xlsxio_read.c 58 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594
  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) {//nesting level for current tag to skip
  950. //start handler to set after skipping
  951. //end handler to set after skipping
  952. //data handler to set after skipping
  953. data->rownr++;
  954. data->colnr = 0;
  955. XML_SetElementHandler(data->xmlparser, data_sheet_expat_callback_find_cell_start, data_sheet_expat_callback_find_row_end);
  956. //for non-calback method suspend here on new row
  957. if (data->flags & XLSXIOREAD_NO_CALLBACK) {
  958. XML_StopParser(data->xmlparser, XML_TRUE);
  959. }
  960. } else {
  961. //skip hidden tow
  962. XML_SetElementHandler(data->xmlparser, NULL, data_sheet_expat_callback_find_row_end);
  963. }
  964. }
  965. }
  966. void data_sheet_expat_callback_find_row_end (void* callbackdata, const XML_Char* name)
  967. {
  968. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  969. if (XML_Char_icmp_ins(name, X("row")) == 0) {
  970. //determine number of columns based on first row
  971. if (data->rownr == 1 && data->cols == 0)
  972. data->cols = data->colnr;
  973. //add empty columns if needed
  974. if (!(data->flags & XLSXIOREAD_NO_CALLBACK) && data->sheet_cell_callback && !(data->flags & XLSXIOREAD_SKIP_EMPTY_CELLS)) {
  975. while (data->colnr < data->cols) {
  976. if ((*data->sheet_cell_callback)(data->rownr, data->colnr + 1, NULL, data->callbackdata)) {
  977. XML_StopParser(data->xmlparser, XML_FALSE);
  978. return;
  979. }
  980. data->colnr++;
  981. }
  982. }
  983. free(data->celldata);
  984. data->celldata = NULL;
  985. XML_SetElementHandler(data->xmlparser, data_sheet_expat_callback_find_row_start, data_sheet_expat_callback_find_sheetdata_end);
  986. //process end of row
  987. if (!(data->flags & XLSXIOREAD_NO_CALLBACK)) {
  988. if (data->sheet_row_callback) {
  989. if ((*data->sheet_row_callback)(data->rownr, data->colnr, data->callbackdata)) {
  990. XML_StopParser(data->xmlparser, XML_FALSE);
  991. return;
  992. }
  993. }
  994. } else {
  995. //for non-calback method suspend here on end of row
  996. if (data->flags & XLSXIOREAD_NO_CALLBACK) {
  997. XML_StopParser(data->xmlparser, XML_TRUE);
  998. }
  999. }
  1000. } else {
  1001. data_sheet_expat_callback_find_sheetdata_end(callbackdata, name);
  1002. }
  1003. }
  1004. void data_sheet_expat_callback_find_cell_start (void* callbackdata, const XML_Char* name, const XML_Char** atts)
  1005. {
  1006. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  1007. if (XML_Char_icmp_ins(name, X("c")) == 0) {
  1008. const XML_Char* t = get_expat_attr_by_name(atts, X("r"));
  1009. size_t cellcolnr = get_col_nr(t);
  1010. //skip everything when out of bounds
  1011. if (cellcolnr && data->cols && (data->flags & XLSXIOREAD_SKIP_EXTRA_CELLS) && cellcolnr > data->cols) {
  1012. data->colnr = cellcolnr - 1;
  1013. return;
  1014. }
  1015. //insert empty rows if needed
  1016. if (data->colnr == 0) {
  1017. size_t cellrownr = get_row_nr(t);
  1018. if (cellrownr) {
  1019. if (!(data->flags & XLSXIOREAD_SKIP_EMPTY_ROWS) && !(data->flags & XLSXIOREAD_NO_CALLBACK)) {
  1020. while (data->rownr < cellrownr) {
  1021. //insert empty columns
  1022. if (!(data->flags & XLSXIOREAD_SKIP_EMPTY_CELLS) && data->sheet_cell_callback) {
  1023. while (data->colnr < data->cols) {
