/src/gdal/frmts/pdf/pdfobject.cpp
Line | Count | Source (jump to first uncovered line) |
1 | | /****************************************************************************** |
2 | | * |
3 | | * Project: PDF driver |
4 | | * Purpose: GDALDataset driver for PDF dataset. |
5 | | * Author: Even Rouault, <even dot rouault at spatialys.com> |
6 | | * |
7 | | ****************************************************************************** |
8 | | * |
9 | | * Support for open-source PDFium library |
10 | | * |
11 | | * Copyright (C) 2015 Klokan Technologies GmbH (http://www.klokantech.com/) |
12 | | * Author: Martin Mikita <martin.mikita@klokantech.com>, xmikit00 @ FIT VUT Brno |
13 | | * |
14 | | ****************************************************************************** |
15 | | * Copyright (c) 2011-2013, Even Rouault <even dot rouault at spatialys.com> |
16 | | * |
17 | | * SPDX-License-Identifier: MIT |
18 | | ****************************************************************************/ |
19 | | |
20 | | #include "gdal_pdf.h" |
21 | | |
22 | | #include <limits> |
23 | | #include <vector> |
24 | | #include "pdfobject.h" |
25 | | |
26 | | /************************************************************************/ |
27 | | /* ROUND_IF_CLOSE() */ |
28 | | /************************************************************************/ |
29 | | |
30 | | double ROUND_IF_CLOSE(double x, double eps) |
31 | 14.4k | { |
32 | 14.4k | if (eps == 0.0) |
33 | 11.7k | eps = fabs(x) < 1 ? 1e-10 : 1e-8; |
34 | 14.4k | const double dfRounded = std::round(x); |
35 | 14.4k | if (fabs(x - dfRounded) < eps) |
36 | 10.8k | return dfRounded; |
37 | 3.61k | else |
38 | 3.61k | return x; |
39 | 14.4k | } |
40 | | |
41 | | /************************************************************************/ |
42 | | /* GDALPDFGetPDFString() */ |
43 | | /************************************************************************/ |
44 | | |
45 | | static CPLString GDALPDFGetPDFString(const char *pszStr) |
46 | 656 | { |
47 | 656 | const GByte *pabyData = reinterpret_cast<const GByte *>(pszStr); |
48 | 656 | GByte ch; |
49 | 86.8k | for (size_t i = 0; (ch = pabyData[i]) != '\0'; i++) |
50 | 86.3k | { |
51 | 86.3k | if (ch < 32 || ch > 127 || ch == '(' || ch == ')' || ch == '\\' || |
52 | 86.3k | ch == '%' || ch == '#') |
53 | 155 | break; |
54 | 86.3k | } |
55 | 656 | CPLString osStr; |
56 | 656 | if (ch == 0) |
57 | 501 | { |
58 | 501 | osStr = "("; |
59 | 501 | osStr += pszStr; |
60 | 501 | osStr += ")"; |
61 | 501 | return osStr; |
62 | 501 | } |
63 | | |
64 | 155 | wchar_t *pwszDest = CPLRecodeToWChar(pszStr, CPL_ENC_UTF8, CPL_ENC_UCS2); |
65 | 155 | osStr = "<FEFF"; |
66 | 260k | for (size_t i = 0; pwszDest[i] != 0; i++) |
67 | 260k | { |
68 | 260k | #ifndef _WIN32 |
69 | 260k | if (pwszDest[i] >= 0x10000 /* && pwszDest[i] <= 0x10FFFF */) |
70 | 0 | { |
71 | | /* Generate UTF-16 surrogate pairs (on Windows, CPLRecodeToWChar |
72 | | * does it for us) */ |
73 | 0 | int nHeadSurrogate = ((pwszDest[i] - 0x10000) >> 10) | 0xd800; |
74 | 0 | int nTrailSurrogate = ((pwszDest[i] - 0x10000) & 0x3ff) | 0xdc00; |
75 | 0 | osStr += CPLSPrintf("%02X", (nHeadSurrogate >> 8) & 0xff); |
76 | 0 | osStr += CPLSPrintf("%02X", (nHeadSurrogate)&0xff); |
77 | 0 | osStr += CPLSPrintf("%02X", (nTrailSurrogate >> 8) & 0xff); |
78 | 0 | osStr += CPLSPrintf("%02X", (nTrailSurrogate)&0xff); |
79 | 0 | } |
80 | 260k | else |
81 | 260k | #endif |
82 | 260k | { |
83 | 260k | osStr += |
84 | 260k | CPLSPrintf("%02X", static_cast<int>(pwszDest[i] >> 8) & 0xff); |
85 | 260k | osStr += CPLSPrintf("%02X", static_cast<int>(pwszDest[i]) & 0xff); |
86 | 260k | } |
87 | 260k | } |
88 | 155 | osStr += ">"; |
89 | 155 | CPLFree(pwszDest); |
90 | 155 | return osStr; |
91 | 656 | } |
92 | | |
93 | | #if defined(HAVE_POPPLER) || defined(HAVE_PDFIUM) |
94 | | |
95 | | /************************************************************************/ |
96 | | /* GDALPDFGetUTF8StringFromBytes() */ |
97 | | /************************************************************************/ |
98 | | |
99 | | static std::string GDALPDFGetUTF8StringFromBytes(const GByte *pabySrc, |
100 | | size_t nLen) |
101 | 384 | { |
102 | 384 | const bool bLEUnicodeMarker = |
103 | 384 | nLen >= 2 && pabySrc[0] == 0xFE && pabySrc[1] == 0xFF; |
104 | 384 | const bool bBEUnicodeMarker = |
105 | 384 | nLen >= 2 && pabySrc[0] == 0xFF && pabySrc[1] == 0xFE; |
106 | 384 | if (!bLEUnicodeMarker && !bBEUnicodeMarker) |
107 | 0 | { |
108 | 0 | std::string osStr; |
109 | 0 | try |
110 | 0 | { |
111 | 0 | osStr.reserve(nLen); |
112 | 0 | } |
113 | 0 | catch (const std::exception &e) |
114 | 0 | { |
115 | 0 | CPLError(CE_Failure, CPLE_OutOfMemory, |
116 | 0 | "GDALPDFGetUTF8StringFromBytes(): %s", e.what()); |
117 | 0 | return osStr; |
118 | 0 | } |
119 | 0 | osStr.assign(reinterpret_cast<const char *>(pabySrc), nLen); |
120 | 0 | const char *pszStr = osStr.c_str(); |
121 | 0 | if (CPLIsUTF8(pszStr, -1)) |
122 | 0 | return osStr; |
123 | 0 | else |
124 | 0 | { |
125 | 0 | char *pszUTF8 = CPLRecode(pszStr, CPL_ENC_ISO8859_1, CPL_ENC_UTF8); |
126 | 0 | std::string osRet = pszUTF8; |
127 | 0 | CPLFree(pszUTF8); |
128 | 0 | return osRet; |
129 | 0 | } |
130 | 0 | } |
131 | | |
132 | | /* This is UTF-16 content */ |
133 | 384 | pabySrc += 2; |
134 | 384 | nLen = (nLen - 2) / 2; |
135 | 384 | std::wstring awszSource; |
136 | 384 | awszSource.resize(nLen + 1); |
137 | 384 | size_t j = 0; |
138 | 704k | for (size_t i = 0; i < nLen; i++, j++) |
139 | 703k | { |
140 | 703k | if (!bBEUnicodeMarker) |
141 | 703k | awszSource[j] = (pabySrc[2 * i] << 8) + pabySrc[2 * i + 1]; |
142 | 0 | else |
143 | 0 | awszSource[j] = (pabySrc[2 * i + 1] << 8) + pabySrc[2 * i]; |
144 | 703k | #ifndef _WIN32 |
145 | | /* Is there a surrogate pair ? See http://en.wikipedia.org/wiki/UTF-16 |
146 | | */ |
147 | | /* On Windows, CPLRecodeFromWChar does this for us, because wchar_t is |
148 | | * only */ |
149 | | /* 2 bytes wide, whereas on Unix it is 32bits */ |
150 | 703k | if (awszSource[j] >= 0xD800 && awszSource[j] <= 0xDBFF && i + 1 < nLen) |
151 | 0 | { |
152 | | /* should be in the range 0xDC00... 0xDFFF */ |
153 | 0 | wchar_t nTrailSurrogate; |
154 | 0 | if (!bBEUnicodeMarker) |
155 | 0 | nTrailSurrogate = |
156 | 0 | (pabySrc[2 * (i + 1)] << 8) + pabySrc[2 * (i + 1) + 1]; |
157 | 0 | else |
158 | 0 | nTrailSurrogate = |
159 | 0 | (pabySrc[2 * (i + 1) + 1] << 8) + pabySrc[2 * (i + 1)]; |
160 | 0 | if (nTrailSurrogate >= 0xDC00 && nTrailSurrogate <= 0xDFFF) |
161 | 0 | { |
162 | 0 | awszSource[j] = ((awszSource[j] - 0xD800) << 10) + |
163 | 0 | (nTrailSurrogate - 0xDC00) + 0x10000; |
164 | 0 | i++; |
165 | 0 | } |
166 | 0 | } |
167 | 703k | #endif |
168 | 703k | } |
169 | 384 | awszSource[j] = 0; |
170 | | |
171 | 384 | char *pszUTF8 = |
172 | 384 | CPLRecodeFromWChar(awszSource.data(), CPL_ENC_UCS2, CPL_ENC_UTF8); |
173 | 384 | awszSource.clear(); |
174 | 384 | std::string osStrUTF8(pszUTF8); |
175 | 384 | CPLFree(pszUTF8); |
176 | 384 | return osStrUTF8; |
177 | 384 | } |
178 | | |
179 | | #endif // defined(HAVE_POPPLER) || defined(HAVE_PDFIUM) |
180 | | |
181 | | /************************************************************************/ |
182 | | /* GDALPDFGetPDFName() */ |
183 | | /************************************************************************/ |
184 | | |
185 | | static std::string GDALPDFGetPDFName(const std::string &osStr) |
186 | 7.52k | { |
187 | 7.52k | std::string osRet; |
188 | 7.52k | for (const char ch : osStr) |
189 | 53.5k | { |
190 | 53.5k | if (!((ch >= 'a' && ch <= 'z') || (ch >= 'A' && ch <= 'Z') || |
191 | 53.5k | (ch >= '0' && ch <= '9') || ch == '-')) |
192 | 0 | osRet += '_'; |
193 | 53.5k | else |
194 | 53.5k | osRet += ch; |
195 | 53.5k | } |
196 | 7.52k | return osRet; |
197 | 7.52k | } |
198 | | |
199 | | /************************************************************************/ |
200 | | /* ==================================================================== */ |
201 | | /* GDALPDFObject */ |
202 | | /* ==================================================================== */ |
203 | | /************************************************************************/ |
204 | | |
205 | | /************************************************************************/ |
206 | | /* ~GDALPDFObject() */ |
207 | | /************************************************************************/ |
208 | | |
209 | | GDALPDFObject::~GDALPDFObject() |
210 | 61.5k | { |
211 | 61.5k | } |
212 | | |
213 | | /************************************************************************/ |
214 | | /* LookupObject() */ |
215 | | /************************************************************************/ |
216 | | |
217 | | GDALPDFObject *GDALPDFObject::LookupObject(const char *pszPath) |
218 | 2.69k | { |
219 | 2.69k | if (GetType() != PDFObjectType_Dictionary) |
220 | 0 | return nullptr; |
221 | 2.69k | return GetDictionary()->LookupObject(pszPath); |
222 | 2.69k | } |
223 | | |
224 | | /************************************************************************/ |
225 | | /* GetTypeName() */ |
226 | | /************************************************************************/ |
227 | | |
228 | | const char *GDALPDFObject::GetTypeName() |
229 | 24 | { |
230 | 24 | switch (GetType()) |
231 | 24 | { |
232 | 24 | case PDFObjectType_Unknown: |
233 | 24 | return GetTypeNameNative(); |
234 | 0 | case PDFObjectType_Null: |
235 | 0 | return "null"; |
236 | 0 | case PDFObjectType_Bool: |
237 | 0 | return "bool"; |
238 | 0 | case PDFObjectType_Int: |
239 | 0 | return "int"; |
240 | 0 | case PDFObjectType_Real: |
241 | 0 | return "real"; |
242 | 0 | case PDFObjectType_String: |
243 | 0 | return "string"; |
244 | 0 | case PDFObjectType_Name: |
245 | 0 | return "name"; |
246 | 0 | case PDFObjectType_Array: |
247 | 0 | return "array"; |
248 | 0 | case PDFObjectType_Dictionary: |
249 | 0 | return "dictionary"; |
250 | 0 | default: |
251 | 0 | return GetTypeNameNative(); |
252 | 24 | } |
253 | 24 | } |
254 | | |
255 | | /************************************************************************/ |
256 | | /* Serialize() */ |
257 | | /************************************************************************/ |
258 | | |
259 | | void GDALPDFObject::Serialize(CPLString &osStr, bool bEmitRef) |
260 | 39.5k | { |
261 | 39.5k | auto nRefNum = GetRefNum(); |
262 | 39.5k | if (bEmitRef && nRefNum.toBool()) |
263 | 8.33k | { |
264 | 8.33k | int nRefGen = GetRefGen(); |
265 | 8.33k | osStr.append(CPLSPrintf("%d %d R", nRefNum.toInt(), nRefGen)); |
266 | 8.33k | return; |
267 | 8.33k | } |
268 | | |
269 | 31.1k | switch (GetType()) |
270 | 31.1k | { |
271 | 0 | case PDFObjectType_Null: |
272 | 0 | osStr.append("null"); |
273 | 0 | return; |
274 | 0 | case PDFObjectType_Bool: |
275 | 0 | osStr.append(GetBool() ? "true" : "false"); |
276 | 0 | return; |
277 | 10.8k | case PDFObjectType_Int: |
278 | 10.8k | osStr.append(CPLSPrintf("%d", GetInt())); |
279 | 10.8k | return; |
280 | 8.36k | case PDFObjectType_Real: |
281 | 8.36k | { |
282 | 8.36k | char szReal[512]; |
283 | 8.36k | double dfRealNonRounded = GetReal(); |
284 | 8.36k | double dfReal = ROUND_IF_CLOSE(dfRealNonRounded); |
285 | 8.36k | if (dfReal >= |
286 | 8.36k | static_cast<double>(std::numeric_limits<GIntBig>::min()) && |
287 | 8.36k | dfReal <= |
288 | 8.34k | static_cast<double>(std::numeric_limits<GIntBig>::max()) && |
289 | 8.36k | dfReal == static_cast<double>(static_cast<GIntBig>(dfReal))) |
290 | 6.68k | { |
291 | 6.68k | snprintf(szReal, sizeof(szReal), CPL_FRMT_GIB, |
292 | 6.68k | static_cast<GIntBig>(dfReal)); |
293 | 6.68k | } |
294 | 1.67k | else if (CanRepresentRealAsString()) |
295 | 0 | { |
296 | | /* Used for OGC BP numeric values */ |
297 | 0 | CPLsnprintf(szReal, sizeof(szReal), "(%.*g)", GetPrecision(), |
298 | 0 | dfReal); |
299 | 0 | } |
300 | 1.67k | else |
301 | 1.67k | { |
302 | 1.67k | CPLsnprintf(szReal, sizeof(szReal), "%.*f", GetPrecision(), |
303 | 1.67k | dfReal); |
304 | | |
305 | | /* Remove non significant trailing zeroes */ |
306 | 1.67k | char *pszDot = strchr(szReal, '.'); |
307 | 1.67k | if (pszDot) |
308 | 1.66k | { |
309 | 1.66k | int iDot = static_cast<int>(pszDot - szReal); |
310 | 1.66k | int nLen = static_cast<int>(strlen(szReal)); |
311 | 3.24k | for (int i = nLen - 1; i > iDot; i--) |
312 | 3.20k | { |
313 | 3.20k | if (szReal[i] == '0') |
314 | 1.58k | szReal[i] = '\0'; |
315 | 1.62k | else |
316 | 1.62k | break; |
317 | 3.20k | } |
318 | 1.66k | } |
319 | 1.67k | } |
320 | 8.36k | osStr.append(szReal); |
321 | 8.36k | return; |
322 | 0 | } |
323 | 656 | case PDFObjectType_String: |
324 | 656 | osStr.append(GDALPDFGetPDFString(GetString().c_str())); |
325 | 656 | return; |
326 | 7.52k | case PDFObjectType_Name: |
327 | 7.52k | osStr.append("/"); |
328 | 7.52k | osStr.append(GDALPDFGetPDFName(GetName())); |
329 | 7.52k | return; |
330 | 3.22k | case PDFObjectType_Array: |
331 | 3.22k | GetArray()->Serialize(osStr); |
332 | 3.22k | return; |
333 | 605 | case PDFObjectType_Dictionary: |
334 | 605 | GetDictionary()->Serialize(osStr); |
335 | 605 | return; |
336 | 0 | case PDFObjectType_Unknown: |
337 | 0 | default: |
338 | 0 | CPLError(CE_Warning, CPLE_AppDefined, |
339 | 0 | "Serializing unknown object !"); |
340 | 0 | return; |
341 | 31.1k | } |
342 | 31.1k | } |
343 | | |
344 | | /************************************************************************/ |
345 | | /* Clone() */ |
346 | | /************************************************************************/ |
