@markitdownjs/pdf 0.1.1 → 0.2.0
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.
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"pdf-converter.d.ts","sourceRoot":"","sources":["../src/pdf-converter.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EACV,SAAS,EACT,eAAe,EACf,gBAAgB,EAIjB,MAAM,sBAAsB,CAAC;AAW9B,8DAA8D;AAC9D,qBAAa,YAAa,YAAW,SAAS;IAC5C,QAAQ,CAAC,EAAE,EAAG,KAAK,CAAU;IAC7B,QAAQ,CAAC,kBAAkB,WAAuB;IAClD,QAAQ,CAAC,mBAAmB,WAAY;IAExC;;;;OAIG;IACG,UAAU,CAAC,KAAK,EAAE,eAAe,GAAG,OAAO,CAAC,OAAO,CAAC;
|
|
1
|
+
{"version":3,"file":"pdf-converter.d.ts","sourceRoot":"","sources":["../src/pdf-converter.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EACV,SAAS,EACT,eAAe,EACf,gBAAgB,EAIjB,MAAM,sBAAsB,CAAC;AAW9B,8DAA8D;AAC9D,qBAAa,YAAa,YAAW,SAAS;IAC5C,QAAQ,CAAC,EAAE,EAAG,KAAK,CAAU;IAC7B,QAAQ,CAAC,kBAAkB,WAAuB;IAClD,QAAQ,CAAC,mBAAmB,WAAY;IAExC;;;;OAIG;IACG,UAAU,CAAC,KAAK,EAAE,eAAe,GAAG,OAAO,CAAC,OAAO,CAAC;IAyB1D;;;;;;;;;;OAUG;IACG,OAAO,CAAC,KAAK,EAAE,eAAe,GAAG,OAAO,CAAC,gBAAgB,CAAC;IAuNhE;;;;;;;;OAQG;IACH,OAAO,CAAC,eAAe;IAOvB;;;;;;;;OAQG;IACH,OAAO,CAAC,eAAe;IAyBvB;;;;OAIG;YACW,WAAW;CAe1B"}
|
package/dist/pdf-converter.js
CHANGED
|
@@ -21,9 +21,7 @@ export class PdfConverter {
|
|
|
21
21
|
}
|
|
22
22
|
}
|
|
23
23
|
if (input.data instanceof Uint8Array || input.data instanceof ArrayBuffer) {
|
|
24
|
-
const bytes = input.data instanceof Uint8Array
|
|
25
|
-
? input.data
|
|
26
|
-
: new Uint8Array(input.data);
|
|
24
|
+
const bytes = input.data instanceof Uint8Array ? input.data : new Uint8Array(input.data);
|
|
27
25
|
if (bytes.length >= 4 &&
|
|
28
26
|
bytes[0] === 0x25 &&
|
|
29
27
|
bytes[1] === 0x50 &&
|
|
@@ -49,18 +47,40 @@ export class PdfConverter {
|
|
|
49
47
|
const startTime = Date.now();
|
|
50
48
|
const data = await this.toByteArray(input.data);
|
|
51
49
|
const renderer = new MarkdownRenderer();
|
|
50
|
+
const opts = input.options ?? {};
|
|
51
|
+
const ratios = opts.headingSizeRatios ?? {};
|
|
52
52
|
let ast;
|
|
53
53
|
let markdown;
|
|
54
54
|
const headings = [];
|
|
55
55
|
const tables = [];
|
|
56
56
|
let pageCount = 0;
|
|
57
|
+
let pdfTitle = "";
|
|
58
|
+
let pdfAuthor = "";
|
|
57
59
|
try {
|
|
58
|
-
|
|
59
|
-
//
|
|
60
|
-
|
|
61
|
-
const
|
|
60
|
+
// Use the legacy entry — runs under Node.js without a separate worker file.
|
|
61
|
+
// The non-legacy entry pulls in a Worker constructor reference that crashes
|
|
62
|
+
// pdfjs's fake-worker bootstrap when Worker is undefined.
