1 | /************* TabFmt C++ Program Source Code File (.CPP) **************/ |
2 | /* PROGRAM NAME: TABFMT */ |
3 | /* ------------- */ |
4 | /* Version 3.9.2 */ |
5 | /* */ |
6 | /* COPYRIGHT: */ |
7 | /* ---------- */ |
8 | /* (C) Copyright to the author Olivier BERTRAND 2001 - 2017 */ |
9 | /* */ |
10 | /* WHAT THIS PROGRAM DOES: */ |
11 | /* ----------------------- */ |
12 | /* This program are the TABFMT classes DB execution routines. */ |
13 | /* The base class CSV is comma separated files. */ |
14 | /* FMT (Formatted) files are those having a complex internal record */ |
15 | /* format described in the Format keyword of their definition. */ |
16 | /***********************************************************************/ |
17 | |
18 | /***********************************************************************/ |
19 | /* Include relevant MariaDB header file. */ |
20 | /***********************************************************************/ |
21 | #include "my_global.h" |
22 | |
23 | #if defined(__WIN__) |
24 | #include <io.h> |
25 | #include <fcntl.h> |
26 | #include <errno.h> |
27 | #include <locale.h> |
28 | #if defined(__BORLANDC__) |
29 | #define __MFC_COMPAT__ // To define min/max as macro |
30 | #endif |
31 | //#include <windows.h> |
32 | #include "osutil.h" |
33 | #else |
34 | #if defined(UNIX) |
35 | #include <errno.h> |
36 | #include <unistd.h> |
37 | #include "osutil.h" |
38 | #else |
39 | #include <io.h> |
40 | #endif |
41 | #include <fcntl.h> |
42 | #endif |
43 | |
44 | /***********************************************************************/ |
45 | /* Include application header files: */ |
46 | /* global.h is header containing all global declarations. */ |
47 | /* plgdbsem.h is header containing the DB application declarations. */ |
48 | /* tabdos.h is header containing the TABDOS class declarations. */ |
49 | /***********************************************************************/ |
50 | #include "global.h" |
51 | #include "plgdbsem.h" |
52 | #include "mycat.h" |
53 | #include "filamap.h" |
54 | #if defined(GZ_SUPPORT) |
55 | #include "filamgz.h" |
56 | #endif // GZ_SUPPORT |
57 | #if defined(ZIP_SUPPORT) |
58 | #include "filamzip.h" |
59 | #endif // ZIP_SUPPORT |
60 | #include "tabfmt.h" |
61 | #include "tabmul.h" |
62 | #define NO_FUNC |
63 | #include "plgcnx.h" // For DB types |
64 | #include "resource.h" |
65 | |
66 | /***********************************************************************/ |
67 | /* This should be an option. */ |
68 | /***********************************************************************/ |
69 | #define MAXCOL 200 /* Default max column nb in result */ |
70 | #define TYPE_UNKNOWN 10 /* Must be greater than other types */ |
71 | |
72 | /***********************************************************************/ |
73 | /* External function. */ |
74 | /***********************************************************************/ |
75 | USETEMP UseTemp(void); |
76 | |
77 | /***********************************************************************/ |
78 | /* CSVColumns: constructs the result blocks containing the description */ |
79 | /* of all the columns of a CSV file that will be retrieved by #GetData.*/ |
80 | /* Note: the algorithm to set the type is based on the internal values */ |
81 | /* of types (TYPE_STRING < TYPE_DOUBLE < TYPE_INT) (1 < 2 < 7). */ |
82 | /* If these values are changed, this will have to be revisited. */ |
83 | /***********************************************************************/ |
84 | PQRYRES CSVColumns(PGLOBAL g, PCSZ dp, PTOS topt, bool info) |
85 | { |
86 | static int buftyp[] = {TYPE_STRING, TYPE_SHORT, TYPE_STRING, |
87 | TYPE_INT, TYPE_INT, TYPE_SHORT}; |
88 | static XFLD fldtyp[] = {FLD_NAME, FLD_TYPE, FLD_TYPENAME, |
89 | FLD_PREC, FLD_LENGTH, FLD_SCALE}; |
90 | static unsigned int length[] = {6, 6, 8, 10, 10, 6}; |
91 | const char *fn; |
92 | char sep, q; |
93 | int rc, mxr; |
94 | bool hdr; |
95 | char *p, *colname[MAXCOL], dechar, buf[8]; |
96 | int i, imax, hmax, n, nerr, phase, blank, digit, dec, type; |
97 | int ncol = sizeof(buftyp) / sizeof(int); |
98 | int num_read = 0, num_max = 10000000; // Statistics |
99 | int len[MAXCOL], typ[MAXCOL], prc[MAXCOL]; |
100 | PCSVDEF tdp; |
101 | PTDBCSV tcvp; |
102 | PTDBASE tdbp; |
103 | PQRYRES qrp; |
104 | PCOLRES crp; |
105 | |
106 | if (info) { |
107 | imax = hmax = 0; |
108 | length[0] = 128; |
109 | goto skipit; |
110 | } // endif info |
111 | |
112 | //if (GetIntegerTableOption(g, topt, "Multiple", 0)) { |
113 | // strcpy(g->Message, "Cannot find column definition for multiple table"); |
114 | // return NULL; |
115 | //} // endif Multiple |
116 | |
117 | // num_max = atoi(p+1); // Max num of record to test |
118 | imax = hmax = nerr = 0; |
119 | |
120 | for (i = 0; i < MAXCOL; i++) { |
121 | colname[i] = NULL; |
122 | len[i] = 0; |
123 | typ[i] = TYPE_UNKNOWN; |
124 | prc[i] = 0; |
125 | } // endfor i |
126 | |
127 | /*********************************************************************/ |
128 | /* Get the CSV table description block. */ |
129 | /*********************************************************************/ |
130 | tdp = new(g) CSVDEF; |
131 | tdp->Database = dp; |
132 | |
133 | if ((tdp->Zipped = GetBooleanTableOption(g, topt, "Zipped" , false))) { |
134 | #if defined(ZIP_SUPPORT) |
135 | tdp->Entry = GetStringTableOption(g, topt, "Entry" , NULL); |
136 | tdp->Mulentries = (tdp->Entry) |
137 | ? strchr(tdp->Entry, '*') || strchr(tdp->Entry, '?') |
138 | : GetBooleanTableOption(g, topt, "Mulentries" , false); |
139 | #else // !ZIP_SUPPORT |
140 | strcpy(g->Message, "ZIP not supported by this version" ); |
141 | return NULL; |
142 | #endif // !ZIP_SUPPORT |
143 | } // endif // Zipped |
144 | |
145 | fn = tdp->Fn = GetStringTableOption(g, topt, "Filename" , NULL); |
146 | |
147 | if (!tdp->Fn) { |
148 | strcpy(g->Message, MSG(MISSING_FNAME)); |
149 | return NULL; |
150 | } // endif Fn |
151 | |
152 | if (!(tdp->Lrecl = GetIntegerTableOption(g, topt, "Lrecl" , 0))) |
153 | tdp->Lrecl = 4096; |
154 | |
155 | tdp->Multiple = GetIntegerTableOption(g, topt, "Multiple" , 0); |
156 | p = (char*)GetStringTableOption(g, topt, "Separator" , "," ); |
157 | tdp->Sep = (strlen(p) == 2 && p[0] == '\\' && p[1] == 't') ? '\t' : *p; |
158 | |
159 | #if defined(__WIN__) |
160 | if (tdp->Sep == ',' || strnicmp(setlocale(LC_NUMERIC, NULL), "French" , 6)) |
161 | dechar = '.'; |
162 | else |
163 | dechar = ','; |
164 | #else // !__WIN__ |
165 | dechar = '.'; |
166 | #endif // !__WIN__ |
167 | |
168 | sep = tdp->Sep; |
169 | tdp->Quoted = GetIntegerTableOption(g, topt, "Quoted" , -1); |
170 | p = (char*)GetStringTableOption(g, topt, "Qchar" , "" ); |
171 | tdp->Qot = *p; |
172 | |
173 | if (tdp->Qot && tdp->Quoted < 0) |
174 | tdp->Quoted = 0; |
