/src/gdal/frmts/pdf/pdfobject.h
Line | Count | Source |
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) 2011-2014, Even Rouault <even dot rouault at spatialys.com> |
16 | | * |
17 | | * SPDX-License-Identifier: MIT |
18 | | ****************************************************************************/ |
19 | | |
20 | | #ifndef PDFOBJECT_H_INCLUDED |
21 | | #define PDFOBJECT_H_INCLUDED |
22 | | |
23 | | #include "pdfsdk_headers_all.h" |
24 | | |
25 | | #include "cpl_string.h" |
26 | | #include <map> |
27 | | #include <vector> |
28 | | |
29 | 651k | #define DEFAULT_DPI (72.0) |
30 | 636k | #define USER_UNIT_IN_INCH (1.0 / DEFAULT_DPI) |
31 | | |
32 | | double ROUND_IF_CLOSE(double x, double eps = 0); |
33 | | |
34 | | typedef enum |
35 | | { |
36 | | PDFObjectType_Unknown, |
37 | | PDFObjectType_Null, |
38 | | PDFObjectType_Bool, |
39 | | PDFObjectType_Int, |
40 | | PDFObjectType_Real, |
41 | | PDFObjectType_String, |
42 | | PDFObjectType_Name, |
43 | | PDFObjectType_Array, |
44 | | PDFObjectType_Dictionary |
45 | | } GDALPDFObjectType; |
46 | | |
47 | | class GDALPDFDictionary; |
48 | | class GDALPDFArray; |
49 | | class GDALPDFStream; |
50 | | |
51 | | class GDALPDFObjectRW; |
52 | | class GDALPDFDictionaryRW; |
53 | | class GDALPDFArrayRW; |
54 | | |
55 | | class GDALPDFObjectNum |
56 | | { |
57 | | int m_nId; |
58 | | |
59 | | public: |
60 | 8.06M | explicit GDALPDFObjectNum(int nId = 0) : m_nId(nId) |
61 | 8.06M | { |
62 | 8.06M | } |
63 | | |
64 | | GDALPDFObjectNum(const GDALPDFObjectNum &other) = default; |
65 | | GDALPDFObjectNum &operator=(const GDALPDFObjectNum &) = default; |
66 | | |
67 | | GDALPDFObjectNum &operator=(int nId) |
68 | 1.04M | { |
69 | 1.04M | m_nId = nId; |
70 | 1.04M | return *this; |
71 | 1.04M | } |
72 | | |
73 | | int toInt() const |
74 | 1.65M | { |
75 | 1.65M | return m_nId; |
76 | 1.65M | } |
77 | | |
78 | | bool toBool() const |
79 | 1.15M | { |
80 | 1.15M | return m_nId > 0; |
81 | 1.15M | } |
82 | | |
83 | | bool operator==(const GDALPDFObjectNum &other) const |
84 | 6.29k | { |
85 | 6.29k | return m_nId == other.m_nId; |
86 | 6.29k | } |
87 | | |
88 | | bool operator<(const GDALPDFObjectNum &other) const |
89 | 0 | { |
90 | 0 | return m_nId < other.m_nId; |
91 | 0 | } |
92 | | }; |
93 | | |
94 | | class GDALPDFObject /* non final */ |
95 | | { |
96 | | protected: |
97 | | virtual const char *GetTypeNameNative() = 0; |
98 | | |
99 | | public: |
100 | | virtual ~GDALPDFObject(); |
101 | | |
102 | | virtual GDALPDFObjectType GetType() = 0; |
103 | | virtual const char *GetTypeName(); |
104 | | virtual int GetBool() = 0; |
105 | | virtual int GetInt() = 0; |
106 | | virtual double GetReal() = 0; |
107 | | |
108 | | virtual int CanRepresentRealAsString() |
109 | 0 | { |
110 | 0 | return FALSE; |
111 | 0 | } |
112 | | |
113 | | virtual const std::string &GetString() = 0; |
114 | | virtual const std::string &GetName() = 0; |
115 | | virtual GDALPDFDictionary *GetDictionary() = 0; |
116 | | virtual GDALPDFArray *GetArray() = 0; |
117 | | virtual GDALPDFStream *GetStream() = 0; |
118 | | virtual GDALPDFObjectNum GetRefNum() = 0; |
119 | | virtual int GetRefGen() = 0; |
