xlsxio_read.c 58 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595
  1. #include "xlsxio_private.h"
  2. #include "xlsxio_read_sharedstrings.h"
  3. #include "xlsxio_read.h"
  4. #include "xlsxio_version.h"
  5. #include <stdlib.h>
  6. #include <stdio.h>
  7. #include <inttypes.h>
  8. #include <string.h>
  9. #include <expat.h>
  10. #ifdef USE_MINIZIP
  11. # include <minizip/unzip.h>
  12. # define ZIPFILETYPE unzFile
  13. # define ZIPFILEENTRYTYPE unzFile
  14. # if defined(_MSC_VER)
  15. # include <io.h>
  16. # define IOSIZETYPE int
  17. # define IOFN(fn) _##fn
  18. # else
  19. # include <unistd.h>
  20. # define IOSIZETYPE ssize_t
  21. # define IOFN(fn) fn
  22. # endif
  23. /*
  24. # if !defined(Z_DEFLATED) && defined(MZ_COMPRESS_METHOD_DEFLATE) // support minizip2 which defines MZ_COMPRESS_METHOD_DEFLATE instead of Z_DEFLATED
  25. # ifndef ZCALLBACK
  26. # define ZCALLBACK
  27. # endif
  28. # define voidpf void*
  29. # define uLong unsigned long
  30. # endif
  31. */
  32. #else
  33. # if (defined(STATIC) || defined(BUILD_XLSXIO_STATIC) || defined(BUILD_XLSXIO_STATIC_DLL) || (defined(BUILD_XLSXIO) && !defined(BUILD_XLSXIO_DLL) && !defined(BUILD_XLSXIO_SHARED))) && !defined(ZIP_STATIC)
  34. # define ZIP_STATIC
  35. # endif
  36. # include <zip.h>
  37. # define ZIPFILETYPE zip_t
  38. # define ZIPFILEENTRYTYPE zip_file_t
  39. # ifndef USE_LIBZIP
  40. # define USE_LIBZIP
  41. # endif
  42. #endif
  43. #if defined(_MSC_VER)
  44. # undef DLL_EXPORT_XLSXIO
  45. # define DLL_EXPORT_XLSXIO
  46. #endif
  47. #define PARSE_BUFFER_SIZE 256
  48. //#define PARSE_BUFFER_SIZE 4
  49. static const XLSXIOCHAR* xlsx_content_type = X("application/vnd.openxmlformats-officedocument.spreadsheetml.sheet.main+xml");
  50. static const XLSXIOCHAR* xlsm_content_type = X("application/vnd.ms-excel.sheet.macroEnabled.main+xml");
  51. static const XLSXIOCHAR* xltx_content_type = X("application/vnd.openxmlformats-officedocument.spreadsheetml.template.main+xml");
  52. static const XLSXIOCHAR* xltm_content_type = X("application/vnd.ms-excel.template.macroEnabled.main+xml");
  53. #if !defined(XML_UNICODE_WCHAR_T) && !defined(XML_UNICODE)
  54. //UTF-8 version
  55. #define XML_Char_dupchar strdup
  56. static ZIPFILEENTRYTYPE* XML_Char_openzip (ZIPFILETYPE* archive, const XML_Char* filename, int flags)
  57. {
  58. if (!filename || !*filename)
  59. return NULL;
  60. #ifdef USE_MINIZIP
  61. if (unzLocateFile(archive, filename, 0) != UNZ_OK)
  62. return NULL;
  63. if (unzOpenCurrentFile(archive) != UNZ_OK)
  64. return NULL;
  65. return archive;
  66. #else
  67. return zip_fopen(archive, filename, flags);
  68. #endif
  69. }
  70. #else
  71. //UTF-16 version
  72. static XML_Char* XML_Char_dupchar(const char* s)
  73. {
  74. size_t len;
  75. XML_Char* result;
  76. if (!s || (len = mbstowcs(NULL, s, 0)) < 0)
  77. return NULL;
  78. if ((result = XML_Char_malloc(len + 1)) != NULL) {
  79. if ((mbstowcs(result, s, len + 1) != len)) {
  80. free(result);
  81. return NULL;
  82. }
  83. }
  84. return result;
  85. }
  86. static char* chardupXML_Char(const XML_Char* s)
  87. {
  88. size_t len;
  89. char* result;
  90. if (!s || (len = wcstombs(NULL, s, 0)) == -1)
  91. return NULL;
  92. if ((result = (char*)malloc(len + 1)) != NULL) {
  93. if ((wcstombs(result, s, len + 1) != len)) {
  94. free(result);
  95. return NULL;
  96. }
  97. }
  98. return result;
  99. }
  100. static ZIPFILEENTRYTYPE* XML_Char_openzip (ZIPFILETYPE* archive, const XML_Char* filename, int flags)
  101. {
  102. ZIPFILEENTRYTYPE* result;
  103. char* s;
  104. if (!filename || !*filename)
  105. return NULL;
  106. if ((s = chardupXML_Char(filename)) == NULL)
  107. return NULL;
  108. #ifdef USE_MINIZIP
  109. if (unzLocateFile(archive, s, 0) != UNZ_OK)
  110. result = NULL;
  111. else if (unzOpenCurrentFile(archive) != UNZ_OK)
  112. result = NULL;
  113. else
  114. result = archive;
  115. #else
  116. result = zip_fopen(archive, s, flags);
  117. #endif
  118. free(s);
  119. return result;
  120. }
  121. #endif
  122. DLL_EXPORT_XLSXIO void xlsxioread_get_version (int* pmajor, int* pminor, int* pmicro)
  123. {
  124. if (pmajor)
  125. *pmajor = XLSXIO_VERSION_MAJOR;
  126. if (pminor)
  127. *pminor = XLSXIO_VERSION_MINOR;
  128. if (pmicro)
  129. *pmicro = XLSXIO_VERSION_MICRO;
  130. }
  131. DLL_EXPORT_XLSXIO const XLSXIOCHAR* xlsxioread_get_version_string ()
  132. {
  133. return (const XLSXIOCHAR*)XLSXIO_VERSION_STRING;
  134. }
  135. ////////////////////////////////////////////////////////////////////////
  136. //process XML file contents
  137. int expat_process_zip_file (ZIPFILETYPE* zip, const XML_Char* filename, XML_StartElementHandler start_handler, XML_EndElementHandler end_handler, XML_CharacterDataHandler data_handler, void* callbackdata, XML_Parser* xmlparser)
  138. {
  139. ZIPFILEENTRYTYPE* zipfile;
  140. XML_Parser parser;
  141. void* buf;
  142. #ifdef USE_MINIZIP
  143. int buflen;
  144. #else
  145. zip_int64_t buflen;
  146. #endif
  147. int done;
  148. enum XML_Status status = XML_STATUS_ERROR;
  149. if ((zipfile = XML_Char_openzip(zip, filename, 0)) == NULL) {
  150. return -1;
  151. }
  152. parser = XML_ParserCreate(NULL);
  153. XML_SetUserData(parser, callbackdata);
  154. XML_SetElementHandler(parser, start_handler, end_handler);
  155. XML_SetCharacterDataHandler(parser, data_handler);
  156. if (xmlparser)
  157. *xmlparser = parser;
  158. buf = XML_GetBuffer(parser, PARSE_BUFFER_SIZE);
  159. #ifdef USE_MINIZIP
  160. while (buf && (buflen = unzReadCurrentFile(zip, buf, PARSE_BUFFER_SIZE)) >= 0) {
  161. #else
  162. while (buf && (buflen = zip_fread(zipfile, buf, PARSE_BUFFER_SIZE)) >= 0) {
  163. #endif
  164. done = buflen < PARSE_BUFFER_SIZE;
  165. if ((status = XML_ParseBuffer(parser, (int)buflen, (done ? 1 : 0))) == XML_STATUS_ERROR) {
  166. break;
  167. }
  168. if (xmlparser && status == XML_STATUS_SUSPENDED)
  169. return 0;
  170. if (done)
  171. break;
  172. buf = XML_GetBuffer(parser, PARSE_BUFFER_SIZE);
  173. }
  174. XML_ParserFree(parser);
  175. #ifdef USE_MINIZIP
  176. unzCloseCurrentFile(zip);
  177. #else
  178. zip_fclose(zipfile);
  179. #endif
  180. //return (status == XML_STATUS_ERROR != XML_ERROR_FINISHED ? 1 : 0);
  181. return 0;
  182. }
  183. XML_Parser expat_process_zip_file_suspendable (ZIPFILEENTRYTYPE* zipfile, XML_StartElementHandler start_handler, XML_EndElementHandler end_handler, XML_CharacterDataHandler data_handler, void* callbackdata)
  184. {
  185. XML_Parser result;
  186. if ((result = XML_ParserCreate(NULL)) != NULL) {
  187. XML_SetUserData(result, callbackdata);
  188. XML_SetElementHandler(result, start_handler, end_handler);
  189. XML_SetCharacterDataHandler(result, data_handler);
  190. }
  191. return result;
  192. }
  193. enum XML_Status expat_process_zip_file_resume (ZIPFILEENTRYTYPE* zipfile, XML_Parser xmlparser)
  194. {
  195. enum XML_Status status;
  196. status = XML_ResumeParser(xmlparser);
  197. if (status == XML_STATUS_SUSPENDED)
  198. return status;
  199. if (status == XML_STATUS_ERROR && XML_GetErrorCode(xmlparser) != XML_ERROR_NOT_SUSPENDED)
  200. return status;
  201. void* buf;
  202. #ifdef USE_MINIZIP
  203. int buflen;
  204. #else
  205. zip_int64_t buflen;
  206. #endif
  207. int done;
  208. buf = XML_GetBuffer(xmlparser, PARSE_BUFFER_SIZE);
  209. #ifdef USE_MINIZIP
  210. while (buf && (buflen = unzReadCurrentFile(zipfile, buf, PARSE_BUFFER_SIZE)) >= 0) {
  211. #else
  212. while (buf && (buflen = zip_fread(zipfile, buf, PARSE_BUFFER_SIZE)) >= 0) {
  213. #endif
  214. done = buflen < PARSE_BUFFER_SIZE;
  215. if ((status = XML_ParseBuffer(xmlparser, (int)buflen, (done ? 1 : 0))) == XML_STATUS_ERROR)
  216. return status;
  217. if (status == XML_STATUS_SUSPENDED)
  218. return status;
  219. if (done)
  220. break;
