nethercote | 5912c81 | 2004-02-15 15:38:08 +0000 | [diff] [blame] | 1 | /* This file is part of hp2ps, a graph drawer for memory profiles. |
| 2 | Copyright (C) 2002 The University Court of the University of Glasgow. |
| 3 | This program is governed by the license contained in the file LICENSE. */ |
| 4 | |
nethercote | c9f3692 | 2004-02-14 16:40:02 +0000 | [diff] [blame] | 5 | #include <ctype.h> |
| 6 | #include <stdio.h> |
| 7 | #include <stdlib.h> |
| 8 | #include <string.h> |
| 9 | #include "Main.h" |
| 10 | #include "Defines.h" |
| 11 | #include "Error.h" |
| 12 | #include "HpFile.h" |
| 13 | #include "Utilities.h" |
| 14 | |
| 15 | #ifndef atof |
| 16 | double atof PROTO((const char *)); |
| 17 | #endif |
| 18 | |
| 19 | /* own stuff already included */ |
| 20 | |
| 21 | #define N_MARKS 50 /* start size of the mark table */ |
| 22 | #define N_SAMPLES 500 /* start size of the sample table */ |
| 23 | |
| 24 | char *theident; |
| 25 | char *thestring; |
| 26 | int theinteger; |
| 27 | floatish thefloatish; |
| 28 | int g_ch; /* last character read */ |
| 29 | token thetok; /* last token */ |
| 30 | int linenum; /* current line number */ |
| 31 | int endfile; /* true at end of file */ |
| 32 | |
| 33 | static boolish gotjob = 0; /* "JOB" read */ |
| 34 | static boolish gotdate = 0; /* "DATE" read */ |
| 35 | static boolish gotvalueunit = 0; /* "VALUE_UNIT" read */ |
| 36 | static boolish gotsampleunit = 0; /* "SAMPLE_UNIT" read */ |
| 37 | static boolish insample = 0; /* true when in sample */ |
| 38 | |
| 39 | static floatish lastsample; /* the last sample time */ |
| 40 | |
| 41 | static void GetHpLine PROTO((FILE *)); /* forward */ |
| 42 | static void GetHpTok PROTO((FILE *)); /* forward */ |
| 43 | |
| 44 | static struct entry *GetEntry PROTO((char *)); /* forward */ |
| 45 | |
| 46 | static void MakeIdentTable PROTO((void)); /* forward */ |
| 47 | |
| 48 | char *jobstring; |
| 49 | char *datestring; |
| 50 | |
| 51 | char *sampleunitstring; |
| 52 | char *valueunitstring; |
| 53 | |
| 54 | floatish *samplemap; /* sample intervals */ |
| 55 | floatish *markmap; /* sample marks */ |
| 56 | |
| 57 | /* |
| 58 | * An extremely simple parser. The input is organised into lines of |
| 59 | * the form |
| 60 | * |
| 61 | * JOB s -- job identifier string |
| 62 | * DATE s -- date string |
| 63 | * SAMPLE_UNIT s -- sample unit eg "seconds" |
| 64 | * VALUE_UNIT s -- value unit eg "bytes" |
| 65 | * MARK i -- sample mark |
| 66 | * BEGIN_SAMPLE i -- start of ith sample |
| 67 | * identifier i -- there are i identifiers in this sample |
| 68 | * END_SAMPLE i -- end of ith sample |
| 69 | * |
| 70 | */ |
| 71 | |
| 72 | void |
| 73 | GetHpFile(infp) |
| 74 | FILE *infp; |
| 75 | { |
| 76 | nsamples = 0; |
| 77 | nmarks = 0; |
| 78 | nidents = 0; |
| 79 | |
| 80 | g_ch = ' '; |
| 81 | endfile = 0; |
| 82 | linenum = 1; |
| 83 | lastsample = 0.0; |
| 84 | |
| 85 | GetHpTok(infp); |
| 86 | |
| 87 | while (endfile == 0) { |
| 88 | GetHpLine(infp); |
| 89 | } |
| 90 | |
| 91 | if (!gotjob) { |
| 92 | Error("%s: JOB missing", hpfile); |
| 93 | } |
| 94 | |
| 95 | if (!gotdate) { |
| 96 | Error("%s: DATE missing", hpfile); |
| 97 | } |
| 98 | |
| 99 | if (!gotvalueunit) { |
| 100 | Error("%s: VALUE_UNIT missing", hpfile); |