|
|
63
|
+
const pdfjsLib = await import("pdfjs-dist/legacy/build/pdf.mjs");
|
|
64
|
+
const loadingTask = pdfjsLib.getDocument({
|
|
65
|
+
data,
|
|
66
|
+
// pdfjs runs `eval()` in some font/cmap paths; disable for safety + Node compat.
|
|
67
|
+
isEvalSupported: false,
|
|
68
|
+
// Inline the worker when no Worker global exists (Node.js).
|
|
69
|
+
// Not in DocumentInitParameters types but accepted at runtime by the legacy build.
|
|
70
|
+
...(typeof Worker === "undefined" ? { disableWorker: true } : {}),
|
|
71
|
+
});
|
|
62
72
|
const pdf = await loadingTask.promise;
|
|
63
73
|
pageCount = pdf.numPages;
|
|
74
|
+
// Extract PDF document metadata when available.
|
|
75
|
+
try {
|
|
76
|
+
const meta = await pdf.getMetadata();
|
|
77
|
+
const info = (meta?.info ?? {});
|
|
78
|
+
pdfTitle = typeof info.Title === "string" ? info.Title : "";
|
|
79
|
+
pdfAuthor = typeof info.Author === "string" ? info.Author : "";
|
|
80
|
+
}
|
|
81
|
+
catch {
|
|
82
|
+
// Metadata is best-effort.
|
|
83
|
+
}
|
|
64
84
|
const pageNodes = [];
|
|
65
85
|
for (let pageNum = 1; pageNum <= pdf.numPages; pageNum++) {
|
|
66
86
|
const page = await pdf.getPage(pageNum);
|
|
@@ -104,24 +124,20 @@ export class PdfConverter {
|
|
|
104
124
|
const trimmed = currentLine.trim();
|
|
105
125
|
if (!trimmed)
|
|
106
126
|
return;
|
|
107
|
-
const heading = this.classifyHeading(trimmed, lastFontSize, bodyFontSize);
|
|
127
|
+
const heading = this.classifyHeading(trimmed, lastFontSize, bodyFontSize, ratios);
|
|
108
128
|
if (heading) {
|
|
109
129
|
headings.push(heading);
|
|
110
130
|
const level = heading.level;
|
|
111
131
|
pageNodes.push(createNode({
|
|
112
132
|
type: "heading",
|
|
113
133
|
level,
|
|
114
|
-
children: [
|
|
115
|
-
createNode({ type: "text", value: heading.text }),
|
|
116
|
-
],
|
|
134
|
+
children: [createNode({ type: "text", value: heading.text })],
|
|
117
135
|
}));
|
|
118
136
|
}
|
|
119
137
|
else {
|
|
120
138
|
pageNodes.push(createNode({
|
|
121
139
|
type: "paragraph",
|
|
122
|
-
children: [
|
|
123
|
-
createNode({ type: "text", value: trimmed }),
|
|
124
|
-
],
|
|
140
|
+
children: [createNode({ type: "text", value: trimmed })],
|
|
125
141
|
}));
|
|
126
142
|
}
|
|
127
143
|
currentLine = "";
|
|
@@ -158,6 +174,21 @@ export class PdfConverter {
|
|
|
158
174
|
children: pageNodes,
|
|
159
175
|
});
|
|
160
176
|
markdown = renderer.render(ast);
|
|
177
|
+
// Apply post-render options.