120 | | |
121 | | virtual int GetPrecision() const |
122 | 0 | { |
123 | 0 | return 16; |
124 | 0 | } |
125 | | |
126 | | GDALPDFObject *LookupObject(const char *pszPath); |
127 | | |
128 | | void Serialize(CPLString &osStr, bool bEmitRef = true); |
129 | | |
130 | | CPLString Serialize() |
131 | 212 | { |
132 | 212 | CPLString osStr; |
133 | 212 | Serialize(osStr); |
134 | 212 | return osStr; |
135 | 212 | } |
136 | | |
137 | | GDALPDFObjectRW *Clone(); |
138 | | }; |
139 | | |
140 | | class GDALPDFDictionary /* non final */ |
141 | | { |
142 | | public: |
143 | | virtual ~GDALPDFDictionary(); |
144 | | |
145 | | virtual GDALPDFObject *Get(const char *pszKey) = 0; |
146 | | virtual std::map<CPLString, GDALPDFObject *> &GetValues() = 0; |
147 | | |
148 | | GDALPDFObject *LookupObject(const char *pszPath); |
149 | | |
150 | | void Serialize(CPLString &osStr); |
151 | | |
152 | | CPLString Serialize() |
153 | 67.4k | { |
154 | 67.4k | CPLString osStr; |
155 | 67.4k | Serialize(osStr); |
156 | 67.4k | return osStr; |
157 | 67.4k | } |
158 | | |
159 | | GDALPDFDictionaryRW *Clone(); |
160 | | }; |
161 | | |
162 | | class GDALPDFArray /* non final */ |
163 | | { |
164 | | public: |
165 | | virtual ~GDALPDFArray(); |
166 | | |
167 | | virtual int GetLength() = 0; |
168 | | virtual GDALPDFObject *Get(int nIndex) = 0; |
169 | | |
170 | | void Serialize(CPLString &osStr); |
171 | | |
172 | | CPLString Serialize() |
173 | 664 | { |
174 | 664 | CPLString osStr; |
175 | 664 | Serialize(osStr); |
176 | 664 | return osStr; |
177 | 664 | } |
178 | | |
179 | | GDALPDFArrayRW *Clone(); |
180 | | }; |
181 | | |
182 | | class GDALPDFStream /* non final */ |
183 | | { |
184 | | public: |
185 | | virtual ~GDALPDFStream(); |
186 | | |
187 | | /** Return the uncompressed stream length, or 0 if empty or error. |
188 | | * If nMaxSize > 0, GetLength() will possibly stop the decompression once |
189 | | * the threshold is reached, and return INT64_MAX */ |
190 | | virtual int64_t GetLength(int64_t nMaxSize = 0) = 0; |
191 | | virtual char *GetBytes() = 0; |
192 | | |
193 | | virtual int64_t GetRawLength() = 0; |
194 | | virtual char *GetRawBytes() = 0; |
195 | | }; |
196 | | |
197 | | class GDALPDFObjectRW final : public GDALPDFObject |
198 | | { |
199 | | private: |
200 | | const GDALPDFObjectType m_eType; |
201 | | int m_nVal = 0; |
202 | | double m_dfVal = 0; |
203 | | CPLString m_osVal{}; |
204 | | GDALPDFDictionaryRW *m_poDict = nullptr; |
205 | | GDALPDFArrayRW *m_poArray = nullptr; |
206 | | GDALPDFObjectNum m_nNum{}; |
207 | | int m_nGen = 0; |
208 | | int m_bCanRepresentRealAsString = FALSE; |
209 | | int m_nPrecision = 16; |
210 | | |
211 | | explicit GDALPDFObjectRW(GDALPDFObjectType eType); |
212 | | |
213 | | CPL_DISALLOW_COPY_ASSIGN(GDALPDFObjectRW) |
214 | | |
215 | | protected: |
216 | | const char *GetTypeNameNative() override; |
217 | | |
218 | | public: |
219 | | static GDALPDFObjectRW *CreateIndirect(const GDALPDFObjectNum &nNum, |
220 | | int nGen); |
221 | | static GDALPDFObjectRW *CreateNull(); |
222 | | static GDALPDFObjectRW *CreateBool(int bVal); |
223 | | static GDALPDFObjectRW *CreateInt(int nVal); |