| 101 | } |
| 102 | |
| 103 | if (!gotsampleunit) { |
| 104 | Error("%s: SAMPLE_UNIT missing", hpfile); |
| 105 | } |
| 106 | |
| 107 | if (nsamples == 0) { |
| 108 | Error("%s: contains no samples", hpfile); |
| 109 | } |
| 110 | |
| 111 | |
| 112 | MakeIdentTable(); |
| 113 | |
| 114 | fclose(hpfp); |
| 115 | } |
| 116 | |
| 117 | |
| 118 | /* |
| 119 | * Read the next line from the input, check the syntax, and perform |
| 120 | * the appropriate action. |
| 121 | */ |
| 122 | |
| 123 | static void |
| 124 | GetHpLine(infp) |
| 125 | FILE* infp; |
| 126 | { |
| 127 | static intish nmarkmax = 0, nsamplemax = 0; |
| 128 | |
| 129 | switch (thetok) { |
| 130 | case JOB_TOK: |
| 131 | GetHpTok(infp); |
| 132 | if (thetok != STRING_TOK) { |
| 133 | Error("%s, line %d: string must follow JOB", hpfile, linenum); |
| 134 | } |
| 135 | jobstring = thestring; |
| 136 | gotjob = 1; |
| 137 | GetHpTok(infp); |
| 138 | break; |
| 139 | |
| 140 | case DATE_TOK: |
| 141 | GetHpTok(infp); |
| 142 | if (thetok != STRING_TOK) { |
| 143 | Error("%s, line %d: string must follow DATE", hpfile, linenum); |
| 144 | } |
| 145 | datestring = thestring; |
| 146 | gotdate = 1; |
| 147 | GetHpTok(infp); |
| 148 | break; |
| 149 | |
| 150 | case SAMPLE_UNIT_TOK: |
| 151 | GetHpTok(infp); |
| 152 | if (thetok != STRING_TOK) { |
| 153 | Error("%s, line %d: string must follow SAMPLE_UNIT", hpfile, |
| 154 | linenum); |
| 155 | } |
| 156 | sampleunitstring = thestring; |
| 157 | gotsampleunit = 1; |
| 158 | GetHpTok(infp); |
| 159 | break; |
| 160 | |
| 161 | case VALUE_UNIT_TOK: |
| 162 | GetHpTok(infp); |
| 163 | if (thetok != STRING_TOK) { |
| 164 | Error("%s, line %d: string must follow VALUE_UNIT", hpfile, |
| 165 | linenum); |
| 166 | } |
| 167 | valueunitstring = thestring; |
| 168 | gotvalueunit = 1; |
| 169 | GetHpTok(infp); |
| 170 | break; |
| 171 | |
| 172 | case MARK_TOK: |
| 173 | GetHpTok(infp); |
| 174 | if (thetok != FLOAT_TOK) { |
| 175 | Error("%s, line %d, floating point number must follow MARK", |
| 176 | hpfile, linenum); |
| 177 | } |
| 178 | if (insample) { |
| 179 | Error("%s, line %d, MARK occurs within sample", hpfile, linenum); |
| 180 | } |
| 181 | if (nmarks >= nmarkmax) { |
| 182 | if (!markmap) { |
| 183 | nmarkmax = N_MARKS; |
| 184 | markmap = (floatish*) xmalloc(nmarkmax * sizeof(floatish)); |
| 185 | } else { |
| 186 | nmarkmax *= 2; |
| 187 | markmap = (floatish*) xrealloc(markmap, nmarkmax * sizeof(floatish)); |
| 188 | } |
| 189 | } |
| 190 | markmap[ nmarks++ ] = thefloatish; |
| 191 | GetHpTok(infp); |
| 192 | break; |
| 193 | |
| 194 | case BEGIN_SAMPLE_TOK: |
| 195 | insample = 1; |
| 196 | GetHpTok(infp); |
| 197 | if (thetok != FLOAT_TOK) { |
| 198 | Error("%s, line %d, floating point number must follow BEGIN_SAMPLE", hpfile, linenum); |
| 199 | } |
| 200 | if (thefloatish < lastsample) { |
| 201 | Error("%s, line %d, samples out of sequence", hpfile, linenum); |
| 202 | } else { |
| 203 | lastsample = thefloatish; |
| 204 | } |
| 205 | if (nsamples >= nsamplemax) { |
| 206 | if (!samplemap) { |
| 207 | nsamplemax = N_SAMPLES; |
| 208 | samplemap = (floatish*) xmalloc(nsamplemax * sizeof(floatish)); |
| 209 | } else { |
| 210 | nsamplemax *= 2; |
| 211 | samplemap = (floatish*) xrealloc(samplemap, |