175 | else if (!tdp->Qot && tdp->Quoted >= 0) |
176 | tdp->Qot = '"'; |
177 | |
178 | q = tdp->Qot; |
179 | hdr = GetBooleanTableOption(g, topt, "Header" , false); |
180 | tdp->Maxerr = GetIntegerTableOption(g, topt, "Maxerr" , 0); |
181 | tdp->Accept = GetBooleanTableOption(g, topt, "Accept" , false); |
182 | |
183 | if (tdp->Accept && tdp->Maxerr == 0) |
184 | tdp->Maxerr = INT_MAX32; // Accept all bad lines |
185 | |
186 | mxr = MY_MAX(0, tdp->Maxerr); |
187 | |
188 | if (trace(1)) |
189 | htrc("File %s Sep=%c Qot=%c Header=%d maxerr=%d\n" , |
190 | SVP(tdp->Fn), tdp->Sep, tdp->Qot, tdp->Header, tdp->Maxerr); |
191 | |
192 | if (tdp->Zipped) |
193 | tcvp = new(g)TDBCSV(tdp, new(g)UNZFAM(tdp)); |
194 | else |
195 | tcvp = new(g) TDBCSV(tdp, new(g) DOSFAM(tdp)); |
196 | |
197 | tcvp->SetMode(MODE_READ); |
198 | |
199 | if (tdp->Multiple) { |
200 | tdbp = new(g)TDBMUL(tcvp); |
201 | tdbp->SetMode(MODE_READ); |
202 | } else |
203 | tdbp = tcvp; |
204 | |
205 | /*********************************************************************/ |
206 | /* Open the CSV file. */ |
207 | /*********************************************************************/ |
208 | if (tdbp->OpenDB(g)) |
209 | return NULL; |
210 | |
211 | if (hdr) { |
212 | /*******************************************************************/ |
213 | /* Make the column names from the first line. */ |
214 | /*******************************************************************/ |
215 | phase = 0; |
216 | |
217 | if ((rc = tdbp->ReadDB(g)) == RC_OK) { |
218 | p = PlgDBDup(g, tcvp->To_Line); |
219 | |
220 | //skip leading blanks |
221 | for (; *p == ' '; p++) ; |
222 | |
223 | if (q && *p == q) { |
224 | // Header is quoted |
225 | p++; |
226 | phase = 1; |
227 | } // endif q |
228 | |
229 | colname[0] = p; |
230 | } else if (rc == RC_EF) { |
231 | sprintf(g->Message, MSG(FILE_IS_EMPTY), fn); |
232 | goto err; |
233 | } else |
234 | goto err; |
235 | |
236 | for (i = 1; *p; p++) |
237 | if (phase == 1 && *p == q) { |
238 | *p = '\0'; |
239 | phase = 0; |
240 | } else if (*p == sep && !phase) { |
241 | *p = '\0'; |
242 | |
243 | //skip leading blanks |
244 | for (; *(p+1) == ' '; p++) ; |
245 | |
246 | if (q && *(p+1) == q) { |
247 | // Header is quoted |
248 | p++; |
249 | phase = 1; |
250 | } // endif q |
251 | |
252 | colname[i++] = p + 1; |
253 | } // endif sep |
254 | |
255 | num_read++; |
256 | imax = hmax = i; |
257 | |
258 | for (i = 0; i < hmax; i++) |
259 | length[0] = MY_MAX(length[0], strlen(colname[i])); |
260 | |
261 | tcvp->Header = true; // In case of multiple table |
262 | } // endif hdr |
263 | |
264 | for (num_read++; num_read <= num_max; num_read++) { |
265 | /*******************************************************************/ |
266 | /* Now start the reading process. Read one line. */ |
267 | /*******************************************************************/ |
268 | if ((rc = tdbp->ReadDB(g)) == RC_OK) { |
269 | } else if (rc == RC_EF) { |
270 | sprintf(g->Message, MSG(EOF_AFTER_LINE), num_read -1); |
271 | break; |
272 | } else { |
273 | sprintf(g->Message, MSG(ERR_READING_REC), num_read, fn); |
274 | goto err; |
275 | } // endif's |
276 | |
277 | /*******************************************************************/ |
278 | /* Make the test for field lengths. */ |
279 | /*******************************************************************/ |
280 | i = n = phase = blank = digit = dec = 0; |
281 | |
282 | for (p = tcvp->To_Line; *p; p++) |
283 | if (*p == sep) { |
284 | if (phase != 1) { |
285 | if (i == MAXCOL - 1) { |
286 | sprintf(g->Message, MSG(TOO_MANY_FIELDS), num_read, fn); |
287 | goto err; |
288 | } // endif i |
289 | |
290 | if (n) { |
291 | len[i] = MY_MAX(len[i], n); |
292 | type = (digit || (dec && n == 1)) ? TYPE_STRING |
293 | : (dec) ? TYPE_DOUBLE : TYPE_INT; |
294 | typ[i] = MY_MIN(type, typ[i]); |
295 | prc[i] = MY_MAX((typ[i] == TYPE_DOUBLE) ? (dec - 1) : 0, prc[i]); |
296 | } // endif n |
297 | |
298 | i++; |
299 | n = phase = blank = digit = dec = 0; |
300 | } else // phase == 1 |
301 | n++; |
302 | |
303 | } else if (*p == ' ') { |
304 | if (phase < 2) |
305 | n++; |
306 | |
307 | if (blank) |
308 | digit = 1; |
309 | |
310 | } else if (*p == q) { |
311 | if (phase == 0) { |
312 | if (blank) |
313 | if (++nerr > mxr) { |
314 | sprintf(g->Message, MSG(MISPLACED_QUOTE), num_read); |
315 | goto err; |
316 | } else |
317 | goto skip; |
318 | |
319 | n = 0; |
320 | phase = digit = 1; |
321 | } else if (phase == 1) { |
322 | if (*(p+1) == q) { |
323 | // This is currently not implemented for CSV tables |
324 | // if (++nerr > mxr) { |
325 | // sprintf(g->Message, MSG(QUOTE_IN_QUOTE), num_read); |
326 | // goto err; |
327 | // } else |
328 | // goto skip; |
329 | |
330 | p++; |
331 | n++; |
332 | } else |
333 | phase = 2; |
334 | |
335 | } else if (++nerr > mxr) { // phase == 2 |
336 | sprintf(g->Message, MSG(MISPLACED_QUOTE), num_read); |
337 | goto err; |
338 | } else |
339 | goto skip; |
340 | |
341 | } else { |
342 | if (phase == 2) |
343 | if (++nerr > mxr) { |
344 | sprintf(g->Message, MSG(MISPLACED_QUOTE), num_read); |
345 | goto err; |
346 | } else |
347 | goto skip; |
348 | |
349 | // isdigit cannot be used here because of debug assert |
350 | if (!strchr("0123456789" , *p)) { |
351 | if (!digit && *p == dechar) |
352 | dec = 1; // Decimal point found |
353 | else if (blank || !(*p == '-' || *p == '+')) |
354 | digit = 1; |
355 | |
356 | } else if (dec) |
357 | dec++; // More decimals |
358 | |
359 | n++; |
360 | blank = 1; |
361 | } // endif's *p |
362 | |
363 | if (phase == 1) |
364 | if (++nerr > mxr) { |
365 | sprintf(g->Message, MSG(UNBALANCE_QUOTE), num_read); |
366 | goto err; |
367 | } else |
368 | goto skip; |
369 | |
370 | if (n) { |
371 | len[i] = MY_MAX(len[i], n); |
372 | type = (digit || n == 0 || (dec && n == 1)) ? TYPE_STRING |
373 | : (dec) ? TYPE_DOUBLE : TYPE_INT; |
374 | typ[i] = MY_MIN(type, typ[i]); |
375 | prc[i] = MY_MAX((typ[i] == TYPE_DOUBLE) ? (dec - 1) : 0, prc[i]); |
376 | } // endif n |
377 | |
378 | imax = MY_MAX(imax, i+1); |
379 | skip: ; // Skip erroneous line |
380 | } // endfor num_read |
381 | |
382 | if (trace(1)) { |
383 | htrc("imax=%d Lengths:" , imax); |
384 | |
385 | for (i = 0; i < imax; i++) |
386 | htrc(" %d" , len[i]); |
387 | |
388 | htrc("\n" ); |
389 | } // endif trace |
390 | |
391 | tdbp->CloseDB(g); |
392 | |
393 | skipit: |
394 | if (trace(1)) |
395 | htrc("CSVColumns: imax=%d hmax=%d len=%d\n" , |
396 | imax, hmax, length[0]); |
397 | |
398 | /*********************************************************************/ |
399 | /* Allocate the structures used to refer to the result set. */ |
400 | /*********************************************************************/ |
401 | qrp = PlgAllocResult(g, ncol, imax, IDS_COLUMNS + 3, |