|
|
178
|
+
if (opts.pageBreakMarker) {
|
|
179
|
+
markdown = markdown.replace(/\n\n---\n\n/g, `\n\n${opts.pageBreakMarker}\n\n`);
|
|
180
|
+
}
|
|
181
|
+
if (opts.injectFrontmatter) {
|
|
182
|
+
const fm = ["---"];
|
|
183
|
+
const yamlEscape = (s) => s.replace(/\\/g, "\\\\").replace(/"/g, '\\"');
|
|
184
|
+
if (pdfTitle)
|
|
185
|
+
fm.push(`title: "${yamlEscape(pdfTitle)}"`);
|
|
186
|
+
if (pdfAuthor)
|
|
187
|
+
fm.push(`author: "${yamlEscape(pdfAuthor)}"`);
|
|
188
|
+
fm.push(`pages: ${pageCount}`);
|
|
189
|
+
fm.push("---");
|
|
190
|
+
markdown = fm.join("\n") + "\n\n" + markdown;
|
|
191
|
+
}
|
|
161
192
|
}
|
|
162
193
|
catch {
|
|
163
194
|
ast = createNode({
|
|
@@ -181,6 +212,8 @@ export class PdfConverter {
|
|
|
181
212
|
markdown,
|
|
182
213
|
metadata: {
|
|
183
214
|
pageCount,
|
|
215
|
+
...(pdfTitle && { title: pdfTitle }),
|
|
216
|
+
...(pdfAuthor && { author: pdfAuthor }),
|
|
184
217
|
format: "application/pdf",
|
|
185
218
|
source: "pdfjs-dist",
|
|
186
219
|
},
|
|
@@ -226,18 +259,18 @@ export class PdfConverter {
|
|
|
226
259
|
* @param bodyFontSize - The most common (body) font size in the document.
|
|
227
260
|
* @returns A HeadingInfo if the line qualifies as a heading, otherwise undefined.
|
|
228
261
|
*/
|
|
229
|
-
classifyHeading(text, fontSize, bodyFontSize) {
|
|
262
|
+
classifyHeading(text, fontSize, bodyFontSize, ratios = {}) {
|
|
230
263
|
const trimmed = text.trim();
|
|
231
264
|
if (!trimmed)
|
|
232
265
|
return undefined;
|
|
233
266
|
// Use relative sizing when a body font size is known.
|
|
234
267
|
if (bodyFontSize > 0) {
|
|
235
268
|
const ratio = fontSize / bodyFontSize;
|
|
236
|
-
if (ratio > 1.8)
|
|
269
|
+
if (ratio > (ratios.h1 ?? 1.8))
|
|
237
270
|
return { level: 1, text: trimmed };
|
|
238
|
-
if (ratio > 1.4)
|
|
271
|
+
if (ratio > (ratios.h2 ?? 1.4))
|
|
239
272
|
return { level: 2, text: trimmed };
|
|
240
|
-
if (ratio > 1.15)
|
|
273
|
+
if (ratio > (ratios.h3 ?? 1.15))
|
|
241
274
|
return { level: 3, text: trimmed };
|
|
242
275
|
return undefined;
|
|
243
276
|
}
|
|
@@ -256,12 +289,19 @@ export class PdfConverter {
|
|
|
256
289
|
* @returns The data as a Uint8Array.
|
|
257
290
|
*/
|
|
258
291
|
async toByteArray(data) {
|
|
259
|
-
if (data instanceof Uint8Array)
|
|
260
|
-
|
|
292
|
+
if (data instanceof Uint8Array) {
|
|
293
|
+
// Node.js Buffer is a Uint8Array subclass, but pdfjs-dist v4+ rejects it with
|
|
294
|
+
// "Please provide binary data as Uint8Array, rather than Buffer." Re-wrap so
|
|
295
|
+
// the constructor is exactly Uint8Array.