  1024. if ((*data->sheet_cell_callback)(data->rownr, data->colnr + 1, NULL, data->callbackdata)) {
  1025. XML_StopParser(data->xmlparser, XML_FALSE);
  1026. return;
  1027. }
  1028. data->colnr++;
  1029. }
  1030. }
  1031. //finish empty row
  1032. if (data->sheet_row_callback) {
  1033. if ((*data->sheet_row_callback)(data->rownr, data->cols, data->callbackdata)) {
  1034. XML_StopParser(data->xmlparser, XML_FALSE);
  1035. return;
  1036. }
  1037. }
  1038. data->rownr++;
  1039. data->colnr = 0;
  1040. }
  1041. } else {
  1042. data->rownr = cellrownr;
  1043. }
  1044. }
  1045. }
  1046. //insert empty columns if needed
  1047. if (cellcolnr) {
  1048. cellcolnr--;
  1049. if (data->flags & XLSXIOREAD_SKIP_EMPTY_CELLS || data->flags & XLSXIOREAD_NO_CALLBACK) {
  1050. data->colnr = cellcolnr;
  1051. } else {
  1052. while (data->colnr < cellcolnr) {
  1053. if (data->sheet_cell_callback) {
  1054. if ((*data->sheet_cell_callback)(data->rownr, data->colnr + 1, NULL, data->callbackdata)) {
  1055. XML_StopParser(data->xmlparser, XML_FALSE);
  1056. return;
  1057. }
  1058. }
  1059. data->colnr++;
  1060. }
  1061. }
  1062. }
  1063. //determing value type
  1064. if ((t = get_expat_attr_by_name(atts, X("t"))) != NULL && XML_Char_icmp(t, X("s")) == 0)
  1065. data->cell_string_type = shared_string;
  1066. else
  1067. data->cell_string_type = value_string;
  1068. //prepare empty value data
  1069. free(data->celldata);
  1070. data->celldata = NULL;
  1071. data->celldatalen = 0;
  1072. XML_SetElementHandler(data->xmlparser, data_sheet_expat_callback_find_value_start, data_sheet_expat_callback_find_cell_end);
  1073. }
  1074. }
  1075. void data_sheet_expat_callback_find_cell_end (void* callbackdata, const XML_Char* name)
  1076. {
  1077. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  1078. if (XML_Char_icmp_ins(name, X("c")) == 0) {
  1079. //determine value
  1080. if (data->celldata) {
  1081. const XML_Char* s = NULL;
  1082. data->celldata[data->celldatalen] = 0;
  1083. if (data->cell_string_type == shared_string) {
  1084. //get shared string
  1085. XML_Char* p = NULL;
  1086. long num = XML_Char_strtol(data->celldata, &p, 10);
  1087. if (!p || (p != data->celldata && *p == 0)) {
  1088. s = sharedstringlist_get(data->sharedstrings, num);
  1089. free(data->celldata);
  1090. data->celldata = (s ? XML_Char_dup(s) : NULL);
  1091. }
  1092. } else if (data->cell_string_type == none) {
  1093. //unknown value type
  1094. free(data->celldata);
  1095. data->celldata = NULL;
  1096. }
  1097. }
  1098. //reset data
  1099. data->colnr++;
  1100. data->cell_string_type = none;
  1101. data->celldatalen = 0;
  1102. XML_SetElementHandler(data->xmlparser, data_sheet_expat_callback_find_cell_start, data_sheet_expat_callback_find_row_end);
  1103. XML_SetCharacterDataHandler(data->xmlparser, NULL);
  1104. //process data if needed
  1105. if (!(data->cols && (data->flags & XLSXIOREAD_SKIP_EXTRA_CELLS) && data->colnr > data->cols)) {
  1106. //process data
  1107. if (!(data->flags & XLSXIOREAD_NO_CALLBACK)) {
  1108. if (data->sheet_cell_callback) {
  1109. if ((*data->sheet_cell_callback)(data->rownr, data->colnr, data->celldata, data->callbackdata)) {
  1110. XML_StopParser(data->xmlparser, XML_FALSE);
  1111. return;
  1112. }
  1113. }
  1114. } else {
  1115. //for non-calback method suspend here with cell data
  1116. if (!data->celldata)