402 | buftyp, fldtyp, length, false, false); |
403 | if (info || !qrp) |
404 | return qrp; |
405 | |
406 | qrp->Nblin = imax; |
407 | |
408 | /*********************************************************************/ |
409 | /* Now get the results into blocks. */ |
410 | /*********************************************************************/ |
411 | for (i = 0; i < imax; i++) { |
412 | if (i >= hmax) { |
413 | sprintf(buf, "COL%.3d" , i+1); |
414 | p = buf; |
415 | } else |
416 | p = colname[i]; |
417 | |
418 | if (typ[i] == TYPE_UNKNOWN) // Void column |
419 | typ[i] = TYPE_STRING; |
420 | |
421 | crp = qrp->Colresp; // Column Name |
422 | crp->Kdata->SetValue(p, i); |
423 | crp = crp->Next; // Data Type |
424 | crp->Kdata->SetValue(typ[i], i); |
425 | crp = crp->Next; // Type Name |
426 | crp->Kdata->SetValue(GetTypeName(typ[i]), i); |
427 | crp = crp->Next; // Precision |
428 | crp->Kdata->SetValue(len[i], i); |
429 | crp = crp->Next; // Length |
430 | crp->Kdata->SetValue(len[i], i); |
431 | crp = crp->Next; // Scale (precision) |
432 | crp->Kdata->SetValue(prc[i], i); |
433 | } // endfor i |
434 | |
435 | /*********************************************************************/ |
436 | /* Return the result pointer for use by GetData routines. */ |
437 | /*********************************************************************/ |
438 | return qrp; |
439 | |
440 | err: |
441 | tdbp->CloseDB(g); |
442 | return NULL; |
443 | } // end of CSVCColumns |
444 | |
445 | /* --------------------------- Class CSVDEF -------------------------- */ |
446 | |
447 | /***********************************************************************/ |
448 | /* CSVDEF constructor. */ |
449 | /***********************************************************************/ |
450 | CSVDEF::CSVDEF(void) |
451 | { |
452 | Fmtd = Header = false; |
453 | //Maxerr = 0; |
454 | Quoted = -1; |
455 | Sep = ','; |
456 | Qot = '\0'; |
457 | } // end of CSVDEF constructor |
458 | |
459 | /***********************************************************************/ |
460 | /* DefineAM: define specific AM block values from XDB file. */ |
461 | /***********************************************************************/ |
462 | bool CSVDEF::DefineAM(PGLOBAL g, LPCSTR am, int poff) |
463 | { |
464 | char buf[8]; |
465 | |
466 | // Double check correctness of offset values |
467 | if (Catfunc == FNC_NO) |
468 | for (PCOLDEF cdp = To_Cols; cdp; cdp = cdp->GetNext()) |
469 | if (cdp->GetOffset() < 1 && !cdp->IsSpecial()) { |
470 | strcpy(g->Message, MSG(BAD_OFFSET_VAL)); |
471 | return true; |
472 | } // endif Offset |
473 | |
474 | // Call DOSDEF DefineAM with am=CSV so FMT is not confused with FIX |
475 | if (DOSDEF::DefineAM(g, "CSV" , poff)) |
476 | return true; |
477 | |
478 | GetCharCatInfo("Separator" , "," , buf, sizeof(buf)); |
479 | Sep = (strlen(buf) == 2 && buf[0] == '\\' && buf[1] == 't') ? '\t' : *buf; |
480 | Quoted = GetIntCatInfo("Quoted" , -1); |
481 | GetCharCatInfo("Qchar" , "" , buf, sizeof(buf)); |
482 | Qot = *buf; |
483 | |
484 | if (Qot && Quoted < 0) |
485 | Quoted = 0; |
486 | else if (!Qot && Quoted >= 0) |
487 | Qot = '"'; |
488 | |
489 | Fmtd = (!Sep || (am && (*am == 'F' || *am == 'f'))); |
490 | Header = GetBoolCatInfo("Header" , false); |
491 | Maxerr = GetIntCatInfo("Maxerr" , 0); |
492 | Accept = GetBoolCatInfo("Accept" , false); |
493 | |
494 | if (Accept && Maxerr == 0) |
495 | Maxerr = INT_MAX32; // Accept all bad lines |
496 | |
497 | return false; |
498 | } // end of DefineAM |
499 | |
500 | /***********************************************************************/ |
501 | /* GetTable: makes a new Table Description Block. */ |
502 | /***********************************************************************/ |
503 | PTDB CSVDEF::GetTable(PGLOBAL g, MODE mode) |
504 | { |
505 | PTDBASE tdbp; |
506 | |
507 | if (Catfunc != FNC_COL) { |
508 | USETEMP tmp = UseTemp(); |
509 | bool map = Mapped && mode != MODE_INSERT && |
510 | !(tmp != TMP_NO && mode == MODE_UPDATE) && |
511 | !(tmp == TMP_FORCE && |
512 | (mode == MODE_UPDATE || mode == MODE_DELETE)); |
513 | PTXF txfp; |
514 | |
515 | /*******************************************************************/ |
516 | /* Allocate a file processing class of the proper type. */ |
517 | /*******************************************************************/ |
518 | if (Zipped) { |
519 | #if defined(ZIP_SUPPORT) |
520 | if (mode == MODE_READ || mode == MODE_ANY || mode == MODE_ALTER) { |
521 | txfp = new(g) UNZFAM(this); |
522 | } else if (mode == MODE_INSERT) { |
523 | txfp = new(g) ZIPFAM(this); |
524 | } else { |
525 | strcpy(g->Message, "UPDATE/DELETE not supported for ZIP" ); |
526 | return NULL; |
527 | } // endif's mode |
528 | #else // !ZIP_SUPPORT |
529 | strcpy(g->Message, "ZIP not supported" ); |
530 | return NULL; |
531 | #endif // !ZIP_SUPPORT |
532 | } else if (map) { |
533 | // Should be now compatible with UNIX |
534 | txfp = new(g) MAPFAM(this); |
535 | } else if (Compressed) { |
536 | #if defined(GZ_SUPPORT) |
537 | if (Compressed == 1) |
538 | txfp = new(g) GZFAM(this); |
539 | else |
540 | txfp = new(g) ZLBFAM(this); |
541 | |
542 | #else // !GZ_SUPPORT |
543 | strcpy(g->Message, "Compress not supported" ); |
544 | return NULL; |
545 | #endif // !GZ_SUPPORT |
546 | } else |
547 | txfp = new(g) DOSFAM(this); |
548 | |
549 | /*******************************************************************/ |
550 | /* Allocate a TDB of the proper type. */ |
551 | /* Column blocks will be allocated only when needed. */ |
552 | /*******************************************************************/ |
553 | if (!Fmtd) |
554 | tdbp = new(g) TDBCSV(this, txfp); |
555 | else |
556 | tdbp = new(g) TDBFMT(this, txfp); |
557 | |
558 | if (Multiple) |
559 | tdbp = new(g) TDBMUL(tdbp); |
560 | else |
561 | /*****************************************************************/ |
562 | /* For block tables, get eventually saved optimization values. */ |
563 | /*****************************************************************/ |
564 | if (tdbp->GetBlockValues(g)) { |
565 | PushWarning(g, tdbp); |
566 | // return NULL; // causes a crash when deleting index |
567 | } else { |
568 | if (IsOptimized()) { |
569 | if (map) { |
570 | txfp = new(g) MBKFAM(this); |
571 | } else if (Compressed) { |
572 | #if defined(GZ_SUPPORT) |
573 | if (Compressed == 1) |
574 | txfp = new(g) ZBKFAM(this); |
575 | else { |
576 | txfp->SetBlkPos(To_Pos); |
577 | ((PZLBFAM)txfp)->SetOptimized(To_Pos != NULL); |
578 | } // endelse |
579 | #else |
580 | sprintf(g->Message, MSG(NO_FEAT_SUPPORT), "GZ" ); |
581 | return NULL; |
582 | #endif |
583 | } else |
584 | txfp = new(g) BLKFAM(this); |
585 | |
586 | ((PTDBDOS)tdbp)->SetTxfp(txfp); |
587 | } // endif Optimized |
588 | |
589 | } // endelse |
590 | |