  221. buf = XML_GetBuffer(xmlparser, PARSE_BUFFER_SIZE);
  222. }
  223. //XML_ParserFree(xmlparser);
  224. return status;
  225. }
  226. //compare XML name ignoring case and ignoring namespace (returns 0 on match)
  227. #ifdef ASSUME_NO_NAMESPACE
  228. #define XML_Char_icmp_ins XML_Char_icmp
  229. #else
  230. int XML_Char_icmp_ins (const XML_Char* value, const XML_Char* name)
  231. {
  232. size_t valuelen = XML_Char_len(value);
  233. size_t namelen = XML_Char_len(name);
  234. if (valuelen == namelen)
  235. return XML_Char_icmp(value, name);
  236. if (valuelen > namelen) {
  237. if (value[valuelen - namelen - 1] != ':')
  238. return 1;
  239. return XML_Char_icmp(value + (valuelen - namelen), name);
  240. }
  241. return -1;
  242. }
  243. #endif
  244. //get expat attribute by name, returns NULL if not found
  245. const XML_Char* get_expat_attr_by_name (const XML_Char** atts, const XML_Char* name)
  246. {
  247. const XML_Char** p = atts;
  248. if (p) {
  249. while (*p) {
  250. //if (XML_Char_icmp(*p++, name) == 0)
  251. if (XML_Char_icmp_ins(*p++, name) == 0)
  252. return *p;
  253. p++;
  254. }
  255. }
  256. return NULL;
  257. }
  258. //generate .rels filename, returns NULL on error, caller must free result
  259. XML_Char* get_relationship_filename (const XML_Char* filename)
  260. {
  261. XML_Char* result;
  262. size_t filenamelen = XML_Char_len(filename);
  263. if ((result = XML_Char_malloc(filenamelen + 12)) != NULL) {
  264. size_t i = filenamelen;
  265. while (i > 0) {
  266. if (filename[i - 1] == '/')
  267. break;
  268. i--;
  269. }
  270. XML_Char_poscpy(result, 0, filename, i);
  271. XML_Char_poscpy(result, i, X("_rels/"), 6);
  272. XML_Char_poscpy(result, i + 6, filename + i, filenamelen - i);
  273. XML_Char_poscpy(result, filenamelen + 6, X(".rels"), 6);
  274. }
  275. return result;
  276. }
  277. //join basepath and filename (caller must free result)
  278. XML_Char* join_basepath_filename (const XML_Char* basepath, const XML_Char* filename)
  279. {
  280. XML_Char* result = NULL;
  281. if (filename && *filename) {
  282. if (filename[0] == '/' && filename[1]) {
  283. //file is absolute: remove leading slash
  284. result = XML_Char_dup(filename + 1);
  285. } else {
  286. //file is relative: prepend base path
  287. size_t basepathlen = (basepath ? XML_Char_len(basepath) : 0);
  288. size_t filenamelen = XML_Char_len(filename);
  289. if ((result = XML_Char_malloc(basepathlen + filenamelen + 1)) != NULL) {
  290. if (basepathlen > 0)
  291. XML_Char_poscpy(result, 0, basepath, basepathlen);
  292. XML_Char_poscpy(result, basepathlen, filename, filenamelen);
  293. result[basepathlen + filenamelen] = 0;
  294. }
  295. }
  296. }
  297. return result;
  298. }
  299. //determine column number based on cell coordinate (e.g. "A1"), returns 1-based column number or 0 on error
  300. size_t get_col_nr (const XML_Char* A1col)
  301. {
  302. const XML_Char* p = A1col;
  303. size_t result = 0;
  304. if (p) {
  305. while (*p) {
  306. if (*p >= 'A' && *p <= 'Z')
  307. result = result * 26 + (*p - 'A') + 1;
  308. else if (*p >= 'a' && *p <= 'z')
  309. result = result * 26 + (*p - 'a') + 1;
  310. else if (*p >= '0' && *p <= '9' && p != A1col)
  311. return result;
  312. else
  313. break;
  314. p++;
  315. }
  316. }
  317. return 0;
  318. }
  319. //determine row number based on cell coordinate (e.g. "A1"), returns 1-based row number or 0 on error
  320. size_t get_row_nr (const XML_Char* A1col)
  321. {
  322. const XML_Char* p = A1col;
  323. size_t result = 0;
  324. if (p) {
  325. while (*p) {
  326. if ((*p >= 'A' && *p <= 'Z') || (*p >= 'a' && *p <= 'z'))
  327. ;
  328. else if (*p >= '0' && *p <= '9' && p != A1col)
  329. result = result * 10 + (*p - '0');
  330. else
  331. return 0;
  332. p++;
  333. }
  334. }
  335. return result;
  336. }
  337. ////////////////////////////////////////////////////////////////////////
  338. struct xlsxio_read_struct {
  339. ZIPFILETYPE* zip;
  340. };
  341. DLL_EXPORT_XLSXIO xlsxioreader xlsxioread_open (const char* filename)
  342. {
  343. xlsxioreader result;
  344. if ((result = (xlsxioreader)malloc(sizeof(struct xlsxio_read_struct))) != NULL) {
  345. #ifdef USE_MINIZIP
  346. if ((result->zip = unzOpen(filename)) == NULL) {
  347. #else
  348. if ((result->zip = zip_open(filename, ZIP_RDONLY, NULL)) == NULL) {
  349. #endif
  350. free(result);
  351. return NULL;
  352. }
  353. }
  354. return result;
  355. }
  356. #ifdef USE_MINIZIP
  357. struct minizip_io_filehandle_data {
  358. int filehandle;
  359. };
  360. voidpf ZCALLBACK minizip_io_filehandle_open_file_fn (voidpf opaque, const char* filename, int mode)
  361. {
  362. if (!opaque || ((struct minizip_io_filehandle_data*)opaque)->filehandle < 0)
  363. return NULL;
  364. return &((struct minizip_io_filehandle_data*)opaque)->filehandle;
  365. }
  366. uLong ZCALLBACK minizip_io_filehandle_read_file_fn (voidpf opaque, voidpf stream, void* buf, uLong size)
  367. {
  368. IOSIZETYPE len;
  369. if (!opaque || !stream || !buf || size == 0)
  370. return 0;
  371. if ((len = IOFN(read)(*(int*)stream, buf, size)) < 0)
  372. return 0;
  373. return len;
  374. }
  375. /*
  376. uLong ZCALLBACK minizip_io_filehandle_write_file_fn (voidpf opaque, voidpf stream, const void* buf, uLong size)
  377. {
  378. return 0;
  379. }
  380. */
  381. int ZCALLBACK minizip_io_filehandle_close_file_fn (voidpf opaque, voidpf stream)
  382. {
  383. if (stream)
  384. close(*(int*)stream);
  385. free(opaque);
  386. return 0;
  387. }
  388. int ZCALLBACK minizip_io_filehandle_testerror_file_fn (voidpf opaque, voidpf stream)
  389. {
  390. return 0;
  391. }
  392. long ZCALLBACK minizip_io_filehandle_tell_file_fn (voidpf opaque, voidpf stream)
  393. {
  394. return IOFN(lseek)(*(int*)stream, 0, SEEK_CUR);
  395. }
  396. long ZCALLBACK minizip_io_filehandle_seek_file_fn (voidpf opaque, voidpf stream, uLong offset, int origin)
  397. {
  398. int whence;
  399. if (!opaque || !stream)
  400. return -1;
  401. switch (origin) {
  402. case ZLIB_FILEFUNC_SEEK_CUR :
  403. whence = SEEK_CUR;
  404. break;
  405. case ZLIB_FILEFUNC_SEEK_END :
  406. whence = SEEK_END;
  407. break;
  408. case ZLIB_FILEFUNC_SEEK_SET :
  409. whence = SEEK_SET;
  410. break;
  411. default :
  412. return -1;
  413. }
  414. return (IOFN(lseek)(*(int*)stream, offset, whence) >= 0 ? 0 : -1);
  415. }
  416. #endif
  417. DLL_EXPORT_XLSXIO xlsxioreader xlsxioread_open_filehandle (int filehandle)
  418. {
  419. xlsxioreader result;
  420. if ((result = (xlsxioreader)malloc(sizeof(struct xlsxio_read_struct))) != NULL) {
  421. #ifdef USE_MINIZIP
  422. zlib_filefunc_def minizip_io_filehandle_functions;
  423. if ((minizip_io_filehandle_functions.opaque = malloc(sizeof(struct minizip_io_filehandle_data))) == NULL) {
  424. free(result);
  425. return NULL;
  426. }
  427. minizip_io_filehandle_functions.zopen_file = minizip_io_filehandle_open_file_fn;
  428. minizip_io_filehandle_functions.zread_file = minizip_io_filehandle_read_file_fn;
  429. minizip_io_filehandle_functions.zwrite_file = /*minizip_io_filehandle_write_file_fn*/NULL;
  430. minizip_io_filehandle_functions.ztell_file = minizip_io_filehandle_tell_file_fn;
  431. minizip_io_filehandle_functions.zseek_file = minizip_io_filehandle_seek_file_fn;
  432. minizip_io_filehandle_functions.zclose_file = minizip_io_filehandle_close_file_fn;
  433. minizip_io_filehandle_functions.zerror_file = minizip_io_filehandle_testerror_file_fn;
  434. ((struct minizip_io_filehandle_data*)minizip_io_filehandle_functions.opaque)->filehandle = filehandle;
  435. if ((result->zip = unzOpen2(NULL, &minizip_io_filehandle_functions)) == NULL) {
  436. free(result);
  437. return NULL;
  438. }
  439. #else
  440. if ((result->zip = zip_fdopen(filehandle, ZIP_RDONLY, NULL)) == NULL) {
  441. free(result);