  1117. data->celldata = XML_Char_dup(X(""));
  1118. XML_StopParser(data->xmlparser, XML_TRUE);
  1119. }
  1120. }
  1121. } else {
  1122. data_sheet_expat_callback_find_row_end(callbackdata, name);
  1123. }
  1124. }
  1125. void data_sheet_expat_callback_find_value_start (void* callbackdata, const XML_Char* name, const XML_Char** atts)
  1126. {
  1127. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  1128. if (XML_Char_icmp_ins(name, X("v")) == 0 || XML_Char_icmp_ins(name, X("t")) == 0) {
  1129. XML_SetElementHandler(data->xmlparser, NULL, data_sheet_expat_callback_find_value_end);
  1130. XML_SetCharacterDataHandler(data->xmlparser, data_sheet_expat_callback_value_data);
  1131. } else if (XML_Char_icmp_ins(name, X("is")) == 0) {
  1132. data->cell_string_type = inline_string;
  1133. } else if (XML_Char_icmp_ins(name, X("rPh")) == 0) {
  1134. data->skiptag = XML_Char_dup(name);
  1135. data->skiptagcount = 1;
  1136. data->skip_start = data_sheet_expat_callback_find_value_start;
  1137. data->skip_end = data_sheet_expat_callback_find_cell_end;
  1138. data->skip_data = NULL;
  1139. XML_SetElementHandler(data->xmlparser, data_sheet_expat_callback_skip_tag_start, data_sheet_expat_callback_skip_tag_end);
  1140. XML_SetCharacterDataHandler(data->xmlparser, NULL);
  1141. }
  1142. }
  1143. void data_sheet_expat_callback_find_value_end (void* callbackdata, const XML_Char* name)
  1144. {
  1145. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  1146. if (XML_Char_icmp_ins(name, X("v")) == 0 || XML_Char_icmp_ins(name, X("t")) == 0) {
  1147. XML_SetElementHandler(data->xmlparser, data_sheet_expat_callback_find_value_start, data_sheet_expat_callback_find_cell_end);
  1148. XML_SetCharacterDataHandler(data->xmlparser, NULL);
  1149. } else if (XML_Char_icmp_ins(name, X("is")) == 0) {
  1150. data->cell_string_type = none;
  1151. } else {
  1152. data_sheet_expat_callback_find_row_end(callbackdata, name);
  1153. }
  1154. }
  1155. void data_sheet_expat_callback_value_data (void* callbackdata, const XML_Char* buf, int buflen)
  1156. {
  1157. struct data_sheet_callback_data* data = (struct data_sheet_callback_data*)callbackdata;
  1158. if (data->cell_string_type != none) {
  1159. if ((data->celldata = XML_Char_realloc(data->celldata, data->celldatalen + buflen + 1)) == NULL) {
  1160. //memory allocation error
  1161. data->celldatalen = 0;
  1162. } else {
  1163. //add new data to value buffer
  1164. XML_Char_poscpy(data->celldata, data->celldatalen, buf, buflen);
  1165. data->celldatalen += buflen;
  1166. }
  1167. }
  1168. }
  1169. ////////////////////////////////////////////////////////////////////////
  1170. struct xlsxio_read_sheet_struct {
  1171. xlsxioreader handle;
  1172. ZIPFILEENTRYTYPE* zipfile;
  1173. struct data_sheet_callback_data processcallbackdata;
  1174. size_t lastrownr;
  1175. size_t paddingrow;
  1176. size_t lastcolnr;
  1177. size_t paddingcol;
  1178. };
  1179. DLL_EXPORT_XLSXIO size_t xlsxioread_sheet_last_row_index (xlsxioreadersheet sheethandle)
  1180. {
  1181. return sheethandle->lastrownr;
  1182. }
  1183. DLL_EXPORT_XLSXIO size_t xlsxioread_sheet_last_column_index (xlsxioreadersheet sheethandle)
  1184. {
  1185. return sheethandle->lastcolnr;
  1186. }
  1187. DLL_EXPORT_XLSXIO unsigned int xlsxioread_sheet_flags (xlsxioreadersheet sheethandle)
  1188. {