591 | } else |
592 | tdbp = new(g)TDBCCL(this); |
593 | |
594 | return tdbp; |
595 | } // end of GetTable |
596 | |
597 | /* -------------------------- Class TDBCSV --------------------------- */ |
598 | |
599 | /***********************************************************************/ |
600 | /* Implementation of the TDBCSV class. */ |
601 | /***********************************************************************/ |
602 | TDBCSV::TDBCSV(PCSVDEF tdp, PTXF txfp) : TDBDOS(tdp, txfp) |
603 | { |
604 | #if defined(_DEBUG) |
605 | assert (tdp); |
606 | #endif |
607 | Field = NULL; |
608 | Offset = NULL; |
609 | Fldlen = NULL; |
610 | Fields = 0; |
611 | Nerr = 0; |
612 | Quoted = tdp->Quoted; |
613 | Maxerr = tdp->Maxerr; |
614 | Accept = tdp->Accept; |
615 | Header = tdp->Header; |
616 | Sep = tdp->GetSep(); |
617 | Qot = tdp->GetQot(); |
618 | } // end of TDBCSV standard constructor |
619 | |
620 | TDBCSV::TDBCSV(PGLOBAL g, PTDBCSV tdbp) : TDBDOS(g, tdbp) |
621 | { |
622 | Fields = tdbp->Fields; |
623 | |
624 | if (Fields) { |
625 | if (tdbp->Offset) |
626 | Offset = (int*)PlugSubAlloc(g, NULL, sizeof(int) * Fields); |
627 | |
628 | if (tdbp->Fldlen) |
629 | Fldlen = (int*)PlugSubAlloc(g, NULL, sizeof(int) * Fields); |
630 | |
631 | Field = (PSZ *)PlugSubAlloc(g, NULL, sizeof(PSZ) * Fields); |
632 | |
633 | for (int i = 0; i < Fields; i++) { |
634 | if (Offset) |
635 | Offset[i] = tdbp->Offset[i]; |
636 | |
637 | if (Fldlen) |
638 | Fldlen[i] = tdbp->Fldlen[i]; |
639 | |
640 | if (Field) { |
641 | assert (Fldlen); |
642 | Field[i] = (PSZ)PlugSubAlloc(g, NULL, Fldlen[i] + 1); |
643 | Field[i][Fldlen[i]] = '\0'; |
644 | } // endif Field |
645 | |
646 | } // endfor i |
647 | |
648 | } else { |
649 | Field = NULL; |
650 | Offset = NULL; |
651 | Fldlen = NULL; |
652 | } // endif Fields |
653 | |
654 | Nerr = tdbp->Nerr; |
655 | Maxerr = tdbp->Maxerr; |
656 | Quoted = tdbp->Quoted; |
657 | Accept = tdbp->Accept; |
658 | Header = tdbp->Header; |
659 | Sep = tdbp->Sep; |
660 | Qot = tdbp->Qot; |
661 | } // end of TDBCSV copy constructor |
662 | |
663 | // Method |
664 | PTDB TDBCSV::Clone(PTABS t) |
665 | { |
666 | PTDB tp; |
667 | PCSVCOL cp1, cp2; |
668 | PGLOBAL g = t->G; // Is this really useful ??? |
669 | |
670 | tp = new(g) TDBCSV(g, this); |
671 | |
672 | for (cp1 = (PCSVCOL)Columns; cp1; cp1 = (PCSVCOL)cp1->GetNext()) { |
673 | cp2 = new(g) CSVCOL(cp1, tp); // Make a copy |
674 | NewPointer(t, cp1, cp2); |
675 | } // endfor cp1 |
676 | |
677 | return tp; |
678 | } // end of Clone |
679 | |
680 | /***********************************************************************/ |
681 | /* Allocate CSV column description block. */ |
682 | /***********************************************************************/ |
683 | PCOL TDBCSV::MakeCol(PGLOBAL g, PCOLDEF cdp, PCOL cprec, int n) |
684 | { |
685 | return new(g) CSVCOL(g, cdp, this, cprec, n); |
686 | } // end of MakeCol |
687 | |
688 | /***********************************************************************/ |
689 | /* Check whether the number of errors is greater than the maximum. */ |
690 | /***********************************************************************/ |
691 | bool TDBCSV::CheckErr(void) |
692 | { |
693 | return (++Nerr) > Maxerr; |
694 | } // end of CheckErr |
695 | |
696 | /***********************************************************************/ |
697 | /* CSV EstimatedLength. Returns an estimated minimum line length. */ |
698 | /***********************************************************************/ |
699 | int TDBCSV::EstimatedLength(void) |
700 | { |
701 | int n = 0; |
702 | PCOLDEF cdp; |
703 | |
704 | if (trace(1)) |
705 | htrc("EstimatedLength: Fields=%d Columns=%p\n" , Fields, Columns); |
706 | |
707 | for (cdp = To_Def->GetCols(); cdp; cdp = cdp->GetNext()) |
708 | if (!cdp->IsSpecial() && !cdp->IsVirtual()) // A true column |
709 | n++; |
710 | |
711 | return --n; // Number of separators if all fields are null |
712 | } // end of Estimated Length |
713 | |
714 | #if 0 |
715 | /***********************************************************************/ |
716 | /* CSV tables needs the use temporary files for Update. */ |
717 | /***********************************************************************/ |
718 | bool TDBCSV::IsUsingTemp(PGLOBAL g) |
719 | { |
720 | return (Use_Temp == TMP_YES || Use_Temp == TMP_FORCE || |
721 | (Use_Temp == TMP_AUTO && Mode == MODE_UPDATE)); |
722 | } // end of IsUsingTemp |
723 | #endif // 0 (Same as TDBDOS one) |
724 | |
725 | /***********************************************************************/ |
726 | /* CSV Access Method opening routine. */ |
727 | /* First allocate the Offset and Fldlen arrays according to the */ |
728 | /* greatest field used in that query. Then call the DOS opening fnc. */ |
729 | /***********************************************************************/ |
730 | bool TDBCSV::OpenDB(PGLOBAL g) |
731 | { |
732 | bool rc = false; |
733 | PCOLDEF cdp; |
734 | PDOSDEF tdp = (PDOSDEF)To_Def; |
735 | |
736 | if (Use != USE_OPEN && (Columns || Mode == MODE_UPDATE)) { |
737 | // Allocate the storage used to read (or write) records |
738 | int i, len; |
739 | PCSVCOL colp; |
740 | |
741 | if (!Fields) // May have been set in TABFMT::OpenDB |
742 | if (Mode != MODE_UPDATE && Mode != MODE_INSERT) { |
743 | for (colp = (PCSVCOL)Columns; colp; colp = (PCSVCOL)colp->Next) |
744 | if (!colp->IsSpecial() && !colp->IsVirtual()) |
745 | Fields = MY_MAX(Fields, (int)colp->Fldnum); |
746 | |
747 | if (Columns) |
748 | Fields++; // Fldnum was 0 based |
749 | |
750 | } else |
751 | for (cdp = tdp->GetCols(); cdp; cdp = cdp->GetNext()) |
752 | if (!cdp->IsSpecial() && !cdp->IsVirtual()) |
753 | Fields++; |
754 | |
755 | Offset = (int*)PlugSubAlloc(g, NULL, sizeof(int) * Fields); |
756 | Fldlen = (int*)PlugSubAlloc(g, NULL, sizeof(int) * Fields); |
757 | |
758 | if (Mode == MODE_INSERT || Mode == MODE_UPDATE) { |
759 | Field = (PSZ*)PlugSubAlloc(g, NULL, sizeof(PSZ) * Fields); |
760 | Fldtyp = (bool*)PlugSubAlloc(g, NULL, sizeof(bool) * Fields); |
761 | } // endif Mode |
762 | |
763 | for (i = 0; i < Fields; i++) { |
764 | Offset[i] = 0; |
765 | Fldlen[i] = 0; |
766 | |
767 | if (Field) { |
768 | Field[i] = NULL; |
769 | Fldtyp[i] = false; |
770 | } // endif Field |
771 | |
772 | } // endfor i |
773 | |
774 | if (Field) |
775 | // Prepare writing fields |
776 | if (Mode != MODE_UPDATE) { |
777 | for (colp = (PCSVCOL)Columns; colp; colp = (PCSVCOL)colp->Next) |
778 | if (!colp->IsSpecial() && !colp->IsVirtual()) { |
779 | i = colp->Fldnum; |
780 | len = colp->GetLength(); |
781 | Field[i] = (PSZ)PlugSubAlloc(g, NULL, len + 1); |
782 | Field[i][len] = '\0'; |
783 | Fldlen[i] = len; |
784 | Fldtyp[i] = IsTypeNum(colp->GetResultType()); |
785 | } // endif colp |
786 | |