| 212 | nsamplemax * sizeof(floatish)); |
| 213 | } |
| 214 | } |
| 215 | samplemap[ nsamples ] = thefloatish; |
| 216 | GetHpTok(infp); |
| 217 | break; |
| 218 | |
| 219 | case END_SAMPLE_TOK: |
| 220 | insample = 0; |
| 221 | GetHpTok(infp); |
| 222 | if (thetok != FLOAT_TOK) { |
| 223 | Error("%s, line %d: floating point number must follow END_SAMPLE", |
| 224 | hpfile, linenum); |
| 225 | } |
| 226 | nsamples++; |
| 227 | GetHpTok(infp); |
| 228 | break; |
| 229 | |
| 230 | case IDENTIFIER_TOK: |
| 231 | GetHpTok(infp); |
| 232 | if (thetok != INTEGER_TOK) { |
| 233 | Error("%s, line %d: integer must follow identifier", hpfile, |
| 234 | linenum); |
| 235 | } |
| 236 | StoreSample(GetEntry(theident), nsamples, (floatish) theinteger); |
| 237 | GetHpTok(infp); |
| 238 | break; |
| 239 | |
| 240 | case EOF_TOK: |
| 241 | endfile = 1; |
| 242 | break; |
| 243 | |
| 244 | default: |
| 245 | Error("%s, line %d: %s unexpected", hpfile, linenum, |
| 246 | TokenToString(thetok)); |
| 247 | break; |
| 248 | } |
| 249 | } |
| 250 | |
| 251 | |
| 252 | char * |
| 253 | TokenToString(t) |
| 254 | token t; |
| 255 | { |
| 256 | switch (t) { |
| 257 | case EOF_TOK: return "EOF"; |
| 258 | case INTEGER_TOK: return "integer"; |
| 259 | case FLOAT_TOK: return "floating point number"; |
| 260 | case IDENTIFIER_TOK: return "identifier"; |
| 261 | case STRING_TOK: return "string"; |
| 262 | case BEGIN_SAMPLE_TOK: return "BEGIN_SAMPLE"; |
| 263 | case END_SAMPLE_TOK: return "END_SAMPLE"; |
| 264 | case JOB_TOK: return "JOB"; |
| 265 | case DATE_TOK: return "DATE"; |
| 266 | case SAMPLE_UNIT_TOK: return "SAMPLE_UNIT"; |
| 267 | case VALUE_UNIT_TOK: return "VALUE_UNIT"; |
| 268 | case MARK_TOK: return "MARK"; |
| 269 | |
| 270 | case X_RANGE_TOK: return "X_RANGE"; |
| 271 | case Y_RANGE_TOK: return "Y_RANGE"; |
| 272 | case ORDER_TOK: return "ORDER"; |
| 273 | case SHADE_TOK: return "SHADE"; |
| 274 | default: return "(strange token)"; |
| 275 | } |
| 276 | } |
| 277 | |
| 278 | /* |
| 279 | * Read the next token from the input and assign its value |
| 280 | * to the global variable "thetok". In the case of numbers, |
| 281 | * the corresponding value is also assigned to "theinteger" |
| 282 | * or "thefloatish" as appropriate; in the case of identifiers |
| 283 | * it is assigned to "theident". |
| 284 | */ |
| 285 | |
| 286 | static void |
| 287 | GetHpTok(infp) |
| 288 | FILE* infp; |
| 289 | { |
| 290 | |
| 291 | while (isspace(g_ch)) { /* skip whitespace */ |
| 292 | if (g_ch == '\n') linenum++; |
| 293 | g_ch = getc(infp); |
| 294 | } |
| 295 | |
| 296 | if (g_ch == EOF) { |
| 297 | thetok = EOF_TOK; |
| 298 | return; |
| 299 | } |
| 300 | |
| 301 | if (isdigit(g_ch)) { |
| 302 | thetok = GetNumber(infp); |
| 303 | return; |
| 304 | } else if (g_ch == '\"') { |
| 305 | GetString(infp); |
| 306 | thetok = STRING_TOK; |
| 307 | return; |
| 308 | } else if (IsIdChar(g_ch)) { |
| 309 | ASSERT(! (isdigit(g_ch))); /* g_ch can't be a digit here */ |
| 310 | GetIdent(infp); |
| 311 | if (!isupper(theident[0])) { |
| 312 | thetok = IDENTIFIER_TOK; |
| 313 | } else if (strcmp(theident, "BEGIN_SAMPLE") == 0) { |
| 314 | thetok = BEGIN_SAMPLE_TOK; |
| 315 | } else if (strcmp(theident, "END_SAMPLE") == 0) { |
| 316 | thetok = END_SAMPLE_TOK; |