  1189. return sheethandle->processcallbackdata.flags;
  1190. }
  1191. 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)
  1192. {
  1193. int result = 0;
  1194. //determine sheet file name
  1195. struct main_sheet_get_rels_callback_data getrelscallbackdata = {
  1196. .sheetname = sheetname,
  1197. .basepath = NULL,
  1198. .sheetrelid = NULL,
  1199. .sheetfile = NULL,
  1200. .sharedstringsfile = NULL,
  1201. .stylesfile = NULL
  1202. };
  1203. iterate_files_by_contenttype(handle->zip, xlsx_content_type, main_sheet_get_sheetfile_callback, &getrelscallbackdata, NULL);
  1204. if (!getrelscallbackdata.sheetrelid)
  1205. iterate_files_by_contenttype(handle->zip, xlsm_content_type, main_sheet_get_sheetfile_callback, &getrelscallbackdata, NULL);
  1206. if (!getrelscallbackdata.sheetrelid)
  1207. iterate_files_by_contenttype(handle->zip, xltx_content_type, main_sheet_get_sheetfile_callback, &getrelscallbackdata, NULL);
  1208. if (!getrelscallbackdata.sheetrelid)
  1209. iterate_files_by_contenttype(handle->zip, xltm_content_type, main_sheet_get_sheetfile_callback, &getrelscallbackdata, NULL);
  1210. //process shared strings
  1211. struct sharedstringlist* sharedstrings = NULL;
  1212. if (getrelscallbackdata.sharedstringsfile && getrelscallbackdata.sharedstringsfile[0]) {
  1213. sharedstrings = sharedstringlist_create();
  1214. struct shared_strings_callback_data sharedstringsdata;
  1215. shared_strings_callback_data_initialize(&sharedstringsdata, sharedstrings);
  1216. if (expat_process_zip_file(handle->zip, getrelscallbackdata.sharedstringsfile, shared_strings_callback_find_sharedstringtable_start, NULL, NULL, &sharedstringsdata, &sharedstringsdata.xmlparser) != 0) {
  1217. //no shared strings found
  1218. sharedstringlist_destroy(sharedstrings);
  1219. sharedstrings = NULL;
  1220. }
  1221. shared_strings_callback_data_cleanup(&sharedstringsdata);
  1222. }
  1223. //process sheet
  1224. if (!(flags & XLSXIOREAD_NO_CALLBACK)) {
  1225. //use callback mechanism
  1226. struct data_sheet_callback_data processcallbackdata;
  1227. data_sheet_callback_data_initialize(&processcallbackdata, sharedstrings, flags, cell_callback, row_callback, callbackdata);
  1228. expat_process_zip_file(handle->zip, getrelscallbackdata.sheetfile, data_sheet_expat_callback_find_worksheet_start, NULL, NULL, &processcallbackdata, &processcallbackdata.xmlparser);
  1229. data_sheet_callback_data_cleanup(&processcallbackdata);
  1230. } else {
  1231. //use simplified interface by suspending the XML parser when data is found
  1232. xlsxioreadersheet sheethandle = (xlsxioreadersheet)callbackdata;
  1233. data_sheet_callback_data_initialize(&sheethandle->processcallbackdata, sharedstrings, flags, NULL, NULL, sheethandle);
  1234. if ((sheethandle->zipfile = XML_Char_openzip(sheethandle->handle->zip, getrelscallbackdata.sheetfile, 0)) == NULL) {
  1235. result = 1;
  1236. }
  1237. if ((sheethandle->processcallbackdata.xmlparser = expat_process_zip_file_suspendable(sheethandle->zipfile, data_sheet_expat_callback_find_worksheet_start, NULL, NULL, &sheethandle->processcallbackdata)) == NULL) {
  1238. result = 2;
  1239. }
  1240. }
  1241. //clean up
  1242. free(getrelscallbackdata.basepath);
  1243. free(getrelscallbackdata.sheetrelid);
  1244. free(getrelscallbackdata.sheetfile);
  1245. free(getrelscallbackdata.sharedstringsfile);
  1246. free(getrelscallbackdata.stylesfile);