347 | | |
348 | | GDALPDFObjectRW *GDALPDFObject::Clone() |
349 | 0 | { |
350 | 0 | auto nRefNum = GetRefNum(); |
351 | 0 | if (nRefNum.toBool()) |
352 | 0 | { |
353 | 0 | int nRefGen = GetRefGen(); |
354 | 0 | return GDALPDFObjectRW::CreateIndirect(nRefNum, nRefGen); |
355 | 0 | } |
356 | | |
357 | 0 | switch (GetType()) |
358 | 0 | { |
359 | 0 | case PDFObjectType_Null: |
360 | 0 | return GDALPDFObjectRW::CreateNull(); |
361 | 0 | case PDFObjectType_Bool: |
362 | 0 | return GDALPDFObjectRW::CreateBool(GetBool()); |
363 | 0 | case PDFObjectType_Int: |
364 | 0 | return GDALPDFObjectRW::CreateInt(GetInt()); |
365 | 0 | case PDFObjectType_Real: |
366 | 0 | return GDALPDFObjectRW::CreateReal(GetReal()); |
367 | 0 | case PDFObjectType_String: |
368 | 0 | return GDALPDFObjectRW::CreateString(GetString().c_str()); |
369 | 0 | case PDFObjectType_Name: |
370 | 0 | return GDALPDFObjectRW::CreateName(GetName().c_str()); |
371 | 0 | case PDFObjectType_Array: |
372 | 0 | return GDALPDFObjectRW::CreateArray(GetArray()->Clone()); |
373 | 0 | case PDFObjectType_Dictionary: |
374 | 0 | return GDALPDFObjectRW::CreateDictionary(GetDictionary()->Clone()); |
375 | 0 | case PDFObjectType_Unknown: |
376 | 0 | default: |
377 | 0 | CPLError(CE_Warning, CPLE_AppDefined, "Cloning unknown object !"); |
378 | 0 | return nullptr; |
379 | 0 | } |
380 | 0 | } |
381 | | |
382 | | /************************************************************************/ |
383 | | /* ==================================================================== */ |
384 | | /* GDALPDFDictionary */ |
385 | | /* ==================================================================== */ |
386 | | /************************************************************************/ |
387 | | |
388 | | /************************************************************************/ |
389 | | /* ~GDALPDFDictionary() */ |
390 | | /************************************************************************/ |
391 | | |
392 | | GDALPDFDictionary::~GDALPDFDictionary() |
393 | 12.8k | { |
394 | 12.8k | } |
395 | | |
396 | | /************************************************************************/ |
397 | | /* LookupObject() */ |
398 | | /************************************************************************/ |
399 | | |
400 | | GDALPDFObject *GDALPDFDictionary::LookupObject(const char *pszPath) |
401 | 4.02k | { |
402 | 4.02k | GDALPDFObject *poCurObj = nullptr; |
403 | 4.02k | char **papszTokens = CSLTokenizeString2(pszPath, ".", 0); |
404 | 9.38k | for (int i = 0; papszTokens[i] != nullptr; i++) |
405 | 6.97k | { |
406 | 6.97k | int iElt = -1; |
407 | 6.97k | char *pszBracket = strchr(papszTokens[i], '['); |
408 | 6.97k | if (pszBracket != nullptr) |
409 | 0 | { |
410 | 0 | iElt = atoi(pszBracket + 1); |
411 | 0 | *pszBracket = '\0'; |
412 | 0 | } |
413 | | |
414 | 6.97k | if (i == 0) |
415 | 4.02k | { |
416 | 4.02k | poCurObj = Get(papszTokens[i]); |
417 | 4.02k | } |
418 | 2.94k | else |
419 | 2.94k | { |
420 | 2.94k | if (poCurObj->GetType() != PDFObjectType_Dictionary) |
421 | 0 | { |
422 | 0 | poCurObj = nullptr; |
423 | 0 | break; |
424 | 0 | } |
425 | 2.94k | poCurObj = poCurObj->GetDictionary()->Get(papszTokens[i]); |
426 | 2.94k | } |
427 | | |
428 | 6.97k | if (poCurObj == nullptr) |
429 | 1.61k | { |
430 | 1.61k | poCurObj = nullptr; |
431 | 1.61k | break; |
432 | 1.61k | } |
433 | | |
434 | 5.35k | if (iElt >= 0) |
435 | 0 | { |
436 | 0 | if (poCurObj->GetType() != PDFObjectType_Array) |
437 | 0 | { |
438 | 0 | poCurObj = nullptr; |
439 | 0 | break; |
440 | 0 | } |
441 | 0 | poCurObj = poCurObj->GetArray()->Get(iElt); |
442 | 0 | } |
443 | 5.35k | } |
444 | 4.02k | CSLDestroy(papszTokens); |
445 | 4.02k | return poCurObj; |
446 | 4.02k | } |
447 | | |
448 | | /************************************************************************/ |
449 | | /* Serialize() */ |
450 | | /************************************************************************/ |
451 | | |
452 | | void GDALPDFDictionary::Serialize(CPLString &osStr) |
453 | 6.55k | { |
454 | 6.55k | osStr.append("<< "); |
455 | 6.55k | for (const auto &oIter : GetValues()) |
456 | 23.7k | { |
457 | 23.7k | const char *pszKey = oIter.first.c_str(); |
458 | 23.7k | GDALPDFObject *poObj = oIter.second; |
459 | 23.7k | osStr.append("/"); |
460 | 23.7k | osStr.append(pszKey); |
461 | 23.7k | osStr.append(" "); |
462 | 23.7k | poObj->Serialize(osStr); |
463 | 23.7k | osStr.append(" "); |
464 | 23.7k | } |
465 | 6.55k | osStr.append(">>"); |
466 | 6.55k | } |
467 | | |
468 | | /************************************************************************/ |
469 | | /* Clone() */ |
470 | | /************************************************************************/ |
471 | | |
472 | | GDALPDFDictionaryRW *GDALPDFDictionary::Clone() |
473 | 0 | { |
474 | 0 | GDALPDFDictionaryRW *poDict = new GDALPDFDictionaryRW(); |
475 | 0 | for (const auto &oIter : GetValues()) |
476 | 0 | { |
477 | 0 | const char *pszKey = oIter.first.c_str(); |
478 | 0 | GDALPDFObject *poObj = oIter.second; |
479 | 0 | poDict->Add(pszKey, poObj->Clone()); |
480 | 0 | } |
481 | 0 | return poDict; |
482 | 0 | } |
483 | | |
484 | | /************************************************************************/ |
485 | | /* ==================================================================== */ |
486 | | /* GDALPDFArray */ |
487 | | /* ==================================================================== */ |
488 | | /************************************************************************/ |
489 | | |
490 | | /************************************************************************/ |
491 | | /* ~GDALPDFArray() */ |
492 | | /************************************************************************/ |
493 | | |
494 | | GDALPDFArray::~GDALPDFArray() |
495 | 5.80k | { |
496 | 5.80k | } |
497 | | |
498 | | /************************************************************************/ |
499 | | /* Serialize() */ |
500 | | /************************************************************************/ |
501 | | |
502 | | void GDALPDFArray::Serialize(CPLString &osStr) |
503 | 3.83k | { |
504 | 3.83k | int nLength = GetLength(); |
505 | 3.83k | int i; |
506 | | |
507 | 3.83k | osStr.append("[ "); |
508 | 17.4k | for (i = 0; i < nLength; i++) |
509 | 13.6k | { |
510 | 13.6k | Get(i)->Serialize(osStr); |
511 | 13.6k | osStr.append(" "); |
512 | 13.6k | } |
513 | 3.83k | osStr.append("]"); |
514 | 3.83k | } |
515 | | |
516 | | /************************************************************************/ |
517 | | /* Clone() */ |
518 | | /************************************************************************/ |
519 | | |
520 | | GDALPDFArrayRW *GDALPDFArray::Clone() |
521 | 0 | { |
522 | 0 | GDALPDFArrayRW *poArray = new GDALPDFArrayRW(); |
523 | 0 | int nLength = GetLength(); |
524 | 0 | int i; |
525 | 0 | for (i = 0; i < nLength; i++) |
526 | 0 | { |
527 | 0 | poArray->Add(Get(i)->Clone()); |
528 | 0 | } |
529 | 0 | return poArray; |
530 | 0 | } |
531 | | |
532 | | /************************************************************************/ |
533 | | /* ==================================================================== */ |
534 | | /* GDALPDFStream */ |
535 | | /* ==================================================================== */ |
536 | | /************************************************************************/ |
537 | | |
538 | | /************************************************************************/ |
539 | | /* ~GDALPDFStream() */ |
540 | | /************************************************************************/ |
541 | | |
542 | 539 | GDALPDFStream::~GDALPDFStream() = default; |
543 | | |
544 | | /************************************************************************/ |
545 | | /* ==================================================================== */ |
546 | | /* GDALPDFObjectRW */ |
547 | | /* ==================================================================== */ |
548 | | /************************************************************************/ |
549 | | |
550 | | /************************************************************************/ |
551 | | /* GDALPDFObjectRW() */ |
552 | | /************************************************************************/ |
553 | | |
554 | 39.5k | GDALPDFObjectRW::GDALPDFObjectRW(GDALPDFObjectType eType) : m_eType(eType) |
555 | 39.5k | { |
556 | 39.5k | } |
557 | | |
558 | | /************************************************************************/ |
559 | | /* ~GDALPDFObjectRW() */ |
560 | | /************************************************************************/ |
561 | | |
562 | | GDALPDFObjectRW::~GDALPDFObjectRW() |
563 | 39.5k | { |
564 | 39.5k | delete m_poDict; |
565 | 39.5k | delete m_poArray; |
566 | 39.5k | } |
567 | | |
568 | | /************************************************************************/ |
569 | | /* CreateIndirect() */ |
570 | | /************************************************************************/ |
571 | | |
572 | | GDALPDFObjectRW *GDALPDFObjectRW::CreateIndirect(const GDALPDFObjectNum &nNum, |
573 | | int nGen) |
574 | 8.33k | { |
575 | 8.33k | GDALPDFObjectRW *poObj = new GDALPDFObjectRW(PDFObjectType_Unknown); |
576 | 8.33k | poObj->m_nNum = nNum; |
577 | 8.33k | poObj->m_nGen = nGen; |
578 | 8.33k | return poObj; |
579 | 8.33k | } |
580 | | |
581 | | /************************************************************************/ |
582 | | /* CreateNull() */ |
583 | | /************************************************************************/ |
584 | | |
585 | | GDALPDFObjectRW *GDALPDFObjectRW::CreateNull() |
586 | 0 | { |
587 | 0 | return new GDALPDFObjectRW(PDFObjectType_Null); |
588 | 0 | } |
589 | | |
590 | | /************************************************************************/ |
591 | | /* CreateBool() */ |
592 | | /************************************************************************/ |
593 | | |
594 | | GDALPDFObjectRW *GDALPDFObjectRW::CreateBool(int bVal) |
595 | 0 | { |
596 | 0 | GDALPDFObjectRW *poObj = new GDALPDFObjectRW(PDFObjectType_Bool); |
597 | 0 | poObj->m_nVal = bVal; |
598 | 0 | return poObj; |
599 | 0 | } |
600 | | |
601 | | /************************************************************************/ |
602 | | /* CreateInt() */ |
603 | | /************************************************************************/ |
604 | | |
605 | | GDALPDFObjectRW *GDALPDFObjectRW::CreateInt(int nVal) |
606 | 10.8k | { |
607 | 10.8k | GDALPDFObjectRW *poObj = new GDALPDFObjectRW(PDFObjectType_Int); |
608 | 10.8k | poObj->m_nVal = nVal; |
609 | 10.8k | return poObj; |
610 | 10.8k | } |
611 | | |
612 | | /************************************************************************/ |
613 | | /* CreateReal() */ |
614 | | /************************************************************************/ |
615 | | |
616 | | GDALPDFObjectRW *GDALPDFObjectRW::CreateReal(double dfVal, |
617 | | int bCanRepresentRealAsString) |
618 | 8.36k | { |
619 | 8.36k | GDALPDFObjectRW *poObj = new GDALPDFObjectRW(PDFObjectType_Real); |
620 | 8.36k | poObj->m_dfVal = dfVal; |
621 | 8.36k | poObj->m_bCanRepresentRealAsString = bCanRepresentRealAsString; |
622 | 8.36k | return poObj; |
623 | 8.36k | } |
624 | | |
625 | | /************************************************************************/ |
626 | | /* CreateRealWithPrecision() */ |
627 | | /************************************************************************/ |
628 | | |
629 | | GDALPDFObjectRW *GDALPDFObjectRW::CreateRealWithPrecision(double dfVal, |
630 | | int nPrecision) |
631 | 0 | { |
632 | 0 | GDALPDFObjectRW *poObj = new GDALPDFObjectRW(PDFObjectType_Real); |
633 | 0 | poObj->m_dfVal = dfVal; |
634 | 0 | poObj->m_nPrecision = nPrecision; |
635 | 0 | return poObj; |
636 | 0 | } |
637 | | |
638 | | /************************************************************************/ |
639 | | /* CreateString() */ |
640 | | /************************************************************************/ |
641 | | |
642 | | GDALPDFObjectRW *GDALPDFObjectRW::CreateString(const char *pszStr) |
643 | 656 | { |
644 | 656 | GDALPDFObjectRW *poObj = new GDALPDFObjectRW(PDFObjectType_String); |
645 | 656 | poObj->m_osVal = pszStr; |
646 | 656 | return poObj; |
647 | 656 | } |
648 | | |
649 | | /************************************************************************/ |
650 | | /* CreateName() */ |
651 | | /************************************************************************/ |
652 | | |
653 | | GDALPDFObjectRW *GDALPDFObjectRW::CreateName(const char *pszName) |
654 | 7.52k | { |
655 | 7.52k | GDALPDFObjectRW *poObj = new GDALPDFObjectRW(PDFObjectType_Name); |
656 | 7.52k | poObj->m_osVal = pszName; |
657 | 7.52k | return poObj; |
658 | 7.52k | } |
659 | | |
660 | | /************************************************************************/ |
661 | | /* CreateDictionary() */ |
662 | | /************************************************************************/ |
663 | | |
664 | | GDALPDFObjectRW *GDALPDFObjectRW::CreateDictionary(GDALPDFDictionaryRW *poDict) |
665 | 605 | { |
666 | 605 | CPLAssert(poDict); |
667 | 605 | GDALPDFObjectRW *poObj = new GDALPDFObjectRW(PDFObjectType_Dictionary); |
668 | 605 | poObj->m_poDict = poDict; |
669 | 605 | return poObj; |
670 | 605 | } |
671 | | |
672 | | /************************************************************************/ |
673 | | /* CreateArray() */ |
674 | | /************************************************************************/ |
675 | | |
676 | | GDALPDFObjectRW *GDALPDFObjectRW::CreateArray(GDALPDFArrayRW *poArray) |
