli/render.d.ts +3 -0
- package/dist/esm/cli/render.js +77 -0
- package/dist/esm/cli/render.js.map +1 -0
- package/dist/esm/cli/types.d.ts +10 -0
- package/dist/esm/cli/types.js +3 -0
- package/dist/esm/cli/types.js.map +1 -0
- package/dist/esm/cli.d.ts +2 -0
- package/dist/esm/cli.js +90 -0
- package/dist/esm/cli.js.map +1 -0
- package/dist/esm/collect.d.ts +6 -0
- package/dist/esm/collect.js +80 -0
- package/dist/esm/collect.js.map +1 -0
- package/dist/esm/dict.d.ts +8 -0
- package/dist/esm/dict.js +28 -0
- package/dist/esm/dict.js.map +1 -0
- package/dist/esm/emit.d.ts +27 -0
- package/dist/esm/emit.js +131 -0
- package/dist/esm/emit.js.map +1 -0
- package/dist/esm/engines/tesseract.d.ts +30 -0
- package/dist/esm/engines/tesseract.js +190 -0
- package/dist/esm/engines/tesseract.js.map +1 -0
- package/dist/esm/errors.d.ts +21 -0
- package/dist/esm/errors.js +9 -0
- package/dist/esm/errors.js.map +1 -0
- package/dist/esm/extract.d.ts +88 -0
- package/dist/esm/extract.js +147 -0
- package/dist/esm/extract.js.map +1 -0
- package/dist/esm/geometry.d.ts +63 -0
- package/dist/esm/geometry.js +279 -0
- package/dist/esm/geometry.js.map +1 -0
- package/dist/esm/index.d.ts +15 -0
- package/dist/esm/index.js +10 -0
- package/dist/esm/index.js.map +1 -0
- package/dist/esm/lines.d.ts +3 -0
- package/dist/esm/lines.js +67 -0
- package/dist/esm/lines.js.map +1 -0
- package/dist/esm/markdown.d.ts +2 -0
- package/dist/esm/markdown.js +153 -0
- package/dist/esm/markdown.js.map +1 -0
- package/dist/esm/package.json +1 -0
- package/dist/esm/page-marker.d.ts +16 -0
- package/dist/esm/page-marker.js +40 -0
- package/dist/esm/page-marker.js.map +1 -0
- package/dist/esm/pdf-open.d.ts +12 -0
- package/dist/esm/pdf-open.js +61 -0
- package/dist/esm/pdf-open.js.map +1 -0
- package/dist/esm/raster.d.ts +26 -0
- package/dist/esm/raster.js +60 -0
- package/dist/esm/raster.js.map +1 -0
- package/dist/esm/render-analysis.d.ts +25 -0
- package/dist/esm/render-analysis.js +115 -0
- package/dist/esm/render-analysis.js.map +1 -0
- package/dist/esm/report.d.ts +12 -0
- package/dist/esm/report.js +182 -0
- package/dist/esm/report.js.map +1 -0
- package/dist/esm/resolve.d.ts +1 -0
- package/dist/esm/resolve.js +7 -0
- package/dist/esm/resolve.js.map +1 -0
- package/dist/esm/types.d.ts +171 -0
- package/dist/esm/types.js +6 -0
- package/dist/esm/types.js.map +1 -0
- package/dist/esm/validate.d.ts +14 -0
- package/dist/esm/validate.js +56 -0
- package/dist/esm/validate.js.map +1 -0
- package/package.json +67 -0
- package/skills/pdf-to-md/SKILL.md +102 -0
|
@@ -0,0 +1,370 @@
|
|
|
1
|
+
// Pass 1: load the PDF and collect positioned glyphs per page, plus the
|
|
2
|
+
// document-wide char-weighted font-height histogram. Heights round to 0.1pt for stable histogram keys; y is flipped so sorting ascending reads top-down.
|
|
3
|
+
"use strict";
|
|
4
|
+
Object.defineProperty(exports, "__esModule", {
|
|
5
|
+
value: true
|
|
6
|
+
});
|
|
7
|
+
function _export(target, all) {
|
|
8
|
+
for(var name in all)Object.defineProperty(target, name, {
|
|
9
|
+
enumerable: true,
|
|
10
|
+
get: Object.getOwnPropertyDescriptor(all, name).get
|
|
11
|
+
});
|
|
12
|
+
}
|
|
13
|
+
_export(exports, {
|
|
14
|
+
get collect () {
|
|
15
|
+
return collect;
|
|
16
|
+
},
|
|
17
|
+
get pageCount () {
|
|
18
|
+
return pageCount;
|
|
19
|
+
}
|
|
20
|
+
});
|
|
21
|
+
var _pdf = require("pdfjs-dist/legacy/build/pdf.mjs");
|
|
22
|
+
var _pdfopents = require("./pdf-open.js");
|
|
23
|
+
function _array_like_to_array(arr, len) {
|
|
24
|
+
if (len == null || len > arr.length) len = arr.length;
|
|
25
|
+
for(var i = 0, arr2 = new Array(len); i < len; i++)arr2[i] = arr[i];
|
|
26
|
+
return arr2;
|
|
27
|
+
}
|
|
28
|
+
function _array_with_holes(arr) {
|
|
29
|
+
if (Array.isArray(arr)) return arr;