| 317 | } else if (strcmp(theident, "JOB") == 0) { |
| 318 | thetok = JOB_TOK; |
| 319 | } else if (strcmp(theident, "DATE") == 0) { |
| 320 | thetok = DATE_TOK; |
| 321 | } else if (strcmp(theident, "SAMPLE_UNIT") == 0) { |
| 322 | thetok = SAMPLE_UNIT_TOK; |
| 323 | } else if (strcmp(theident, "VALUE_UNIT") == 0) { |
| 324 | thetok = VALUE_UNIT_TOK; |
| 325 | } else if (strcmp(theident, "MARK") == 0) { |
| 326 | thetok = MARK_TOK; |
| 327 | } else { |
| 328 | thetok = IDENTIFIER_TOK; |
| 329 | } |
| 330 | return; |
| 331 | } else { |
| 332 | Error("%s, line %d: strange character (%c)", hpfile, linenum, g_ch); |
| 333 | } |
| 334 | } |
| 335 | |
| 336 | |
| 337 | /* |
| 338 | * Read a sequence of digits and convert the result to an integer |
| 339 | * or floating point value (assigned to the "theinteger" or |
| 340 | * "thefloatish"). |
| 341 | */ |
| 342 | |
njn | a288784 | 2005-05-25 21:26:35 +0000 | [diff] [blame] | 343 | static char numberstring[ NUMBER_LENGTH ]; |
nethercote | c9f3692 | 2004-02-14 16:40:02 +0000 | [diff] [blame] | 344 | |
| 345 | token |
| 346 | GetNumber(infp) |
| 347 | FILE* infp; |
| 348 | { |
| 349 | int i; |
| 350 | int containsdot; |
| 351 | |
| 352 | ASSERT(isdigit(ch)); /* we must have a digit to start with */ |
| 353 | |
| 354 | containsdot = 0; |
| 355 | |
| 356 | for (i = 0; i < NUMBER_LENGTH && (isdigit(g_ch) || g_ch == '.'); i++) { |
| 357 | numberstring[ i ] = g_ch; |
| 358 | containsdot |= (g_ch == '.'); |
| 359 | g_ch = getc(infp); |
| 360 | } |
| 361 | |
| 362 | ASSERT(i < NUMBER_LENGTH); /* did not overflow */ |
| 363 | |
| 364 | numberstring[ i ] = '\0'; |
| 365 | |
| 366 | if (containsdot) { |
| 367 | thefloatish = (floatish) atof(numberstring); |
| 368 | return FLOAT_TOK; |
| 369 | } else { |
| 370 | theinteger = atoi(numberstring); |
| 371 | return INTEGER_TOK; |
| 372 | } |
| 373 | } |
| 374 | |
| 375 | /* |
| 376 | * Read a sequence of identifier characters and assign the result |
| 377 | * to the string "theident". |
| 378 | */ |
| 379 | |
| 380 | void |
| 381 | GetIdent(infp) |
| 382 | FILE *infp; |
| 383 | { |
| 384 | unsigned int i; |
| 385 | char idbuffer[5000]; |
| 386 | |
| 387 | for (i = 0; i < (sizeof idbuffer)-1 && IsIdChar(g_ch); i++) { |
| 388 | idbuffer[ i ] = g_ch; |
| 389 | g_ch = getc(infp); |
| 390 | } |
| 391 | |
| 392 | idbuffer[ i ] = '\0'; |
| 393 | |
| 394 | if (theident) |
| 395 | free(theident); |
| 396 | |
| 397 | theident = copystring(idbuffer); |
| 398 | } |
| 399 | |
| 400 | |
| 401 | /* |
| 402 | * Read a sequence of characters that make up a string and |
| 403 | * assign the result to "thestring". |
| 404 | */ |
| 405 | |
| 406 | void |
| 407 | GetString(infp) |
| 408 | FILE *infp; |
| 409 | { |
| 410 | unsigned int i; |
| 411 | char stringbuffer[5000]; |
| 412 | |
| 413 | ASSERT(ch == '\"'); |
| 414 | |
| 415 | g_ch = getc(infp); /* skip the '\"' that begins the string */ |
| 416 | |
| 417 | for (i = 0; i < (sizeof stringbuffer)-1 && g_ch != '\"'; i++) { |
| 418 | stringbuffer[ i ] = g_ch; |
| 419 | g_ch = getc(infp); |
| 420 | } |
| 421 | |
| 422 | stringbuffer[i] = '\0'; |
| 423 | thestring = copystring(stringbuffer); |
| 424 | |
| 425 | ASSERT(g_ch == '\"'); |
| 426 | |
| 427 | g_ch = getc(infp); /* skip the '\"' that terminates the string */ |