677 | 3.22k | { |
678 | 3.22k | CPLAssert(poArray); |
679 | 3.22k | GDALPDFObjectRW *poObj = new GDALPDFObjectRW(PDFObjectType_Array); |
680 | 3.22k | poObj->m_poArray = poArray; |
681 | 3.22k | return poObj; |
682 | 3.22k | } |
683 | | |
684 | | /************************************************************************/ |
685 | | /* GetTypeNameNative() */ |
686 | | /************************************************************************/ |
687 | | |
688 | | const char *GDALPDFObjectRW::GetTypeNameNative() |
689 | 0 | { |
690 | 0 | CPLError(CE_Failure, CPLE_AppDefined, "Should not go here"); |
691 | 0 | return ""; |
692 | 0 | } |
693 | | |
694 | | /************************************************************************/ |
695 | | /* GetType() */ |
696 | | /************************************************************************/ |
697 | | |
698 | | GDALPDFObjectType GDALPDFObjectRW::GetType() |
699 | 31.1k | { |
700 | 31.1k | return m_eType; |
701 | 31.1k | } |
702 | | |
703 | | /************************************************************************/ |
704 | | /* GetBool() */ |
705 | | /************************************************************************/ |
706 | | |
707 | | int GDALPDFObjectRW::GetBool() |
708 | 0 | { |
709 | 0 | if (m_eType == PDFObjectType_Bool) |
710 | 0 | return m_nVal; |
711 | | |
712 | 0 | return FALSE; |
713 | 0 | } |
714 | | |
715 | | /************************************************************************/ |
716 | | /* GetInt() */ |
717 | | /************************************************************************/ |
718 | | |
719 | | int GDALPDFObjectRW::GetInt() |
720 | 10.8k | { |
721 | 10.8k | if (m_eType == PDFObjectType_Int) |
722 | 10.8k | return m_nVal; |
723 | | |
724 | 0 | return 0; |
725 | 10.8k | } |
726 | | |
727 | | /************************************************************************/ |
728 | | /* GetReal() */ |
729 | | /************************************************************************/ |
730 | | |
731 | | double GDALPDFObjectRW::GetReal() |
732 | 8.36k | { |
733 | 8.36k | return m_dfVal; |
734 | 8.36k | } |
735 | | |
736 | | /************************************************************************/ |
737 | | /* GetString() */ |
738 | | /************************************************************************/ |
739 | | |
740 | | const CPLString &GDALPDFObjectRW::GetString() |
741 | 656 | { |
742 | 656 | return m_osVal; |
743 | 656 | } |
744 | | |
745 | | /************************************************************************/ |
746 | | /* GetName() */ |
747 | | /************************************************************************/ |
748 | | |
749 | | const CPLString &GDALPDFObjectRW::GetName() |
750 | 7.52k | { |
751 | 7.52k | return m_osVal; |
752 | 7.52k | } |
753 | | |
754 | | /************************************************************************/ |
755 | | /* GetDictionary() */ |
756 | | /************************************************************************/ |
757 | | |
758 | | GDALPDFDictionary *GDALPDFObjectRW::GetDictionary() |
759 | 605 | { |
760 | 605 | return m_poDict; |
761 | 605 | } |
762 | | |
763 | | /************************************************************************/ |
764 | | /* GetArray() */ |
765 | | /************************************************************************/ |
766 | | |
767 | | GDALPDFArray *GDALPDFObjectRW::GetArray() |
768 | 3.22k | { |
769 | 3.22k | return m_poArray; |
770 | 3.22k | } |
771 | | |
772 | | /************************************************************************/ |
773 | | /* GetStream() */ |
774 | | /************************************************************************/ |
775 | | |
776 | | GDALPDFStream *GDALPDFObjectRW::GetStream() |
777 | 0 | { |
778 | 0 | return nullptr; |
779 | 0 | } |
780 | | |
781 | | /************************************************************************/ |
782 | | /* GetRefNum() */ |
783 | | /************************************************************************/ |
784 | | |
785 | | GDALPDFObjectNum GDALPDFObjectRW::GetRefNum() |
786 | 39.5k | { |
787 | 39.5k | return m_nNum; |
788 | 39.5k | } |
789 | | |
790 | | /************************************************************************/ |
791 | | /* GetRefGen() */ |
792 | | /************************************************************************/ |
793 | | |
794 | | int GDALPDFObjectRW::GetRefGen() |
795 | 8.33k | { |
796 | 8.33k | return m_nGen; |
797 | 8.33k | } |
798 | | |
799 | | /************************************************************************/ |
800 | | /* ==================================================================== */ |
801 | | /* GDALPDFDictionaryRW */ |
802 | | /* ==================================================================== */ |
803 | | /************************************************************************/ |
804 | | |
805 | | /************************************************************************/ |
806 | | /* GDALPDFDictionaryRW() */ |
807 | | /************************************************************************/ |
808 | | |
809 | 6.55k | GDALPDFDictionaryRW::GDALPDFDictionaryRW() = default; |
810 | | |
811 | | /************************************************************************/ |
812 | | /* ~GDALPDFDictionaryRW() */ |
813 | | /************************************************************************/ |
814 | | |
815 | | GDALPDFDictionaryRW::~GDALPDFDictionaryRW() |
816 | 6.55k | { |
817 | 6.55k | std::map<CPLString, GDALPDFObject *>::iterator oIter = m_map.begin(); |
818 | 6.55k | std::map<CPLString, GDALPDFObject *>::iterator oEnd = m_map.end(); |
819 | 30.2k | for (; oIter != oEnd; ++oIter) |
820 | 23.7k | delete oIter->second; |
821 | 6.55k | } |
822 | | |
823 | | /************************************************************************/ |
824 | | /* Get() */ |
825 | | /************************************************************************/ |
826 | | |
827 | | GDALPDFObject *GDALPDFDictionaryRW::Get(const char *pszKey) |
828 | 0 | { |
829 | 0 | std::map<CPLString, GDALPDFObject *>::iterator oIter = m_map.find(pszKey); |
830 | 0 | if (oIter != m_map.end()) |
831 | 0 | return oIter->second; |
832 | 0 | return nullptr; |
833 | 0 | } |
834 | | |
835 | | /************************************************************************/ |
836 | | /* GetValues() */ |
837 | | /************************************************************************/ |
838 | | |
839 | | std::map<CPLString, GDALPDFObject *> &GDALPDFDictionaryRW::GetValues() |
840 | 6.55k | { |
841 | 6.55k | return m_map; |
842 | 6.55k | } |
843 | | |
844 | | /************************************************************************/ |
845 | | /* Add() */ |
846 | | /************************************************************************/ |
847 | | |
848 | | GDALPDFDictionaryRW &GDALPDFDictionaryRW::Add(const char *pszKey, |
849 | | GDALPDFObject *poVal) |
850 | 23.7k | { |
851 | 23.7k | std::map<CPLString, GDALPDFObject *>::iterator oIter = m_map.find(pszKey); |
852 | 23.7k | if (oIter != m_map.end()) |
853 | 0 | { |
854 | 0 | delete oIter->second; |
855 | 0 | oIter->second = poVal; |
856 | 0 | } |
857 | 23.7k | else |
858 | 23.7k | m_map[pszKey] = poVal; |
859 | | |
860 | 23.7k | return *this; |
861 | 23.7k | } |
862 | | |
863 | | /************************************************************************/ |
864 | | /* Remove() */ |
865 | | /************************************************************************/ |
866 | | |
867 | | GDALPDFDictionaryRW &GDALPDFDictionaryRW::Remove(const char *pszKey) |
868 | 0 | { |
869 | 0 | std::map<CPLString, GDALPDFObject *>::iterator oIter = m_map.find(pszKey); |
870 | 0 | if (oIter != m_map.end()) |
871 | 0 | { |
872 | 0 | delete oIter->second; |
873 | 0 | m_map.erase(pszKey); |
874 | 0 | } |
875 | |
|
876 | 0 | return *this; |
877 | 0 | } |
878 | | |
879 | | /************************************************************************/ |
880 | | /* ==================================================================== */ |
881 | | /* GDALPDFArrayRW */ |
882 | | /* ==================================================================== */ |
883 | | /************************************************************************/ |
884 | | |
885 | | /************************************************************************/ |
886 | | /* GDALPDFArrayRW() */ |
887 | | /************************************************************************/ |
888 | | |
889 | | GDALPDFArrayRW::GDALPDFArrayRW() |
890 | 3.83k | { |
891 | 3.83k | } |
892 | | |
893 | | /************************************************************************/ |
894 | | /* ~GDALPDFArrayRW() */ |
895 | | /************************************************************************/ |
896 | | |
897 | | GDALPDFArrayRW::~GDALPDFArrayRW() |
898 | 3.83k | { |
899 | 17.4k | for (size_t i = 0; i < m_array.size(); i++) |
900 | 13.6k | delete m_array[i]; |
901 | 3.83k | } |
902 | | |
903 | | /************************************************************************/ |
904 | | /* GetLength() */ |
905 | | /************************************************************************/ |
906 | | |
907 | | int GDALPDFArrayRW::GetLength() |
908 | 17.4k | { |
909 | 17.4k | return static_cast<int>(m_array.size()); |
910 | 17.4k | } |
911 | | |
912 | | /************************************************************************/ |
913 | | /* Get() */ |
914 | | /************************************************************************/ |
915 | | |
916 | | GDALPDFObject *GDALPDFArrayRW::Get(int nIndex) |
917 | 13.6k | { |
918 | 13.6k | if (nIndex < 0 || nIndex >= GetLength()) |
919 | 0 | return nullptr; |
920 | 13.6k | return m_array[nIndex]; |
921 | 13.6k | } |
922 | | |
923 | | /************************************************************************/ |
924 | | /* Add() */ |
925 | | /************************************************************************/ |
926 | | |
927 | | GDALPDFArrayRW &GDALPDFArrayRW::Add(GDALPDFObject *poObj) |
928 | 13.6k | { |
929 | 13.6k | m_array.push_back(poObj); |
930 | 13.6k | return *this; |
931 | 13.6k | } |
932 | | |
933 | | /************************************************************************/ |
934 | | /* Add() */ |
935 | | /************************************************************************/ |
936 | | |
937 | | GDALPDFArrayRW &GDALPDFArrayRW::Add(double *padfVal, int nCount, |
938 | | int bCanRepresentRealAsString) |
939 | 0 | { |
940 | 0 | for (int i = 0; i < nCount; i++) |
941 | 0 | m_array.push_back( |
942 | 0 | GDALPDFObjectRW::CreateReal(padfVal[i], bCanRepresentRealAsString)); |
943 | 0 | return *this; |
944 | 0 | } |
945 | | |
946 | | #ifdef HAVE_POPPLER |
947 | | |
948 | | /************************************************************************/ |
949 | | /* ==================================================================== */ |
950 | | /* GDALPDFDictionaryPoppler */ |
951 | | /* ==================================================================== */ |
952 | | /************************************************************************/ |
953 | | |
954 | | class GDALPDFDictionaryPoppler : public GDALPDFDictionary |
955 | | { |
956 | | private: |
957 | | Dict *m_poDict; |
958 | | std::map<CPLString, GDALPDFObject *> m_map{}; |
959 | | |
960 | | CPL_DISALLOW_COPY_ASSIGN(GDALPDFDictionaryPoppler) |
961 | | |
962 | | public: |
963 | 6.24k | GDALPDFDictionaryPoppler(Dict *poDict) : m_poDict(poDict) |
964 | 6.24k | { |
965 | 6.24k | } |
966 | | |
967 | | virtual ~GDALPDFDictionaryPoppler(); |
968 | | |
969 | | virtual GDALPDFObject *Get(const char *pszKey) override; |
970 | | virtual std::map<CPLString, GDALPDFObject *> &GetValues() override; |
971 | | }; |
972 | | |
973 | | /************************************************************************/ |
974 | | /* ==================================================================== */ |
975 | | /* GDALPDFArrayPoppler */ |
976 | | /* ==================================================================== */ |
977 | | /************************************************************************/ |
978 | | |
979 | | class GDALPDFArrayPoppler : public GDALPDFArray |
980 | | { |
981 | | private: |
982 | | const Array *m_poArray; |
983 | | std::vector<std::unique_ptr<GDALPDFObject>> m_v{}; |
984 | | |
985 | | CPL_DISALLOW_COPY_ASSIGN(GDALPDFArrayPoppler) |
986 | | |
987 | | public: |
988 | 1.97k | GDALPDFArrayPoppler(const Array *poArray) : m_poArray(poArray) |
989 | 1.97k | { |
990 | 1.97k | } |
991 | | |
992 | | virtual int GetLength() override; |
993 | | virtual GDALPDFObject *Get(int nIndex) override; |
994 | | }; |
995 | | |
996 | | /************************************************************************/ |
997 | | /* ==================================================================== */ |
998 | | /* GDALPDFStreamPoppler */ |
999 | | /* ==================================================================== */ |
1000 | | /************************************************************************/ |
1001 | | |
1002 | | class GDALPDFStreamPoppler : public GDALPDFStream |
1003 | | { |
1004 | | private: |
1005 | | int64_t m_nLength = -1; |
1006 | | Stream *m_poStream; |
1007 | | int64_t m_nRawLength = -1; |
1008 | | |
1009 | | CPL_DISALLOW_COPY_ASSIGN(GDALPDFStreamPoppler) |
1010 | | |
1011 | | public: |
1012 | 539 | GDALPDFStreamPoppler(Stream *poStream) : m_poStream(poStream) |
1013 | 539 | { |
1014 | 539 | } |
1015 | | |
1016 | | virtual int64_t GetLength(int64_t nMaxSize = 0) override; |