  1247. return result;
  1248. }
  1249. ////////////////////////////////////////////////////////////////////////
  1250. struct xlsxio_read_sheetlist_struct {
  1251. xlsxioreader handle;
  1252. ZIPFILEENTRYTYPE* zipfile;
  1253. struct main_sheet_list_callback_data sheetcallbackdata;
  1254. XML_Parser xmlparser;
  1255. XML_Char* nextsheetname;
  1256. };
  1257. int xlsxioread_list_sheets_resumable_callback (const XLSXIOCHAR* name, void* callbackdata)
  1258. {
  1259. //struct main_sheet_list_callback_data* data = (struct main_sheet_list_callback_data*)callbackdata;
  1260. xlsxioreadersheetlist data = (xlsxioreadersheetlist)callbackdata;
  1261. data->nextsheetname = XML_Char_dup(name);
  1262. XML_StopParser(data->xmlparser, XML_TRUE);
  1263. return 0;
  1264. }
  1265. void xlsxioread_find_main_sheet_file_callback (ZIPFILETYPE* zip, const XML_Char* filename, const XML_Char* contenttype, void* callbackdata)
  1266. {
  1267. XML_Char** data = (XML_Char**)callbackdata;
  1268. *data = XML_Char_dup(filename);
  1269. }
  1270. DLL_EXPORT_XLSXIO xlsxioreadersheetlist xlsxioread_sheetlist_open (xlsxioreader handle)
  1271. {
  1272. //determine main sheet name
  1273. XML_Char* mainsheetfile = NULL;
  1274. iterate_files_by_contenttype(handle->zip, xlsx_content_type, xlsxioread_find_main_sheet_file_callback, &mainsheetfile, NULL);
  1275. if (!mainsheetfile)
  1276. iterate_files_by_contenttype(handle->zip, xlsm_content_type, xlsxioread_find_main_sheet_file_callback, &mainsheetfile, NULL);
  1277. if (!mainsheetfile)
  1278. iterate_files_by_contenttype(handle->zip, xltx_content_type, xlsxioread_find_main_sheet_file_callback, &mainsheetfile, NULL);
  1279. if (!mainsheetfile)
  1280. iterate_files_by_contenttype(handle->zip, xltm_content_type, xlsxioread_find_main_sheet_file_callback, &mainsheetfile, NULL);
  1281. if (!mainsheetfile)
  1282. return NULL;
  1283. //process contents of main sheet
  1284. xlsxioreadersheetlist result;
  1285. if ((result = (xlsxioreadersheetlist)malloc(sizeof(struct xlsxio_read_sheetlist_struct))) == NULL)
  1286. return NULL;
  1287. result->handle = handle;
  1288. result->sheetcallbackdata.xmlparser = NULL;
  1289. result->sheetcallbackdata.callback = xlsxioread_list_sheets_resumable_callback;
  1290. result->sheetcallbackdata.callbackdata = result;
  1291. result->nextsheetname = NULL;
  1292. if ((result->zipfile = XML_Char_openzip(handle->zip, mainsheetfile, 0)) != NULL) {
  1293. result->xmlparser = expat_process_zip_file_suspendable(result->zipfile, main_sheet_list_expat_callback_element_start, NULL, NULL, &result->sheetcallbackdata);
  1294. }
  1295. //clean up
  1296. free(mainsheetfile);
  1297. return result;
  1298. }
  1299. DLL_EXPORT_XLSXIO void xlsxioread_sheetlist_close (xlsxioreadersheetlist sheetlisthandle)
  1300. {
  1301. if (!sheetlisthandle)
  1302. return;
  1303. if (sheetlisthandle->xmlparser)
  1304. XML_ParserFree(sheetlisthandle->xmlparser);
  1305. if (sheetlisthandle->zipfile)
  1306. #ifdef USE_MINIZIP
  1307. unzCloseCurrentFile(sheetlisthandle->zipfile);
  1308. #else
  1309. zip_fclose(sheetlisthandle->zipfile);
  1310. #endif
  1311. free(sheetlisthandle->nextsheetname);
  1312. free(sheetlisthandle);
  1313. }
  1314. DLL_EXPORT_XLSXIO const XLSXIOCHAR* xlsxioread_sheetlist_next (xlsxioreadersheetlist sheetlisthandle)
  1315. {
  1316. if (!sheetlisthandle->zipfile || !sheetlisthandle->xmlparser)