|
|
30
|
+
}
|
|
31
|
+
function asyncGeneratorStep(gen, resolve, reject, _next, _throw, key, arg) {
|
|
32
|
+
try {
|
|
33
|
+
var info = gen[key](arg);
|
|
34
|
+
var value = info.value;
|
|
35
|
+
} catch (error) {
|
|
36
|
+
reject(error);
|
|
37
|
+
return;
|
|
38
|
+
}
|
|
39
|
+
if (info.done) resolve(value);
|
|
40
|
+
else Promise.resolve(value).then(_next, _throw);
|
|
41
|
+
}
|
|
42
|
+
function _async_to_generator(fn) {
|
|
43
|
+
return function() {
|
|
44
|
+
var self = this, args = arguments;
|
|
45
|
+
return new Promise(function(resolve, reject) {
|
|
46
|
+
var gen = fn.apply(self, args);
|
|
47
|
+
function _next(value) {
|
|
48
|
+
asyncGeneratorStep(gen, resolve, reject, _next, _throw, "next", value);
|
|
49
|
+
}
|
|
50
|
+
function _throw(err) {
|
|
51
|
+
asyncGeneratorStep(gen, resolve, reject, _next, _throw, "throw", err);
|
|
52
|
+
}
|
|
53
|
+
_next(undefined);
|
|
54
|
+
});
|
|
55
|
+
};
|
|
56
|
+
}
|
|
57
|
+
function _iterable_to_array_limit(arr, i) {
|
|
58
|
+
var _i = arr == null ? null : typeof Symbol !== "undefined" && arr[Symbol.iterator] || arr["@@iterator"];
|
|
59
|
+
if (_i == null) return;
|
|
60
|
+
var _arr = [];
|
|
61
|
+
var _n = true;
|
|
62
|
+
var _d = false;
|
|
63
|
+
var _s, _e;
|
|
64
|
+
try {
|
|
65
|
+
for(_i = _i.call(arr); !(_n = (_s = _i.next()).done); _n = true){
|
|
66
|
+
_arr.push(_s.value);
|
|
67
|
+
if (i && _arr.length === i) break;
|
|
68
|
+
}
|
|
69
|
+
} catch (err) {
|
|
70
|
+
_d = true;
|
|
71
|
+
_e = err;
|
|
72
|
+
} finally{
|
|
73
|
+
try {
|
|
74
|
+
if (!_n && _i["return"] != null) _i["return"]();
|
|
75
|
+
} finally{
|
|
76
|
+
if (_d) throw _e;
|
|
77
|
+
}
|
|
78
|
+
}
|
|
79
|
+
return _arr;
|
|
80
|
+
}
|
|
81
|
+
function _non_iterable_rest() {
|
|
82
|
+
throw new TypeError("Invalid attempt to destructure non-iterable instance.\nIn order to be iterable, non-array objects must have a [Symbol.iterator]() method.");
|
|
83
|
+
}
|
|
84
|
+
function _sliced_to_array(arr, i) {
|
|
85
|
+
return _array_with_holes(arr) || _iterable_to_array_limit(arr, i) || _unsupported_iterable_to_array(arr, i) || _non_iterable_rest();
|
|
86
|
+
}
|
|
87
|
+
function _ts_generator(thisArg, body) {
|
|
88
|
+
var f, y, t, _ = {
|
|
89
|
+
label: 0,
|
|
90
|
+
sent: function() {
|
|
91
|
+
if (t[0] & 1) throw t[1];
|
|
92
|
+
return t[1];
|
|
93
|
+
},
|
|
94
|
+
trys: [],
|
|
95
|
+
ops: []
|
|
96
|
+
}, g = Object.create((typeof Iterator === "function" ? Iterator : Object).prototype), d = Object.defineProperty;
|
|
97
|
+
return d(g, "next", {
|
|
98
|
+
value: verb(0)
|
|
99
|
+
}), d(g, "throw", {
|
|
100
|
+
value: verb(1)
|
|
101
|
+
}), d(g, "return", {
|
|
102
|
+
value: verb(2)
|
|
103
|
+
}), typeof Symbol === "function" && d(g, Symbol.iterator, {
|
|
104
|
+
value: function() {
|
|
105
|
+
return this;
|
|
106
|
+
}
|
|
107
|
+
}), g;
|
|
108
|
+
function verb(n) {
|
|
109
|
+
return function(v) {
|
|
110
|
+
return step([
|
|
111
|
+
n,
|
|
112
|
+
v
|
|
113
|
+
]);
|
|
114
|
+
};
|
|
115
|
+
}
|
|
116
|
+
function step(op) {
|
|
117
|
+
if (f) throw new TypeError("Generator is already executing.");
|
|
118
|
+
while(g && (g = 0, op[0] && (_ = 0)), _)try {
|
|
119
|
+
if (f = 1, y && (t = op[0] & 2 ? y["return"] : op[0] ? y["throw"] || ((t = y["return"]) && t.call(y), 0) : y.next) && !(t = t.call(y, op[1])).done) return t;
|
|
120
|
+
if (y = 0, t) op = [
|
|
121
|
+
op[0] & 2,
|
|
122
|
+
t.value
|
|
123
|
+
];
|
|
124
|
+
switch(op[0]){
|
|
125
|
+
case 0:
|
|
126
|
+
case 1:
|
|
127
|
+
t = op;
|
|
128
|
+
break;
|
|
129
|
+
case 4:
|
|
130
|
+
_.label++;
|
|
131
|
+
return {
|
|
132
|
+
value: op[1],
|
|
133
|
+
done: false
|
|
134
|
+
};
|
|
135
|
+
case 5:
|
|
136
|
+
_.label++;
|
|
137
|
+
y = op[1];
|
|
138
|
+
op = [
|
|
139