787 | } else // MODE_UPDATE |
788 | for (cdp = tdp->GetCols(); cdp; cdp = cdp->GetNext()) |
789 | if (!cdp->IsSpecial() && !cdp->IsVirtual()) { |
790 | i = cdp->GetOffset() - 1; |
791 | len = cdp->GetLength(); |
792 | Field[i] = (PSZ)PlugSubAlloc(g, NULL, len + 1); |
793 | Field[i][len] = '\0'; |
794 | Fldlen[i] = len; |
795 | Fldtyp[i] = IsTypeNum(cdp->GetType()); |
796 | } // endif cdp |
797 | |
798 | } // endif Use |
799 | |
800 | if (Header) { |
801 | // Check that the Lrecl is at least equal to the header line length |
802 | int headlen = 0; |
803 | PCOLDEF cdp; |
804 | PDOSDEF tdp = (PDOSDEF)To_Def; |
805 | |
806 | for (cdp = tdp->GetCols(); cdp; cdp = cdp->GetNext()) |
807 | headlen += strlen(cdp->GetName()) + 3; // 3 if names are quoted |
808 | |
809 | if (headlen > Lrecl) { |
810 | Lrecl = headlen; |
811 | Txfp->Lrecl = headlen; |
812 | } // endif headlen |
813 | |
814 | } // endif Header |
815 | |
816 | Nerr = 0; |
817 | rc = TDBDOS::OpenDB(g); |
818 | |
819 | if (!rc && Mode == MODE_UPDATE && To_Kindex) |
820 | // Because KINDEX::Init is executed in mode READ, we must restore |
821 | // the Fldlen array that was modified when reading the table file. |
822 | for (cdp = tdp->GetCols(); cdp; cdp = cdp->GetNext()) |
823 | Fldlen[cdp->GetOffset() - 1] = cdp->GetLength(); |
824 | |
825 | return rc; |
826 | } // end of OpenDB |
827 | |
828 | /***********************************************************************/ |
829 | /* SkipHeader: Physically skip first header line if applicable. */ |
830 | /* This is called from TDBDOS::OpenDB and must be executed before */ |
831 | /* Kindex construction if the file is accessed using an index. */ |
832 | /***********************************************************************/ |
833 | bool TDBCSV::(PGLOBAL g) |
834 | { |
835 | int len = GetFileLength(g); |
836 | bool rc = false; |
837 | |
838 | #if defined(_DEBUG) |
839 | if (len < 0) |
840 | return true; |
841 | #endif // _DEBUG |
842 | |
843 | if (Header) { |
844 | if (Mode == MODE_INSERT) { |
845 | if (!len) { |
846 | // New file, the header line must be constructed and written |
847 | int i, n = 0; |
848 | int hlen = 0; |
849 | bool q = Qot && Quoted > 0; |
850 | PCOLDEF cdp; |
851 | |
852 | // Estimate the length of the header list |
853 | for (cdp = To_Def->GetCols(); cdp; cdp = cdp->GetNext()) { |
854 | hlen += (1 + strlen(cdp->GetName())); |
855 | hlen += ((q) ? 2 : 0); |
856 | n++; // Calculate the number of columns |
857 | } // endfor cdp |
858 | |
859 | if (hlen > Lrecl) { |
860 | sprintf(g->Message, MSG(LRECL_TOO_SMALL), hlen); |
861 | return true; |
862 | } // endif hlen |
863 | |
864 | // File is empty, write a header record |
865 | memset(To_Line, 0, Lrecl); |
866 | |
867 | // The column order in the file is given by the offset value |
868 | for (i = 1; i <= n; i++) |
869 | for (cdp = To_Def->GetCols(); cdp; cdp = cdp->GetNext()) |
870 | if (cdp->GetOffset() == i) { |
871 | if (q) |
872 | To_Line[strlen(To_Line)] = Qot; |
873 | |
874 | strcat(To_Line, cdp->GetName()); |
875 | |
876 | if (q) |
877 | To_Line[strlen(To_Line)] = Qot; |
878 | |
879 | if (i < n) |
880 | To_Line[strlen(To_Line)] = Sep; |
881 | |
882 | } // endif Offset |
883 | |
884 | rc = (Txfp->WriteBuffer(g) == RC_FX); |
885 | } // endif !FileLength |
886 | |
887 | } else if (Mode == MODE_DELETE) { |
888 | if (len) |
889 | rc = (Txfp->SkipRecord(g, true) == RC_FX); |
890 | |
891 | } else if (len) // !Insert && !Delete |
892 | rc = (Txfp->SkipRecord(g, false) == RC_FX || Txfp->RecordPos(g)); |
893 | |
894 | } // endif Header |
895 | |
896 | return rc; |
897 | } // end of SkipHeader |
898 | |
899 | /***********************************************************************/ |
900 | /* ReadBuffer: Physical read routine for the CSV access method. */ |
901 | /***********************************************************************/ |
902 | int TDBCSV::ReadBuffer(PGLOBAL g) |
903 | { |
904 | //char *p1, *p2, *p = NULL; |
905 | char *p2, *p = NULL; |
906 | int i, n, len, rc = Txfp->ReadBuffer(g); |
907 | bool bad = false; |
908 | |
909 | if (trace(2)) |
910 | htrc("CSV: Row is '%s' rc=%d\n" , To_Line, rc); |
911 | |
912 | if (rc != RC_OK || !Fields) |
913 | return rc; |
914 | else |
915 | p2 = To_Line; |
916 | |
917 | // Find the offsets and lengths of the columns for this row |
918 | for (i = 0; i < Fields; i++) { |
919 | if (!bad) { |
920 | if (Qot && *p2 == Qot) { // Quoted field |
921 | //for (n = 0, p1 = ++p2; (p = strchr(p1, Qot)); p1 = p + 2) |
922 | // if (*(p + 1) == Qot) |
923 | // n++; // Doubled internal quotes |
924 | // else |
925 | // break; // Final quote |
926 | |
927 | for (n = 0, p = ++p2; p; p++) |
928 | if (*p == Qot || *p == '\\') { |
929 | if (*(++p) == Qot) |
930 | n++; // Escaped internal quotes |
931 | else if (*(p - 1) == Qot) |
932 | break; // Final quote |
933 | } // endif *p |
934 | |
935 | if (p) { |
936 | //len = p++ - p2; |
937 | len = (int)(p - p2 - 1); |
938 | |
939 | // if (Sep != ' ') |
940 | // for (; *p == ' '; p++) ; // Skip blanks |
941 | |
942 | if (*p != Sep && i != Fields - 1) { // Should be the separator |
943 | if (CheckErr()) { |
944 | sprintf(g->Message, MSG(MISSING_FIELD), |
945 | i+1, Name, RowNumber(g)); |
946 | return RC_FX; |
947 | } else if (Accept) |
948 | bad = true; |
949 | else |
950 | return RC_NF; |
951 | |
952 | } // endif p |
953 | |
954 | if (n) { |
955 | int j, k; |
956 | |
957 | // Suppress the escape of internal quotes |
958 | for (j = k = 0; j < len; j++, k++) { |
959 | if (p2[j] == Qot || (p2[j] == '\\' && p2[j + 1] == Qot)) |
960 | j++; // skip escape char |
961 | else if (p2[j] == '\\') |
962 | p2[k++] = p2[j++]; // avoid \\Qot |
963 | |
964 | p2[k] = p2[j]; |
965 | } // endfor i, j |
966 | |
967 | len -= n; |
968 | } // endif n |
969 | |
970 | } else if (CheckErr()) { |
971 | sprintf(g->Message, MSG(BAD_QUOTE_FIELD), |
972 | Name, i+1, RowNumber(g)); |
973 | return RC_FX; |
974 | } else if (Accept) { |
975 | len = strlen(p2); |
976 | bad = true; |
977 | } else |
978 | return RC_NF; |
979 | |
980 | } else if ((p = strchr(p2, Sep))) |
981 | len = (int)(p - p2); |
982 | else if (i == Fields - 1) |
983 | len = strlen(p2); |
984 | else if (Accept && Maxerr == 0) { |
985 | len = strlen(p2); |
986 | bad = true; |
987 | } else if (CheckErr()) { |
988 | sprintf(g->Message, MSG(MISSING_FIELD), i+1, Name, RowNumber(g)); |
989 | return RC_FX; |
990 | } else if (Accept) { |
991 | len = strlen(p2); |
992 | bad = true; |
993 | } else |
994 | return RC_NF; |
995 | |
996 | } else |
997 | len = 0; |
998 | |
999 | Offset[i] = (int)(p2 - To_Line); |
1000 | |
1001 | if (Mode != MODE_UPDATE) |
1002 | Fldlen[i] = len; |
1003 | else if (len > Fldlen[i]) { |
1004 | sprintf(g->Message, MSG(FIELD_TOO_LONG), i+1, RowNumber(g)); |
1005 | return RC_FX; |
1006 | } else { |