  1317. return NULL;
  1318. free(sheetlisthandle->nextsheetname);
  1319. sheetlisthandle->nextsheetname = NULL;
  1320. enum XML_Status status;
  1321. if ((status = expat_process_zip_file_resume(sheetlisthandle->zipfile, sheetlisthandle->xmlparser)) != XML_STATUS_SUSPENDED) {
  1322. return NULL;
  1323. }
  1324. return sheetlisthandle->nextsheetname;
  1325. }
  1326. ////////////////////////////////////////////////////////////////////////
  1327. DLL_EXPORT_XLSXIO xlsxioreadersheet xlsxioread_sheet_open (xlsxioreader handle, const XLSXIOCHAR* sheetname, unsigned int flags)
  1328. {
  1329. xlsxioreadersheet result;
  1330. if ((result = (xlsxioreadersheet)malloc(sizeof(struct xlsxio_read_sheet_struct))) == NULL)
  1331. return NULL;
  1332. result->handle = handle;
  1333. result->zipfile = NULL;
  1334. result->lastrownr = 0;
  1335. result->paddingrow = 0;
  1336. result->lastcolnr = 0;
  1337. result->paddingcol = 0;
  1338. xlsxioread_process(handle, sheetname, flags | XLSXIOREAD_NO_CALLBACK, NULL, NULL, result);
  1339. return result;
  1340. }
  1341. DLL_EXPORT_XLSXIO void xlsxioread_sheet_close (xlsxioreadersheet sheethandle)
  1342. {
  1343. if (!sheethandle)
  1344. return;
  1345. if (sheethandle->processcallbackdata.xmlparser)
  1346. XML_ParserFree(sheethandle->processcallbackdata.xmlparser);
  1347. data_sheet_callback_data_cleanup(&sheethandle->processcallbackdata);
  1348. if (sheethandle->zipfile)
  1349. #ifdef USE_MINIZIP
  1350. unzCloseCurrentFile(sheethandle->zipfile);
  1351. #else
  1352. zip_fclose(sheethandle->zipfile);
  1353. #endif
  1354. free(sheethandle);
  1355. }
  1356. DLL_EXPORT_XLSXIO int xlsxioread_sheet_next_row (xlsxioreadersheet sheethandle)
  1357. {
  1358. enum XML_Status status;
  1359. if (!sheethandle) {
  1360. return 0;
  1361. }
  1362. sheethandle->lastcolnr = 0;
  1363. //when padding rows don't retrieve new data
  1364. if (sheethandle->paddingrow) {
  1365. if (sheethandle->paddingrow < sheethandle->processcallbackdata.rownr) {
  1366. return 3;
  1367. } else {
  1368. sheethandle->paddingrow = 0;
  1369. return 2;
  1370. }
  1371. }
  1372. sheethandle->paddingcol = 0;
  1373. //go to beginning of next row
  1374. while ((status = expat_process_zip_file_resume(sheethandle->zipfile, sheethandle->processcallbackdata.xmlparser)) == XML_STATUS_SUSPENDED && sheethandle->processcallbackdata.colnr != 0) {
  1375. }
  1376. return (status == XML_STATUS_SUSPENDED ? 1 : 0);
  1377. }
  1378. DLL_EXPORT_XLSXIO XLSXIOCHAR* xlsxioread_sheet_next_cell (xlsxioreadersheet sheethandle)
  1379. {
  1380. XML_Char* result;
  1381. if (!sheethandle)
  1382. return NULL;
  1383. //append empty column if needed
  1384. if (sheethandle->paddingcol) {
  1385. if (sheethandle->paddingcol > sheethandle->processcallbackdata.cols) {
  1386. //last empty column added, finish row
  1387. sheethandle->paddingcol = 0;
  1388. //when padding rows prepare for the next one
  1389. if (sheethandle->paddingrow) {
  1390. sheethandle->lastrownr++;
  1391. sheethandle->paddingrow++;
  1392. if (sheethandle->paddingrow + 1 < sheethandle->processcallbackdata.rownr) {
  1393. sheethandle->paddingcol = 1;
  1394. }
  1395. }
  1396. return NULL;
  1397. } else {
  1398. //add another empty column
  1399. sheethandle->paddingcol++;
  1400. return XML_Char_dup(X(""));
  1401. }
  1402. }