|
+
0
|
|
140
|
+
];
|
|
141
|
+
continue;
|
|
142
|
+
case 7:
|
|
143
|
+
op = _.ops.pop();
|
|
144
|
+
_.trys.pop();
|
|
145
|
+
continue;
|
|
146
|
+
default:
|
|
147
|
+
if (!(t = _.trys, t = t.length > 0 && t[t.length - 1]) && (op[0] === 6 || op[0] === 2)) {
|
|
148
|
+
_ = 0;
|
|
149
|
+
continue;
|
|
150
|
+
}
|
|
151
|
+
if (op[0] === 3 && (!t || op[1] > t[0] && op[1] < t[3])) {
|
|
152
|
+
_.label = op[1];
|
|
153
|
+
break;
|
|
154
|
+
}
|
|
155
|
+
if (op[0] === 6 && _.label < t[1]) {
|
|
156
|
+
_.label = t[1];
|
|
157
|
+
t = op;
|
|
158
|
+
break;
|
|
159
|
+
}
|
|
160
|
+
if (t && _.label < t[2]) {
|
|
161
|
+
_.label = t[2];
|
|
162
|
+
_.ops.push(op);
|
|
163
|
+
break;
|
|
164
|
+
}
|
|
165
|
+
if (t[2]) _.ops.pop();
|
|
166
|
+
_.trys.pop();
|
|
167
|
+
continue;
|
|
168
|
+
}
|
|
169
|
+
op = body.call(thisArg, _);
|
|
170
|
+
} catch (e) {
|
|
171
|
+
op = [
|
|
172
|
+
6,
|
|
173
|
+
e
|
|
174
|
+
];
|
|
175
|
+
y = 0;
|
|
176
|
+
} finally{
|
|
177
|
+
f = t = 0;
|
|
178
|
+
}
|
|
179
|
+
if (op[0] & 5) throw op[1];
|
|
180
|
+
return {
|
|
181
|
+
value: op[0] ? op[1] : void 0,
|
|
182
|
+
done: true
|
|
183
|
+
};
|
|
184
|
+
}
|
|
185
|
+
}
|
|
186
|
+
function _unsupported_iterable_to_array(o, minLen) {
|
|
187
|
+
if (!o) return;
|
|
188
|
+
if (typeof o === "string") return _array_like_to_array(o, minLen);
|
|
189
|
+
var n = Object.prototype.toString.call(o).slice(8, -1);
|
|
190
|
+
if (n === "Object" && o.constructor) n = o.constructor.name;
|
|
191
|
+
if (n === "Map" || n === "Set") return Array.from(n);
|
|
192
|
+
if (n === "Arguments" || /^(?:Ui|I)nt(?:8|16|32)(?:Clamped)?Array$/.test(n)) return _array_like_to_array(o, minLen);
|
|
193
|
+
}
|
|
194
|
+
// HYBRID-PAGE DETECTION (README's "Design contract"): a page can carry a
|
|
195
|
+
// real text layer AND a separately-painted large image with its OWN baked-in text that `page.getTextContent()` never sees.
|
|
196
|
+
//
|
|
197
|
+
// Signal: scan the operator list for `OPS.paintImageXObject`, whose args are
|
|
198
|
+
// `[objId, width, height]` — available synchronously, `large` is >=200x200px. Masks/nested Form XObjects aren't walked (accepted gap); missing dims fall back to "op present" as the signal.
|
|
199
|
+
var LARGE_IMAGE_MIN_PX = 200;
|
|
200
|
+
function hasLargeImage(page) {
|
|
201
|
+
return _async_to_generator(function() {
|
|
202
|
+
var _ref, fnArray, argsArray, i, _argsArray_i, w, h;
|
|
203
|
+
return _ts_generator(this, function(_state) {
|
|
204
|
+
switch(_state.label){
|
|
205
|
+
case 0:
|
|
206
|
+
return [
|
|
207
|
+
4,
|
|
208
|
+
page.getOperatorList()
|
|
209
|
+
];
|
|
210
|
+
case 1:
|
|
211
|
+
_ref = _state.sent(), fnArray = _ref.fnArray, argsArray = _ref.argsArray;
|
|
212
|
+
for(i = 0; i < fnArray.length; i++){
|
|
213
|
+
if (fnArray[i] !== _pdf.OPS.paintImageXObject) continue;
|
|
214
|
+
_argsArray_i = _sliced_to_array(argsArray[i], 3), w = _argsArray_i[1], h = _argsArray_i[2];
|
|
215
|
+
if (typeof w === 'number' && typeof h === 'number') {
|
|
216
|
+
if (w >= LARGE_IMAGE_MIN_PX && h >= LARGE_IMAGE_MIN_PX) return [
|
|
217
|
+
2,
|
|
218
|
+
true
|
|
219
|
+
];
|
|
220
|
+
} else return [
|
|
221
|
+
2,
|
|
222
|
+
true
|
|
223
|
+
]; // dims unavailable: coarser "op present" signal
|
|
224
|
+
}
|
|
225
|
+
return [
|
|
226
|
+
2,
|
|
227
|
+
false
|
|
228
|
+
];
|
|
229
|
+
}
|
|
230
|
+
});
|
|
231
|
+
})();
|
|
232
|
+
}
|
|
233
|
+
function pageCount(src) {
|
|
234
|
+
return _async_to_generator(function() {
|
|
235
|
+
var doc;
|
|
236
|
+
return _ts_generator(this, function(_state) {
|
|
237
|
+
switch(_state.label){
|
|
238
|
+
case 0:
|
|
239
|
+
return [
|
|
240
|
+
4,
|
|
241
|
+
(0, _pdfopents.openDocument)(src)
|
|
242
|
+
];
|
|
243
|
+
case 1:
|
|
244
|
+
doc = _state.sent();
|