  442. return NULL;
  443. }
  444. #endif
  445. }
  446. return result;
  447. }
  448. #ifdef USE_MINIZIP
  449. struct minizip_io_memory_data {
  450. void* data;
  451. uint64_t datalen;
  452. int freedata;
  453. };
  454. struct minizip_io_memory_handle {
  455. uint64_t pos;
  456. };
  457. voidpf ZCALLBACK minizip_io_memory_open_file_fn (voidpf opaque, const char* filename, int mode)
  458. {
  459. struct minizip_io_memory_handle* result;
  460. if (!opaque || !((struct minizip_io_memory_data*)opaque)->data)
  461. return NULL;
  462. if ((result = (struct minizip_io_memory_handle*)malloc(sizeof(struct minizip_io_memory_handle))) != NULL) {
  463. result->pos = 0;
  464. }
  465. return result;
  466. }
  467. uLong ZCALLBACK minizip_io_memory_read_file_fn (voidpf opaque, voidpf stream, void* buf, uLong size)
  468. {
  469. uLong len;
  470. if (!opaque || !stream || !buf || size == 0)
  471. return 0;
  472. if (((struct minizip_io_memory_handle*)stream)->pos + size <= ((struct minizip_io_memory_data*)opaque)->datalen)
  473. len = size;
  474. else
  475. len = ((struct minizip_io_memory_data*)opaque)->datalen - ((struct minizip_io_memory_handle*)stream)->pos;
  476. memcpy(buf, (char *)(((struct minizip_io_memory_data*)opaque)->data) + ((struct minizip_io_memory_handle*)stream)->pos, len);
  477. ((struct minizip_io_memory_handle*)stream)->pos += len;
  478. return len;
  479. }
  480. /*
  481. uLong ZCALLBACK minizip_io_memory_write_file_fn (voidpf opaque, voidpf stream, const void* buf, uLong size)
  482. {
  483. return 0;
  484. }
  485. */
  486. int ZCALLBACK minizip_io_memory_close_file_fn (voidpf opaque, voidpf stream)
  487. {
  488. free(stream);
  489. if (opaque && ((struct minizip_io_memory_data*)opaque)->freedata)
  490. free(((struct minizip_io_memory_data*)opaque)->data);
  491. free(opaque);
  492. return 0;
  493. }
  494. int ZCALLBACK minizip_io_memory_testerror_file_fn (voidpf opaque, voidpf stream)
  495. {
  496. return 0;
  497. }
  498. long ZCALLBACK minizip_io_memory_tell_file_fn (voidpf opaque, voidpf stream)
  499. {
  500. if (!opaque || !stream)
  501. return 0;
  502. return ((struct minizip_io_memory_handle*)stream)->pos;
  503. }
  504. long ZCALLBACK minizip_io_memory_seek_file_fn (voidpf opaque, voidpf stream, uLong offset, int origin)
  505. {
  506. switch (origin) {
  507. case ZLIB_FILEFUNC_SEEK_CUR :
  508. /*if (offset < 0) {
  509. if (((struct minizip_io_memory_handle*)stream)->pos < -offset)
  510. ((struct minizip_io_memory_handle*)stream)->pos = 0;
  511. else
  512. ((struct minizip_io_memory_handle*)stream)->pos += offset;
  513. } else*/ {
  514. if (((struct minizip_io_memory_handle*)stream)->pos + offset > ((struct minizip_io_memory_data*)opaque)->datalen)
  515. ((struct minizip_io_memory_handle*)stream)->pos = ((struct minizip_io_memory_data*)opaque)->datalen;
  516. else
  517. ((struct minizip_io_memory_handle*)stream)->pos += offset;
  518. }
  519. break;
  520. case ZLIB_FILEFUNC_SEEK_END :
  521. /*if (offset < 0) {
  522. if (((struct minizip_io_memory_data*)opaque)->datalen < -offset)
  523. ((struct minizip_io_memory_handle*)stream)->pos = 0;
  524. else
  525. ((struct minizip_io_memory_handle*)stream)->pos = ((struct minizip_io_memory_data*)opaque)->datalen + offset;
  526. } else*/ {
  527. ((struct minizip_io_memory_handle*)stream)->pos = ((struct minizip_io_memory_data*)opaque)->datalen;
  528. }
  529. break;
  530. case ZLIB_FILEFUNC_SEEK_SET :
  531. /*if (offset < 0) {
  532. ((struct minizip_io_memory_handle*)stream)->pos = 0;
  533. } else*/ {
  534. if (offset > ((struct minizip_io_memory_data*)opaque)->datalen)
  535. ((struct minizip_io_memory_handle*)stream)->pos = ((struct minizip_io_memory_data*)opaque)->datalen;
  536. else
  537. ((struct minizip_io_memory_handle*)stream)->pos = offset;
  538. }
  539. ((struct minizip_io_memory_handle*)stream)->pos = offset;
  540. break;
  541. default :
  542. return -1;
  543. }
  544. return 0;
  545. }
  546. #endif
  547. DLL_EXPORT_XLSXIO xlsxioreader xlsxioread_open_memory (void* data, uint64_t datalen, int freedata)
  548. {
  549. xlsxioreader result;
  550. #ifdef USE_MINIZIP
  551. if ((result = (xlsxioreader)malloc(sizeof(struct xlsxio_read_struct))) != NULL) {
  552. zlib_filefunc_def minizip_io_memory_functions;
  553. if ((minizip_io_memory_functions.opaque = malloc(sizeof(struct minizip_io_memory_data))) == NULL) {
  554. free(result);
  555. return NULL;
  556. }
  557. minizip_io_memory_functions.zopen_file = minizip_io_memory_open_file_fn;
  558. minizip_io_memory_functions.zread_file = minizip_io_memory_read_file_fn;
  559. minizip_io_memory_functions.zwrite_file = /*minizip_io_memory_write_file_fn*/NULL;
  560. minizip_io_memory_functions.ztell_file = minizip_io_memory_tell_file_fn;
  561. minizip_io_memory_functions.zseek_file = minizip_io_memory_seek_file_fn;
  562. minizip_io_memory_functions.zclose_file = minizip_io_memory_close_file_fn;
  563. minizip_io_memory_functions.zerror_file = minizip_io_memory_testerror_file_fn;
  564. ((struct minizip_io_memory_data*)minizip_io_memory_functions.opaque)->data = data;
  565. ((struct minizip_io_memory_data*)minizip_io_memory_functions.opaque)->datalen = datalen;
  566. ((struct minizip_io_memory_data*)minizip_io_memory_functions.opaque)->freedata = freedata;
  567. if ((result->zip = unzOpen2(NULL, &minizip_io_memory_functions)) == NULL) {
  568. free(result);
  569. return NULL;
  570. }
  571. }
  572. #else
  573. zip_source_t* zipsrc;
  574. if ((zipsrc = zip_source_buffer_create(data, datalen, freedata, NULL)) == NULL) {
  575. return NULL;
  576. }
  577. if ((result = (xlsxioreader)malloc(sizeof(struct xlsxio_read_struct))) != NULL) {
  578. if ((result->zip = zip_open_from_source(zipsrc, ZIP_RDONLY, NULL)) == NULL) {
  579. zip_source_free(zipsrc);
  580. free(result);
  581. return NULL;
  582. }
  583. }
  584. #endif
  585. return result;
  586. }
  587. DLL_EXPORT_XLSXIO void xlsxioread_close (xlsxioreader handle)
  588. {
  589. if (handle) {
  590. //note: no need to call zip_source_free() after successful use in zip_open_from_source()
  591. #ifdef USE_MINIZIP
  592. unzClose(handle->zip);
  593. #else
  594. zip_close(handle->zip);
  595. #endif
  596. free(handle);
  597. }
  598. }
  599. ////////////////////////////////////////////////////////////////////////
  600. //callback function definition for use with iterate_files_by_contenttype
  601. typedef void (*contenttype_file_callback_fn)(ZIPFILETYPE* zip, const XML_Char* filename, const XML_Char* contenttype, void* callbackdata);
  602. struct iterate_files_by_contenttype_callback_data {
  603. ZIPFILETYPE* zip;
  604. const XML_Char* contenttype;
  605. contenttype_file_callback_fn filecallbackfn;
  606. void* filecallbackdata;
  607. };
  608. //expat callback function for element start used by iterate_files_by_contenttype
  609. void iterate_files_by_contenttype_expat_callback_element_start (void* callbackdata, const XML_Char* name, const XML_Char** atts)
  610. {
  611. struct iterate_files_by_contenttype_callback_data* data = (struct iterate_files_by_contenttype_callback_data*)callbackdata;
  612. if (XML_Char_icmp_ins(name, X("Override")) == 0) {
  613. //explicitly specified file
  614. const XML_Char* contenttype;
  615. const XML_Char* partname;
  616. if ((contenttype = get_expat_attr_by_name(atts, X("ContentType"))) != NULL && XML_Char_icmp(contenttype, data->contenttype) == 0) {
  617. if ((partname = get_expat_attr_by_name(atts, X("PartName"))) != NULL) {
  618. if (partname[0] == '/')
  619. partname++;
  620. data->filecallbackfn(data->zip, partname, contenttype, data->filecallbackdata);
  621. }
  622. }
  623. } else if (XML_Char_icmp_ins(name, X("Default")) == 0) {
  624. //by extension
  625. const XML_Char* contenttype;
  626. const XML_Char* extension;