224 | | static GDALPDFObjectRW *CreateReal(double dfVal, |
225 | | int bCanRepresentRealAsString = FALSE); |
226 | | static GDALPDFObjectRW *CreateRealWithPrecision(double dfVal, |
227 | | int nPrecision); |
228 | | static GDALPDFObjectRW *CreateString(const char *pszStr); |
229 | | static GDALPDFObjectRW *CreateName(const char *pszName); |
230 | | static GDALPDFObjectRW *CreateDictionary(GDALPDFDictionaryRW *poDict); |
231 | | static GDALPDFObjectRW *CreateArray(GDALPDFArrayRW *poArray); |
232 | | ~GDALPDFObjectRW() override; |
233 | | |
234 | | GDALPDFObjectType GetType() override; |
235 | | int GetBool() override; |
236 | | int GetInt() override; |
237 | | double GetReal() override; |
238 | | |
239 | | int CanRepresentRealAsString() override |
240 | 25.7k | { |
241 | 25.7k | return m_bCanRepresentRealAsString; |
242 | 25.7k | } |
243 | | |
244 | | const CPLString &GetString() override; |
245 | | const CPLString &GetName() override; |
246 | | GDALPDFDictionary *GetDictionary() override; |
247 | | GDALPDFArray *GetArray() override; |
248 | | GDALPDFStream *GetStream() override; |
249 | | GDALPDFObjectNum GetRefNum() override; |
250 | | int GetRefGen() override; |
251 | | |
252 | | int GetPrecision() const override |
253 | 25.7k | { |
254 | 25.7k | return m_nPrecision; |
255 | 25.7k | } |
256 | | }; |
257 | | |
258 | | class GDALPDFDictionaryRW final : public GDALPDFDictionary |
259 | | { |
260 | | private: |
261 | | std::map<CPLString, GDALPDFObject *> m_map{}; |
262 | | |
263 | | CPL_DISALLOW_COPY_ASSIGN(GDALPDFDictionaryRW) |
264 | | |
265 | | public: |
266 | | GDALPDFDictionaryRW(); |
267 | | ~GDALPDFDictionaryRW() override; |
268 | | |
269 | | GDALPDFObject *Get(const char *pszKey) override; |
270 | | std::map<CPLString, GDALPDFObject *> &GetValues() override; |
271 | | |
272 | | GDALPDFDictionaryRW &Add(const char *pszKey, GDALPDFObject *poVal); |
273 | | GDALPDFDictionaryRW &Remove(const char *pszKey); |
274 | | |
275 | | GDALPDFDictionaryRW &Add(const char *pszKey, GDALPDFArrayRW *poArray) |
276 | 59.4k | { |
277 | 59.4k | return Add(pszKey, GDALPDFObjectRW::CreateArray(poArray)); |
278 | 59.4k | } |
279 | | |
280 | | GDALPDFDictionaryRW &Add(const char *pszKey, GDALPDFDictionaryRW *poDict) |
281 | 107k | { |
282 | 107k | return Add(pszKey, GDALPDFObjectRW::CreateDictionary(poDict)); |
283 | 107k | } |
284 | | |
285 | | GDALPDFDictionaryRW &Add(const char *pszKey, const char *pszVal) |
286 | 136k | { |
287 | 136k | return Add(pszKey, GDALPDFObjectRW::CreateString(pszVal)); |
288 | 136k | } |
289 | | |
290 | | GDALPDFDictionaryRW &Add(const char *pszKey, int nVal) |
291 | 26.7k | { |
292 | 26.7k | return Add(pszKey, GDALPDFObjectRW::CreateInt(nVal)); |
293 | 26.7k | } |
294 | | |
295 | | GDALPDFDictionaryRW &Add(const char *pszKey, double dfVal, |
296 | | int bCanRepresentRealAsString = FALSE) |
297 | 25.0k | { |
298 | 25.0k | return Add(pszKey, GDALPDFObjectRW::CreateReal( |
299 | 25.0k | dfVal, bCanRepresentRealAsString)); |
300 | 25.0k | } |
301 | | |
302 | | GDALPDFDictionaryRW &Add(const char *pszKey, const GDALPDFObjectNum &nNum, |
303 | | int nGen) |
304 | 117k | { |