|
245
|
+
return [
|
|
246
|
+
2,
|
|
247
|
+
doc.numPages
|
|
248
|
+
];
|
|
249
|
+
}
|
|
250
|
+
});
|
|
251
|
+
})();
|
|
252
|
+
}
|
|
253
|
+
function collect(_0) {
|
|
254
|
+
return _async_to_generator(function(src) {
|
|
255
|
+
var _ref, first, last, doc, from, to, pages, heightChars, p, page, items, vp, glyphs, textChars, _iteratorNormalCompletion, _didIteratorError, _iteratorError, _iterator, _step, it, h, largeImage, _tmp;
|
|
256
|
+
var _arguments = arguments;
|
|
257
|
+
return _ts_generator(this, function(_state) {
|
|
258
|
+
switch(_state.label){
|
|
259
|
+
case 0:
|
|
260
|
+
_ref = _arguments.length > 1 && _arguments[1] !== void 0 ? _arguments[1] : {}, first = _ref.first, last = _ref.last;
|
|
261
|
+
return [
|
|
262
|
+
4,
|
|
263
|
+
(0, _pdfopents.openDocument)(src)
|
|
264
|
+
];
|
|
265
|
+
case 1:
|
|
266
|
+
doc = _state.sent();
|
|
267
|
+
from = Math.min(Math.max(1, first !== null && first !== void 0 ? first : 1), Math.max(1, doc.numPages));
|
|
268
|
+
to = Math.min(Math.max(from, last !== null && last !== void 0 ? last : doc.numPages), doc.numPages);
|
|
269
|
+
pages = [];
|
|
270
|
+
heightChars = new Map();
|
|
271
|
+
p = from;
|
|
272
|
+
_state.label = 2;
|
|
273
|
+
case 2:
|
|
274
|
+
if (!(p <= to)) return [
|
|
275
|
+
3,
|
|
276
|
+
8
|
|
277
|
+
];
|
|
278
|
+
return [
|
|
279
|
+
4,
|
|
280
|
+
doc.getPage(p)
|
|
281
|
+
];
|
|
282
|
+
case 3:
|
|
283
|
+
page = _state.sent();
|
|
284
|
+
return [
|
|
285
|
+
4,
|
|
286
|
+
page.getTextContent()
|
|
287
|
+
];
|
|
288
|
+
case 4:
|
|
289
|
+
items = _state.sent().items;
|
|
290
|
+
vp = page.getViewport({
|
|
291
|
+
scale: 1
|
|
292
|
+
});
|
|
293
|
+
glyphs = [];
|
|
294
|
+
textChars = 0;
|
|
295
|
+
_iteratorNormalCompletion = true, _didIteratorError = false, _iteratorError = undefined;
|
|
296
|
+
try {
|
|
297
|
+
for(_iterator = items[Symbol.iterator](); !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true){
|
|
298
|
+
it = _step.value;
|
|
299
|
+
if (!('str' in it) || typeof it.str !== 'string') continue;
|
|
300
|
+
textChars += it.str.length;
|
|
301
|
+
if (!it.str.trim()) continue;
|
|
302
|
+
h = Math.round(Math.abs(it.transform[3]) * 10) / 10;
|
|
303
|
+
glyphs.push({
|
|
304
|
+
s: it.str,
|
|
305
|
+
x: it.transform[4],
|
|
306
|
+
y: vp.height - it.transform[5],
|
|
307
|
+
w: it.width,
|
|
308
|
+
h: h
|
|
309
|
+
});
|
|
310
|
+
heightChars.set(h, (heightChars.get(h) || 0) + it.str.length);
|
|
311
|
+
}
|
|
312
|
+
} catch (err) {
|
|
313
|
+
_didIteratorError = true;
|
|
314
|
+
_iteratorError = err;
|
|
315
|
+
} finally{
|
|
316
|
+
try {
|
|
317
|
+
if (!_iteratorNormalCompletion && _iterator.return != null) {
|
|
318
|
+
_iterator.return();
|
|
319
|
+
}
|
|
320
|
+
} finally{
|
|
321
|
+
if (_didIteratorError) {
|
|
322
|
+
throw _iteratorError;
|
|
323
|
+
}
|
|
324
|
+
}
|
|
325
|
+
}
|
|
326
|
+
_tmp = textChars > 80;
|
|
327
|
+
if (!_tmp) return [
|
|
328
|
+
3,
|
|
329
|
+
6
|
|
330
|
+
];
|
|
331
|
+
return [
|
|
332
|
+
4,
|
|
333
|
+
hasLargeImage(page)
|
|
334
|
+
];
|
|
335
|
+
case 5:
|
|
336
|
+
_tmp = _state.sent();
|
|
337
|
+
_state.label = 6;
|
|
338
|
+
case 6:
|
|
339
|
+
largeImage = _tmp;
|
|
340
|
+
pages.push({
|
|
341
|
+
page: p,
|
|
342
|
+
glyphs: glyphs,
|
|
343
|
+
width: vp.width,
|
|
344
|
+
height: vp.height,
|
|
345
|
+
textChars: textChars,
|
|
346
|
+
largeImage: largeImage
|
|
347
|
+
});
|
|
348
|
+
_state.label = 7;
|
|
349
|
+
case 7:
|
|
350
|
+
p++;
|
|
351
|
+
return [
|
|
352
|
+
3,
|
|
353
|
+
2
|
|
354
|
+
];
|
|
355
|
+
case 8:
|
|
356
|
+
// `numPages` is the SELECTED count, not doc.numPages: the auto-OCR
|
|
357
|
+
// chars/page floor must judge a `--pages` subset against its own page count, or a genuinely image-only page reads as merely "sparse".