1017 | | virtual char *GetBytes() override; |
1018 | | |
1019 | | virtual int64_t GetRawLength() override; |
1020 | | virtual char *GetRawBytes() override; |
1021 | | }; |
1022 | | |
1023 | | /************************************************************************/ |
1024 | | /* ==================================================================== */ |
1025 | | /* GDALPDFObjectPoppler */ |
1026 | | /* ==================================================================== */ |
1027 | | /************************************************************************/ |
1028 | | |
1029 | | /************************************************************************/ |
1030 | | /* ~GDALPDFObjectPoppler() */ |
1031 | | /************************************************************************/ |
1032 | | |
1033 | | GDALPDFObjectPoppler::~GDALPDFObjectPoppler() |
1034 | 22.0k | { |
1035 | 22.0k | delete m_poToDestroy; |
1036 | 22.0k | delete m_poDict; |
1037 | 22.0k | delete m_poArray; |
1038 | 22.0k | delete m_poStream; |
1039 | 22.0k | } |
1040 | | |
1041 | | /************************************************************************/ |
1042 | | /* GetType() */ |
1043 | | /************************************************************************/ |
1044 | | |
1045 | | GDALPDFObjectType GDALPDFObjectPoppler::GetType() |
1046 | 94.9k | { |
1047 | 94.9k | switch (m_poConst->getType()) |
1048 | 94.9k | { |
1049 | 539 | case objNull: |
1050 | 539 | return PDFObjectType_Null; |
1051 | 0 | case objBool: |
1052 | 0 | return PDFObjectType_Bool; |
1053 | 30.0k | case objInt: |
1054 | 30.0k | return PDFObjectType_Int; |
1055 | 5.98k | case objReal: |
1056 | 5.98k | return PDFObjectType_Real; |
1057 | 1.96k | case objString: |
1058 | 1.96k | return PDFObjectType_String; |
1059 | 5.19k | case objName: |
1060 | 5.19k | return PDFObjectType_Name; |
1061 | 24.0k | case objArray: |
1062 | 24.0k | return PDFObjectType_Array; |
1063 | 21.1k | case objDict: |
1064 | 21.1k | return PDFObjectType_Dictionary; |
1065 | 5.89k | case objStream: |
1066 | 5.89k | return PDFObjectType_Dictionary; |
1067 | 120 | default: |
1068 | 120 | return PDFObjectType_Unknown; |
1069 | 94.9k | } |
1070 | 94.9k | } |
1071 | | |
1072 | | /************************************************************************/ |
1073 | | /* GetTypeNameNative() */ |
1074 | | /************************************************************************/ |
1075 | | |
1076 | | const char *GDALPDFObjectPoppler::GetTypeNameNative() |
1077 | 24 | { |
1078 | 24 | return m_poConst->getTypeName(); |
1079 | 24 | } |
1080 | | |
1081 | | /************************************************************************/ |
1082 | | /* GetBool() */ |
1083 | | /************************************************************************/ |
1084 | | |
1085 | | int GDALPDFObjectPoppler::GetBool() |
1086 | 0 | { |
1087 | 0 | if (GetType() == PDFObjectType_Bool) |
1088 | 0 | return m_poConst->getBool(); |
1089 | 0 | else |
1090 | 0 | return 0; |
1091 | 0 | } |
1092 | | |
1093 | | /************************************************************************/ |
1094 | | /* GetInt() */ |
1095 | | /************************************************************************/ |
1096 | | |
1097 | | int GDALPDFObjectPoppler::GetInt() |
1098 | 9.85k | { |
1099 | 9.85k | if (GetType() == PDFObjectType_Int) |
1100 | 9.85k | return m_poConst->getInt(); |
1101 | 0 | else |
1102 | 0 | return 0; |
1103 | 9.85k | } |
1104 | | |
1105 | | /************************************************************************/ |
1106 | | /* GetReal() */ |
1107 | | /************************************************************************/ |
1108 | | |
1109 | | double GDALPDFObjectPoppler::GetReal() |
1110 | 1.49k | { |
1111 | 1.49k | if (GetType() == PDFObjectType_Real) |
1112 | 1.49k | return m_poConst->getReal(); |
1113 | 0 | else |
1114 | 0 | return 0.0; |
1115 | 1.49k | } |
1116 | | |
1117 | | /************************************************************************/ |
1118 | | /* GetString() */ |
1119 | | /************************************************************************/ |
1120 | | |
1121 | | const std::string &GDALPDFObjectPoppler::GetString() |
1122 | 1.39k | { |
1123 | 1.39k | if (GetType() == PDFObjectType_String) |
1124 | 1.39k | { |
1125 | 1.39k | const GooString *gooString = m_poConst->getString(); |
1126 | 1.39k | const std::string &osStdStr = gooString->toStr(); |
1127 | 1.39k | const bool bLEUnicodeMarker = |
1128 | 1.39k | osStdStr.size() >= 2 && static_cast<uint8_t>(osStdStr[0]) == 0xFE && |
1129 | 1.39k | static_cast<uint8_t>(osStdStr[1]) == 0xFF; |
1130 | 1.39k | const bool bBEUnicodeMarker = |
1131 | 1.39k | osStdStr.size() >= 2 && static_cast<uint8_t>(osStdStr[0]) == 0xFF && |
1132 | 1.39k | static_cast<uint8_t>(osStdStr[1]) == 0xFE; |
1133 | 1.39k | if (!bLEUnicodeMarker && !bBEUnicodeMarker) |
1134 | 1.00k | { |
1135 | 1.00k | if (CPLIsUTF8(osStdStr.c_str(), -1)) |
1136 | 1.00k | { |
1137 | 1.00k | return osStdStr; |
1138 | 1.00k | } |
1139 | 0 | else |
1140 | 0 | { |
1141 | 0 | char *pszUTF8 = |
1142 | 0 | CPLRecode(osStdStr.data(), CPL_ENC_ISO8859_1, CPL_ENC_UTF8); |
1143 | 0 | osStr = pszUTF8; |
1144 | 0 | CPLFree(pszUTF8); |
1145 | 0 | return osStr; |
1146 | 0 | } |
1147 | 1.00k | } |
1148 | 384 | return (osStr = GDALPDFGetUTF8StringFromBytes( |
1149 | 384 | reinterpret_cast<const GByte *>(osStdStr.data()), |
1150 | 384 | osStdStr.size())); |
1151 | 1.39k | } |
1152 | 0 | else |
1153 | 0 | return (osStr = ""); |
1154 | 1.39k | } |
1155 | | |
1156 | | /************************************************************************/ |
1157 | | /* GetName() */ |
1158 | | /************************************************************************/ |
1159 | | |
1160 | | const std::string &GDALPDFObjectPoppler::GetName() |
1161 | 3.09k | { |
1162 | 3.09k | if (GetType() == PDFObjectType_Name) |
1163 | 3.09k | return (osStr = m_poConst->getName()); |
1164 | 0 | else |
1165 | 0 | return (osStr = ""); |
1166 | 3.09k | } |
1167 | | |
1168 | | /************************************************************************/ |
1169 | | /* GetDictionary() */ |
1170 | | /************************************************************************/ |
1171 | | |
1172 | | GDALPDFDictionary *GDALPDFObjectPoppler::GetDictionary() |
1173 | 12.4k | { |
1174 | 12.4k | if (GetType() != PDFObjectType_Dictionary) |
1175 | 0 | return nullptr; |
1176 | | |
1177 | 12.4k | if (m_poDict) |
1178 | 6.18k | return m_poDict; |
1179 | | |
1180 | 6.24k | Dict *poDict = (m_poConst->getType() == objStream) |
1181 | 6.24k | ? m_poConst->getStream()->getDict() |
1182 | 6.24k | : m_poConst->getDict(); |
1183 | 6.24k | if (poDict == nullptr) |
1184 | 0 | return nullptr; |
1185 | 6.24k | m_poDict = new GDALPDFDictionaryPoppler(poDict); |
1186 | 6.24k | return m_poDict; |
1187 | 6.24k | } |
1188 | | |
1189 | | /************************************************************************/ |
1190 | | /* GetArray() */ |
1191 | | /************************************************************************/ |
1192 | | |
1193 | | GDALPDFArray *GDALPDFObjectPoppler::GetArray() |
1194 | 11.9k | { |
1195 | 11.9k | if (GetType() != PDFObjectType_Array) |
1196 | 0 | return nullptr; |
1197 | | |
1198 | 11.9k | if (m_poArray) |
1199 | 10.0k | return m_poArray; |
1200 | | |
1201 | 1.97k | Array *poArray = m_poConst->getArray(); |
1202 | 1.97k | if (poArray == nullptr) |
1203 | 0 | return nullptr; |
1204 | 1.97k | m_poArray = new GDALPDFArrayPoppler(poArray); |
1205 | 1.97k | return m_poArray; |
1206 | 1.97k | } |
1207 | | |
1208 | | /************************************************************************/ |
1209 | | /* GetStream() */ |
1210 | | /************************************************************************/ |
1211 | | |
1212 | | GDALPDFStream *GDALPDFObjectPoppler::GetStream() |
1213 | 539 | { |
1214 | 539 | if (m_poConst->getType() != objStream) |
1215 | 0 | return nullptr; |
1216 | | |
1217 | 539 | if (m_poStream) |
1218 | 0 | return m_poStream; |
1219 | 539 | m_poStream = new GDALPDFStreamPoppler(m_poConst->getStream()); |
1220 | 539 | return m_poStream; |
1221 | 539 | } |
1222 | | |
1223 | | /************************************************************************/ |
1224 | | /* SetRefNumAndGen() */ |
1225 | | /************************************************************************/ |
1226 | | |
1227 | | void GDALPDFObjectPoppler::SetRefNumAndGen(const GDALPDFObjectNum &nNum, |
1228 | | int nGen) |
1229 | 20.9k | { |
1230 | 20.9k | m_nRefNum = nNum; |
1231 | 20.9k | m_nRefGen = nGen; |
1232 | 20.9k | } |
1233 | | |
1234 | | /************************************************************************/ |
1235 | | /* GetRefNum() */ |
1236 | | /************************************************************************/ |
1237 | | |
1238 | | GDALPDFObjectNum GDALPDFObjectPoppler::GetRefNum() |
1239 | 539 | { |
1240 | 539 | return m_nRefNum; |
1241 | 539 | } |
1242 | | |
1243 | | /************************************************************************/ |
1244 | | /* GetRefGen() */ |
1245 | | /************************************************************************/ |
1246 | | |
1247 | | int GDALPDFObjectPoppler::GetRefGen() |
1248 | 539 | { |
1249 | 539 | return m_nRefGen; |
1250 | 539 | } |
1251 | | |
1252 | | /************************************************************************/ |
1253 | | /* ==================================================================== */ |
1254 | | /* GDALPDFDictionaryPoppler */ |
1255 | | /* ==================================================================== */ |
1256 | | /************************************************************************/ |
1257 | | |
1258 | | /************************************************************************/ |
1259 | | /* ~GDALPDFDictionaryPoppler() */ |
1260 | | /************************************************************************/ |
1261 | | |
1262 | | GDALPDFDictionaryPoppler::~GDALPDFDictionaryPoppler() |
1263 | 6.24k | { |
1264 | 6.24k | std::map<CPLString, GDALPDFObject *>::iterator oIter = m_map.begin(); |
1265 | 6.24k | std::map<CPLString, GDALPDFObject *>::iterator oEnd = m_map.end(); |
1266 | 17.8k | for (; oIter != oEnd; ++oIter) |
1267 | 11.5k | delete oIter->second; |
1268 | 6.24k | } |
1269 | | |
1270 | | /************************************************************************/ |
1271 | | /* Get() */ |
1272 | | /************************************************************************/ |
1273 | | |
1274 | | GDALPDFObject *GDALPDFDictionaryPoppler::Get(const char *pszKey) |
1275 | 23.1k | { |
1276 | 23.1k | std::map<CPLString, GDALPDFObject *>::iterator oIter = m_map.find(pszKey); |
1277 | 23.1k | if (oIter != m_map.end()) |
1278 | 5.92k | return oIter->second; |
1279 | | |
1280 | 17.1k | auto &&o(m_poDict->lookupNF(pszKey)); |
1281 | 17.1k | if (!o.isNull()) |
1282 | 11.5k | { |
1283 | 11.5k | GDALPDFObjectNum nRefNum; |
1284 | 11.5k | int nRefGen = 0; |
1285 | 11.5k | if (o.isRef()) |
1286 | 3.89k | { |
1287 | 3.89k | nRefNum = o.getRefNum(); |
1288 | 3.89k | nRefGen = o.getRefGen(); |
1289 | 3.89k | Object o2(m_poDict->lookup(pszKey)); |
1290 | 3.89k | if (!o2.isNull()) |
1291 | 3.89k | { |
1292 | 3.89k | GDALPDFObjectPoppler *poObj = |
1293 | 3.89k | new GDALPDFObjectPoppler(new Object(std::move(o2)), TRUE); |
1294 | 3.89k | poObj->SetRefNumAndGen(nRefNum, nRefGen); |
1295 | 3.89k | m_map[pszKey] = poObj; |
1296 | 3.89k | return poObj; |
1297 | 3.89k | } |
1298 | 3.89k | } |
1299 | 7.67k | else |
1300 | 7.67k | { |
1301 | 7.67k | GDALPDFObjectPoppler *poObj = |
1302 | 7.67k | new GDALPDFObjectPoppler(new Object(o.copy()), TRUE); |
1303 | 7.67k | poObj->SetRefNumAndGen(nRefNum, nRefGen); |
1304 | 7.67k | m_map[pszKey] = poObj; |
1305 | 7.67k | return poObj; |
1306 | 7.67k | } |
1307 | 11.5k | } |
1308 | 5.62k | return nullptr; |
1309 | 17.1k | } |
1310 | | |
1311 | | /************************************************************************/ |
1312 | | /* GetValues() */ |
1313 | | /************************************************************************/ |
1314 | | |
1315 | | std::map<CPLString, GDALPDFObject *> &GDALPDFDictionaryPoppler::GetValues() |
1316 | 1.33k | { |
1317 | 1.33k | int i = 0; |
1318 | 1.33k | int nLength = m_poDict->getLength(); |
1319 | 2.66k | for (i = 0; i < nLength; i++) |
1320 | 1.33k | { |
1321 | 1.33k | const char *pszKey = m_poDict->getKey(i); |
1322 | 1.33k | Get(pszKey); |
1323 | 1.33k | } |
1324 | 1.33k | return m_map; |
1325 | 1.33k | } |
1326 | | |
1327 | | /************************************************************************/ |
1328 | | /* ==================================================================== */ |
1329 | | /* GDALPDFArrayPoppler */ |
1330 | | /* ==================================================================== */ |
1331 | | /************************************************************************/ |
1332 | | |
1333 | | /************************************************************************/ |
1334 | | /* GDALPDFCreateArray() */ |
1335 | | /************************************************************************/ |
1336 | | |
1337 | | GDALPDFArray *GDALPDFCreateArray(const Array *array) |
1338 | 0 | { |
1339 | 0 | return new GDALPDFArrayPoppler(array); |
1340 | 0 | } |
1341 | | |
1342 | | /************************************************************************/ |
1343 | | /* GetLength() */ |
1344 | | /************************************************************************/ |
1345 | | |
1346 | | int GDALPDFArrayPoppler::GetLength() |