|
|
296
|
+
return data.constructor === Uint8Array
|
|
297
|
+
? data
|
|
298
|
+
: new Uint8Array(data.buffer, data.byteOffset, data.byteLength);
|
|
299
|
+
}
|
|
261
300
|
if (data instanceof ArrayBuffer)
|
|
262
301
|
return new Uint8Array(data);
|
|
263
|
-
if (data instanceof Blob)
|
|
302
|
+
if (typeof Blob !== "undefined" && data instanceof Blob) {
|
|
264
303
|
return new Uint8Array(await data.arrayBuffer());
|
|
304
|
+
}
|
|
265
305
|
return new TextEncoder().encode(data);
|
|
266
306
|
}
|
|
267
307
|
}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"pdf-converter.js","sourceRoot":"","sources":["../src/pdf-converter.ts"],"names":[],"mappings":"AAQA,OAAO,EACL,UAAU,GAMX,MAAM,sBAAsB,CAAC;AAC9B,OAAO,EAAE,gBAAgB,EAAE,MAAM,mBAAmB,CAAC;AAErD,8DAA8D;AAC9D,MAAM,OAAO,YAAY;IACd,EAAE,GAAG,KAAc,CAAC;IACpB,kBAAkB,GAAG,CAAC,iBAAiB,CAAC,CAAC;IACzC,mBAAmB,GAAG,CAAC,MAAM,CAAC,CAAC;IAExC;;;;OAIG;IACH,KAAK,CAAC,UAAU,CAAC,KAAsB;QACrC,IAAI,KAAK,CAAC,QAAQ,IAAI,IAAI,CAAC,kBAAkB,CAAC,QAAQ,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC;YACvE,OAAO,IAAI,CAAC;QACd,CAAC;QACD,IAAI,KAAK,CAAC,QAAQ,EAAE,CAAC;YACnB,MAAM,GAAG,GAAG,KAAK,CAAC,QAAQ,CAAC,WAAW,EAAE,CAAC;YACzC,IAAI,IAAI,CAAC,mBAAmB,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;gBAC1D,OAAO,IAAI,CAAC;YACd,CAAC;QACH,CAAC;QACD,IAAI,KAAK,CAAC,IAAI,YAAY,UAAU,IAAI,KAAK,CAAC,IAAI,YAAY,WAAW,EAAE,CAAC;YAC1E,MAAM,KAAK,
|
|
1
|
+
{"version":3,"file":"pdf-converter.js","sourceRoot":"","sources":["../src/pdf-converter.ts"],"names":[],"mappings":"AAQA,OAAO,EACL,UAAU,GAMX,MAAM,sBAAsB,CAAC;AAC9B,OAAO,EAAE,gBAAgB,EAAE,MAAM,mBAAmB,CAAC;AAErD,8DAA8D;AAC9D,MAAM,OAAO,YAAY;IACd,EAAE,GAAG,KAAc,CAAC;IACpB,kBAAkB,GAAG,CAAC,iBAAiB,CAAC,CAAC;IACzC,mBAAmB,GAAG,CAAC,MAAM,CAAC,CAAC;IAExC;;;;OAIG;IACH,KAAK,CAAC,UAAU,CAAC,KAAsB;QACrC,IAAI,KAAK,CAAC,QAAQ,IAAI,IAAI,CAAC,kBAAkB,CAAC,QAAQ,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC;YACvE,OAAO,IAAI,CAAC;QACd,CAAC;QACD,IAAI,KAAK,CAAC,QAAQ,EAAE,CAAC;YACnB,MAAM,GAAG,GAAG,KAAK,CAAC,QAAQ,CAAC,WAAW,EAAE,CAAC;YACzC,IAAI,IAAI,CAAC,mBAAmB,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;gBAC1D,OAAO,IAAI,CAAC;YACd,CAAC;QACH,CAAC;QACD,IAAI,KAAK,CAAC,IAAI,YAAY,UAAU,IAAI,KAAK,CAAC,IAAI,YAAY,WAAW,EAAE,CAAC;YAC1E,MAAM,KAAK,GAAG,KAAK,CAAC,IAAI,YAAY,UAAU,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