1007 | strncpy(Field[i], p2, len); |
1008 | Field[i][len] = '\0'; |
1009 | } // endif Mode |
1010 | |
1011 | if (p) |
1012 | p2 = p + 1; |
1013 | |
1014 | } // endfor i |
1015 | |
1016 | return rc; |
1017 | } // end of ReadBuffer |
1018 | |
1019 | /***********************************************************************/ |
1020 | /* Prepare the line to write. */ |
1021 | /***********************************************************************/ |
1022 | bool TDBCSV::PrepareWriting(PGLOBAL g) |
1023 | { |
1024 | char sep[2], qot[2]; |
1025 | int i, nlen, oldlen = strlen(To_Line); |
1026 | |
1027 | if (trace(2)) |
1028 | htrc("CSV WriteDB: R%d Mode=%d key=%p link=%p\n" , |
1029 | Tdb_No, Mode, To_Key_Col, To_Link); |
1030 | |
1031 | // Before writing the line we must check its length |
1032 | if ((nlen = CheckWrite(g)) < 0) |
1033 | return true; |
1034 | |
1035 | // Before writing the line we must make it |
1036 | sep[0] = Sep; |
1037 | sep[1] = '\0'; |
1038 | qot[0] = Qot; |
1039 | qot[1] = '\0'; |
1040 | *To_Line = '\0'; |
1041 | |
1042 | for (i = 0; i < Fields; i++) { |
1043 | if (i) |
1044 | strcat(To_Line, sep); |
1045 | |
1046 | if (Field[i]) |
1047 | if (!strlen(Field[i])) { |
1048 | // Generally null fields are not quoted |
1049 | if (Quoted > 2) |
1050 | // Except if explicitely required |
1051 | strcat(strcat(To_Line, qot), qot); |
1052 | |
1053 | } else if (Qot && (strchr(Field[i], Sep) || *Field[i] == Qot |
1054 | || Quoted > 1 || (Quoted == 1 && !Fldtyp[i]))) |
1055 | if (strchr(Field[i], Qot)) { |
1056 | // Field contains quotes that must be doubled |
1057 | int j, k = strlen(To_Line), n = strlen(Field[i]); |
1058 | |
1059 | To_Line[k++] = Qot; |
1060 | |
1061 | for (j = 0; j < n; j++) { |
1062 | if (Field[i][j] == Qot) |
1063 | To_Line[k++] = Qot; |
1064 | |
1065 | To_Line[k++] = Field[i][j]; |
1066 | } // endfor j |
1067 | |
1068 | To_Line[k++] = Qot; |
1069 | To_Line[k] = '\0'; |
1070 | } else |
1071 | strcat(strcat(strcat(To_Line, qot), Field[i]), qot); |
1072 | |
1073 | else |
1074 | strcat(To_Line, Field[i]); |
1075 | |
1076 | } // endfor i |
1077 | |
1078 | #if defined(_DEBUG) |
1079 | assert ((unsigned)nlen == strlen(To_Line)); |
1080 | #endif |
1081 | |
1082 | if (Mode == MODE_UPDATE && nlen < oldlen |
1083 | && !((PDOSFAM)Txfp)->GetUseTemp()) { |
1084 | // In Update mode with no temp file, line length must not change |
1085 | To_Line[nlen] = Sep; |
1086 | |
1087 | for (nlen++; nlen < oldlen; nlen++) |
1088 | To_Line[nlen] = ' '; |
1089 | |
1090 | To_Line[nlen] = '\0'; |
1091 | } // endif |
1092 | |
1093 | if (trace(2)) |
1094 | htrc("Write: line is=%s" , To_Line); |
1095 | |
1096 | return false; |
1097 | } // end of PrepareWriting |
1098 | |
1099 | /***********************************************************************/ |
1100 | /* Data Base write routine CSV file access method. */ |
1101 | /***********************************************************************/ |
1102 | int TDBCSV::WriteDB(PGLOBAL g) |
1103 | { |
1104 | // Before writing the line we must check and prepare it |
1105 | if (PrepareWriting(g)) |
1106 | return RC_FX; |
1107 | |
1108 | /*********************************************************************/ |
1109 | /* Now start the writing process. */ |
1110 | /*********************************************************************/ |
1111 | return Txfp->WriteBuffer(g); |
1112 | } // end of WriteDB |
1113 | |
1114 | /***********************************************************************/ |
1115 | /* Check whether a new line fit in the file lrecl size. */ |
1116 | /***********************************************************************/ |
1117 | int TDBCSV::CheckWrite(PGLOBAL g) |
1118 | { |
1119 | int maxlen, n, nlen = (Fields - 1); |
1120 | |
1121 | if (trace(2)) |
1122 | htrc("CheckWrite: R%d Mode=%d\n" , Tdb_No, Mode); |
1123 | |
1124 | // Before writing the line we must check its length |
1125 | maxlen = (Mode == MODE_UPDATE && !Txfp->GetUseTemp()) |
1126 | ? strlen(To_Line) : Lrecl; |
1127 | |
1128 | // Check whether record is too int |
1129 | for (int i = 0; i < Fields; i++) |
1130 | if (Field[i]) { |
1131 | if (!(n = strlen(Field[i]))) |
1132 | n += (Quoted > 2 ? 2 : 0); |
1133 | else if (strchr(Field[i], Sep) || (Qot && *Field[i] == Qot) |
1134 | || Quoted > 1 || (Quoted == 1 && !Fldtyp[i])) |
1135 | if (!Qot) { |
1136 | sprintf(g->Message, MSG(SEP_IN_FIELD), i + 1); |
1137 | return -1; |
1138 | } else { |
1139 | // Quotes inside a quoted field must be doubled |
1140 | char *p1, *p2; |
1141 | |
1142 | for (p1 = Field[i]; (p2 = strchr(p1, Qot)); p1 = p2 + 1) |
1143 | n++; |
1144 | |
1145 | n += 2; // Outside quotes |
1146 | } // endif |
1147 | |
1148 | if ((nlen += n) > maxlen) { |
1149 | strcpy(g->Message, MSG(LINE_TOO_LONG)); |
1150 | return -1; |
1151 | } // endif nlen |
1152 | |
1153 | } // endif Field |
1154 | |
1155 | return nlen; |
1156 | } // end of CheckWrite |
1157 | |
1158 | /* ------------------------------------------------------------------- */ |
1159 | |
1160 | /***********************************************************************/ |
1161 | /* Implementation of the TDBFMT class. */ |
1162 | /***********************************************************************/ |
1163 | TDBFMT::TDBFMT(PGLOBAL g, PTDBFMT tdbp) : TDBCSV(g, tdbp) |
1164 | { |
1165 | FldFormat = tdbp->FldFormat; |
1166 | To_Fld = tdbp->To_Fld; |
1167 | FmtTest = tdbp->FmtTest; |
1168 | Linenum = tdbp->Linenum; |
1169 | } // end of TDBFMT copy constructor |
1170 | |
1171 | // Method |
1172 | PTDB TDBFMT::Clone(PTABS t) |
1173 | { |
1174 | PTDB tp; |
1175 | PCSVCOL cp1, cp2; |
1176 | //PFMTCOL cp1, cp2; |
1177 | PGLOBAL g = t->G; // Is this really useful ??? |
1178 | |
1179 | tp = new(g) TDBFMT(g, this); |
1180 | |
1181 | for (cp1 = (PCSVCOL)Columns; cp1; cp1 = (PCSVCOL)cp1->GetNext()) { |
1182 | //for (cp1 = (PFMTCOL)Columns; cp1; cp1 = (PFMTCOL)cp1->GetNext()) { |
1183 | cp2 = new(g) CSVCOL(cp1, tp); // Make a copy |
1184 | // cp2 = new(g) FMTCOL(cp1, tp); // Make a copy |
1185 | NewPointer(t, cp1, cp2); |
1186 | } // endfor cp1 |
1187 | |
1188 | return tp; |
1189 | } // end of Clone |
1190 | |
1191 | /***********************************************************************/ |
1192 | /* Allocate FMT column description block. */ |
1193 | /***********************************************************************/ |
1194 | PCOL TDBFMT::MakeCol(PGLOBAL g, PCOLDEF cdp, PCOL cprec, int n) |
1195 | { |
1196 | return new(g) CSVCOL(g, cdp, this, cprec, n); |
1197 | //return new(g) FMTCOL(cdp, this, cprec, n); |
1198 | } // end of MakeCol |
1199 | |
1200 | /***********************************************************************/ |
1201 | /* FMT EstimatedLength. Returns an estimated minimum line length. */ |
1202 | /* The big problem here is how can we astimated that minimum ? */ |