  1403. //get value
  1404. if (!sheethandle->processcallbackdata.celldata)
  1405. if (expat_process_zip_file_resume(sheethandle->zipfile, sheethandle->processcallbackdata.xmlparser) != XML_STATUS_SUSPENDED)
  1406. sheethandle->processcallbackdata.celldata = NULL;
  1407. //insert empty rows if needed
  1408. if (!(sheethandle->processcallbackdata.flags & XLSXIOREAD_SKIP_EMPTY_ROWS) && sheethandle->lastrownr + 1 < sheethandle->processcallbackdata.rownr) {
  1409. sheethandle->paddingrow = sheethandle->lastrownr + 1;
  1410. sheethandle->paddingcol = sheethandle->processcallbackdata.colnr*0 + 1;
  1411. return xlsxioread_sheet_next_cell(sheethandle);
  1412. }
  1413. //insert empty column before if needed
  1414. if (!(sheethandle->processcallbackdata.flags & XLSXIOREAD_SKIP_EMPTY_CELLS)) {
  1415. if (sheethandle->lastcolnr + 1 < sheethandle->processcallbackdata.colnr) {
  1416. sheethandle->lastcolnr++;
  1417. return XML_Char_dup(X(""));
  1418. }
  1419. }
  1420. result = sheethandle->processcallbackdata.celldata;
  1421. sheethandle->processcallbackdata.celldata = NULL;
  1422. //end of row
  1423. if (!result) {
  1424. sheethandle->lastrownr = sheethandle->processcallbackdata.rownr;
  1425. //insert empty column at end if row if needed
  1426. if (!result && !(sheethandle->processcallbackdata.flags & XLSXIOREAD_SKIP_EMPTY_CELLS) && sheethandle->processcallbackdata.colnr < sheethandle->processcallbackdata.cols) {
  1427. sheethandle->paddingcol = sheethandle->lastcolnr + 1;
  1428. return xlsxioread_sheet_next_cell(sheethandle);
  1429. }
  1430. }
  1431. sheethandle->lastcolnr = sheethandle->processcallbackdata.colnr;
  1432. return result;
  1433. }
  1434. DLL_EXPORT_XLSXIO int xlsxioread_sheet_next_cell_string (xlsxioreadersheet sheethandle, XLSXIOCHAR** pvalue)
  1435. {
  1436. XML_Char* result;
  1437. if (!sheethandle)
  1438. return -1;
  1439. if ((result = xlsxioread_sheet_next_cell(sheethandle)) == NULL)
  1440. return 0;
  1441. if (pvalue)
  1442. *pvalue = result;
  1443. return 1;
  1444. }
  1445. DLL_EXPORT_XLSXIO int xlsxioread_sheet_next_cell_int (xlsxioreadersheet sheethandle, int64_t* pvalue)
  1446. {
  1447. XML_Char* result;
  1448. int status;
  1449. if ((result = xlsxioread_sheet_next_cell(sheethandle)) == NULL)
  1450. return 0;
  1451. if (pvalue) {
  1452. status = XML_Char_sscanf(result, X("%" PRIi64), pvalue);
  1453. if (status == EOF || status == 0)
  1454. *pvalue = 0;
  1455. //alternative: use strtoimax()
  1456. }
  1457. free(result);
  1458. return 1;
  1459. }
  1460. DLL_EXPORT_XLSXIO int xlsxioread_sheet_next_cell_float (xlsxioreadersheet sheethandle, double* pvalue)
  1461. {
  1462. XML_Char* result;
  1463. if ((result = xlsxioread_sheet_next_cell(sheethandle)) == NULL)
  1464. return 0;
  1465. if (pvalue)
  1466. *pvalue = XML_Char_tod(result);
  1467. free(result);
  1468. return 1;
  1469. }
  1470. DLL_EXPORT_XLSXIO int xlsxioread_sheet_next_cell_datetime (xlsxioreadersheet sheethandle, time_t* pvalue)
  1471. {
  1472. XML_Char* result;
  1473. if ((result = xlsxioread_sheet_next_cell(sheethandle)) == NULL)
  1474. return 0;
  1475. if (pvalue) {
  1476. double value = XML_Char_tod(result);
  1477. if (value != 0) {
  1478. value = (value - 25569) * 86400; //converstion from Excel to Unix timestamp
  1479. }
  1480. *pvalue = value;
  1481. }
  1482. free(result);
  1483. return 1;
  1484. }