1347 | 15.0k | { |
1348 | 15.0k | return m_poArray->getLength(); |
1349 | 15.0k | } |
1350 | | |
1351 | | /************************************************************************/ |
1352 | | /* Get() */ |
1353 | | /************************************************************************/ |
1354 | | |
1355 | | GDALPDFObject *GDALPDFArrayPoppler::Get(int nIndex) |
1356 | 10.2k | { |
1357 | 10.2k | if (nIndex < 0 || nIndex >= GetLength()) |
1358 | 0 | return nullptr; |
1359 | | |
1360 | 10.2k | if (m_v.empty()) |
1361 | 1.97k | m_v.resize(GetLength()); |
1362 | | |
1363 | 10.2k | if (m_v[nIndex] != nullptr) |
1364 | 1.43k | return m_v[nIndex].get(); |
1365 | | |
1366 | 8.86k | auto &&o(m_poArray->getNF(nIndex)); |
1367 | 8.86k | if (!o.isNull()) |
1368 | 8.86k | { |
1369 | 8.86k | GDALPDFObjectNum nRefNum; |
1370 | 8.86k | int nRefGen = 0; |
1371 | 8.86k | if (o.isRef()) |
1372 | 826 | { |
1373 | 826 | nRefNum = o.getRefNum(); |
1374 | 826 | nRefGen = o.getRefGen(); |
1375 | 826 | Object o2(m_poArray->get(nIndex)); |
1376 | 826 | if (!o2.isNull()) |
1377 | 826 | { |
1378 | 826 | auto poObj = std::make_unique<GDALPDFObjectPoppler>( |
1379 | 826 | new Object(std::move(o2)), TRUE); |
1380 | 826 | poObj->SetRefNumAndGen(nRefNum, nRefGen); |
1381 | 826 | m_v[nIndex] = std::move(poObj); |
1382 | 826 | return m_v[nIndex].get(); |
1383 | 826 | } |
1384 | 826 | } |
1385 | 8.03k | else |
1386 | 8.03k | { |
1387 | 8.03k | auto poObj = std::make_unique<GDALPDFObjectPoppler>( |
1388 | 8.03k | new Object(o.copy()), TRUE); |
1389 | 8.03k | poObj->SetRefNumAndGen(nRefNum, nRefGen); |
1390 | 8.03k | m_v[nIndex] = std::move(poObj); |
1391 | 8.03k | return m_v[nIndex].get(); |
1392 | 8.03k | } |
1393 | 8.86k | } |
1394 | 0 | return nullptr; |
1395 | 8.86k | } |
1396 | | |
1397 | | /************************************************************************/ |
1398 | | /* ==================================================================== */ |
1399 | | /* GDALPDFStreamPoppler */ |
1400 | | /* ==================================================================== */ |
1401 | | /************************************************************************/ |
1402 | | |
1403 | | /************************************************************************/ |
1404 | | /* GetLength() */ |
1405 | | /************************************************************************/ |
1406 | | |
1407 | | int64_t GDALPDFStreamPoppler::GetLength(int64_t nMaxSize) |
1408 | 539 | { |
1409 | 539 | if (m_nLength >= 0) |
1410 | 0 | return m_nLength; |
1411 | | |
1412 | | #if POPPLER_MAJOR_VERSION > 25 || \ |
1413 | | (POPPLER_MAJOR_VERSION == 25 && POPPLER_MINOR_VERSION >= 2) |
1414 | | if (!m_poStream->reset()) |
1415 | | return 0; |
1416 | | #else |
1417 | 539 | m_poStream->reset(); |
1418 | 539 | #endif |
1419 | 539 | m_nLength = 0; |
1420 | 539 | unsigned char readBuf[4096]; |
1421 | 539 | int readChars; |
1422 | 1.07k | while ((readChars = m_poStream->doGetChars(4096, readBuf)) != 0) |
1423 | 539 | { |
1424 | 539 | m_nLength += readChars; |
1425 | 539 | if (nMaxSize != 0 && m_nLength > nMaxSize) |
1426 | 0 | { |
1427 | 0 | m_nLength = -1; |
1428 | 0 | return std::numeric_limits<int64_t>::max(); |
1429 | 0 | } |
1430 | 539 | } |
1431 | 539 | return m_nLength; |
1432 | 539 | } |
1433 | | |
1434 | | /************************************************************************/ |
1435 | | /* GooStringToCharStart() */ |
1436 | | /************************************************************************/ |
1437 | | |
1438 | | static char *GooStringToCharStart(GooString &gstr) |
1439 | 539 | { |
1440 | 539 | auto nLength = gstr.getLength(); |
1441 | 539 | if (nLength) |
1442 | 539 | { |
1443 | 539 | char *pszContent = static_cast<char *>(VSI_MALLOC_VERBOSE(nLength + 1)); |
1444 | 539 | if (pszContent) |
1445 | 539 | { |
1446 | 539 | const char *srcStr = gstr.c_str(); |
1447 | 539 | memcpy(pszContent, srcStr, nLength); |
1448 | 539 | pszContent[nLength] = '\0'; |
1449 | 539 | } |
1450 | 539 | return pszContent; |
1451 | 539 | } |
1452 | 0 | return nullptr; |
1453 | 539 | } |
1454 | | |
1455 | | /************************************************************************/ |
1456 | | /* GetBytes() */ |
1457 | | /************************************************************************/ |
1458 | | |
1459 | | char *GDALPDFStreamPoppler::GetBytes() |
1460 | 539 | { |
1461 | 539 | GooString gstr; |
1462 | 539 | try |
1463 | 539 | { |
1464 | 539 | m_poStream->fillGooString(&gstr); |
1465 | 539 | } |
1466 | 539 | catch (const std::exception &e) |
1467 | 539 | { |
1468 | 0 | CPLError(CE_Failure, CPLE_OutOfMemory, |
1469 | 0 | "GDALPDFStreamPoppler::GetBytes(): %s", e.what()); |
1470 | 0 | return nullptr; |
1471 | 0 | } |
1472 | 539 | m_nLength = static_cast<int64_t>(gstr.toStr().size()); |
1473 | 539 | return GooStringToCharStart(gstr); |
1474 | 539 | } |
1475 | | |
1476 | | /************************************************************************/ |
1477 | | /* GetRawLength() */ |
1478 | | /************************************************************************/ |
1479 | | |
1480 | | int64_t GDALPDFStreamPoppler::GetRawLength() |
1481 | 0 | { |
1482 | 0 | if (m_nRawLength >= 0) |
1483 | 0 | return m_nRawLength; |
1484 | | |
1485 | 0 | auto undecodeStream = m_poStream->getUndecodedStream(); |
1486 | | #if POPPLER_MAJOR_VERSION > 25 || \ |
1487 | | (POPPLER_MAJOR_VERSION == 25 && POPPLER_MINOR_VERSION >= 2) |
1488 | | if (!undecodeStream->reset()) |
1489 | | return 0; |
1490 | | #else |
1491 | 0 | undecodeStream->reset(); |
1492 | 0 | #endif |
1493 | 0 | m_nRawLength = 0; |
1494 | 0 | while (undecodeStream->getChar() != EOF) |
1495 | 0 | m_nRawLength++; |
1496 | 0 | return m_nRawLength; |
1497 | 0 | } |
1498 | | |
1499 | | /************************************************************************/ |
1500 | | /* GetRawBytes() */ |
1501 | | /************************************************************************/ |
1502 | | |
1503 | | char *GDALPDFStreamPoppler::GetRawBytes() |
1504 | 0 | { |
1505 | 0 | GooString gstr; |
1506 | 0 | auto undecodeStream = m_poStream->getUndecodedStream(); |
1507 | 0 | try |
1508 | 0 | { |
1509 | 0 | undecodeStream->fillGooString(&gstr); |
1510 | 0 | } |
1511 | 0 | catch (const std::exception &e) |
1512 | 0 | { |
1513 | 0 | CPLError(CE_Failure, CPLE_OutOfMemory, |
1514 | 0 | "GDALPDFStreamPoppler::GetRawBytes(): %s", e.what()); |
1515 | 0 | return nullptr; |
1516 | 0 | } |
1517 | 0 | m_nRawLength = gstr.getLength(); |
1518 | 0 | return GooStringToCharStart(gstr); |
1519 | 0 | } |
1520 | | |
1521 | | #endif // HAVE_POPPLER |
1522 | | |
1523 | | #ifdef HAVE_PODOFO |
1524 | | |
1525 | | /************************************************************************/ |
1526 | | /* ==================================================================== */ |
1527 | | /* GDALPDFDictionaryPodofo */ |
1528 | | /* ==================================================================== */ |
1529 | | /************************************************************************/ |
1530 | | |
1531 | | class GDALPDFDictionaryPodofo : public GDALPDFDictionary |
1532 | | { |
1533 | | private: |
1534 | | const PoDoFo::PdfDictionary *m_poDict; |
1535 | | const PoDoFo::PdfVecObjects &m_poObjects; |
1536 | | std::map<CPLString, GDALPDFObject *> m_map{}; |
1537 | | |
1538 | | CPL_DISALLOW_COPY_ASSIGN(GDALPDFDictionaryPodofo) |
1539 | | |
1540 | | public: |
1541 | | GDALPDFDictionaryPodofo(const PoDoFo::PdfDictionary *poDict, |
1542 | | const PoDoFo::PdfVecObjects &poObjects) |
1543 | | : m_poDict(poDict), m_poObjects(poObjects) |
1544 | | { |
1545 | | } |
1546 | | |
1547 | | virtual ~GDALPDFDictionaryPodofo(); |
1548 | | |
1549 | | virtual GDALPDFObject *Get(const char *pszKey) override; |
1550 | | virtual std::map<CPLString, GDALPDFObject *> &GetValues() override; |
1551 | | }; |
1552 | | |
1553 | | /************************************************************************/ |
1554 | | /* ==================================================================== */ |
1555 | | /* GDALPDFArrayPodofo */ |
1556 | | /* ==================================================================== */ |
1557 | | /************************************************************************/ |
1558 | | |
1559 | | class GDALPDFArrayPodofo : public GDALPDFArray |
1560 | | { |
1561 | | private: |
1562 | | const PoDoFo::PdfArray *m_poArray; |
1563 | | const PoDoFo::PdfVecObjects &m_poObjects; |
1564 | | std::vector<std::unique_ptr<GDALPDFObject>> m_v{}; |
1565 | | |
1566 | | CPL_DISALLOW_COPY_ASSIGN(GDALPDFArrayPodofo) |
1567 | | |
1568 | | public: |
1569 | | GDALPDFArrayPodofo(const PoDoFo::PdfArray *poArray, |
1570 | | const PoDoFo::PdfVecObjects &poObjects) |
1571 | | : m_poArray(poArray), m_poObjects(poObjects) |
1572 | | { |
1573 | | } |
1574 | | |
1575 | | virtual int GetLength() override; |
1576 | | virtual GDALPDFObject *Get(int nIndex) override; |
1577 | | }; |
1578 | | |
1579 | | /************************************************************************/ |
1580 | | /* ==================================================================== */ |
1581 | | /* GDALPDFStreamPodofo */ |
1582 | | /* ==================================================================== */ |
1583 | | /************************************************************************/ |
1584 | | |
1585 | | class GDALPDFStreamPodofo : public GDALPDFStream |
1586 | | { |
1587 | | private: |
1588 | | #if PODOFO_VERSION_MAJOR > 0 || \ |
1589 | | (PODOFO_VERSION_MAJOR == 0 && PODOFO_VERSION_MINOR >= 10) |
1590 | | const PoDoFo::PdfObjectStream *m_pStream; |
1591 | | #else |
1592 | | const PoDoFo::PdfStream *m_pStream; |
1593 | | #endif |
1594 | | |
1595 | | CPL_DISALLOW_COPY_ASSIGN(GDALPDFStreamPodofo) |
1596 | | |
1597 | | public: |
1598 | | GDALPDFStreamPodofo( |
1599 | | #if PODOFO_VERSION_MAJOR > 0 || \ |
1600 | | (PODOFO_VERSION_MAJOR == 0 && PODOFO_VERSION_MINOR >= 10) |
1601 | | const PoDoFo::PdfObjectStream * |
1602 | | #else |
1603 | | const PoDoFo::PdfStream * |
1604 | | #endif |
1605 | | pStream) |
1606 | | : m_pStream(pStream) |
1607 | | { |
1608 | | } |
1609 | | |
1610 | | virtual ~GDALPDFStreamPodofo() |
1611 | | { |
1612 | | } |
1613 | | |
1614 | | virtual int64_t GetLength(int64_t nMaxSize = 0) override; |
1615 | | virtual char *GetBytes() override; |
1616 | | |
1617 | | virtual int64_t GetRawLength() override; |
1618 | | virtual char *GetRawBytes() override; |
1619 | | }; |
1620 | | |
1621 | | /************************************************************************/ |
1622 | | /* ==================================================================== */ |
1623 | | /* GDALPDFObjectPodofo */ |
1624 | | /* ==================================================================== */ |
1625 | | /************************************************************************/ |
1626 | | |
1627 | | /************************************************************************/ |
1628 | | /* GDALPDFObjectPodofo() */ |
1629 | | /************************************************************************/ |
1630 | | |
1631 | | GDALPDFObjectPodofo::GDALPDFObjectPodofo(const PoDoFo::PdfObject *po, |
1632 | | const PoDoFo::PdfVecObjects &poObjects) |
1633 | | : m_po(po), m_poObjects(poObjects) |
1634 | | { |
1635 | | try |
1636 | | { |
1637 | | #if PODOFO_VERSION_MAJOR > 0 || \ |
1638 | | (PODOFO_VERSION_MAJOR == 0 && PODOFO_VERSION_MINOR >= 10) |
1639 | | if (m_po->GetDataType() == PoDoFo::PdfDataType::Reference) |
1640 | | { |
1641 | | PoDoFo::PdfObject *poObj = |
1642 | | m_poObjects.GetObject(m_po->GetReference()); |
1643 | | if (poObj) |
1644 | | m_po = poObj; |
1645 | | } |
1646 | | #else |
1647 | | if (m_po->GetDataType() == PoDoFo::ePdfDataType_Reference) |
1648 | | { |
1649 | | PoDoFo::PdfObject *poObj = |
1650 | | m_poObjects.GetObject(m_po->GetReference()); |
1651 | | if (poObj) |
1652 | | m_po = poObj; |
1653 | | } |
1654 | | #endif |
1655 | | } |
1656 | | catch (PoDoFo::PdfError &oError) |
1657 | | { |
1658 | | CPLError(CE_Failure, CPLE_AppDefined, "Invalid PDF : %s", |
1659 | | oError.what()); |
1660 | | } |
1661 | | } |
1662 | | |
1663 | | /************************************************************************/ |
1664 | | /* ~GDALPDFObjectPodofo() */ |
1665 | | /************************************************************************/ |
1666 | | |
1667 | | GDALPDFObjectPodofo::~GDALPDFObjectPodofo() |
1668 | | { |
1669 | | delete m_poDict; |
1670 | | delete m_poArray; |
1671 | | delete m_poStream; |
1672 | | } |
1673 | | |
1674 | | /************************************************************************/ |
1675 | | /* GetType() */ |
1676 | | /************************************************************************/ |
1677 | | |
1678 | | GDALPDFObjectType GDALPDFObjectPodofo::GetType() |
1679 | | { |
1680 | | try |
1681 | | { |
1682 | | switch (m_po->GetDataType()) |
1683 | | { |
1684 | | #if PODOFO_VERSION_MAJOR > 0 || \ |
1685 | | (PODOFO_VERSION_MAJOR == 0 && PODOFO_VERSION_MINOR >= 10) |
1686 | | case PoDoFo::PdfDataType::Null: |
1687 | | return PDFObjectType_Null; |
1688 | | case PoDoFo::PdfDataType::Bool: |
1689 | | return PDFObjectType_Bool; |
1690 | | case PoDoFo::PdfDataType::Number: |
1691 | | return PDFObjectType_Int; |
1692 | | case PoDoFo::PdfDataType::Real: |
1693 | | return PDFObjectType_Real; |
1694 | | case PoDoFo::PdfDataType::String: |
1695 | | return PDFObjectType_String; |
1696 | | case PoDoFo::PdfDataType::Name: |
1697 | | return PDFObjectType_Name; |
1698 | | case PoDoFo::PdfDataType::Array: |