,UAAU,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;YACzF,IACE,KAAK,CAAC,MAAM,IAAI,CAAC;gBACjB,KAAK,CAAC,CAAC,CAAC,KAAK,IAAI;gBACjB,KAAK,CAAC,CAAC,CAAC,KAAK,IAAI;gBACjB,KAAK,CAAC,CAAC,CAAC,KAAK,IAAI;gBACjB,KAAK,CAAC,CAAC,CAAC,KAAK,IAAI,EACjB,CAAC;gBACD,OAAO,IAAI,CAAC;YACd,CAAC;QACH,CAAC;QACD,OAAO,KAAK,CAAC;IACf,CAAC;IAED;;;;;;;;;;OAUG;IACH,KAAK,CAAC,OAAO,CAAC,KAAsB;QAClC,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAC7B,MAAM,IAAI,GAAG,MAAM,IAAI,CAAC,WAAW,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QAChD,MAAM,QAAQ,GAAG,IAAI,gBAAgB,EAAE,CAAC;QACxC,MAAM,IAAI,GAAG,KAAK,CAAC,OAAO,IAAI,EAAE,CAAC;QACjC,MAAM,MAAM,GAAG,IAAI,CAAC,iBAAiB,IAAI,EAAE,CAAC;QAE5C,IAAI,GAAiB,CAAC;QACtB,IAAI,QAAgB,CAAC;QACrB,MAAM,QAAQ,GAAkB,EAAE,CAAC;QACnC,MAAM,MAAM,GAAgB,EAAE,CAAC;QAC/B,IAAI,SAAS,GAAG,CAAC,CAAC;QAClB,IAAI,QAAQ,GAAG,EAAE,CAAC;QAClB,IAAI,SAAS,GAAG,EAAE,CAAC;QAEnB,IAAI,CAAC;YACH,4EAA4E;YAC5E,4EAA4E;YAC5E,0DAA0D;YAC1D,MAAM,QAAQ,GAAG,MAAM,MAAM,CAAC,iCAAiC,CAAC,CAAC;YAEjE,MAAM,WAAW,GAAG,QAAQ,CAAC,WAAW,CAAC;gBACvC,IAAI;gBACJ,iFAAiF;gBACjF,eAAe,EAAE,KAAK;gBACtB,4DAA4D;gBAC5D,mFAAmF;gBACnF,GAAG,CAAC,OAAO,MAAM,KAAK,WAAW,CAAC,CAAC,CAAC,EAAE,aAAa,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;aACpB,CAAC,CAAC;YACjD,MAAM,GAAG,GAAG,MAAM,WAAW,CAAC,OAAO,CAAC;YACtC,SAAS,GAAG,GAAG,CAAC,QAAQ,CAAC;YAEzB,gDAAgD;YAChD,IAAI,CAAC;gBACH,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,WAAW,EAAE,CAAC;gBAErC,MAAM,IAAI,GAAG,CAAC,IAAI,EAAE,IAAI,IAAI,EAAE,CAA4B,CAAC;gBAC3D,QAAQ,GAAG,OAAO,IAAI,CAAC,KAAK,KAAK,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,CAAC;gBAC5D,SAAS,GAAG,OAAO,IAAI,CAAC,MAAM,KAAK,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC;YACjE,CAAC;YAAC,MAAM,CAAC;gBACP,2BAA2B;YAC7B,CAAC;YAED,MAAM,SAAS,GAAc,EAAE,CAAC;YAEhC,KAAK,IAAI,OAAO,GAAG,CAAC,EAAE,OAAO,IAAI,GAAG,CAAC,QAAQ,EAAE,OAAO,EAAE,EAAE,CAAC;gBACzD,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC;gBACxC,MAAM,WAAW,GAAG,MAAM,IAAI,CAAC,cAAc,EAAE,CAAC;gBAChD,MAAM,WAAW,GAAG,MAAM,IAAI,CAAC,cAAc,EAAE,CAAC;gBAEhD,iDAAiD;gBACjD,MAAM,SAAS,GAAG,IAAI,GAAG,EAAkB,CAAC;gBAC5C,KAAK,MAAM,IAAI,IAAI,WAAW,CAAC,KAAK,EAAE,CAAC;oBACrC,IAAI,CAAC,CAAC,KAAK,IAAI,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG;wBAAE,SAAS;oBAC5C,MAAM,QAAQ,GAAG,IAAI,CAAC,eAAe,CAAC,IAAI,CAAC,CAAC;oBAC5C,IAAI,QAAQ,GAAG,CAAC,EAAE,CAAC;wBACjB,