|
|
358
|
+
return [
|
|
359
|
+
2,
|
|
360
|
+
{
|
|
361
|
+
numPages: pages.length,
|
|
362
|
+
pages: pages,
|
|
363
|
+
heightChars: heightChars
|
|
364
|
+
}
|
|
365
|
+
];
|
|
366
|
+
}
|
|
367
|
+
});
|
|
368
|
+
}).apply(this, arguments);
|
|
369
|
+
}
|
|
370
|
+
/* CJS INTEROP */ if (exports.__esModule && exports.default) { try { Object.defineProperty(exports.default, '__esModule', { value: true }); for (var key in exports) { exports.default[key] = exports[key]; } } catch (_) {}; module.exports = exports.default; }
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"sources":["/Users/kevin/Projects/pdf-to-md/src/collect.ts"],"sourcesContent":["// Pass 1: load the PDF and collect positioned glyphs per page, plus the\n// document-wide char-weighted font-height histogram. Heights round to 0.1pt for stable histogram keys; y is flipped so sorting ascending reads top-down.\nimport type { PDFPageProxy } from 'pdfjs-dist/legacy/build/pdf.mjs';\nimport { OPS } from 'pdfjs-dist/legacy/build/pdf.mjs';\nimport { openDocument } from './pdf-open.ts';\nimport type { CollectedPage, CollectResult, Glyph } from './types.ts';\n\n// HYBRID-PAGE DETECTION (README's \"Design contract\"): a page can carry a\n// real text layer AND a separately-painted large image with its OWN baked-in text that `page.getTextContent()` never sees.\n//\n// Signal: scan the operator list for `OPS.paintImageXObject`, whose args are\n// `[objId, width, height]` — available synchronously, `large` is >=200x200px. Masks/nested Form XObjects aren't walked (accepted gap); missing dims fall back to \"op present\" as the signal.\nconst LARGE_IMAGE_MIN_PX = 200;\nasync function hasLargeImage(page: PDFPageProxy): Promise<boolean> {\n const { fnArray, argsArray } = await page.getOperatorList();\n for (let i = 0; i < fnArray.length; i++) {\n if (fnArray[i] !== OPS.paintImageXObject) continue;\n const [, w, h] = argsArray[i];\n if (typeof w === 'number' && typeof h === 'number') {\n if (w >= LARGE_IMAGE_MIN_PX && h >= LARGE_IMAGE_MIN_PX) return true;\n } else return true; // dims unavailable: coarser \"op present\" signal\n }\n return false;\n}\n\n// { first, last }: 1-based, inclusive page range (the CLI's --pages).\n// Out-of-range values CLAMP to the document's bounds; `page` on each returned entry is always the SOURCE page number, so markers/splicing stay correct.\n//\n// CAVEAT: `heightChars` is built ONLY from the pages actually collected, so\n// a `--pages` subset's heading levels can differ from a full run's — accepted since re-extracted pages get spliced back in for their prose, not diffed.\n\n// Cheap page-count-only load (no per-page text extraction) — the OCR path\n// needs it to resolve its own [first, last] clamping and print a \"### pN\" marker for every page in range, including ones with no recognized words.\nexport async function pageCount(src: string): Promise<number> {\n const doc = await openDocument(src);\n return doc.numPages;\n}\n\nexport async function collect(src: string, { first, last }: { first?: number; last?: number } = {}): Promise<CollectResult> {\n const doc = await openDocument(src);\n const from = Math.min(Math.max(1, first ?? 1), Math.max(1, doc.numPages));\n const to = Math.min(Math.max(from, last ?? doc.numPages), doc.numPages);\n const pages: CollectedPage[] = [];\n const heightChars = new Map<number, number>();\n for (let p = from; p <= to; p++) {\n const page = await doc.getPage(p);\n const { items } = await page.getTextContent();\n const vp = page.getViewport({ scale: 1 });\n const glyphs: Glyph[] = [];\n let textChars = 0;\n for (const it of items) {\n if (!('str' in it) || typeof it.str !== 'string') continue;\n textChars += it.str.length;\n if (!it.str.trim()) continue;\n const h = Math.round(Math.abs(it.transform[3]) * 10) / 10;\n glyphs.push({\n s: it.str,\n x: it.transform[4],\n y: vp.height - it.transform[5],\n w: it.width,\n h,\n });\n heightChars.set(h, (heightChars.get(h) || 0) + it.str.length);\n }\n // \"text layer used\" for hybrid-page purposes (page text > 80 chars) — a\n // page with only a couple of stray scraps isn't \"using\" the text layer even if it also carries a large image.