1203 | /***********************************************************************/ |
1204 | int TDBFMT::EstimatedLength(void) |
1205 | { |
1206 | // This is rather stupid !!! |
1207 | return ((PDOSDEF)To_Def)->GetEnding() + (int)((Lrecl / 10) + 1); |
1208 | } // end of EstimatedLength |
1209 | |
1210 | /***********************************************************************/ |
1211 | /* FMT Access Method opening routine. */ |
1212 | /***********************************************************************/ |
1213 | bool TDBFMT::OpenDB(PGLOBAL g) |
1214 | { |
1215 | Linenum = 0; |
1216 | |
1217 | if (Mode == MODE_INSERT || Mode == MODE_UPDATE) { |
1218 | sprintf(g->Message, MSG(FMT_WRITE_NIY), "FMT" ); |
1219 | return true; // NIY |
1220 | } // endif Mode |
1221 | |
1222 | if (Use != USE_OPEN && Columns) { |
1223 | // Make the formats used to read records |
1224 | PSZ pfm; |
1225 | int i, n; |
1226 | PCSVCOL colp; |
1227 | PCOLDEF cdp; |
1228 | PDOSDEF tdp = (PDOSDEF)To_Def; |
1229 | |
1230 | for (colp = (PCSVCOL)Columns; colp; colp = (PCSVCOL)colp->Next) |
1231 | if (!colp->IsSpecial() && !colp->IsVirtual()) // a true column |
1232 | Fields = MY_MAX(Fields, (int)colp->Fldnum); |
1233 | |
1234 | if (Columns) |
1235 | Fields++; // Fldnum was 0 based |
1236 | |
1237 | To_Fld = PlugSubAlloc(g, NULL, Lrecl + 1); |
1238 | FldFormat = (PSZ*)PlugSubAlloc(g, NULL, sizeof(PSZ) * Fields); |
1239 | memset(FldFormat, 0, sizeof(PSZ) * Fields); |
1240 | FmtTest = (int*)PlugSubAlloc(g, NULL, sizeof(int) * Fields); |
1241 | memset(FmtTest, 0, sizeof(int) * Fields); |
1242 | |
1243 | // Get the column formats |
1244 | for (cdp = tdp->GetCols(); cdp; cdp = cdp->GetNext()) |
1245 | if (!cdp->IsSpecial() && !cdp->IsVirtual() |
1246 | && (i = cdp->GetOffset() - 1) < Fields) { |
1247 | if (!(pfm = cdp->GetFmt())) { |
1248 | sprintf(g->Message, MSG(NO_FLD_FORMAT), i + 1, Name); |
1249 | return true; |
1250 | } // endif pfm |
1251 | |
1252 | // Roughly check the Fmt format |
1253 | if ((n = strlen(pfm) - 2) < 4) { |
1254 | sprintf(g->Message, MSG(BAD_FLD_FORMAT), i + 1, Name); |
1255 | return true; |
1256 | } // endif n |
1257 | |
1258 | FldFormat[i] = (PSZ)PlugSubAlloc(g, NULL, n + 5); |
1259 | strcpy(FldFormat[i], pfm); |
1260 | |
1261 | if (!strcmp(pfm + n, "%m" )) { |
1262 | // This is a field that can be missing. Flag it so it can |
1263 | // be handled with special processing. |
1264 | FldFormat[i][n+1] = 'n'; // To have sscanf normal processing |
1265 | FmtTest[i] = 2; |
1266 | } else if (i+1 < Fields && strcmp(pfm + n, "%n" )) { |
1267 | // There are trailing characters after the field contents |
1268 | // add a marker for the next field start position. |
1269 | strcat(FldFormat[i], "%n" ); |
1270 | FmtTest[i] = 1; |
1271 | } // endif's |
1272 | |
1273 | } // endif i |
1274 | |
1275 | } // endif Use |
1276 | |
1277 | return TDBCSV::OpenDB(g); |
1278 | } // end of OpenDB |
1279 | |
1280 | /***********************************************************************/ |
1281 | /* ReadBuffer: Physical read routine for the FMT access method. */ |
1282 | /***********************************************************************/ |
1283 | int TDBFMT::ReadBuffer(PGLOBAL g) |
1284 | { |
1285 | int i, len, n, deb, fin, nwp, pos = 0, rc; |
1286 | bool bad = false; |
1287 | |
1288 | if ((rc = Txfp->ReadBuffer(g)) != RC_OK || !Fields) |
1289 | return rc; |
1290 | else |
1291 | ++Linenum; |
1292 | |
1293 | if (trace(2)) |
1294 | htrc("FMT: Row %d is '%s' rc=%d\n" , Linenum, To_Line, rc); |
1295 | |
1296 | // Find the offsets and lengths of the columns for this row |
1297 | for (i = 0; i < Fields; i++) { |
1298 | if (!bad) { |
1299 | deb = fin = -1; |
1300 | |
1301 | if (!FldFormat[i]) { |
1302 | n = 0; |
1303 | } else if (FmtTest[i] == 1) { |
1304 | nwp = -1; |
1305 | n = sscanf(To_Line + pos, FldFormat[i], &deb, To_Fld, &fin, &nwp); |
1306 | } else { |
1307 | n = sscanf(To_Line + pos, FldFormat[i], &deb, To_Fld, &fin); |
1308 | |
1309 | if (n != 1 && (deb >= 0 || i == Fields - 1) && FmtTest[i] == 2) { |
1310 | // Missing optional field, not an error |
1311 | n = 1; |
1312 | |
1313 | if (i == Fields - 1) |
1314 | fin = deb = 0; |
1315 | else |
1316 | fin = deb; |
1317 | |
1318 | } // endif n |
1319 | |
1320 | nwp = fin; |
1321 | } // endif i |
1322 | |
1323 | if (n != 1 || deb < 0 || fin < 0 || nwp < 0) { |
1324 | // This is to avoid a very strange sscanf bug occuring |
1325 | // with fields that ends with a null character. |
1326 | // This bug causes subsequent sscanf to return in error, |
1327 | // so next lines are not parsed correctly. |
1328 | sscanf("a" , "%*c" ); // Seems to reset things Ok |
1329 | |
1330 | if (CheckErr()) { |
1331 | sprintf(g->Message, MSG(BAD_LINEFLD_FMT), Linenum, i + 1, Name); |
1332 | return RC_FX; |
1333 | } else if (Accept) |
1334 | bad = true; |
1335 | else |
1336 | return RC_NF; |
1337 | |
1338 | } // endif n... |
1339 | |
1340 | } // endif !bad |
1341 | |
1342 | if (!bad) { |
1343 | Offset[i] = pos + deb; |
1344 | len = fin - deb; |
1345 | } else { |
1346 | nwp = 0; |
1347 | Offset[i] = pos; |
1348 | len = 0; |
1349 | } // endif bad |
1350 | |
1351 | // if (Mode != MODE_UPDATE) |
1352 | Fldlen[i] = len; |
1353 | // else if (len > Fldlen[i]) { |
1354 | // sprintf(g->Message, MSG(FIELD_TOO_LONG), i+1, To_Tdb->RowNumber(g)); |
1355 | // return RC_FX; |
1356 | // } else { |
1357 | // strncpy(Field[i], To_Line + pos, len); |
1358 | // Field[i][len] = '\0'; |
1359 | // } // endif Mode |
1360 | |
1361 | pos += nwp; |
1362 | } // endfor i |
1363 | |
1364 | if (bad) |
1365 | Nerr++; |
1366 | else |
1367 | sscanf("a" , "%*c" ); // Seems to reset things Ok |
1368 | |
1369 | return rc; |
1370 | } // end of ReadBuffer |
1371 | |
1372 | /***********************************************************************/ |
1373 | /* Data Base write routine FMT file access method. */ |
1374 | /***********************************************************************/ |
1375 | int TDBFMT::WriteDB(PGLOBAL g) |
1376 | { |
1377 | sprintf(g->Message, MSG(FMT_WRITE_NIY), "FMT" ); |
1378 | return RC_FX; // NIY |
1379 | } // end of WriteDB |
1380 | |
1381 | // ------------------------ CSVCOL functions ---------------------------- |
1382 | |
1383 | /***********************************************************************/ |
1384 | /* CSVCOL public constructor */ |
1385 | /***********************************************************************/ |
1386 | CSVCOL::CSVCOL(PGLOBAL g, PCOLDEF cdp, PTDB tdbp, PCOL cprec, int i) |
1387 | : DOSCOL(g, cdp, tdbp, cprec, i, "CSV" ) |
1388 | { |
1389 | Fldnum = Deplac - 1; |
1390 | Deplac = 0; |
1391 | } // end of CSVCOL constructor |
1392 | |
1393 | /***********************************************************************/ |