  627. if ((contenttype = get_expat_attr_by_name(atts, X("ContentType"))) != NULL && XML_Char_icmp(contenttype, data->contenttype) == 0) {
  628. if ((extension = get_expat_attr_by_name(atts, X("Extension"))) != NULL) {
  629. XML_Char* filename;
  630. size_t filenamelen;
  631. size_t extensionlen = XML_Char_len(extension);
  632. #ifdef USE_MINIZIP
  633. #define UNZIP_FILENAME_BUFFER_STEP 32
  634. char* buf;
  635. size_t buflen;
  636. int status;
  637. unz_global_info zipglobalinfo;
  638. unzGetGlobalInfo(data->zip, &zipglobalinfo);
  639. buf = (char*)malloc(buflen = UNZIP_FILENAME_BUFFER_STEP);
  640. status = unzGoToFirstFile(data->zip);
  641. while (status == UNZ_OK) {
  642. buf[buflen - 1] = 0;
  643. while ((status = unzGetCurrentFileInfo(data->zip, NULL, buf, buflen, NULL, 0, NULL, 0)) == UNZ_OK && buf[buflen - 1] != 0) {
  644. buflen += UNZIP_FILENAME_BUFFER_STEP;
  645. buf = (char*)realloc(buf, buflen);
  646. buf[buflen - 1] = 0;
  647. }
  648. if (status != UNZ_OK)
  649. break;
  650. filename = XML_Char_dupchar(buf);
  651. status = unzGoToNextFile(data->zip);
  652. #else
  653. zip_int64_t i;
  654. zip_int64_t zipnumfiles = zip_get_num_entries(data->zip, 0);
  655. for (i = 0; i < zipnumfiles; i++) {
  656. filename = XML_Char_dupchar(zip_get_name(data->zip, i, ZIP_FL_ENC_GUESS));
  657. #endif
  658. filenamelen = XML_Char_len(filename);
  659. if (filenamelen > extensionlen && filename[filenamelen - extensionlen - 1] == '.' && XML_Char_icmp(filename + filenamelen - extensionlen, extension) == 0) {
  660. data->filecallbackfn(data->zip, filename, contenttype, data->filecallbackdata);
  661. }
  662. free(filename);
  663. }
  664. #ifdef USE_MINIZIP
  665. free(buf);
  666. #endif
  667. }
  668. }
  669. }
  670. }
  671. //list file names by content type
  672. int iterate_files_by_contenttype (ZIPFILETYPE* zip, const XML_Char* contenttype, contenttype_file_callback_fn filecallbackfn, void* filecallbackdata, XML_Parser* xmlparser)
  673. {
  674. struct iterate_files_by_contenttype_callback_data callbackdata = {
  675. .zip = zip,
  676. .contenttype = contenttype,
  677. .filecallbackfn = filecallbackfn,
  678. .filecallbackdata = filecallbackdata
  679. };
  680. return expat_process_zip_file(zip, X("[Content_Types].xml"), iterate_files_by_contenttype_expat_callback_element_start, NULL, NULL, &callbackdata, xmlparser);
  681. }
  682. ////////////////////////////////////////////////////////////////////////
  683. //callback structure used by main_sheet_list_expat_callback_element_start
  684. struct main_sheet_list_callback_data {
  685. XML_Parser xmlparser;
  686. xlsxioread_list_sheets_callback_fn callback;
  687. void* callbackdata;
  688. };
  689. //callback used by xlsxioread_list_sheets
  690. void main_sheet_list_expat_callback_element_start (void* callbackdata, const XML_Char* name, const XML_Char** atts)
  691. {
  692. struct main_sheet_list_callback_data* data = (struct main_sheet_list_callback_data*)callbackdata;
  693. if (data && data->callback) {
  694. if (XML_Char_icmp_ins(name, X("sheet")) == 0) {
  695. const XML_Char* sheetname;
  696. //const XML_Char* relid = get_expat_attr_by_name(atts, X("r:id"));
  697. if ((sheetname = get_expat_attr_by_name(atts, X("name"))) != NULL) {
  698. if (data->callback) {
  699. if ((*data->callback)(sheetname, data->callbackdata) != 0) {
  700. XML_StopParser(data->xmlparser, XML_FALSE);
  701. return;
  702. }
  703. /*
  704. } else {
  705. //for non-calback method suspend here
  706. XML_StopParser(data->xmlparser, XML_TRUE);
  707. */
  708. }
  709. }
  710. }
  711. }
  712. }
  713. //process contents each sheet listed in main sheet
  714. void xlsxioread_list_sheets_callback (ZIPFILETYPE* zip, const XML_Char* filename, const XML_Char* contenttype, void* callbackdata)
  715. {
  716. //get sheet information from file
  717. expat_process_zip_file(zip, filename, main_sheet_list_expat_callback_element_start, NULL, NULL, callbackdata, &((struct main_sheet_list_callback_data*)callbackdata)->xmlparser);
  718. }
  719. //list all worksheets
  720. DLL_EXPORT_XLSXIO void xlsxioread_list_sheets (xlsxioreader handle, xlsxioread_list_sheets_callback_fn callback, void* callbackdata)
  721. {
  722. if (!handle || !callback)
  723. return;
  724. //process contents of main sheet
  725. struct main_sheet_list_callback_data sheetcallbackdata = {
  726. .xmlparser = NULL,
  727. .callback = callback,
  728. .callbackdata = callbackdata
  729. };
  730. iterate_files_by_contenttype(handle->zip, xlsx_content_type, xlsxioread_list_sheets_callback, &sheetcallbackdata, &sheetcallbackdata.xmlparser);
  731. iterate_files_by_contenttype(handle->zip, xlsm_content_type, xlsxioread_list_sheets_callback, &sheetcallbackdata, &sheetcallbackdata.xmlparser);
  732. iterate_files_by_contenttype(handle->zip, xltx_content_type, xlsxioread_list_sheets_callback, &sheetcallbackdata, &sheetcallbackdata.xmlparser);
  733. iterate_files_by_contenttype(handle->zip, xltm_content_type, xlsxioread_list_sheets_callback, &sheetcallbackdata, &sheetcallbackdata.xmlparser);
  734. }
  735. ////////////////////////////////////////////////////////////////////////
  736. //callback data structure used by main_sheet_get_sheetfile_callback
  737. struct main_sheet_get_rels_callback_data {
  738. XML_Parser xmlparser;
  739. const XML_Char* sheetname;
  740. XML_Char* basepath;
  741. XML_Char* sheetrelid;
  742. XML_Char* sheetfile;
  743. XML_Char* sharedstringsfile;
  744. XML_Char* stylesfile;
  745. };
  746. //determine relationship id for specific sheet name
  747. void main_sheet_get_relid_expat_callback_element_start (void* callbackdata, const XML_Char* name, const XML_Char** atts)
  748. {
  749. struct main_sheet_get_rels_callback_data* data = (struct main_sheet_get_rels_callback_data*)callbackdata;
  750. if (XML_Char_icmp_ins(name, X("sheet")) == 0) {
  751. const XML_Char* sheetname = get_expat_attr_by_name(atts, X("name"));
  752. if (!data->sheetname || XML_Char_icmp(sheetname, data->sheetname) == 0) {
  753. const XML_Char* relid = get_expat_attr_by_name(atts, X("r:id"));
  754. if (relid && *relid) {
  755. data->sheetrelid = XML_Char_dup(relid);
  756. XML_StopParser(data->xmlparser, XML_FALSE);
  757. return;
  758. }
  759. }
  760. }
  761. }
  762. //determine file names for specific relationship id
  763. void main_sheet_get_sheetfile_expat_callback_element_start (void* callbackdata, const XML_Char* name, const XML_Char** atts)
  764. {
  765. struct main_sheet_get_rels_callback_data* data = (struct main_sheet_get_rels_callback_data*)callbackdata;
  766. if (data->sheetrelid) {
  767. if (XML_Char_icmp_ins(name, X("Relationship")) == 0) {
  768. const XML_Char* reltype;
  769. if ((reltype = get_expat_attr_by_name(atts, X("Type"))) != NULL) {
  770. if (XML_Char_icmp(reltype, X("http://schemas.openxmlformats.org/officeDocument/2006/relationships/worksheet")) == 0) {
  771. const XML_Char* relid = get_expat_attr_by_name(atts, X("Id"));
  772. if (XML_Char_icmp(relid, data->sheetrelid) == 0) {
  773. const XML_Char* filename = get_expat_attr_by_name(atts, X("Target"));
  774. if (filename && *filename) {
  775. data->sheetfile = join_basepath_filename(data->basepath, filename);
  776. }
  777. }
  778. } else if (XML_Char_icmp(reltype, X("http://schemas.openxmlformats.org/officeDocument/2006/relationships/sharedStrings")) == 0) {
  779. const XML_Char* filename = get_expat_attr_by_name(atts, X("Target"));
  780. if (filename && *filename) {
  781. data->sharedstringsfile = join_basepath_filename(data->basepath, filename);
  782. }
  783. } else if (XML_Char_icmp(reltype, X("http://schemas.openxmlformats.org/officeDocument/2006/relationships/styles")) == 0) {
  784. const XML_Char* filename = get_expat_attr_by_name(atts, X("Target"));
  785. if (filename && *filename) {
  786. data->stylesfile = join_basepath_filename(data->basepath, filename);