\n const largeImage = textChars > 80 && (await hasLargeImage(page));\n pages.push({\n page: p,\n glyphs,\n width: vp.width,\n height: vp.height,\n textChars,\n largeImage,\n });\n }\n // `numPages` is the SELECTED count, not doc.numPages: the auto-OCR\n // chars/page floor must judge a `--pages` subset against its own page count, or a genuinely image-only page reads as merely \"sparse\".\n return { numPages: pages.length, pages, heightChars };\n}\n"],"names":["collect","pageCount","LARGE_IMAGE_MIN_PX","hasLargeImage","page","fnArray","argsArray","i","w","h","getOperatorList","length","OPS","paintImageXObject","src","doc","openDocument","numPages","first","last","from","to","pages","heightChars","p","items","vp","glyphs","textChars","it","largeImage","Math","min","max","Map","getPage","getTextContent","getViewport","scale","str","trim","round","abs","transform","push","s","x","y","height","width","set","get"],"mappings":"AAAA,wEAAwE;AACxE,yJAAyJ;;;;;;;;;;;;QAqCnIA;eAAAA;;QALAC;eAAAA;;;mBA9BF;yBACS;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAG7B,yEAAyE;AACzE,2HAA2H;AAC3H,EAAE;AACF,6EAA6E;AAC7E,6LAA6L;AAC7L,IAAMC,qBAAqB;AAC3B,SAAeC,cAAcC,IAAkB;;YACd,MAAvBC,SAASC,WACRC,GAEUD,cAARE,GAAGC;;;;oBAHiB;;wBAAML,KAAKM,eAAe;;;oBAA1B,OAAA,eAAvBL,UAAuB,KAAvBA,SAASC,YAAc,KAAdA;oBACjB,IAASC,IAAI,GAAGA,IAAIF,QAAQM,MAAM,EAAEJ,IAAK;wBACvC,IAAIF,OAAO,CAACE,EAAE,KAAKK,QAAG,CAACC,iBAAiB,EAAE;wBACzBP,gCAAAA,SAAS,CAACC,EAAE,MAApBC,IAAQF,iBAALG,IAAKH;wBACjB,IAAI,OAAOE,MAAM,YAAY,OAAOC,MAAM,UAAU;4BAClD,IAAID,KAAKN,sBAAsBO,KAAKP,oBAAoB;;gCAAO;;wBACjE,OAAO;;4BAAO;2BAAM,gDAAgD;oBACtE;oBACA;;wBAAO;;;;IACT;;AAUO,SAAeD,UAAUa,GAAW;;YACnCC;;;;oBAAM;;wBAAMC,IAAAA,uBAAY,EAACF;;;oBAAzBC,MAAM;oBACZ;;wBAAOA,IAAIE,QAAQ;;;;IACrB;;AAEO,SAAejB;wCAAQc,GAAW;YAAE,MAAEI,OAAOC,MAC5CJ,KACAK,MACAC,IACAC,OACAC,aACGC,GACDpB,MACEqB,OACFC,IACAC,QACFC,WACC,2BAAA,mBAAA,gBAAA,WAAA,OAAMC,IAIHpB,GAYFqB;;;;;oBA5BiC,OAAA,oEAAqD,CAAC,GAApDZ,QAAF,KAAEA,OAAOC,OAAT,KAASA;oBACtC;;wBAAMH,IAAAA,uBAAY,EAACF;;;oBAAzBC,MAAM;oBACNK,OAAOW,KAAKC,GAAG,CAACD,KAAKE,GAAG,CAAC,GAAGf,kBAAAA,mBAAAA,QAAS,IAAIa,KAAKE,GAAG,CAAC,GAAGlB,IAAIE,QAAQ;oBACjEI,KAAKU,KAAKC,GAAG,CAACD,KAAKE,GAAG,CAACb,MAAMD,iBAAAA,kBAAAA,OAAQJ,IAAIE,QAAQ,GAAGF,IAAIE,QAAQ;oBAChEK;oBACAC,cAAc,IAAIW;oBACfV,IAAIJ;;;yBAAMI,CAAAA,KAAKH,EAAC;;;;oBACV;;wBAAMN,IAAIoB,OAAO,CAACX;;;oBAAzBpB,OAAO;oBACK;;wBAAMA,KAAKgC,cAAc;;;oBAAnCX,QAAU,cAAVA;oBACFC,KAAKtB,KAAKiC,WAAW,CAAC;wBAAEC,OAAO;oBAAE;oBACjCX;oBACFC,YAAY;oBACX,kCAAA,2BAAA;;wBAAL,IAAK,YAAYH,4BAAZ,6BAAA,QAAA,yBAAA,iCAAmB;4BAAbI,KAAN;4BACH,IAAI,CAAE,CAAA,SAASA,EAAC,KAAM,OAAOA,GAAGU,GAAG,KAAK,UAAU;4BAClDX,aAAaC,GAAGU,GAAG,CAAC5B,MAAM;4BAC1B,IAAI,CAACkB,GAAGU,GAAG,CAACC,IAAI,IAAI;4BACd/B,IAAIsB,KAAKU,KAAK,CAACV,KAAKW,GAAG,CAACb,GAAGc,SAAS,CAAC,EAAE,IAAI,MAAM;4BACvDhB,OAAOiB,IAAI,CAAC;gCACVC,GAAGhB,GAAGU,GAAG;gCACTO,GAAGjB,GAAGc,SAAS,CAAC,EAAE;gCAClBI,GAAGrB,GAAGsB,MAAM,GAAGnB,GAAGc,SAAS,CAAC,EAAE;gCAC9BnC,GAAGqB,GAAGoB,KAAK;gCACXxC,GAAAA;4BACF;4BACAc,YAAY2B,GAAG,CAACzC,GAAG,AAACc,CAAAA,YAAY4B,GAAG,CAAC1C,MAAM,CAAA,IAAKoB,GAAGU,GAAG,CAAC5B,MAAM;wBAC9D;;wBAbK;wBAAA;;;iCAAA,6BAAA;gCAAA;;;gCAAA;sCAAA;;;;2BAgBciB,YAAY;+BAAZA;;;;oBAAmB;;wBAAMzB,cAAcC;;;2BAApB;;;oBAAhC0B;oBACNR,MAAMsB,IAAI,CAAC;wBACTxC,MAAMoB;wBACNG,QAAAA;wBACAsB,OAAOvB,GAAGuB,KAAK;wBACfD,QAAQtB,GAAGsB,MAAM;wBACjBpB,WAAAA;wBACAE,YAAAA;oBACF;;;oBA9B0BN;;;;;;oBAgC5B,mEAAmE;oBACnE,sIAAsI;oBACtI;;wBAAO;4BAAEP,UAAUK,MAAMX,MAAM;4BAAEW,OAAAA;4BAAOC,aAAAA;wBAAY;;;;IACtD"}
|
|
@@ -0,0 +1,8 @@
|
|
|
1
|
+
export declare const DICT_PATH: string;
|
|
2
|
+
/** The hyphen-rejoin dictionary; an empty set if none was found. */
|
|
3
|
+
export declare function getDict(): Set<string>;
|
|
4
|
+