1699 | | return PDFObjectType_Array; |
1700 | | case PoDoFo::PdfDataType::Dictionary: |
1701 | | return PDFObjectType_Dictionary; |
1702 | | default: |
1703 | | return PDFObjectType_Unknown; |
1704 | | #else |
1705 | | case PoDoFo::ePdfDataType_Null: |
1706 | | return PDFObjectType_Null; |
1707 | | case PoDoFo::ePdfDataType_Bool: |
1708 | | return PDFObjectType_Bool; |
1709 | | case PoDoFo::ePdfDataType_Number: |
1710 | | return PDFObjectType_Int; |
1711 | | case PoDoFo::ePdfDataType_Real: |
1712 | | return PDFObjectType_Real; |
1713 | | case PoDoFo::ePdfDataType_HexString: |
1714 | | return PDFObjectType_String; |
1715 | | case PoDoFo::ePdfDataType_String: |
1716 | | return PDFObjectType_String; |
1717 | | case PoDoFo::ePdfDataType_Name: |
1718 | | return PDFObjectType_Name; |
1719 | | case PoDoFo::ePdfDataType_Array: |
1720 | | return PDFObjectType_Array; |
1721 | | case PoDoFo::ePdfDataType_Dictionary: |
1722 | | return PDFObjectType_Dictionary; |
1723 | | default: |
1724 | | return PDFObjectType_Unknown; |
1725 | | #endif |
1726 | | } |
1727 | | } |
1728 | | catch (PoDoFo::PdfError &oError) |
1729 | | { |
1730 | | CPLError(CE_Failure, CPLE_AppDefined, "Invalid PDF : %s", |
1731 | | oError.what()); |
1732 | | return PDFObjectType_Unknown; |
1733 | | } |
1734 | | } |
1735 | | |
1736 | | /************************************************************************/ |
1737 | | /* GetTypeNameNative() */ |
1738 | | /************************************************************************/ |
1739 | | |
1740 | | const char *GDALPDFObjectPodofo::GetTypeNameNative() |
1741 | | { |
1742 | | try |
1743 | | { |
1744 | | return m_po->GetDataTypeString(); |
1745 | | } |
1746 | | catch (PoDoFo::PdfError &oError) |
1747 | | { |
1748 | | CPLError(CE_Failure, CPLE_AppDefined, "Invalid PDF : %s", |
1749 | | oError.what()); |
1750 | | return "unknown"; |
1751 | | } |
1752 | | } |
1753 | | |
1754 | | /************************************************************************/ |
1755 | | /* GetBool() */ |
1756 | | /************************************************************************/ |
1757 | | |
1758 | | int GDALPDFObjectPodofo::GetBool() |
1759 | | { |
1760 | | #if PODOFO_VERSION_MAJOR > 0 || \ |
1761 | | (PODOFO_VERSION_MAJOR == 0 && PODOFO_VERSION_MINOR >= 10) |
1762 | | if (m_po->GetDataType() == PoDoFo::PdfDataType::Bool) |
1763 | | #else |
1764 | | if (m_po->GetDataType() == PoDoFo::ePdfDataType_Bool) |
1765 | | #endif |
1766 | | return m_po->GetBool(); |
1767 | | else |
1768 | | return 0; |
1769 | | } |
1770 | | |
1771 | | /************************************************************************/ |
1772 | | /* GetInt() */ |
1773 | | /************************************************************************/ |
1774 | | |
1775 | | int GDALPDFObjectPodofo::GetInt() |
1776 | | { |
1777 | | #if PODOFO_VERSION_MAJOR > 0 || \ |
1778 | | (PODOFO_VERSION_MAJOR == 0 && PODOFO_VERSION_MINOR >= 10) |
1779 | | if (m_po->GetDataType() == PoDoFo::PdfDataType::Number) |
1780 | | #else |
1781 | | if (m_po->GetDataType() == PoDoFo::ePdfDataType_Number) |
1782 | | #endif |
1783 | | return static_cast<int>(m_po->GetNumber()); |
1784 | | else |
1785 | | return 0; |
1786 | | } |
1787 | | |
1788 | | /************************************************************************/ |
1789 | | /* GetReal() */ |
1790 | | /************************************************************************/ |
1791 | | |
1792 | | double GDALPDFObjectPodofo::GetReal() |
1793 | | { |
1794 | | if (GetType() == PDFObjectType_Real) |
1795 | | return m_po->GetReal(); |
1796 | | else |
1797 | | return 0.0; |
1798 | | } |
1799 | | |
1800 | | /************************************************************************/ |
1801 | | /* GetString() */ |
1802 | | /************************************************************************/ |
1803 | | |
1804 | | const std::string &GDALPDFObjectPodofo::GetString() |
1805 | | { |
1806 | | if (GetType() == PDFObjectType_String) |
1807 | | #if PODOFO_VERSION_MAJOR > 0 || \ |
1808 | | (PODOFO_VERSION_MAJOR == 0 && PODOFO_VERSION_MINOR >= 10) |
1809 | | return (osStr = m_po->GetString().GetString()); |
1810 | | #else |
1811 | | return (osStr = m_po->GetString().GetStringUtf8()); |
1812 | | #endif |
1813 | | else |
1814 | | return (osStr = ""); |
1815 | | } |
1816 | | |
1817 | | /************************************************************************/ |
1818 | | /* GetName() */ |
1819 | | /************************************************************************/ |
1820 | | |
1821 | | const std::string &GDALPDFObjectPodofo::GetName() |
1822 | | { |
1823 | | if (GetType() == PDFObjectType_Name) |
1824 | | #if PODOFO_VERSION_MAJOR > 0 || \ |
1825 | | (PODOFO_VERSION_MAJOR == 0 && PODOFO_VERSION_MINOR >= 10) |
1826 | | return (osStr = m_po->GetName().GetString()); |
1827 | | #else |
1828 | | return (osStr = m_po->GetName().GetName()); |
1829 | | #endif |
1830 | | else |
1831 | | return (osStr = ""); |
1832 | | } |
1833 | | |
1834 | | /************************************************************************/ |
1835 | | /* GetDictionary() */ |
1836 | | /************************************************************************/ |
1837 | | |
1838 | | GDALPDFDictionary *GDALPDFObjectPodofo::GetDictionary() |
1839 | | { |
1840 | | if (GetType() != PDFObjectType_Dictionary) |
1841 | | return nullptr; |
1842 | | |
1843 | | if (m_poDict) |
1844 | | return m_poDict; |
1845 | | |
1846 | | m_poDict = new GDALPDFDictionaryPodofo(&m_po->GetDictionary(), m_poObjects); |
1847 | | return m_poDict; |
1848 | | } |
1849 | | |
1850 | | /************************************************************************/ |
1851 | | /* GetArray() */ |
1852 | | /************************************************************************/ |
1853 | | |
1854 | | GDALPDFArray *GDALPDFObjectPodofo::GetArray() |
1855 | | { |
1856 | | if (GetType() != PDFObjectType_Array) |
1857 | | return nullptr; |
1858 | | |
1859 | | if (m_poArray) |
1860 | | return m_poArray; |
1861 | | |
1862 | | m_poArray = new GDALPDFArrayPodofo(&m_po->GetArray(), m_poObjects); |
1863 | | return m_poArray; |
1864 | | } |
1865 | | |
1866 | | /************************************************************************/ |
1867 | | /* GetStream() */ |
1868 | | /************************************************************************/ |
1869 | | |
1870 | | GDALPDFStream *GDALPDFObjectPodofo::GetStream() |
1871 | | { |
1872 | | try |
1873 | | { |
1874 | | if (!m_po->HasStream()) |
1875 | | return nullptr; |
1876 | | } |
1877 | | catch (PoDoFo::PdfError &oError) |
1878 | | { |
1879 | | CPLError(CE_Failure, CPLE_AppDefined, "Invalid object : %s", |
1880 | | oError.what()); |
1881 | | return nullptr; |
1882 | | } |
1883 | | catch (...) |
1884 | | { |
1885 | | CPLError(CE_Failure, CPLE_AppDefined, "Invalid object"); |
1886 | | return nullptr; |
1887 | | } |
1888 | | |
1889 | | if (m_poStream == nullptr) |
1890 | | m_poStream = new GDALPDFStreamPodofo(m_po->GetStream()); |
1891 | | return m_poStream; |
1892 | | } |
1893 | | |
1894 | | /************************************************************************/ |
1895 | | /* GetRefNum() */ |
1896 | | /************************************************************************/ |
1897 | | |
1898 | | GDALPDFObjectNum GDALPDFObjectPodofo::GetRefNum() |
1899 | | { |
1900 | | #if PODOFO_VERSION_MAJOR > 0 || \ |
1901 | | (PODOFO_VERSION_MAJOR == 0 && PODOFO_VERSION_MINOR >= 10) |
1902 | | return GDALPDFObjectNum(m_po->GetIndirectReference().ObjectNumber()); |
1903 | | #else |
1904 | | return GDALPDFObjectNum(m_po->Reference().ObjectNumber()); |
1905 | | #endif |
1906 | | } |
1907 | | |
1908 | | /************************************************************************/ |
1909 | | /* GetRefGen() */ |
1910 | | /************************************************************************/ |
1911 | | |
1912 | | int GDALPDFObjectPodofo::GetRefGen() |
1913 | | { |
1914 | | #if PODOFO_VERSION_MAJOR > 0 || \ |
1915 | | (PODOFO_VERSION_MAJOR == 0 && PODOFO_VERSION_MINOR >= 10) |
1916 | | return m_po->GetIndirectReference().GenerationNumber(); |
1917 | | #else |
1918 | | return m_po->Reference().GenerationNumber(); |
1919 | | #endif |
1920 | | } |
1921 | | |
1922 | | /************************************************************************/ |
1923 | | /* ==================================================================== */ |
1924 | | /* GDALPDFDictionaryPodofo */ |
1925 | | /* ==================================================================== */ |
1926 | | /************************************************************************/ |
1927 | | |
1928 | | /************************************************************************/ |
1929 | | /* ~GDALPDFDictionaryPodofo() */ |
1930 | | /************************************************************************/ |
1931 | | |
1932 | | GDALPDFDictionaryPodofo::~GDALPDFDictionaryPodofo() |
1933 | | { |
1934 | | std::map<CPLString, GDALPDFObject *>::iterator oIter = m_map.begin(); |
1935 | | std::map<CPLString, GDALPDFObject *>::iterator oEnd = m_map.end(); |
1936 | | for (; oIter != oEnd; ++oIter) |
1937 | | delete oIter->second; |
1938 | | } |
1939 | | |
1940 | | /************************************************************************/ |
1941 | | /* Get() */ |
1942 | | /************************************************************************/ |
1943 | | |
1944 | | GDALPDFObject *GDALPDFDictionaryPodofo::Get(const char *pszKey) |
1945 | | { |
1946 | | std::map<CPLString, GDALPDFObject *>::iterator oIter = m_map.find(pszKey); |
1947 | | if (oIter != m_map.end()) |
1948 | | return oIter->second; |
1949 | | |
1950 | | const PoDoFo::PdfObject *poVal = m_poDict->GetKey(PoDoFo::PdfName(pszKey)); |
1951 | | if (poVal) |
1952 | | { |
1953 | | GDALPDFObjectPodofo *poObj = |
1954 | | new GDALPDFObjectPodofo(poVal, m_poObjects); |
1955 | | m_map[pszKey] = poObj; |
1956 | | return poObj; |
1957 | | } |
1958 | | else |
1959 | | { |
1960 | | return nullptr; |
1961 | | } |
1962 | | } |
1963 | | |
1964 | | /************************************************************************/ |
1965 | | /* GetValues() */ |
1966 | | /************************************************************************/ |
1967 | | |
1968 | | std::map<CPLString, GDALPDFObject *> &GDALPDFDictionaryPodofo::GetValues() |
1969 | | { |
1970 | | #if PODOFO_VERSION_MAJOR > 0 || \ |
1971 | | (PODOFO_VERSION_MAJOR == 0 && PODOFO_VERSION_MINOR >= 10) |
1972 | | for (const auto &oIter : *m_poDict) |
1973 | | { |
1974 | | Get(oIter.first.GetString().c_str()); |
1975 | | } |
1976 | | #else |
1977 | | for (const auto &oIter : m_poDict->GetKeys()) |
1978 | | { |
1979 | | Get(oIter.first.GetName().c_str()); |
1980 | | } |
1981 | | #endif |
1982 | | return m_map; |
1983 | | } |
1984 | | |
1985 | | /************************************************************************/ |
1986 | | /* ==================================================================== */ |
1987 | | /* GDALPDFArrayPodofo */ |
1988 | | /* ==================================================================== */ |
1989 | | /************************************************************************/ |
1990 | | |
1991 | | /************************************************************************/ |
1992 | | /* GetLength() */ |
1993 | | /************************************************************************/ |
1994 | | |
1995 | | int GDALPDFArrayPodofo::GetLength() |
1996 | | { |
1997 | | return static_cast<int>(m_poArray->GetSize()); |
1998 | | } |
1999 | | |
2000 | | /************************************************************************/ |
2001 | | /* Get() */ |
2002 | | /************************************************************************/ |
2003 | | |
2004 | | GDALPDFObject *GDALPDFArrayPodofo::Get(int nIndex) |
2005 | | { |
2006 | | if (nIndex < 0 || nIndex >= GetLength()) |
2007 | | return nullptr; |
2008 | | |
2009 | | if (m_v.empty()) |
2010 | | m_v.resize(GetLength()); |
2011 | | |
2012 | | if (m_v[nIndex] != nullptr) |
2013 | | return m_v[nIndex].get(); |
2014 | | |
2015 | | const PoDoFo::PdfObject &oVal = (*m_poArray)[nIndex]; |
2016 | | m_v[nIndex] = std::make_unique<GDALPDFObjectPodofo>(&oVal, m_poObjects); |
2017 | | return m_v[nIndex].get(); |
2018 | | } |
2019 | | |
2020 | | /************************************************************************/ |
2021 | | /* ==================================================================== */ |
2022 | | /* GDALPDFStreamPodofo */ |
2023 | | /* ==================================================================== */ |
2024 | | /************************************************************************/ |
2025 | | |
2026 | | /************************************************************************/ |
2027 | | /* GetLength() */ |
2028 | | /************************************************************************/ |
2029 | | |
2030 | | int64_t GDALPDFStreamPodofo::GetLength(int64_t /* nMaxSize */) |
2031 | | { |
2032 | | #if PODOFO_VERSION_MAJOR > 0 || \ |
2033 | | (PODOFO_VERSION_MAJOR == 0 && PODOFO_VERSION_MINOR >= 10) |
2034 | | PoDoFo::charbuff str; |
2035 | | try |
2036 | | { |
2037 | | m_pStream->CopyToSafe(str); |
2038 | | } |
2039 | | catch (PoDoFo::PdfError &e) |
2040 | | { |
2041 | | CPLError(CE_Failure, CPLE_AppDefined, "CopyToSafe() failed: %s", |
2042 | | e.what()); |
2043 | | return 0; |
2044 | | } |
2045 | | return static_cast<int64_t>(str.size()); |
2046 | | #else |
2047 | | char *pBuffer = nullptr; |
2048 | | PoDoFo::pdf_long nLen = 0; |
2049 | | try |