  787. }
  788. }
  789. }
  790. }
  791. }
  792. }
  793. //determine the file name for a specified sheet name
  794. void main_sheet_get_sheetfile_callback (ZIPFILETYPE* zip, const XML_Char* filename, const XML_Char* contenttype, void* callbackdata)
  795. {
  796. struct main_sheet_get_rels_callback_data* data = (struct main_sheet_get_rels_callback_data*)callbackdata;
  797. if (!data->sheetrelid) {
  798. expat_process_zip_file(zip, filename, main_sheet_get_relid_expat_callback_element_start, NULL, NULL, callbackdata, &data->xmlparser);
  799. }
  800. if (data->sheetrelid) {
  801. XML_Char* relfilename;
  802. //determine base name (including trailing slash)
  803. size_t i = XML_Char_len(filename);
  804. while (i > 0) {
  805. if (filename[i - 1] == '/')
  806. break;
  807. i--;
  808. }
  809. if (data->basepath)
  810. free(data->basepath);
  811. if ((data->basepath = XML_Char_malloc(i + 1)) != NULL) {
  812. XML_Char_poscpy(data->basepath, 0, filename, i);
  813. data->basepath[i] = 0;
  814. }
  815. //find sheet filename in relationship contents
  816. if ((relfilename = get_relationship_filename(filename)) != NULL) {
  817. expat_process_zip_file(zip, relfilename, main_sheet_get_sheetfile_expat_callback_element_start, NULL, NULL, callbackdata, &data->xmlparser);
  818. free(relfilename);
  819. } else {
  820. free(data->sheetrelid);
  821. data->sheetrelid = NULL;
  822. if (data->basepath) {
  823. free(data->basepath);
  824. data->basepath = NULL;
  825. }
  826. }
  827. }
  828. }
  829. ////////////////////////////////////////////////////////////////////////
  830. typedef enum {
  831. none,
  832. value_string,
  833. inline_string,
  834. shared_string
  835. } cell_string_type_enum;
  836. #define XLSXIOREAD_NO_CALLBACK 0x80
  837. struct data_sheet_callback_data {
  838. XML_Parser xmlparser;
  839. struct sharedstringlist* sharedstrings;
  840. size_t rownr;
  841. size_t colnr;
  842. size_t cols;
  843. XML_Char* celldata;
  844. size_t celldatalen;
  845. cell_string_type_enum cell_string_type;
  846. unsigned int flags;
  847. XML_Char* skiptag; //tag to skip
  848. size_t skiptagcount; //nesting level for current tag to skip
  849. XML_StartElementHandler skip_start; //start handler to set after skipping
  850. XML_EndElementHandler skip_end; //end handler to set after skipping
  851. XML_CharacterDataHandler skip_data; //data handler to set after skipping
  852. xlsxioread_process_row_callback_fn sheet_row_callback;
  853. xlsxioread_process_cell_callback_fn sheet_cell_callback;
  854. void* callbackdata;
  855. };
  856. void data_sheet_callback_data_initialize (struct data_sheet_callback_data* data, struct sharedstringlist* sharedstrings, unsigned int flags, xlsxioread_process_cell_callback_fn cell_callback, xlsxioread_process_row_callback_fn row_callback, void* callbackdata)
  857. {
  858. data->xmlparser = NULL;
  859. data->sharedstrings = sharedstrings;
  860. data->rownr = 0;
  861. data->colnr = 0;
  862. data->cols = 0;
  863. data->celldata = NULL;
  864. data->celldatalen = 0;
  865. data->cell_string_type = none;
  866. data->flags = flags;
  867. data->skiptag = NULL;
  868. data->skiptagcount = 0;
  869. data->skip_start = NULL;
  870. data->skip_end = NULL;
  871. data->skip_data = NULL;
  872. data->sheet_cell_callback = cell_callback;
  873. data->sheet_row_callback = row_callback;
  874. data->callbackdata = callbackdata;
  875. }
  876. void data_sheet_callback_data_cleanup (struct data_sheet_callback_data* data)
  877. {
  878. sharedstringlist_destroy(data->sharedstrings);
  879. free(data->celldata);
  880. free(data->skiptag);
  881. }
  882. void data_sheet_expat_callback_skip_tag_start (void* callbackdata, const XML_Char* name, const XML_Char** atts)
  883. {
  884. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  885. if (name && XML_Char_icmp_ins(name, data->skiptag) == 0) {
  886. //increment nesting level
  887. data->skiptagcount++;
  888. }
  889. }
  890. void data_sheet_expat_callback_skip_tag_end (void* callbackdata, const XML_Char* name)
  891. {
  892. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  893. if (!name || XML_Char_icmp_ins(name, data->skiptag) == 0) {
  894. if (--data->skiptagcount == 0) {
  895. //restore handlers when done skipping
  896. XML_SetElementHandler(data->xmlparser, data->skip_start, data->skip_end);
  897. XML_SetCharacterDataHandler(data->xmlparser, data->skip_data);
  898. free(data->skiptag);
  899. data->skiptag = NULL;
  900. }
  901. }
  902. }
  903. void data_sheet_expat_callback_find_worksheet_start (void* callbackdata, const XML_Char* name, const XML_Char** atts);
  904. void data_sheet_expat_callback_find_worksheet_end (void* callbackdata, const XML_Char* name);
  905. void data_sheet_expat_callback_find_sheetdata_start (void* callbackdata, const XML_Char* name, const XML_Char** atts);
  906. void data_sheet_expat_callback_find_sheetdata_end (void* callbackdata, const XML_Char* name);
  907. void data_sheet_expat_callback_find_row_start (void* callbackdata, const XML_Char* name, const XML_Char** atts);
  908. void data_sheet_expat_callback_find_row_end (void* callbackdata, const XML_Char* name);
  909. void data_sheet_expat_callback_find_cell_start (void* callbackdata, const XML_Char* name, const XML_Char** atts);
  910. void data_sheet_expat_callback_find_cell_end (void* callbackdata, const XML_Char* name);
  911. void data_sheet_expat_callback_find_value_start (void* callbackdata, const XML_Char* name, const XML_Char** atts);
  912. void data_sheet_expat_callback_find_value_end (void* callbackdata, const XML_Char* name);
  913. void data_sheet_expat_callback_value_data (void* callbackdata, const XML_Char* buf, int buflen);
  914. void data_sheet_expat_callback_find_worksheet_start (void* callbackdata, const XML_Char* name, const XML_Char** atts)
  915. {
  916. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  917. if (XML_Char_icmp_ins(name, X("worksheet")) == 0) {
  918. XML_SetElementHandler(data->xmlparser, data_sheet_expat_callback_find_sheetdata_start, NULL);
  919. }
  920. }
  921. void data_sheet_expat_callback_find_worksheet_end (void* callbackdata, const XML_Char* name)
  922. {
  923. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  924. if (XML_Char_icmp_ins(name, X("worksheet")) == 0) {
  925. XML_SetElementHandler(data->xmlparser, data_sheet_expat_callback_find_worksheet_start, NULL);
  926. }
  927. }
  928. void data_sheet_expat_callback_find_sheetdata_start (void* callbackdata, const XML_Char* name, const XML_Char** atts)
  929. {
  930. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  931. if (XML_Char_icmp_ins(name, X("sheetData")) == 0) {
  932. XML_SetElementHandler(data->xmlparser, data_sheet_expat_callback_find_row_start, data_sheet_expat_callback_find_sheetdata_end);
  933. }
  934. }
  935. void data_sheet_expat_callback_find_sheetdata_end (void* callbackdata, const XML_Char* name)
  936. {
  937. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  938. if (XML_Char_icmp_ins(name, X("sheetData")) == 0) {
  939. XML_SetElementHandler(data->xmlparser, data_sheet_expat_callback_find_sheetdata_start, data_sheet_expat_callback_find_worksheet_end);
  940. } else {
  941. data_sheet_expat_callback_find_worksheet_end(callbackdata, name);
  942. }
  943. }
  944. void data_sheet_expat_callback_find_row_start (void* callbackdata, const XML_Char* name, const XML_Char** atts)
  945. {
  946. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  947. if (XML_Char_icmp_ins(name, X("row")) == 0) {
  948. const XML_Char* hidden = get_expat_attr_by_name(atts, X("hidden"));
  949. if (!hidden || XML_Char_tol(hidden) == 0 || !(data->flags & XLSXIOREAD_SKIP_HIDDEN_ROWS)) {
  950. //nesting level for current tag to skip
  951. //start handler to set after skipping
  952. //end handler to set after skipping