/** False when no dictionary file exists at DICT_PATH — every line-final
|
|
5
|
+
* hyphen then stays unresolved. */
|
|
6
|
+
export declare function dictAvailable(): boolean;
|
|
7
|
+
/** The notice callers surface via warn() when dictAvailable() is false. */
|
|
8
|
+
export declare function dictUnavailableWarning(): string;
|
|
@@ -0,0 +1,8 @@
|
|
|
1
|
+
export declare const DICT_PATH: string;
|
|
2
|
+
/** The hyphen-rejoin dictionary; an empty set if none was found. */
|
|
3
|
+
export declare function getDict(): Set<string>;
|
|
4
|
+
/** False when no dictionary file exists at DICT_PATH — every line-final
|
|
5
|
+
* hyphen then stays unresolved. */
|
|
6
|
+
export declare function dictAvailable(): boolean;
|
|
7
|
+
/** The notice callers surface via warn() when dictAvailable() is false. */
|
|
8
|
+
export declare function dictUnavailableWarning(): string;
|
package/dist/cjs/dict.js
ADDED
|
@@ -0,0 +1,55 @@
|
|
|
1
|
+
// The hyphen-rejoin dictionary, shared by src/emit.ts and src/markdown.ts.
|
|
2
|
+
// Loaded once, lazily, on first use — never at import — so importing the library does no file IO on its own.
|
|
3
|
+
"use strict";
|
|
4
|
+
Object.defineProperty(exports, "__esModule", {
|
|
5
|
+
value: true
|
|
6
|
+
});
|
|
7
|
+
function _export(target, all) {
|
|
8
|
+
for(var name in all)Object.defineProperty(target, name, {
|
|
9
|
+
enumerable: true,
|
|
10
|
+
get: Object.getOwnPropertyDescriptor(all, name).get
|
|
11
|
+
});
|
|
12
|
+
}
|
|
13
|
+
_export(exports, {
|
|
14
|
+
get DICT_PATH () {
|
|
15
|
+
return DICT_PATH;
|
|
16
|
+
},
|
|
17
|
+
get dictAvailable () {
|
|
18
|
+
return dictAvailable;
|
|
19
|
+
},
|
|
20
|
+
get dictUnavailableWarning () {
|
|
21
|
+
return dictUnavailableWarning;
|
|
22
|
+
},
|
|
23
|
+
get getDict () {
|
|
24
|
+
return getDict;
|
|
25
|
+
}
|
|
26
|
+
});
|
|
27
|
+
var _nodefs = require("node:fs");
|
|
28
|
+
var _wordlist = /*#__PURE__*/ _interop_require_default(require("word-list"));
|
|
29
|
+
function _interop_require_default(obj) {
|
|
30
|
+
return obj && obj.__esModule ? obj : {
|
|
31
|
+
default: obj
|
|
32
|
+
};
|
|
33
|
+
}
|
|
34
|
+
var DICT_PATH = process.env.PDF_TO_MD_DICT_PATH || _wordlist.default;
|
|
35
|
+
var cache = null;
|
|
36
|
+
function load() {
|
|
37
|
+
if (!cache) {
|
|
38
|
+
var available = (0, _nodefs.existsSync)(DICT_PATH);
|
|
39
|
+
cache = {
|
|
40
|
+
words: available ? new Set((0, _nodefs.readFileSync)(DICT_PATH, 'utf8').toLowerCase().split('\n')) : new Set(),
|
|
41
|
+
available: available
|
|
42
|
+
};
|
|
43
|
+
}
|
|
44
|
+
return cache;
|
|
45
|
+
}
|
|
46
|
+
function getDict() {
|
|
47
|
+
return load().words;
|
|
48
|
+
}
|
|
49
|
+
function dictAvailable() {
|
|
50
|
+
return load().available;
|
|
51
|
+
}
|
|
52
|
+
function dictUnavailableWarning() {
|
|
53
|
+
return "no dictionary at ".concat(DICT_PATH, " — line-final hyphens kept unresolved");
|
|
54
|
+
}
|
|
55
|
+
/* CJS INTEROP */ if (exports.__esModule && exports.default) { try { Object.defineProperty(exports.default, '__esModule', { value: true }); for (var key in exports) { exports.default[key] = exports[key]; } } catch (_) {}; module.exports = exports.default; }
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"sources":["/Users/kevin/Projects/pdf-to-md/src/dict.ts"],"sourcesContent":["// The hyphen-rejoin dictionary, shared by src/emit.ts and src/markdown.ts.\n// Loaded once, lazily, on first use — never at import — so importing the library does no file IO on its own.\nimport { existsSync, readFileSync } from 'node:fs';\nimport wordListPath from 'word-list';\n\n// Bundled list (not the OS's /usr/share/dict/words) → identical rejoin decisions on every\n// OS/machine. PDF_TO_MD_DICT_PATH overrides it — the test seam for simulating no dictionary.