SAAS,CAAC,GAAG,CAAC,QAAQ,EAAE,CAAC,SAAS,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;oBAC9D,CAAC;gBACH,CAAC;gBAED,8CAA8C;gBAC9C,IAAI,YAAY,GAAG,CAAC,CAAC;gBACrB,IAAI,QAAQ,GAAG,CAAC,CAAC;gBACjB,KAAK,MAAM,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,SAAS,EAAE,CAAC;oBACtC,IAAI,KAAK,GAAG,QAAQ,EAAE,CAAC;wBACrB,QAAQ,GAAG,KAAK,CAAC;wBACjB,YAAY,GAAG,IAAI,CAAC;oBACtB,CAAC;gBACH,CAAC;gBAED,qCAAqC;gBACrC,MAAM,OAAO,GAAG,IAAI,GAAG,EAAkB,CAAC;gBAC1C,KAAK,MAAM,KAAK,IAAI,WAAW,EAAE,CAAC;oBAChC,IAAI,KAAK,CAAC,OAAO,KAAK,MAAM,IAAI,KAAK,CAAC,GAAG,EAAE,CAAC;wBAC1C,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC;wBACxB,gDAAgD;wBAChD,MAAM,GAAG,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;wBAC1G,OAAO,CAAC,GAAG,CAAC,GAAG,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC;oBAC9B,CAAC;gBACH,CAAC;gBAED,oDAAoD;gBACpD,MAAM,KAAK,GAAG,WAAW,CAAC,KAAK,CAAC;gBAChC,IAAI,WAAW,GAAG,EAAE,CAAC;gBACrB,IAAI,YAAY,GAAG,CAAC,CAAC;gBACrB,IAAI,KAAK,GAAG,CAAC,CAAC;gBAEd,MAAM,SAAS,GAAG,GAAG,EAAE;oBACrB,MAAM,OAAO,GAAG,WAAW,CAAC,IAAI,EAAE,CAAC;oBACnC,IAAI,CAAC,OAAO;wBAAE,OAAO;oBAErB,MAAM,OAAO,GAAG,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,YAAY,EAAE,YAAY,EAAE,MAAM,CAAC,CAAC;oBAClF,IAAI,OAAO,EAAE,CAAC;wBACZ,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;wBACvB,MAAM,KAAK,GAAG,OAAO,CAAC,KAA8B,CAAC;wBACrD,SAAS,CAAC,IAAI,CACZ,UAAU,CAAc;4BACtB,IAAI,EAAE,SAAS;4BACf,KAAK;4BACL,QAAQ,EAAE,CAAC,UAAU,CAAW,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC;yBACxE,CAAC,CACH,CAAC;oBACJ,CAAC;yBAAM,CAAC;wBACN,SAAS,CAAC,IAAI,CACZ,UAAU,CAAgB;4BACxB,IAAI,EAAE,WAAW;4BACjB,QAAQ,EAAE,CAAC,UAAU,CAAW,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,OAAO,EAAE,CAAC,CAAC;yBACnE,CAAC,CACH,CAAC;oBACJ,CAAC;oBACD,WAAW,GAAG,EAAE,CAAC;gBACnB,CAAC,CAAC;gBAEF,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;oBACzB,IAAI,CAAC,CAAC,KAAK,IAAI,IAAI,CAAC;wBAAE,SAAS;oBAC/B,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,CAAC;oBACrB,IAAI,CAAC,GAAG;wBAAE,SAAS;oBAEnB,MAAM,QAAQ,GAAG,IAAI,CAAC,eAAe,CAAC,IAAI,CAAC,CAAC;oBAC5C,MAAM,CAAC,GAAG,WAAW,IAAI,IAAI,IAAI,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;oBAExE,uEAAuE;oBACvE,IACE,WAAW;wBACX,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,KAAK,CAAC,GAAG,QAAQ,GAAG,GAAG;4BACnC,CAAC,QAAQ,KAAK,YAAY,IAAI,YAAY,GAAG,CAAC,CAAC,CAAC,EAClD,CAAC;wBACD,SAAS,EAAE,CAAC;oBACd,CAAC;oBAED,WAAW,IAAI,GAAG,CAAC;oBACnB,YAAY,GAAG,QAAQ,CAAC;oBACxB,KAAK,GAAG,CAAC,CAAC;gBACZ,CAAC;gBAED,SAAS,EAAE,CAAC;gBAEZ,0DAA0D;gBAC1D,IAAI,OAAO,GAAG,GAAG,CAAC,QAAQ,EAAE,CAAC;oBAC3B,SAAS,CAAC,IAAI,CACZ,UAAU,CAAgB;wBACxB,IAAI,EAAE,YAAY;wBAClB,UAAU,EAAE,OAAO;qBACpB,CAAC,CACH,CAAC;gBACJ,CAAC;YACH,CAAC;YAED,GAAG,GAAG,UAAU,CAAe;gBAC7B,IAAI,EAAE,UAAU;gBAChB,QAAQ,EAAE,SAAS;aACpB,CAAC,CAAC;YAEH,QAAQ,GAAG,QAAQ,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;YAEhC,6BAA6B;YAC7B,IAAI,IAAI,CAAC,eAAe,EAAE,CAAC;gBACzB,QAAQ,GAAG,QAAQ,CAAC,OAAO,CAAC,cAAc,EAAE,OAAO,IAAI,CAAC,eAAe,MAAM,CAAC,CAAC;YACjF,CAAC;YACD,IAAI,IAAI,CAAC,iBAAiB,EAAE,CAAC;gBAC3B,MAAM,EAAE,GAAa,CAAC,KAAK,CAAC,CAAC;gBAC7B,MAAM,UAAU,GAAG,CAAC,CAAS,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;gBAChF,IAAI,QAAQ;oBAAE,EAAE,CAAC,IAAI,CAAC,WAAW,UAAU,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC;gBAC1D,IAAI,SAAS;oBAAE,EAAE,CAAC,IAAI,CAAC,YAAY,UAAU,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC;gBAC7D,EAAE,CAAC,IAAI,CAAC,UAAU,SAAS,EAAE,CAAC,CAAC;gBAC/B,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;gBACf,QAAQ,GAAG,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,MAAM,GAAG,QAAQ,CAAC;YAC/C,CAAC;QACH,CAAC;QAAC,MAAM,CAAC;YACP,GAAG,GAAG,UAAU,CAAe;gBAC7B,IAAI,EAAE,UAAU;gBAChB,QAAQ,EAAE;oBACR,UAAU,CAAgB;wBACxB,IAAI,EAAE,WAAW;wBACjB,QAAQ,EAAE;4BACR,UAAU,CAAW;gCACnB,IAAI,EAAE,MAAM;gCACZ,KAAK,EAAE,sCAAsC;6BAC9C,CAAC;yBACH;qBACF,CAAC;iBACH;aACF,CAAC,CAAC;YACH,QAAQ,GAAG,QAAQ,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAClC,CAAC;QAED,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAE3B,OAAO;YACL,QAAQ;YACR,QAAQ,EAAE;gBACR,SAAS;gBACT,GAAG,CAAC,QAAQ,IAAI,EAAE,KAAK,EAAE,QAAQ,EAAE,CAAC;gBACpC,GAAG,CAAC,SAAS,IAAI,EAAE,MAAM,EAAE,SAAS,EAAE,CAAC;gBACvC,MAAM,EAAE,iBAAiB;gBACzB,MAAM,EAAE,YAAY;aACrB;YACD,MAAM,EAAE,EAAE;YACV,MAAM;YACN,MAAM,EAAE,EAAE;YACV,QAAQ;YACR,GAAG;YACH,MAAM,EAAE,UAAU;YAClB,WAAW,EAAE,IAAI,CAAC,EAAE;YACpB,KAAK,EAAE;gBACL,SAAS;gBACT,OAAO;gBACP,QAAQ,EAAE,OAAO,GAAG,SAAS;gBAC7B,SAAS,EAAE,IAAI,CAAC,MAAM;gBACtB,UAAU,EAAE,QAAQ,CAAC,MAAM;gBAC3B,cAAc,EAAE,SAAS;aAC1B;SACF,CAAC;IACJ,CAAC;IAED;;;;;;;;OAQG;IACK,eAAe,CAAC,IAA6B;QACnD,MAAM,SAAS,GAAG,IAAI,CAAC,SAAiC,CAAC;QACzD,IAAI,CAAC,SAAS,IAAI,SAAS,CAAC,MAAM,GAAG,CAAC;YAAE,OAAO,CAAC,CAAC;QACjD,MAAM,GAAG,