  953. //data handler to set after skipping
  954. data->rownr++;
  955. data->colnr = 0;
  956. XML_SetElementHandler(data->xmlparser, data_sheet_expat_callback_find_cell_start, data_sheet_expat_callback_find_row_end);
  957. //for non-calback method suspend here on new row
  958. if (data->flags & XLSXIOREAD_NO_CALLBACK) {
  959. XML_StopParser(data->xmlparser, XML_TRUE);
  960. }
  961. } else {
  962. //skip hidden tow
  963. XML_SetElementHandler(data->xmlparser, NULL, data_sheet_expat_callback_find_row_end);
  964. }
  965. }
  966. }
  967. void data_sheet_expat_callback_find_row_end (void* callbackdata, const XML_Char* name)
  968. {
  969. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  970. if (XML_Char_icmp_ins(name, X("row")) == 0) {
  971. //determine number of columns based on first row
  972. if (data->rownr == 1 && data->cols == 0)
  973. data->cols = data->colnr;
  974. //add empty columns if needed
  975. if (!(data->flags & XLSXIOREAD_NO_CALLBACK) && data->sheet_cell_callback && !(data->flags & XLSXIOREAD_SKIP_EMPTY_CELLS)) {
  976. while (data->colnr < data->cols) {
  977. if ((*data->sheet_cell_callback)(data->rownr, data->colnr + 1, NULL, data->callbackdata)) {
  978. XML_StopParser(data->xmlparser, XML_FALSE);
  979. return;
  980. }
  981. data->colnr++;
  982. }
  983. }
  984. free(data->celldata);
  985. data->celldata = NULL;
  986. XML_SetElementHandler(data->xmlparser, data_sheet_expat_callback_find_row_start, data_sheet_expat_callback_find_sheetdata_end);
  987. //process end of row
  988. if (!(data->flags & XLSXIOREAD_NO_CALLBACK)) {
  989. if (data->sheet_row_callback) {
  990. if ((*data->sheet_row_callback)(data->rownr, data->colnr, data->callbackdata)) {
  991. XML_StopParser(data->xmlparser, XML_FALSE);
  992. return;
  993. }
  994. }
  995. } else {
  996. //for non-calback method suspend here on end of row
  997. if (data->flags & XLSXIOREAD_NO_CALLBACK) {
  998. XML_StopParser(data->xmlparser, XML_TRUE);
  999. }
  1000. }
  1001. } else {
  1002. data_sheet_expat_callback_find_sheetdata_end(callbackdata, name);
  1003. }
  1004. }
  1005. void data_sheet_expat_callback_find_cell_start (void* callbackdata, const XML_Char* name, const XML_Char** atts)
  1006. {
  1007. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  1008. if (XML_Char_icmp_ins(name, X("c")) == 0) {
  1009. const XML_Char* t = get_expat_attr_by_name(atts, X("r"));
  1010. size_t cellcolnr = get_col_nr(t);
  1011. //skip everything when out of bounds
  1012. if (cellcolnr && data->cols && (data->flags & XLSXIOREAD_SKIP_EXTRA_CELLS) && cellcolnr > data->cols) {
  1013. data->colnr = cellcolnr - 1;
  1014. return;
  1015. }
  1016. //insert empty rows if needed
  1017. if (data->colnr == 0) {
  1018. size_t cellrownr = get_row_nr(t);
  1019. if (cellrownr) {
  1020. if (!(data->flags & XLSXIOREAD_SKIP_EMPTY_ROWS) && !(data->flags & XLSXIOREAD_NO_CALLBACK)) {
  1021. while (data->rownr < cellrownr) {
  1022. //insert empty columns
  1023. if (!(data->flags & XLSXIOREAD_SKIP_EMPTY_CELLS) && data->sheet_cell_callback) {
  1024. while (data->colnr < data->cols) {
  1025. if ((*data->sheet_cell_callback)(data->rownr, data->colnr + 1, NULL, data->callbackdata)) {
  1026. XML_StopParser(data->xmlparser, XML_FALSE);
  1027. return;
  1028. }
  1029. data->colnr++;
  1030. }
  1031. }
  1032. //finish empty row
  1033. if (data->sheet_row_callback) {
  1034. if ((*data->sheet_row_callback)(data->rownr, data->cols, data->callbackdata)) {
  1035. XML_StopParser(data->xmlparser, XML_FALSE);
  1036. return;
  1037. }
  1038. }
  1039. data->rownr++;
  1040. data->colnr = 0;
  1041. }
  1042. } else {
  1043. data->rownr = cellrownr;
  1044. }
  1045. }
  1046. }
  1047. //insert empty columns if needed
  1048. if (cellcolnr) {
  1049. cellcolnr--;
  1050. if (data->flags & XLSXIOREAD_SKIP_EMPTY_CELLS || data->flags & XLSXIOREAD_NO_CALLBACK) {
  1051. data->colnr = cellcolnr;
  1052. } else {
  1053. while (data->colnr < cellcolnr) {
  1054. if (data->sheet_cell_callback) {
  1055. if ((*data->sheet_cell_callback)(data->rownr, data->colnr + 1, NULL, data->callbackdata)) {
  1056. XML_StopParser(data->xmlparser, XML_FALSE);
  1057. return;
  1058. }
  1059. }
  1060. data->colnr++;
  1061. }
  1062. }
  1063. }
  1064. //determing value type
  1065. if ((t = get_expat_attr_by_name(atts, X("t"))) != NULL && XML_Char_icmp(t, X("s")) == 0)
  1066. data->cell_string_type = shared_string;
  1067. else
  1068. data->cell_string_type = value_string;
  1069. //prepare empty value data
  1070. free(data->celldata);
  1071. data->celldata = NULL;
  1072. data->celldatalen = 0;
  1073. XML_SetElementHandler(data->xmlparser, data_sheet_expat_callback_find_value_start, data_sheet_expat_callback_find_cell_end);
  1074. }
  1075. }
  1076. void data_sheet_expat_callback_find_cell_end (void* callbackdata, const XML_Char* name)
  1077. {
  1078. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  1079. if (XML_Char_icmp_ins(name, X("c")) == 0) {
  1080. //determine value
  1081. if (data->celldata) {
  1082. const XML_Char* s = NULL;
  1083. data->celldata[data->celldatalen] = 0;
  1084. if (data->cell_string_type == shared_string) {
  1085. //get shared string
  1086. XML_Char* p = NULL;
  1087. long num = XML_Char_strtol(data->celldata, &p, 10);
  1088. if (!p || (p != data->celldata && *p == 0)) {
  1089. s = sharedstringlist_get(data->sharedstrings, num);
  1090. free(data->celldata);
  1091. data->celldata = (s ? XML_Char_dup(s) : NULL);
  1092. }
  1093. } else if (data->cell_string_type == none) {
  1094. //unknown value type
  1095. free(data->celldata);
  1096. data->celldata = NULL;
  1097. }
  1098. }
  1099. //reset data
  1100. data->colnr++;
  1101. data->cell_string_type = none;
  1102. data->celldatalen = 0;
  1103. XML_SetElementHandler(data->xmlparser, data_sheet_expat_callback_find_cell_start, data_sheet_expat_callback_find_row_end);
  1104. XML_SetCharacterDataHandler(data->xmlparser, NULL);
  1105. //process data if needed
  1106. if (!(data->cols && (data->flags & XLSXIOREAD_SKIP_EXTRA_CELLS) && data->colnr > data->cols)) {
  1107. //process data
  1108. if (!(data->flags & XLSXIOREAD_NO_CALLBACK)) {
  1109. if (data->sheet_cell_callback) {
  1110. if ((*data->sheet_cell_callback)(data->rownr, data->colnr, data->celldata, data->callbackdata)) {
  1111. XML_StopParser(data->xmlparser, XML_FALSE);
  1112. return;
  1113. }
  1114. }
  1115. } else {
  1116. //for non-calback method suspend here with cell data
  1117. if (!data->celldata)
  1118. data->celldata = XML_Char_dup(X(""));
  1119. XML_StopParser(data->xmlparser, XML_TRUE);
  1120. }
  1121. }
  1122. } else {
  1123. data_sheet_expat_callback_find_row_end(callbackdata, name);
  1124. }
  1125. }
  1126. void data_sheet_expat_callback_find_value_start (void* callbackdata, const XML_Char* name, const XML_Char** atts)
  1127. {
  1128. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  1129. if (XML_Char_icmp_ins(name, X("v")) == 0 || XML_Char_icmp_ins(name, X("t")) == 0) {
  1130. XML_SetElementHandler(data->xmlparser, NULL, data_sheet_expat_callback_find_value_end);
  1131. XML_SetCharacterDataHandler(data->xmlparser, data_sheet_expat_callback_value_data);
  1132. } else if (XML_Char_icmp_ins(name, X("is")) == 0) {
  1133. data->cell_string_type = inline_string;
  1134. } else if (XML_Char_icmp_ins(name, X("rPh")) == 0) {
  1135. data->skiptag = XML_Char_dup(name);
  1136. data->skiptagcount = 1;
  1137. data->skip_start = data_sheet_expat_callback_find_value_start;
  1138. data->skip_end = data_sheet_expat_callback_find_cell_end;
  1139. data->skip_data = NULL;
  1140. XML_SetElementHandler(data->xmlparser, data_sheet_expat_callback_skip_tag_start, data_sheet_expat_callback_skip_tag_end);
  1141. XML_SetCharacterDataHandler(data->xmlparser, NULL);
  1142. }
  1143. }