\nexport const DICT_PATH = process.env.PDF_TO_MD_DICT_PATH || wordListPath;\n\nlet cache: { words: Set<string>; available: boolean } | null = null;\n\nfunction load(): { words: Set<string>; available: boolean } {\n if (!cache) {\n const available = existsSync(DICT_PATH);\n cache = { words: available ? new Set(readFileSync(DICT_PATH, 'utf8').toLowerCase().split('\\n')) : new Set(), available };\n }\n return cache;\n}\n\n/** The hyphen-rejoin dictionary; an empty set if none was found. */\nexport function getDict(): Set<string> {\n return load().words;\n}\n\n/** False when no dictionary file exists at DICT_PATH — every line-final\n * hyphen then stays unresolved. */\nexport function dictAvailable(): boolean {\n return load().available;\n}\n\n/** The notice callers surface via warn() when dictAvailable() is false. */\nexport function dictUnavailableWarning(): string {\n return `no dictionary at ${DICT_PATH} — line-final hyphens kept unresolved`;\n}\n"],"names":["DICT_PATH","dictAvailable","dictUnavailableWarning","getDict","process","env","PDF_TO_MD_DICT_PATH","wordListPath","cache","load","available","existsSync","words","Set","readFileSync","toLowerCase","split"],"mappings":"AAAA,2EAA2E;AAC3E,6GAA6G;;;;;;;;;;;;QAMhGA;eAAAA;;QAmBGC;eAAAA;;QAKAC;eAAAA;;QAXAC;eAAAA;;;sBAlByB;+DAChB;;;;;;AAIlB,IAAMH,YAAYI,QAAQC,GAAG,CAACC,mBAAmB,IAAIC,iBAAY;AAExE,IAAIC,QAA2D;AAE/D,SAASC;IACP,IAAI,CAACD,OAAO;QACV,IAAME,YAAYC,IAAAA,kBAAU,EAACX;QAC7BQ,QAAQ;YAAEI,OAAOF,YAAY,IAAIG,IAAIC,IAAAA,oBAAY,EAACd,WAAW,QAAQe,WAAW,GAAGC,KAAK,CAAC,SAAS,IAAIH;YAAOH,WAAAA;QAAU;IACzH;IACA,OAAOF;AACT;AAGO,SAASL;IACd,OAAOM,OAAOG,KAAK;AACrB;AAIO,SAASX;IACd,OAAOQ,OAAOC,SAAS;AACzB;AAGO,SAASR;IACd,OAAO,AAAC,oBAA6B,OAAVF,WAAU;AACvC"}
|
|
@@ -0,0 +1,27 @@
|
|
|
1
|
+
import type { PageOCR, Tuning, Word } from './geometry.js';
|
|
2
|
+
/** Share of word-like tokens (3+ letter run) below which a block counts as
|
|
3
|
+
* unstructured. Measured (`review-blocks.test.ts`): junk 0.000-0.100 vs prose 0.743-0.857, cut 0.4. */
|
|
4
|
+
export declare const UNSTRUCTURED_WORDLIKE_FRACTION = 0.4;
|
|
5
|
+
export type EmitBlock = {
|
|
6
|
+
kind: 'heading';
|
|
7
|
+
text: string;
|
|
8
|
+
level: number;
|
|
9
|
+
words: Word[];
|
|
10
|
+
} | {
|
|
11
|
+
kind: 'paragraph';
|
|
12
|
+
text: string;
|
|
13
|
+
words: Word[];
|
|
14
|
+
} | {
|
|
15
|
+
kind: 'quote';
|
|
16
|
+
text: string;
|
|
17
|
+
words: Word[];
|
|
18
|
+
} | {
|
|
19
|
+
kind: 'review';
|
|
20
|
+
text: string;
|
|
21
|
+
/** The parseable reason inside `> [review: ...]`, carried on the block
|
|
22
|
+
* so more reasons can be added without changing the render format. */
|
|
23
|
+
reason: string;
|
|
24
|
+
words: Word[];
|
|
25
|
+
};
|
|
26
|
+
export declare function renderEmitBlock(b: EmitBlock): string;
|
|
27
|
+
export declare function emitPage(p: PageOCR, levelOf: Map<number, number>, tuning: Tuning): EmitBlock[];
|
|
@@ -0,0 +1,27 @@
|
|
|
1
|
+
import type { PageOCR, Tuning, Word } from './geometry.js';
|
|
2
|
+
/** Share of word-like tokens (3+ letter run) below which a block counts as
|
|
3
|
+
* unstructured. Measured (`review-blocks.test.ts`): junk 0.000-0.100 vs prose 0.743-0.857, cut 0.4. */
|
|
4
|
+
export declare const UNSTRUCTURED_WORDLIKE_FRACTION = 0.4;
|
|
5
|
+
export type EmitBlock = {
|
|
6
|
+
kind: 'heading';
|
|
7
|
+
text: string;
|
|
8
|
+
level: number;
|
|
9
|
+
words: Word[];
|
|
10
|
+
} | {
|
|
11
|
+
kind: 'paragraph';
|
|
12
|
+
text: string;
|
|
13
|
+
words: Word[];
|
|
14
|
+
} | {
|
|
15
|
+
kind: 'quote';
|
|
16
|
+
text: string;
|
|
17
|
+
words: Word[];
|
|
18
|
+
} | {
|
|
19
|
+
kind: 'review';
|
|
20
|
+
text: string;
|
|
21
|
+
/** The parseable reason inside `> [review: ...]`, carried on the block
|
|
22
|
+
* so more reasons can be added without changing the render format. */
|
|
23
|
+
reason: string;
|
|
24
|
+
words: Word[];
|
|
25
|
+
};
|
|
26
|
+
export declare function renderEmitBlock(b: EmitBlock): string;
|
|
27
|
+
export declare function emitPage(p: PageOCR, levelOf: Map<number, number>, tuning: Tuning): EmitBlock[];
|