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC;QACzB,OAAO,GAAG,KAAK,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC/C,CAAC;IAED;;;;;;;;OAQG;IACK,eAAe,CACrB,IAAY,EACZ,QAAgB,EAChB,YAAoB,EACpB,SAAoD,EAAE;QAEtD,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;QAC5B,IAAI,CAAC,OAAO;YAAE,OAAO,SAAS,CAAC;QAE/B,sDAAsD;QACtD,IAAI,YAAY,GAAG,CAAC,EAAE,CAAC;YACrB,MAAM,KAAK,GAAG,QAAQ,GAAG,YAAY,CAAC;YACtC,IAAI,KAAK,GAAG,CAAC,MAAM,CAAC,EAAE,IAAI,GAAG,CAAC;gBAAE,OAAO,EAAE,KAAK,EAAE,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC;YACnE,IAAI,KAAK,GAAG,CAAC,MAAM,CAAC,EAAE,IAAI,GAAG,CAAC;gBAAE,OAAO,EAAE,KAAK,EAAE,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC;YACnE,IAAI,KAAK,GAAG,CAAC,MAAM,CAAC,EAAE,IAAI,IAAI,CAAC;gBAAE,OAAO,EAAE,KAAK,EAAE,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC;YACpE,OAAO,SAAS,CAAC;QACnB,CAAC;QAED,mCAAmC;QACnC,IAAI,QAAQ,GAAG,EAAE;YAAE,OAAO,EAAE,KAAK,EAAE,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC;QACtD,IAAI,QAAQ,GAAG,EAAE;YAAE,OAAO,EAAE,KAAK,EAAE,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC;QACtD,IAAI,QAAQ,GAAG,EAAE;YAAE,OAAO,EAAE,KAAK,EAAE,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC;QACtD,OAAO,SAAS,CAAC;IACnB,CAAC;IAED;;;;OAIG;IACK,KAAK,CAAC,WAAW,CAAC,IAA8C;QACtE,IAAI,IAAI,YAAY,UAAU,EAAE,CAAC;YAC/B,8EAA8E;YAC9E,6EAA6E;YAC7E,yCAAyC;YACzC,OAAO,IAAI,CAAC,WAAW,KAAK,UAAU;gBACpC,CAAC,CAAC,IAAI;gBACN,CAAC,CAAC,IAAI,UAAU,CAAC,IAAI,CAAC,MAAM,EAAE,IAAI,CAAC,UAAU,EAAE,IAAI,CAAC,UAAU,CAAC,CAAC;QACpE,CAAC;QACD,IAAI,IAAI,YAAY,WAAW;YAAE,OAAO,IAAI,UAAU,CAAC,IAAI,CAAC,CAAC;QAC7D,IAAI,OAAO,IAAI,KAAK,WAAW,IAAI,IAAI,YAAY,IAAI,EAAE,CAAC;YACxD,OAAO,IAAI,UAAU,CAAC,MAAM,IAAI,CAAC,WAAW,EAAE,CAAC,CAAC;QAClD,CAAC;QACD,OAAO,IAAI,WAAW,EAAE,CAAC,MAAM,CAAC,IAAc,CAAC,CAAC;IAClD,CAAC;CACF"}
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@markitdownjs/pdf",
|
|
3
|
-
"version": "0.
|
|
3
|
+
"version": "0.2.0",
|
|
4
4
|
"description": "PDF converter for MarkItDownJS",
|
|
5
5
|
"license": "MIT",
|
|
6
6
|
"type": "module",
|
|
@@ -18,8 +18,8 @@
|
|
|
18
18
|
],
|
|
19
19
|
"dependencies": {
|
|
20
20
|
"pdfjs-dist": "^4.0.0",
|
|
21
|
-
"@markitdownjs/
|
|
22
|
-
"@markitdownjs/
|
|
21
|
+
"@markitdownjs/shared": "0.2.0",
|
|
22
|
+
"@markitdownjs/ast": "0.1.2"
|
|
23
23
|
},
|
|
24
24
|
"devDependencies": {
|
|
25
25
|
"typescript": "^5.5.0"
|