  1144. void data_sheet_expat_callback_find_value_end (void* callbackdata, const XML_Char* name)
  1145. {
  1146. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  1147. if (XML_Char_icmp_ins(name, X("v")) == 0 || XML_Char_icmp_ins(name, X("t")) == 0) {
  1148. XML_SetElementHandler(data->xmlparser, data_sheet_expat_callback_find_value_start, data_sheet_expat_callback_find_cell_end);
  1149. XML_SetCharacterDataHandler(data->xmlparser, NULL);
  1150. } else if (XML_Char_icmp_ins(name, X("is")) == 0) {
  1151. data->cell_string_type = none;
  1152. } else {
  1153. data_sheet_expat_callback_find_row_end(callbackdata, name);
  1154. }
  1155. }
  1156. void data_sheet_expat_callback_value_data (void* callbackdata, const XML_Char* buf, int buflen)
  1157. {
  1158. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  1159. if (data->cell_string_type != none) {
  1160. if ((data->celldata = XML_Char_realloc(data->celldata, data->celldatalen + buflen + 1)) == NULL) {
  1161. //memory allocation error
  1162. data->celldatalen = 0;
  1163. } else {
  1164. //add new data to value buffer
  1165. XML_Char_poscpy(data->celldata, data->celldatalen, buf, buflen);
  1166. data->celldatalen += buflen;
  1167. }
  1168. }
  1169. }
  1170. ////////////////////////////////////////////////////////////////////////
  1171. struct xlsxio_read_sheet_struct {
  1172. xlsxioreader handle;
  1173. ZIPFILEENTRYTYPE* zipfile;
  1174. struct data_sheet_callback_data processcallbackdata;
  1175. size_t lastrownr;
  1176. size_t paddingrow;
  1177. size_t lastcolnr;
  1178. size_t paddingcol;
  1179. };
  1180. DLL_EXPORT_XLSXIO size_t xlsxioread_sheet_last_row_index (xlsxioreadersheet sheethandle)
  1181. {
  1182. return sheethandle->lastrownr;
  1183. }
  1184. DLL_EXPORT_XLSXIO size_t xlsxioread_sheet_last_column_index (xlsxioreadersheet sheethandle)
  1185. {
  1186. return sheethandle->lastcolnr;
  1187. }
  1188. DLL_EXPORT_XLSXIO unsigned int xlsxioread_sheet_flags (xlsxioreadersheet sheethandle)
  1189. {
  1190. return sheethandle->processcallbackdata.flags;
  1191. }
  1192. DLL_EXPORT_XLSXIO int xlsxioread_process (xlsxioreader handle, const XLSXIOCHAR* sheetname, unsigned int flags, xlsxioread_process_cell_callback_fn cell_callback, xlsxioread_process_row_callback_fn row_callback, void* callbackdata)
  1193. {
  1194. int result = 0;
  1195. //determine sheet file name
  1196. struct main_sheet_get_rels_callback_data getrelscallbackdata = {
  1197. .sheetname = sheetname,
  1198. .basepath = NULL,
  1199. .sheetrelid = NULL,
  1200. .sheetfile = NULL,
  1201. .sharedstringsfile = NULL,
  1202. .stylesfile = NULL
  1203. };
  1204. iterate_files_by_contenttype(handle->zip, xlsx_content_type, main_sheet_get_sheetfile_callback, &getrelscallbackdata, NULL);
  1205. if (!getrelscallbackdata.sheetrelid)
  1206. iterate_files_by_contenttype(handle->zip, xlsm_content_type, main_sheet_get_sheetfile_callback, &getrelscallbackdata, NULL);
  1207. if (!getrelscallbackdata.sheetrelid)
  1208. iterate_files_by_contenttype(handle->zip, xltx_content_type, main_sheet_get_sheetfile_callback, &getrelscallbackdata, NULL);
  1209. if (!getrelscallbackdata.sheetrelid)
  1210. iterate_files_by_contenttype(handle->zip, xltm_content_type, main_sheet_get_sheetfile_callback, &getrelscallbackdata, NULL);
  1211. //process shared strings
  1212. struct sharedstringlist* sharedstrings = NULL;
  1213. if (getrelscallbackdata.sharedstringsfile && getrelscallbackdata.sharedstringsfile[0]) {
  1214. sharedstrings = sharedstringlist_create();
  1215. struct shared_strings_callback_data sharedstringsdata;
  1216. shared_strings_callback_data_initialize(&sharedstringsdata, sharedstrings);
  1217. if (expat_process_zip_file(handle->zip, getrelscallbackdata.sharedstringsfile, shared_strings_callback_find_sharedstringtable_start, NULL, NULL, &sharedstringsdata, &sharedstringsdata.xmlparser) != 0) {
  1218. //no shared strings found
  1219. sharedstringlist_destroy(sharedstrings);
  1220. sharedstrings = NULL;
  1221. }
  1222. shared_strings_callback_data_cleanup(&sharedstringsdata);
  1223. }
  1224. //process sheet
  1225. if (!(flags & XLSXIOREAD_NO_CALLBACK)) {
  1226. //use callback mechanism
  1227. struct data_sheet_callback_data processcallbackdata;
  1228. data_sheet_callback_data_initialize(&processcallbackdata, sharedstrings, flags, cell_callback, row_callback, callbackdata);
  1229. expat_process_zip_file(handle->zip, getrelscallbackdata.sheetfile, data_sheet_expat_callback_find_worksheet_start, NULL, NULL, &processcallbackdata, &processcallbackdata.xmlparser);
  1230. data_sheet_callback_data_cleanup(&processcallbackdata);
  1231. } else {
  1232. //use simplified interface by suspending the XML parser when data is found
  1233. xlsxioreadersheet sheethandle = (xlsxioreadersheet)callbackdata;
  1234. data_sheet_callback_data_initialize(&sheethandle->processcallbackdata, sharedstrings, flags, NULL, NULL, sheethandle);
  1235. if ((sheethandle->zipfile = XML_Char_openzip(sheethandle->handle->zip, getrelscallbackdata.sheetfile, 0)) == NULL) {
  1236. result = 1;
  1237. }
  1238. if ((sheethandle->processcallbackdata.xmlparser = expat_process_zip_file_suspendable(sheethandle->zipfile, data_sheet_expat_callback_find_worksheet_start, NULL, NULL, &sheethandle->processcallbackdata)) == NULL) {
  1239. result = 2;
  1240. }
  1241. }
  1242. //clean up
  1243. free(getrelscallbackdata.basepath);
  1244. free(getrelscallbackdata.sheetrelid);
  1245. free(getrelscallbackdata.sheetfile);
  1246. free(getrelscallbackdata.sharedstringsfile);
  1247. free(getrelscallbackdata.stylesfile);
  1248. return result;
  1249. }
  1250. ////////////////////////////////////////////////////////////////////////
  1251. struct xlsxio_read_sheetlist_struct {
  1252. xlsxioreader handle;
  1253. ZIPFILEENTRYTYPE* zipfile;
  1254. struct main_sheet_list_callback_data sheetcallbackdata;
  1255. XML_Parser xmlparser;
  1256. XML_Char* nextsheetname;
  1257. };
  1258. int xlsxioread_list_sheets_resumable_callback (const XLSXIOCHAR* name, void* callbackdata)
  1259. {
  1260. //struct main_sheet_list_callback_data* data = (struct main_sheet_list_callback_data*)callbackdata;
  1261. xlsxioreadersheetlist data = (xlsxioreadersheetlist)callbackdata;
  1262. data->nextsheetname = XML_Char_dup(name);
  1263. XML_StopParser(data->xmlparser, XML_TRUE);
  1264. return 0;
  1265. }
  1266. void xlsxioread_find_main_sheet_file_callback (ZIPFILETYPE* zip, const XML_Char* filename, const XML_Char* contenttype, void* callbackdata)
  1267. {
  1268. XML_Char** data = (XML_Char**)callbackdata;
  1269. *data = XML_Char_dup(filename);
  1270. }
  1271. DLL_EXPORT_XLSXIO xlsxioreadersheetlist xlsxioread_sheetlist_open (xlsxioreader handle)
  1272. {
  1273. //determine main sheet name
  1274. XML_Char* mainsheetfile = NULL;
  1275. iterate_files_by_contenttype(handle->zip, xlsx_content_type, xlsxioread_find_main_sheet_file_callback, &mainsheetfile, NULL);
  1276. if (!mainsheetfile)
  1277. iterate_files_by_contenttype(handle->zip, xlsm_content_type, xlsxioread_find_main_sheet_file_callback, &mainsheetfile, NULL);
  1278. if (!mainsheetfile)
  1279. iterate_files_by_contenttype(handle->zip, xltx_content_type, xlsxioread_find_main_sheet_file_callback, &mainsheetfile, NULL);
  1280. if (!mainsheetfile)
  1281. iterate_files_by_contenttype(handle->zip, xltm_content_type, xlsxioread_find_main_sheet_file_callback, &mainsheetfile, NULL);
  1282. if (!mainsheetfile)
  1283. return NULL;
  1284. //process contents of main sheet
  1285. xlsxioreadersheetlist result;
  1286. if ((result = (xlsxioreadersheetlist)malloc(sizeof(struct xlsxio_read_sheetlist_struct))) == NULL)
  1287. return NULL;
  1288. result->handle = handle;
  1289. result->sheetcallbackdata.xmlparser = NULL;
  1290. result->sheetcallbackdata.callback = xlsxioread_list_sheets_resumable_callback;