305 | 117k | return Add(pszKey, GDALPDFObjectRW::CreateIndirect(nNum, nGen)); |
306 | 117k | } |
307 | | }; |
308 | | |
309 | | class GDALPDFArrayRW final : public GDALPDFArray |
310 | | { |
311 | | private: |
312 | | std::vector<GDALPDFObject *> m_array{}; |
313 | | |
314 | | CPL_DISALLOW_COPY_ASSIGN(GDALPDFArrayRW) |
315 | | |
316 | | public: |
317 | | GDALPDFArrayRW(); |
318 | | ~GDALPDFArrayRW() override; |
319 | | |
320 | | int GetLength() override; |
321 | | GDALPDFObject *Get(int nIndex) override; |
322 | | |
323 | | GDALPDFArrayRW &Add(GDALPDFObject *poObj); |
324 | | |
325 | | GDALPDFArrayRW &Add(GDALPDFArrayRW *poArray) |
326 | 7 | { |
327 | 7 | return Add(GDALPDFObjectRW::CreateArray(poArray)); |
328 | 7 | } |
329 | | |
330 | | GDALPDFArrayRW &Add(GDALPDFDictionaryRW *poDict) |
331 | 48.7k | { |
332 | 48.7k | return Add(GDALPDFObjectRW::CreateDictionary(poDict)); |
333 | 48.7k | } |
334 | | |
335 | | GDALPDFArrayRW &Add(const char *pszVal) |
336 | 0 | { |
337 | 0 | return Add(GDALPDFObjectRW::CreateString(pszVal)); |
338 | 0 | } |
339 | | |
340 | | GDALPDFArrayRW &Add(int nVal) |
341 | 102k | { |
342 | 102k | return Add(GDALPDFObjectRW::CreateInt(nVal)); |
343 | 102k | } |
344 | | |
345 | | GDALPDFArrayRW &Add(double dfVal, int bCanRepresentRealAsString = FALSE) |
346 | 4.39k | { |
347 | 4.39k | return Add( |
348 | 4.39k | GDALPDFObjectRW::CreateReal(dfVal, bCanRepresentRealAsString)); |
349 | 4.39k | } |
350 | | |
351 | | GDALPDFArrayRW &AddWithPrecision(double dfVal, int nPrecision) |
352 | 0 | { |
353 | 0 | return Add(GDALPDFObjectRW::CreateRealWithPrecision(dfVal, nPrecision)); |
354 | 0 | } |
355 | | |
356 | | GDALPDFArrayRW &Add(double *padfVal, int nCount, |
357 | | int bCanRepresentRealAsString = FALSE); |
358 | | |
359 | | GDALPDFArrayRW &Add(const GDALPDFObjectNum &nNum, int nGen) |
360 | 39.1k | { |
361 | 39.1k | return Add(GDALPDFObjectRW::CreateIndirect(nNum, nGen)); |
362 | 39.1k | } |
363 | | }; |
364 | | |
365 | | #ifdef HAVE_POPPLER |
366 | | |
367 | | class GDALPDFObjectPoppler final : public GDALPDFObject |
368 | | { |
369 | | private: |
370 | | Object *m_poToDestroy; |
371 | | const Object *m_poConst; |
372 | | GDALPDFDictionary *m_poDict = nullptr; |
373 | | GDALPDFArray *m_poArray = nullptr; |
374 | | GDALPDFStream *m_poStream = nullptr; |
375 | | std::string osStr{}; |
376 | | GDALPDFObjectNum m_nRefNum{}; |
377 | | int m_nRefGen = 0; |
378 | | |
379 | | CPL_DISALLOW_COPY_ASSIGN(GDALPDFObjectPoppler) |
380 | | |
381 | | protected: |
382 | | const char *GetTypeNameNative() override; |
383 | | |
384 | | public: |
385 | | GDALPDFObjectPoppler(Object *po, bool bDestroy) |
386 | 3.41M | : m_poToDestroy(bDestroy ? po : nullptr), m_poConst(po) |
387 | 3.41M | { |
388 | 3.41M | } |
389 | | |
390 | | GDALPDFObjectPoppler(const Object *po) |
391 | 32.6k | : m_poToDestroy(nullptr), m_poConst(po) |
392 | 32.6k | { |
393 | 32.6k | } |
394 | | |
395 | | void SetRefNumAndGen(const GDALPDFObjectNum &nNum, int nGen); |
396 | | |
397 | | ~GDALPDFObjectPoppler() override; |
398 | | |
399 | | GDALPDFObjectType GetType() override; |
400 | | int GetBool() override; |
401 | | int GetInt() override; |