1394 | /* CSVCOL constructor used for copying columns. */ |
1395 | /* tdbp is the pointer to the new table descriptor. */ |
1396 | /***********************************************************************/ |
1397 | CSVCOL::CSVCOL(CSVCOL *col1, PTDB tdbp) : DOSCOL(col1, tdbp) |
1398 | { |
1399 | Fldnum = col1->Fldnum; |
1400 | } // end of CSVCOL copy constructor |
1401 | |
1402 | /***********************************************************************/ |
1403 | /* VarSize: This function tells UpdateDB whether or not the block */ |
1404 | /* optimization file must be redone if this column is updated, even */ |
1405 | /* it is not sorted or clustered. This applies to a blocked table, */ |
1406 | /* because if it is updated using a temporary file, the block size */ |
1407 | /* may be modified. */ |
1408 | /***********************************************************************/ |
1409 | bool CSVCOL::VarSize(void) |
1410 | { |
1411 | PTXF txfp = ((PTDBCSV)To_Tdb)->Txfp; |
1412 | |
1413 | if (txfp->IsBlocked() && txfp->GetUseTemp()) |
1414 | // Blocked table using a temporary file |
1415 | return true; |
1416 | else |
1417 | return false; |
1418 | |
1419 | } // end VarSize |
1420 | |
1421 | /***********************************************************************/ |
1422 | /* ReadColumn: call DOSCOL::ReadColumn after having set the offet */ |
1423 | /* and length of the field to read as calculated by TDBCSV::ReadDB. */ |
1424 | /***********************************************************************/ |
1425 | void CSVCOL::ReadColumn(PGLOBAL g) |
1426 | { |
1427 | int rc; |
1428 | PTDBCSV tdbp = (PTDBCSV)To_Tdb; |
1429 | |
1430 | /*********************************************************************/ |
1431 | /* If physical reading of the line was deferred, do it now. */ |
1432 | /*********************************************************************/ |
1433 | if (!tdbp->IsRead()) |
1434 | if ((rc = tdbp->ReadBuffer(g)) != RC_OK) { |
1435 | if (rc == RC_EF) |
1436 | sprintf(g->Message, MSG(INV_DEF_READ), rc); |
1437 | |
1438 | throw 34; |
1439 | } // endif |
1440 | |
1441 | if (tdbp->Mode != MODE_UPDATE) { |
1442 | int colen = Long; // Column length |
1443 | |
1444 | // Set the field offset and length for this row |
1445 | Deplac = tdbp->Offset[Fldnum]; // Field offset |
1446 | Long = tdbp->Fldlen[Fldnum]; // Field length |
1447 | |
1448 | if (trace(2)) |
1449 | htrc("CSV ReadColumn %s Fldnum=%d offset=%d fldlen=%d\n" , |
1450 | Name, Fldnum, Deplac, Long); |
1451 | |
1452 | if (Long > colen && tdbp->CheckErr()) { |
1453 | Long = colen; // Restore column length |
1454 | sprintf(g->Message, MSG(FLD_TOO_LNG_FOR), |
1455 | Fldnum + 1, Name, To_Tdb->RowNumber(g), tdbp->GetFile(g)); |
1456 | throw 34; |
1457 | } // endif Long |
1458 | |
1459 | // Now do the reading |
1460 | DOSCOL::ReadColumn(g); |
1461 | |
1462 | // Restore column length |
1463 | Long = colen; |
1464 | } else { // Mode Update |
1465 | // Field have been copied in TDB Field array |
1466 | PSZ fp = tdbp->Field[Fldnum]; |
1467 | |
1468 | if (Dsp) |
1469 | for (int i = 0; fp[i]; i++) |
1470 | if (fp[i] == Dsp) |
1471 | fp[i] = '.'; |
1472 | |
1473 | Value->SetValue_psz(fp); |
1474 | |
1475 | // Set null when applicable |
1476 | if (Nullable) |
1477 | Value->SetNull(Value->IsZero()); |
1478 | |
1479 | } // endif Mode |
1480 | |
1481 | } // end of ReadColumn |
1482 | |
1483 | /***********************************************************************/ |
1484 | /* WriteColumn: The column is written in TDBCSV matching Field. */ |
1485 | /***********************************************************************/ |
1486 | void CSVCOL::WriteColumn(PGLOBAL g) |
1487 | { |
1488 | char *p, buf[64]; |
1489 | int flen; |
1490 | PTDBCSV tdbp = (PTDBCSV)To_Tdb; |
1491 | |
1492 | if (trace(2)) |
1493 | htrc("CSV WriteColumn: col %s R%d coluse=%.4X status=%.4X\n" , |
1494 | Name, tdbp->GetTdb_No(), ColUse, Status); |
1495 | |
1496 | flen = GetLength(); |
1497 | |
1498 | if (trace(2)) |
1499 | htrc("Lrecl=%d Long=%d field=%d coltype=%d colval=%p\n" , |
1500 | tdbp->Lrecl, Long, flen, Buf_Type, Value); |
1501 | |
1502 | /*********************************************************************/ |
1503 | /* Check whether the new value has to be converted to Buf_Type. */ |
1504 | /*********************************************************************/ |
1505 | if (Value != To_Val) |
1506 | Value->SetValue_pval(To_Val, false); // Convert the updated value |
1507 | |
1508 | /*********************************************************************/ |
1509 | /* Get the string representation of the column value. */ |
1510 | /*********************************************************************/ |
1511 | p = Value->ShowValue(buf); |
1512 | |
1513 | if (trace(2)) |
1514 | htrc("new length(%p)=%d\n" , p, strlen(p)); |
1515 | |
1516 | if ((signed)strlen(p) > flen) { |
1517 | sprintf(g->Message, MSG(BAD_FLD_LENGTH), Name, p, flen, |
1518 | tdbp->RowNumber(g), tdbp->GetFile(g)); |
1519 | throw 34; |
1520 | } else if (Dsp) |
1521 | for (int i = 0; p[i]; i++) |
1522 | if (p[i] == '.') |
1523 | p[i] = Dsp; |
1524 | |
1525 | if (trace(2)) |
1526 | htrc("buffer=%s\n" , p); |
1527 | |
1528 | /*********************************************************************/ |
1529 | /* Updating must be done also during the first pass so writing the */ |
1530 | /* updated record can be checked for acceptable record length. */ |
1531 | /*********************************************************************/ |
1532 | if (Fldnum < 0) { |
1533 | // This can happen for wrong offset value in XDB files |
1534 | sprintf(g->Message, MSG(BAD_FIELD_RANK), Fldnum + 1, Name); |
1535 | throw 34; |
1536 | } else |
1537 | strncpy(tdbp->Field[Fldnum], p, flen); |
1538 | |
1539 | if (trace(2)) |
1540 | htrc(" col written: '%s'\n" , p); |
1541 | |
1542 | } // end of WriteColumn |
1543 | |
1544 | /* ---------------------------TDBCCL class --------------------------- */ |
1545 | |
1546 | /***********************************************************************/ |
1547 | /* TDBCCL class constructor. */ |
1548 | /***********************************************************************/ |
1549 | TDBCCL::TDBCCL(PCSVDEF tdp) : TDBCAT(tdp) |
1550 | { |
1551 | Topt = tdp->GetTopt(); |
1552 | } // end of TDBCCL constructor |
1553 | |
1554 | /***********************************************************************/ |
1555 | /* GetResult: Get the list the CSV file columns. */ |
1556 | /***********************************************************************/ |
1557 | PQRYRES TDBCCL::GetResult(PGLOBAL g) |
1558 | { |
1559 | return CSVColumns(g, ((PTABDEF)To_Def)->GetPath(), Topt, false); |
1560 | } // end of GetResult |
1561 | |
1562 | /* ------------------------ End of TabFmt ---------------------------- */ |
1563 | |