2050 | | { |
2051 | | m_pStream->GetFilteredCopy(&pBuffer, &nLen); |
2052 | | PoDoFo::podofo_free(pBuffer); |
2053 | | return static_cast<int64_t>(nLen); |
2054 | | } |
2055 | | catch (PoDoFo::PdfError &e) |
2056 | | { |
2057 | | } |
2058 | | return 0; |
2059 | | #endif |
2060 | | } |
2061 | | |
2062 | | /************************************************************************/ |
2063 | | /* GetBytes() */ |
2064 | | /************************************************************************/ |
2065 | | |
2066 | | char *GDALPDFStreamPodofo::GetBytes() |
2067 | | { |
2068 | | #if PODOFO_VERSION_MAJOR > 0 || \ |
2069 | | (PODOFO_VERSION_MAJOR == 0 && PODOFO_VERSION_MINOR >= 10) |
2070 | | PoDoFo::charbuff str; |
2071 | | try |
2072 | | { |
2073 | | m_pStream->CopyToSafe(str); |
2074 | | } |
2075 | | catch (PoDoFo::PdfError &e) |
2076 | | { |
2077 | | CPLError(CE_Failure, CPLE_AppDefined, "CopyToSafe() failed: %s", |
2078 | | e.what()); |
2079 | | return nullptr; |
2080 | | } |
2081 | | char *pszContent = static_cast<char *>(VSI_MALLOC_VERBOSE(str.size() + 1)); |
2082 | | if (!pszContent) |
2083 | | { |
2084 | | return nullptr; |
2085 | | } |
2086 | | memcpy(pszContent, str.data(), str.size()); |
2087 | | pszContent[str.size()] = '\0'; |
2088 | | return pszContent; |
2089 | | #else |
2090 | | char *pBuffer = nullptr; |
2091 | | PoDoFo::pdf_long nLen = 0; |
2092 | | try |
2093 | | { |
2094 | | m_pStream->GetFilteredCopy(&pBuffer, &nLen); |
2095 | | } |
2096 | | catch (PoDoFo::PdfError &e) |
2097 | | { |
2098 | | return nullptr; |
2099 | | } |
2100 | | char *pszContent = static_cast<char *>(VSI_MALLOC_VERBOSE(nLen + 1)); |
2101 | | if (!pszContent) |
2102 | | { |
2103 | | PoDoFo::podofo_free(pBuffer); |
2104 | | return nullptr; |
2105 | | } |
2106 | | memcpy(pszContent, pBuffer, nLen); |
2107 | | PoDoFo::podofo_free(pBuffer); |
2108 | | pszContent[nLen] = '\0'; |
2109 | | return pszContent; |
2110 | | #endif |
2111 | | } |
2112 | | |
2113 | | /************************************************************************/ |
2114 | | /* GetRawLength() */ |
2115 | | /************************************************************************/ |
2116 | | |
2117 | | int64_t GDALPDFStreamPodofo::GetRawLength() |
2118 | | { |
2119 | | try |
2120 | | { |
2121 | | auto nLen = m_pStream->GetLength(); |
2122 | | return static_cast<int64_t>(nLen); |
2123 | | } |
2124 | | catch (PoDoFo::PdfError &e) |
2125 | | { |
2126 | | } |
2127 | | return 0; |
2128 | | } |
2129 | | |
2130 | | /************************************************************************/ |
2131 | | /* GetRawBytes() */ |
2132 | | /************************************************************************/ |
2133 | | |
2134 | | char *GDALPDFStreamPodofo::GetRawBytes() |
2135 | | { |
2136 | | #if PODOFO_VERSION_MAJOR > 0 || \ |
2137 | | (PODOFO_VERSION_MAJOR == 0 && PODOFO_VERSION_MINOR >= 10) |
2138 | | PoDoFo::charbuff str; |
2139 | | try |
2140 | | { |
2141 | | PoDoFo::StringStreamDevice stream(str); |
2142 | | #ifdef USE_HACK_BECAUSE_PdfInputStream_constructor_is_not_exported_in_podofo_0_11 |
2143 | | auto *poNonConstStream = |
2144 | | const_cast<PoDoFo::PdfObjectStream *>(m_pStream); |
2145 | | auto inputStream = poNonConstStream->GetProvider().GetInputStream( |
2146 | | poNonConstStream->GetParent()); |
2147 | | inputStream->CopyTo(stream); |
2148 | | #else |
2149 | | // Should work but fails to link because PdfInputStream destructor |
2150 | | // is not exported |
2151 | | auto inputStream = m_pStream->GetInputStream(/*raw=*/true); |
2152 | | inputStream.CopyTo(stream); |
2153 | | #endif |
2154 | | stream.Flush(); |
2155 | | } |
2156 | | catch (PoDoFo::PdfError &e) |
2157 | | { |
2158 | | CPLError(CE_Failure, CPLE_AppDefined, "CopyToSafe() failed: %s", |
2159 | | e.what()); |
2160 | | return nullptr; |
2161 | | } |
2162 | | char *pszContent = static_cast<char *>(VSI_MALLOC_VERBOSE(str.size() + 1)); |
2163 | | if (!pszContent) |
2164 | | { |
2165 | | return nullptr; |
2166 | | } |
2167 | | memcpy(pszContent, str.data(), str.size()); |
2168 | | pszContent[str.size()] = '\0'; |
2169 | | return pszContent; |
2170 | | #else |
2171 | | char *pBuffer = nullptr; |
2172 | | PoDoFo::pdf_long nLen = 0; |
2173 | | try |
2174 | | { |
2175 | | m_pStream->GetCopy(&pBuffer, &nLen); |
2176 | | } |
2177 | | catch (PoDoFo::PdfError &e) |
2178 | | { |
2179 | | return nullptr; |
2180 | | } |
2181 | | char *pszContent = static_cast<char *>(VSI_MALLOC_VERBOSE(nLen + 1)); |
2182 | | if (!pszContent) |
2183 | | { |
2184 | | PoDoFo::podofo_free(pBuffer); |
2185 | | return nullptr; |
2186 | | } |
2187 | | memcpy(pszContent, pBuffer, nLen); |
2188 | | PoDoFo::podofo_free(pBuffer); |
2189 | | pszContent[nLen] = '\0'; |
2190 | | return pszContent; |
2191 | | #endif |
2192 | | } |
2193 | | |
2194 | | #endif // HAVE_PODOFO |
2195 | | |
2196 | | #ifdef HAVE_PDFIUM |
2197 | | |
2198 | | /************************************************************************/ |
2199 | | /* ==================================================================== */ |
2200 | | /* GDALPDFDictionaryPdfium */ |
2201 | | /* ==================================================================== */ |
2202 | | /************************************************************************/ |
2203 | | |
2204 | | class GDALPDFDictionaryPdfium : public GDALPDFDictionary |
2205 | | { |
2206 | | private: |
2207 | | RetainPtr<const CPDF_Dictionary> m_poDict; |
2208 | | std::map<CPLString, GDALPDFObject *> m_map{}; |
2209 | | |
2210 | | public: |
2211 | | GDALPDFDictionaryPdfium(RetainPtr<const CPDF_Dictionary> poDict) |
2212 | | : m_poDict(std::move(poDict)) |
2213 | | { |
2214 | | } |
2215 | | |
2216 | | virtual ~GDALPDFDictionaryPdfium(); |
2217 | | |
2218 | | virtual GDALPDFObject *Get(const char *pszKey) override; |
2219 | | virtual std::map<CPLString, GDALPDFObject *> &GetValues() override; |
2220 | | }; |
2221 | | |
2222 | | /************************************************************************/ |
2223 | | /* ==================================================================== */ |
2224 | | /* GDALPDFArrayPdfium */ |
2225 | | /* ==================================================================== */ |
2226 | | /************************************************************************/ |
2227 | | |
2228 | | class GDALPDFArrayPdfium : public GDALPDFArray |
2229 | | { |
2230 | | private: |
2231 | | const CPDF_Array *m_poArray; |
2232 | | std::vector<std::unique_ptr<GDALPDFObject>> m_v{}; |
2233 | | |
2234 | | CPL_DISALLOW_COPY_ASSIGN(GDALPDFArrayPdfium) |
2235 | | |
2236 | | public: |
2237 | | GDALPDFArrayPdfium(const CPDF_Array *poArray) : m_poArray(poArray) |
2238 | | { |
2239 | | } |
2240 | | |
2241 | | virtual int GetLength() override; |
2242 | | virtual GDALPDFObject *Get(int nIndex) override; |
2243 | | }; |
2244 | | |
2245 | | /************************************************************************/ |
2246 | | /* ==================================================================== */ |
2247 | | /* GDALPDFStreamPdfium */ |
2248 | | /* ==================================================================== */ |
2249 | | /************************************************************************/ |
2250 | | |
2251 | | class GDALPDFStreamPdfium : public GDALPDFStream |
2252 | | { |
2253 | | private: |
2254 | | RetainPtr<const CPDF_Stream> m_pStream; |
2255 | | int64_t m_nSize = 0; |
2256 | | std::unique_ptr<uint8_t, VSIFreeReleaser> m_pData = nullptr; |
2257 | | int64_t m_nRawSize = 0; |
2258 | | std::unique_ptr<uint8_t, VSIFreeReleaser> m_pRawData = nullptr; |
2259 | | |
2260 | | void Decompress(); |
2261 | | void FillRaw(); |
2262 | | |
2263 | | public: |
2264 | | GDALPDFStreamPdfium(RetainPtr<const CPDF_Stream> pStream) |
2265 | | : m_pStream(std::move(pStream)) |
2266 | | { |
2267 | | } |
2268 | | |
2269 | | virtual int64_t GetLength(int64_t nMaxSize = 0) override; |
2270 | | virtual char *GetBytes() override; |
2271 | | |
2272 | | virtual int64_t GetRawLength() override; |
2273 | | virtual char *GetRawBytes() override; |
2274 | | }; |
2275 | | |
2276 | | /************************************************************************/ |
2277 | | /* ==================================================================== */ |
2278 | | /* GDALPDFObjectPdfium */ |
2279 | | /* ==================================================================== */ |
2280 | | /************************************************************************/ |
2281 | | |
2282 | | /************************************************************************/ |
2283 | | /* GDALPDFObjectPdfium() */ |
2284 | | /************************************************************************/ |
2285 | | |
2286 | | GDALPDFObjectPdfium::GDALPDFObjectPdfium(RetainPtr<const CPDF_Object> obj) |
2287 | | : m_obj(std::move(obj)) |
2288 | | { |
2289 | | CPLAssert(m_obj != nullptr); |
2290 | | } |
2291 | | |
2292 | | /************************************************************************/ |
2293 | | /* ~GDALPDFObjectPdfium() */ |
2294 | | /************************************************************************/ |
2295 | | |
2296 | | GDALPDFObjectPdfium::~GDALPDFObjectPdfium() |
2297 | | { |
2298 | | delete m_poDict; |
2299 | | delete m_poArray; |
2300 | | delete m_poStream; |
2301 | | } |
2302 | | |
2303 | | /************************************************************************/ |
2304 | | /* Build() */ |
2305 | | /************************************************************************/ |
2306 | | |
2307 | | GDALPDFObjectPdfium * |
2308 | | GDALPDFObjectPdfium::Build(RetainPtr<const CPDF_Object> obj) |
2309 | | { |
2310 | | if (obj == nullptr) |
2311 | | return nullptr; |
2312 | | if (obj->GetType() == CPDF_Object::Type::kReference) |
2313 | | { |
2314 | | obj = obj->GetDirect(); |
2315 | | if (obj == nullptr) |
2316 | | { |
2317 | | CPLError(CE_Failure, CPLE_AppDefined, |
2318 | | "Cannot resolve indirect object"); |
2319 | | return nullptr; |
2320 | | } |
2321 | | } |
2322 | | return new GDALPDFObjectPdfium(std::move(obj)); |
2323 | | } |
2324 | | |
2325 | | /************************************************************************/ |
2326 | | /* GetType() */ |
2327 | | /************************************************************************/ |
2328 | | |
2329 | | GDALPDFObjectType GDALPDFObjectPdfium::GetType() |
2330 | | { |
2331 | | switch (m_obj->GetType()) |
2332 | | { |
2333 | | case CPDF_Object::Type::kNullobj: |
2334 | | return PDFObjectType_Null; |
2335 | | case CPDF_Object::Type::kBoolean: |
2336 | | return PDFObjectType_Bool; |
2337 | | case CPDF_Object::Type::kNumber: |
2338 | | return (cpl::down_cast<const CPDF_Number *>(m_obj.Get())) |
2339 | | ->IsInteger() |
2340 | | ? PDFObjectType_Int |
2341 | | : PDFObjectType_Real; |
2342 | | case CPDF_Object::Type::kString: |
2343 | | return PDFObjectType_String; |
2344 | | case CPDF_Object::Type::kName: |
2345 | | return PDFObjectType_Name; |
2346 | | case CPDF_Object::Type::kArray: |
2347 | | return PDFObjectType_Array; |
2348 | | case CPDF_Object::Type::kDictionary: |
2349 | | return PDFObjectType_Dictionary; |
2350 | | case CPDF_Object::Type::kStream: |
2351 | | return PDFObjectType_Dictionary; |
2352 | | case CPDF_Object::Type::kReference: |
2353 | | // unresolved reference |
2354 | | return PDFObjectType_Unknown; |
2355 | | default: |
2356 | | CPLAssert(false); |
2357 | | return PDFObjectType_Unknown; |
2358 | | } |
2359 | | } |
2360 | | |
2361 | | /************************************************************************/ |
2362 | | /* GetTypeNameNative() */ |
2363 | | /************************************************************************/ |
2364 | | |
2365 | | const char *GDALPDFObjectPdfium::GetTypeNameNative() |
2366 | | { |
2367 | | if (m_obj->GetType() == CPDF_Object::Type::kStream) |
2368 | | return "stream"; |
2369 | | else |
2370 | | return ""; |
2371 | | } |
2372 | | |
2373 | | /************************************************************************/ |
2374 | | /* GetBool() */ |
2375 | | /************************************************************************/ |
2376 | | |
2377 | | int GDALPDFObjectPdfium::GetBool() |
2378 | | { |
2379 | | return m_obj->GetInteger(); |
2380 | | } |
2381 | | |
2382 | | /************************************************************************/ |
2383 | | /* GetInt() */ |
2384 | | /************************************************************************/ |
2385 | | |
2386 | | int GDALPDFObjectPdfium::GetInt() |
2387 | | { |
2388 | | return m_obj->GetInteger(); |
2389 | | } |
2390 | | |
2391 | | /************************************************************************/ |
2392 | | /* CPLRoundToMoreLikelyDouble() */ |
2393 | | /************************************************************************/ |
2394 | | |
2395 | | // We try to compensate for rounding errors when converting the number |
2396 | | // in the PDF expressed as a string (e.g 297.84) to float32 by pdfium : |
2397 | | // 297.8399963378906 Which is technically correct per the PDF spec, but in |
2398 | | // practice poppler or podofo use double and Geospatial PDF are often encoded |
2399 | | // with double precision. |
2400 | | |
2401 | | static double CPLRoundToMoreLikelyDouble(float f) |
2402 | | { |
2403 | | if (std::round(f) == f) |
2404 | | return f; |
2405 | | |
2406 | | char szBuffer[80]; |
2407 | | CPLsnprintf(szBuffer, 80, "%f\n", f); |
2408 | | double d = f; |
2409 | | char *pszDot = strchr(szBuffer, '.'); |
2410 | | if (pszDot == nullptr) |
2411 | | return d; |
2412 | | pszDot++; |