402 | | double GetReal() override; |
403 | | const std::string &GetString() override; |
404 | | const std::string &GetName() override; |
405 | | GDALPDFDictionary *GetDictionary() override; |
406 | | GDALPDFArray *GetArray() override; |
407 | | GDALPDFStream *GetStream() override; |
408 | | GDALPDFObjectNum GetRefNum() override; |
409 | | int GetRefGen() override; |
410 | | }; |
411 | | |
412 | | GDALPDFArray *GDALPDFCreateArray(const Array *array); |
413 | | |
414 | | #endif // HAVE_POPPLER |
415 | | |
416 | | #ifdef HAVE_PODOFO |
417 | | |
418 | | class GDALPDFObjectPodofo final : public GDALPDFObject |
419 | | { |
420 | | private: |
421 | | const PoDoFo::PdfObject *m_po; |
422 | | const PoDoFo::PdfVecObjects &m_poObjects; |
423 | | GDALPDFDictionary *m_poDict = nullptr; |
424 | | GDALPDFArray *m_poArray = nullptr; |
425 | | GDALPDFStream *m_poStream = nullptr; |
426 | | std::string osStr{}; |
427 | | |
428 | | CPL_DISALLOW_COPY_ASSIGN(GDALPDFObjectPodofo) |
429 | | |
430 | | protected: |
431 | | const char *GetTypeNameNative() override; |
432 | | |
433 | | public: |
434 | | GDALPDFObjectPodofo(const PoDoFo::PdfObject *po, |
435 | | const PoDoFo::PdfVecObjects &poObjects); |
436 | | |
437 | | ~GDALPDFObjectPodofo() override; |
438 | | |
439 | | GDALPDFObjectType GetType() override; |
440 | | int GetBool() override; |
441 | | int GetInt() override; |
442 | | double GetReal() override; |
443 | | const std::string &GetString() override; |
444 | | const std::string &GetName() override; |
445 | | GDALPDFDictionary *GetDictionary() override; |
446 | | GDALPDFArray *GetArray() override; |
447 | | GDALPDFStream *GetStream() override; |
448 | | GDALPDFObjectNum GetRefNum() override; |
449 | | int GetRefGen() override; |
450 | | }; |
451 | | |
452 | | #endif // HAVE_PODOFO |
453 | | |
454 | | #ifdef HAVE_PDFIUM |
455 | | |
456 | | class GDALPDFObjectPdfium final : public GDALPDFObject |
457 | | { |
458 | | private: |
459 | | RetainPtr<const CPDF_Object> m_obj; |
460 | | GDALPDFDictionary *m_poDict = nullptr; |
461 | | GDALPDFArray *m_poArray = nullptr; |
462 | | GDALPDFStream *m_poStream = nullptr; |
463 | | std::string osStr{}; |
464 | | |
465 | | GDALPDFObjectPdfium(RetainPtr<const CPDF_Object> obj); |
466 | | |
467 | | CPL_DISALLOW_COPY_ASSIGN(GDALPDFObjectPdfium) |
468 | | |
469 | | protected: |
470 | | const char *GetTypeNameNative() override; |
471 | | |
472 | | public: |
473 | | static GDALPDFObjectPdfium *Build(RetainPtr<const CPDF_Object> obj); |
474 | | |
475 | | ~GDALPDFObjectPdfium() override; |
476 | | |
477 | | GDALPDFObjectType GetType() override; |
478 | | int GetBool() override; |
479 | | int GetInt() override; |
480 | | double GetReal() override; |
481 | | const std::string &GetString() override; |
482 | | const std::string &GetName() override; |
483 | | GDALPDFDictionary *GetDictionary() override; |
484 | | GDALPDFArray *GetArray() override; |
485 | | GDALPDFStream *GetStream() override; |
486 | | GDALPDFObjectNum GetRefNum() override; |
487 | | int GetRefGen() override; |
488 | | }; |
489 | | |
490 | | #endif // HAVE_PDFIUM |
491 | | |
492 | | #endif // PDFOBJECT_H_INCLUDED |