  1291. result->sheetcallbackdata.callbackdata = result;
  1292. result->nextsheetname = NULL;
  1293. if ((result->zipfile = XML_Char_openzip(handle->zip, mainsheetfile, 0)) != NULL) {
  1294. result->xmlparser = expat_process_zip_file_suspendable(result->zipfile, main_sheet_list_expat_callback_element_start, NULL, NULL, &result->sheetcallbackdata);
  1295. }
  1296. //clean up
  1297. free(mainsheetfile);
  1298. return result;
  1299. }
  1300. DLL_EXPORT_XLSXIO void xlsxioread_sheetlist_close (xlsxioreadersheetlist sheetlisthandle)
  1301. {
  1302. if (!sheetlisthandle)
  1303. return;
  1304. if (sheetlisthandle->xmlparser)
  1305. XML_ParserFree(sheetlisthandle->xmlparser);
  1306. if (sheetlisthandle->zipfile)
  1307. #ifdef USE_MINIZIP
  1308. unzCloseCurrentFile(sheetlisthandle->zipfile);
  1309. #else
  1310. zip_fclose(sheetlisthandle->zipfile);
  1311. #endif
  1312. free(sheetlisthandle->nextsheetname);
  1313. free(sheetlisthandle);
  1314. }
  1315. DLL_EXPORT_XLSXIO const XLSXIOCHAR* xlsxioread_sheetlist_next (xlsxioreadersheetlist sheetlisthandle)
  1316. {
  1317. if (!sheetlisthandle->zipfile || !sheetlisthandle->xmlparser)
  1318. return NULL;
  1319. free(sheetlisthandle->nextsheetname);
  1320. sheetlisthandle->nextsheetname = NULL;
  1321. enum XML_Status status;
  1322. if ((status = expat_process_zip_file_resume(sheetlisthandle->zipfile, sheetlisthandle->xmlparser)) != XML_STATUS_SUSPENDED) {
  1323. return NULL;
  1324. }
  1325. return sheetlisthandle->nextsheetname;
  1326. }
  1327. ////////////////////////////////////////////////////////////////////////
  1328. DLL_EXPORT_XLSXIO xlsxioreadersheet xlsxioread_sheet_open (xlsxioreader handle, const XLSXIOCHAR* sheetname, unsigned int flags)
  1329. {
  1330. xlsxioreadersheet result;
  1331. if ((result = (xlsxioreadersheet)malloc(sizeof(struct xlsxio_read_sheet_struct))) == NULL)
  1332. return NULL;
  1333. result->handle = handle;
  1334. result->zipfile = NULL;
  1335. result->lastrownr = 0;
  1336. result->paddingrow = 0;
  1337. result->lastcolnr = 0;
  1338. result->paddingcol = 0;
  1339. xlsxioread_process(handle, sheetname, flags | XLSXIOREAD_NO_CALLBACK, NULL, NULL, result);
  1340. return result;
  1341. }
  1342. DLL_EXPORT_XLSXIO void xlsxioread_sheet_close (xlsxioreadersheet sheethandle)
  1343. {
  1344. if (!sheethandle)
  1345. return;
  1346. if (sheethandle->processcallbackdata.xmlparser)
  1347. XML_ParserFree(sheethandle->processcallbackdata.xmlparser);
  1348. data_sheet_callback_data_cleanup(&sheethandle->processcallbackdata);
  1349. if (sheethandle->zipfile)
  1350. #ifdef USE_MINIZIP
  1351. unzCloseCurrentFile(sheethandle->zipfile);
  1352. #else
  1353. zip_fclose(sheethandle->zipfile);
  1354. #endif
  1355. free(sheethandle);
  1356. }
  1357. DLL_EXPORT_XLSXIO int xlsxioread_sheet_next_row (xlsxioreadersheet sheethandle)
  1358. {
  1359. enum XML_Status status;
  1360. if (!sheethandle) {
  1361. return 0;
  1362. }
  1363. sheethandle->lastcolnr = 0;
  1364. //when padding rows don't retrieve new data
  1365. if (sheethandle->paddingrow) {
  1366. if (sheethandle->paddingrow < sheethandle->processcallbackdata.rownr) {
  1367. return 3;
  1368. } else {
  1369. sheethandle->paddingrow = 0;
  1370. return 2;
  1371. }
  1372. }
  1373. sheethandle->paddingcol = 0;
  1374. //go to beginning of next row
  1375. while ((status = expat_process_zip_file_resume(sheethandle->zipfile, sheethandle->processcallbackdata.xmlparser)) == XML_STATUS_SUSPENDED && sheethandle->processcallbackdata.colnr != 0) {
  1376. }
  1377. return (status == XML_STATUS_SUSPENDED ? 1 : 0);
  1378. }
  1379. DLL_EXPORT_XLSXIO XLSXIOCHAR* xlsxioread_sheet_next_cell (xlsxioreadersheet sheethandle)
  1380. {
  1381. XML_Char* result;
  1382. if (!sheethandle)
  1383. return NULL;
  1384. //append empty column if needed
  1385. if (sheethandle->paddingcol) {
  1386. if (sheethandle->paddingcol > sheethandle->processcallbackdata.cols) {
  1387. //last empty column added, finish row
  1388. sheethandle->paddingcol = 0;
  1389. //when padding rows prepare for the next one
  1390. if (sheethandle->paddingrow) {
  1391. sheethandle->lastrownr++;
  1392. sheethandle->paddingrow++;
  1393. if (sheethandle->paddingrow + 1 < sheethandle->processcallbackdata.rownr) {
  1394. sheethandle->paddingcol = 1;
  1395. }
  1396. }
  1397. return NULL;
  1398. } else {
  1399. //add another empty column
  1400. sheethandle->paddingcol++;
  1401. return XML_Char_dup(X(""));
  1402. }
  1403. }
  1404. //get value
  1405. if (!sheethandle->processcallbackdata.celldata)
  1406. if (expat_process_zip_file_resume(sheethandle->zipfile, sheethandle->processcallbackdata.xmlparser) != XML_STATUS_SUSPENDED)
  1407. sheethandle->processcallbackdata.celldata = NULL;
  1408. //insert empty rows if needed
  1409. if (!(sheethandle->processcallbackdata.flags & XLSXIOREAD_SKIP_EMPTY_ROWS) && sheethandle->lastrownr + 1 < sheethandle->processcallbackdata.rownr) {
  1410. sheethandle->paddingrow = sheethandle->lastrownr + 1;
  1411. sheethandle->paddingcol = sheethandle->processcallbackdata.colnr*0 + 1;
  1412. return xlsxioread_sheet_next_cell(sheethandle);
  1413. }
  1414. //insert empty column before if needed
  1415. if (!(sheethandle->processcallbackdata.flags & XLSXIOREAD_SKIP_EMPTY_CELLS)) {
  1416. if (sheethandle->lastcolnr + 1 < sheethandle->processcallbackdata.colnr) {
  1417. sheethandle->lastcolnr++;
  1418. return XML_Char_dup(X(""));
  1419. }
  1420. }
  1421. result = sheethandle->processcallbackdata.celldata;
  1422. sheethandle->processcallbackdata.celldata = NULL;
  1423. //end of row
  1424. if (!result) {
  1425. sheethandle->lastrownr = sheethandle->processcallbackdata.rownr;
  1426. //insert empty column at end if row if needed
  1427. if (!result && !(sheethandle->processcallbackdata.flags & XLSXIOREAD_SKIP_EMPTY_CELLS) && sheethandle->processcallbackdata.colnr < sheethandle->processcallbackdata.cols) {
  1428. sheethandle->paddingcol = sheethandle->lastcolnr + 1;
  1429. return xlsxioread_sheet_next_cell(sheethandle);
  1430. }
  1431. }
  1432. sheethandle->lastcolnr = sheethandle->processcallbackdata.colnr;
  1433. return result;
  1434. }
  1435. DLL_EXPORT_XLSXIO int xlsxioread_sheet_next_cell_string (xlsxioreadersheet sheethandle, XLSXIOCHAR** pvalue)
  1436. {
  1437. XML_Char* result;
  1438. if (!sheethandle)
  1439. return -1;
  1440. if ((result = xlsxioread_sheet_next_cell(sheethandle)) == NULL)
  1441. return 0;
  1442. if (pvalue)
  1443. *pvalue = result;
  1444. return 1;
  1445. }
  1446. DLL_EXPORT_XLSXIO int xlsxioread_sheet_next_cell_int (xlsxioreadersheet sheethandle, int64_t* pvalue)
  1447. {
  1448. XML_Char* result;
  1449. int status;
  1450. if ((result = xlsxioread_sheet_next_cell(sheethandle)) == NULL)
  1451. return 0;
  1452. if (pvalue) {
  1453. status = XML_Char_sscanf(result, X("%" PRIi64), pvalue);
  1454. if (status == EOF || status == 0)
  1455. *pvalue = 0;
  1456. //alternative: use strtoimax()
  1457. }
  1458. free(result);
  1459. return 1;
  1460. }
  1461. DLL_EXPORT_XLSXIO int xlsxioread_sheet_next_cell_float (xlsxioreadersheet sheethandle, double* pvalue)
  1462. {
  1463. XML_Char* result;
  1464. if ((result = xlsxioread_sheet_next_cell(sheethandle)) == NULL)
  1465. return 0;
  1466. if (pvalue)
  1467. *pvalue = XML_Char_tod(result);
  1468. free(result);
  1469. return 1;
  1470. }
  1471. DLL_EXPORT_XLSXIO int xlsxioread_sheet_next_cell_datetime (xlsxioreadersheet sheethandle, time_t* pvalue)
  1472. {
  1473. XML_Char* result;
  1474. if ((result = xlsxioread_sheet_next_cell(sheethandle)) == NULL)
  1475. return 0;
  1476. if (pvalue) {
  1477. double value = XML_Char_tod(result);
  1478. if (value != 0) {
  1479. value = (value - 25569) * 86400; //converstion from Excel to Unix timestamp
  1480. }
  1481. *pvalue = value;
  1482. }
  1483. free(result);
  1484. return 1;
  1485. }