2413 | | if (pszDot[0] == 0 || pszDot[1] == 0) |
2414 | | return d; |
2415 | | if (STARTS_WITH(pszDot + 2, "99")) |
2416 | | { |
2417 | | pszDot[2] = 0; |
2418 | | double d2 = CPLAtof(szBuffer) + 0.01; |
2419 | | float f2 = static_cast<float>(d2); |
2420 | | if (f == f2 || nextafterf(f, f + 1.0f) == f2 || |
2421 | | nextafterf(f, f - 1.0f) == f2) |
2422 | | d = d2; |
2423 | | } |
2424 | | else if (STARTS_WITH(pszDot + 2, "00")) |
2425 | | { |
2426 | | pszDot[2] = 0; |
2427 | | double d2 = CPLAtof(szBuffer); |
2428 | | float f2 = static_cast<float>(d2); |
2429 | | if (f == f2 || nextafterf(f, f + 1.0f) == f2 || |
2430 | | nextafterf(f, f - 1.0f) == f2) |
2431 | | d = d2; |
2432 | | } |
2433 | | return d; |
2434 | | } |
2435 | | |
2436 | | /************************************************************************/ |
2437 | | /* GetReal() */ |
2438 | | /************************************************************************/ |
2439 | | |
2440 | | double GDALPDFObjectPdfium::GetReal() |
2441 | | { |
2442 | | return CPLRoundToMoreLikelyDouble(m_obj->GetNumber()); |
2443 | | } |
2444 | | |
2445 | | /************************************************************************/ |
2446 | | /* GetString() */ |
2447 | | /************************************************************************/ |
2448 | | |
2449 | | const std::string &GDALPDFObjectPdfium::GetString() |
2450 | | { |
2451 | | if (GetType() == PDFObjectType_String) |
2452 | | { |
2453 | | const auto bs = m_obj->GetString(); |
2454 | | // If empty string, code crashes |
2455 | | if (bs.IsEmpty()) |
2456 | | return (osStr = ""); |
2457 | | return (osStr = GDALPDFGetUTF8StringFromBytes( |
2458 | | reinterpret_cast<const GByte *>(bs.c_str()), |
2459 | | static_cast<int>(bs.GetLength()))); |
2460 | | } |
2461 | | else |
2462 | | return (osStr = ""); |
2463 | | } |
2464 | | |
2465 | | /************************************************************************/ |
2466 | | /* GetName() */ |
2467 | | /************************************************************************/ |
2468 | | |
2469 | | const std::string &GDALPDFObjectPdfium::GetName() |
2470 | | { |
2471 | | if (GetType() == PDFObjectType_Name) |
2472 | | return (osStr = m_obj->GetString().c_str()); |
2473 | | else |
2474 | | return (osStr = ""); |
2475 | | } |
2476 | | |
2477 | | /************************************************************************/ |
2478 | | /* GetDictionary() */ |
2479 | | /************************************************************************/ |
2480 | | |
2481 | | GDALPDFDictionary *GDALPDFObjectPdfium::GetDictionary() |
2482 | | { |
2483 | | if (GetType() != PDFObjectType_Dictionary) |
2484 | | return nullptr; |
2485 | | |
2486 | | if (m_poDict) |
2487 | | return m_poDict; |
2488 | | |
2489 | | m_poDict = new GDALPDFDictionaryPdfium(m_obj->GetDict()); |
2490 | | return m_poDict; |
2491 | | } |
2492 | | |
2493 | | /************************************************************************/ |
2494 | | /* GetArray() */ |
2495 | | /************************************************************************/ |
2496 | | |
2497 | | GDALPDFArray *GDALPDFObjectPdfium::GetArray() |
2498 | | { |
2499 | | if (GetType() != PDFObjectType_Array) |
2500 | | return nullptr; |
2501 | | |
2502 | | if (m_poArray) |
2503 | | return m_poArray; |
2504 | | |
2505 | | m_poArray = |
2506 | | new GDALPDFArrayPdfium(cpl::down_cast<const CPDF_Array *>(m_obj.Get())); |
2507 | | return m_poArray; |
2508 | | } |
2509 | | |
2510 | | /************************************************************************/ |
2511 | | /* GetStream() */ |
2512 | | /************************************************************************/ |
2513 | | |
2514 | | GDALPDFStream *GDALPDFObjectPdfium::GetStream() |
2515 | | { |
2516 | | if (m_obj->GetType() != CPDF_Object::Type::kStream) |
2517 | | return nullptr; |
2518 | | |
2519 | | if (m_poStream) |
2520 | | return m_poStream; |
2521 | | auto pStream = pdfium::WrapRetain(m_obj->AsStream()); |
2522 | | if (pStream) |
2523 | | { |
2524 | | m_poStream = new GDALPDFStreamPdfium(std::move(pStream)); |
2525 | | return m_poStream; |
2526 | | } |
2527 | | else |
2528 | | return nullptr; |
2529 | | } |
2530 | | |
2531 | | /************************************************************************/ |
2532 | | /* GetRefNum() */ |
2533 | | /************************************************************************/ |
2534 | | |
2535 | | GDALPDFObjectNum GDALPDFObjectPdfium::GetRefNum() |
2536 | | { |
2537 | | return GDALPDFObjectNum(m_obj->GetObjNum()); |
2538 | | } |
2539 | | |
2540 | | /************************************************************************/ |
2541 | | /* GetRefGen() */ |
2542 | | /************************************************************************/ |
2543 | | |
2544 | | int GDALPDFObjectPdfium::GetRefGen() |
2545 | | { |
2546 | | return m_obj->GetGenNum(); |
2547 | | } |
2548 | | |
2549 | | /************************************************************************/ |
2550 | | /* ==================================================================== */ |
2551 | | /* GDALPDFDictionaryPdfium */ |
2552 | | /* ==================================================================== */ |
2553 | | /************************************************************************/ |
2554 | | |
2555 | | /************************************************************************/ |
2556 | | /* ~GDALPDFDictionaryPdfium() */ |
2557 | | /************************************************************************/ |
2558 | | |
2559 | | GDALPDFDictionaryPdfium::~GDALPDFDictionaryPdfium() |
2560 | | { |
2561 | | std::map<CPLString, GDALPDFObject *>::iterator oIter = m_map.begin(); |
2562 | | std::map<CPLString, GDALPDFObject *>::iterator oEnd = m_map.end(); |
2563 | | for (; oIter != oEnd; ++oIter) |
2564 | | delete oIter->second; |
2565 | | } |
2566 | | |
2567 | | /************************************************************************/ |
2568 | | /* Get() */ |
2569 | | /************************************************************************/ |
2570 | | |
2571 | | GDALPDFObject *GDALPDFDictionaryPdfium::Get(const char *pszKey) |
2572 | | { |
2573 | | std::map<CPLString, GDALPDFObject *>::iterator oIter = m_map.find(pszKey); |
2574 | | if (oIter != m_map.end()) |
2575 | | return oIter->second; |
2576 | | |
2577 | | ByteString pdfiumKey(pszKey); |
2578 | | GDALPDFObjectPdfium *poObj = |
2579 | | GDALPDFObjectPdfium::Build(m_poDict->GetObjectFor(pdfiumKey)); |
2580 | | if (poObj) |
2581 | | { |
2582 | | m_map[pszKey] = poObj; |
2583 | | return poObj; |
2584 | | } |
2585 | | else |
2586 | | { |
2587 | | return nullptr; |
2588 | | } |
2589 | | } |
2590 | | |
2591 | | /************************************************************************/ |
2592 | | /* GetValues() */ |
2593 | | /************************************************************************/ |
2594 | | |
2595 | | std::map<CPLString, GDALPDFObject *> &GDALPDFDictionaryPdfium::GetValues() |
2596 | | { |
2597 | | CPDF_DictionaryLocker dictIterator(m_poDict); |
2598 | | for (const auto &iter : dictIterator) |
2599 | | { |
2600 | | // No object for this key |
2601 | | if (!iter.second) |
2602 | | continue; |
2603 | | |
2604 | | const char *pszKey = iter.first.c_str(); |
2605 | | // Objects exists in the map |
2606 | | if (m_map.find(pszKey) != m_map.end()) |
2607 | | continue; |
2608 | | GDALPDFObjectPdfium *poObj = GDALPDFObjectPdfium::Build(iter.second); |
2609 | | if (poObj == nullptr) |
2610 | | continue; |
2611 | | m_map[pszKey] = poObj; |
2612 | | } |
2613 | | |
2614 | | return m_map; |
2615 | | } |
2616 | | |
2617 | | /************************************************************************/ |
2618 | | /* ==================================================================== */ |
2619 | | /* GDALPDFArrayPdfium */ |
2620 | | /* ==================================================================== */ |
2621 | | /************************************************************************/ |
2622 | | |
2623 | | /************************************************************************/ |
2624 | | /* GetLength() */ |
2625 | | /************************************************************************/ |
2626 | | |
2627 | | int GDALPDFArrayPdfium::GetLength() |
2628 | | { |
2629 | | return static_cast<int>(m_poArray->size()); |
2630 | | } |
2631 | | |
2632 | | /************************************************************************/ |
2633 | | /* Get() */ |
2634 | | /************************************************************************/ |
2635 | | |
2636 | | GDALPDFObject *GDALPDFArrayPdfium::Get(int nIndex) |
2637 | | { |
2638 | | if (nIndex < 0 || nIndex >= GetLength()) |
2639 | | return nullptr; |
2640 | | |
2641 | | if (m_v.empty()) |
2642 | | m_v.resize(GetLength()); |
2643 | | |
2644 | | if (m_v[nIndex] != nullptr) |
2645 | | return m_v[nIndex].get(); |
2646 | | |
2647 | | auto poObj = std::unique_ptr<GDALPDFObjectPdfium>( |
2648 | | GDALPDFObjectPdfium::Build(m_poArray->GetObjectAt(nIndex))); |
2649 | | if (poObj == nullptr) |
2650 | | return nullptr; |
2651 | | m_v[nIndex] = std::move(poObj); |
2652 | | return m_v[nIndex].get(); |
2653 | | } |
2654 | | |
2655 | | /************************************************************************/ |
2656 | | /* ==================================================================== */ |
2657 | | /* GDALPDFStreamPdfium */ |
2658 | | /* ==================================================================== */ |
2659 | | /************************************************************************/ |
2660 | | |
2661 | | void GDALPDFStreamPdfium::Decompress() |
2662 | | { |
2663 | | if (m_pData != nullptr) |
2664 | | return; |
2665 | | auto acc(pdfium::MakeRetain<CPDF_StreamAcc>(m_pStream)); |
2666 | | acc->LoadAllDataFiltered(); |
2667 | | m_nSize = static_cast<int64_t>(acc->GetSize()); |
2668 | | m_pData.reset(); |
2669 | | const auto nSize = static_cast<size_t>(m_nSize); |
2670 | | if (static_cast<int64_t>(nSize) != m_nSize) |
2671 | | { |
2672 | | m_nSize = 0; |
2673 | | } |
2674 | | if (m_nSize) |
2675 | | { |
2676 | | m_pData.reset(static_cast<uint8_t *>(VSI_MALLOC_VERBOSE(nSize))); |
2677 | | if (!m_pData) |
2678 | | m_nSize = 0; |
2679 | | else |
2680 | | memcpy(&m_pData.get()[0], acc->DetachData().data(), nSize); |
2681 | | } |
2682 | | } |
2683 | | |
2684 | | /************************************************************************/ |
2685 | | /* GetLength() */ |
2686 | | /************************************************************************/ |
2687 | | |
2688 | | int64_t GDALPDFStreamPdfium::GetLength(int64_t /* nMaxSize */) |
2689 | | { |
2690 | | Decompress(); |
2691 | | return m_nSize; |
2692 | | } |
2693 | | |
2694 | | /************************************************************************/ |
2695 | | /* GetBytes() */ |
2696 | | /************************************************************************/ |
2697 | | |
2698 | | char *GDALPDFStreamPdfium::GetBytes() |
2699 | | { |
2700 | | size_t nLength = static_cast<size_t>(GetLength()); |
2701 | | if (nLength == 0) |
2702 | | return nullptr; |
2703 | | char *pszContent = static_cast<char *>(VSI_MALLOC_VERBOSE(nLength + 1)); |
2704 | | if (!pszContent) |
2705 | | return nullptr; |
2706 | | memcpy(pszContent, m_pData.get(), nLength); |
2707 | | pszContent[nLength] = '\0'; |
2708 | | return pszContent; |
2709 | | } |
2710 | | |
2711 | | /************************************************************************/ |
2712 | | /* FillRaw() */ |
2713 | | /************************************************************************/ |
2714 | | |
2715 | | void GDALPDFStreamPdfium::FillRaw() |
2716 | | { |
2717 | | if (m_pRawData != nullptr) |
2718 | | return; |
2719 | | auto acc(pdfium::MakeRetain<CPDF_StreamAcc>(m_pStream)); |
2720 | | acc->LoadAllDataRaw(); |
2721 | | m_nRawSize = static_cast<int64_t>(acc->GetSize()); |
2722 | | m_pRawData.reset(); |
2723 | | const auto nSize = static_cast<size_t>(m_nRawSize); |
2724 | | if (static_cast<int64_t>(nSize) != m_nRawSize) |
2725 | | { |
2726 | | m_nRawSize = 0; |
2727 | | } |
2728 | | if (m_nRawSize) |
2729 | | { |
2730 | | m_pRawData.reset( |
2731 | | static_cast<uint8_t *>(VSI_MALLOC_VERBOSE(m_nRawSize))); |
2732 | | if (!m_pRawData) |
2733 | | m_nRawSize = 0; |
2734 | | else |
2735 | | memcpy(&m_pRawData.get()[0], acc->DetachData().data(), m_nRawSize); |
2736 | | } |
2737 | | } |
2738 | | |
2739 | | /************************************************************************/ |
2740 | | /* GetRawLength() */ |
2741 | | /************************************************************************/ |
2742 | | |
2743 | | int64_t GDALPDFStreamPdfium::GetRawLength() |
2744 | | { |
2745 | | FillRaw(); |
2746 | | return m_nRawSize; |
2747 | | } |
2748 | | |
2749 | | /************************************************************************/ |
2750 | | /* GetRawBytes() */ |
2751 | | /************************************************************************/ |
2752 | | |
2753 | | char *GDALPDFStreamPdfium::GetRawBytes() |
2754 | | { |
2755 | | size_t nLength = static_cast<size_t>(GetRawLength()); |
2756 | | if (nLength == 0) |
2757 | | return nullptr; |
2758 | | char *pszContent = |
2759 | | static_cast<char *>(VSI_MALLOC_VERBOSE(sizeof(char) * (nLength + 1))); |
2760 | | if (!pszContent) |
2761 | | return nullptr; |
2762 | | memcpy(pszContent, m_pRawData.get(), nLength); |
2763 | | pszContent[nLength] = '\0'; |
2764 | | return pszContent; |
2765 | | } |
2766 | | |
2767 | | #endif // HAVE_PDFIUM |