| 428 | } |
| 429 | |
| 430 | boolish |
| 431 | IsIdChar(ch) |
| 432 | int ch; |
| 433 | { |
| 434 | return (!isspace(ch)); |
| 435 | } |
| 436 | |
| 437 | |
| 438 | /* |
| 439 | * The information associated with each identifier is stored |
| 440 | * in a linked list of chunks. The table below allows the list |
| 441 | * of chunks to be retrieved given an identifier name. |
| 442 | */ |
| 443 | |
| 444 | #define N_HASH 513 |
| 445 | |
| 446 | static struct entry* hashtable[ N_HASH ]; |
| 447 | |
| 448 | static intish |
| 449 | Hash(s) |
| 450 | char *s; |
| 451 | { |
| 452 | int r; |
| 453 | |
| 454 | for (r = 0; *s; s++) { |
| 455 | r = r + r + r + *s; |
| 456 | } |
| 457 | |
| 458 | if (r < 0) r = -r; |
| 459 | |
| 460 | return r % N_HASH; |
| 461 | } |
| 462 | |
| 463 | /* |
| 464 | * Get space for a new chunk. Initialise it, and return a pointer |
| 465 | * to the new chunk. |
| 466 | */ |
| 467 | |
| 468 | static struct chunk* |
| 469 | MakeChunk() |
| 470 | { |
| 471 | struct chunk* ch; |
| 472 | struct datapoint* d; |
| 473 | |
| 474 | ch = (struct chunk*) xmalloc( sizeof(struct chunk) ); |
| 475 | |
| 476 | d = (struct datapoint*) xmalloc (sizeof(struct datapoint) * N_CHUNK); |
| 477 | |
| 478 | ch->nd = 0; |
| 479 | ch->d = d; |
| 480 | ch->next = 0; |
| 481 | return ch; |
| 482 | } |
| 483 | |
| 484 | |
| 485 | /* |
| 486 | * Get space for a new entry. Initialise it, and return a pointer |
| 487 | * to the new entry. |
| 488 | */ |
| 489 | |
| 490 | struct entry * |
| 491 | MakeEntry(name) |
| 492 | char *name; |
| 493 | { |
| 494 | struct entry* e; |
| 495 | |
| 496 | e = (struct entry *) xmalloc(sizeof(struct entry)); |
| 497 | e->chk = MakeChunk(); |
| 498 | e->name = copystring(name); |
| 499 | return e; |
| 500 | } |
| 501 | |
| 502 | /* |
| 503 | * Get the entry associated with "name", creating a new entry if |
| 504 | * necessary. |
| 505 | */ |
| 506 | |
| 507 | static struct entry * |
| 508 | GetEntry(name) |
| 509 | char* name; |
| 510 | { |
| 511 | intish h; |
| 512 | struct entry* e; |
| 513 | |
| 514 | h = Hash(name); |
| 515 | |
| 516 | for (e = hashtable[ h ]; e; e = e->next) { |
| 517 | if (strcmp(e->name, name) == 0) { |
| 518 | break; |
| 519 | } |
| 520 | } |
| 521 | |
| 522 | if (e) { |
| 523 | return (e); |
| 524 | } else { |
| 525 | nidents++; |
| 526 | e = MakeEntry(name); |
| 527 | e->next = hashtable[ h ]; |
| 528 | hashtable[ h ] = e; |
| 529 | return (e); |
| 530 | } |
| 531 | } |
| 532 | |
| 533 | |
| 534 | /* |
| 535 | * Store information from a sample. |
| 536 | */ |
| 537 | |
| 538 | void |
| 539 | StoreSample(en, bucket, value) |
| 540 | struct entry* en; intish bucket; floatish value; |
| 541 | { |
| 542 | struct chunk* chk; |
| 543 | |
| 544 | for (chk = en->chk; chk->next != 0; chk = chk->next) |
| 545 | ; |
| 546 | |
| 547 | if (chk->nd < N_CHUNK) { |
| 548 | chk->d[ chk->nd ].bucket = bucket; |
| 549 | chk->d[ chk->nd ].value = value; |
| 550 | chk->nd += 1; |
| 551 | } else { |
| 552 | struct chunk* t; |
| 553 | t = chk->next = MakeChunk(); |
| 554 | t->d[ 0 ].bucket = bucket; |
| 555 | t->d[ 0 ].value = value; |
| 556 | t->nd += 1; |
| 557 | } |
| 558 | } |
| 559 | |
| 560 | |
| 561 | struct entry** identtable; |
| 562 | |
| 563 | /* |
| 564 | * The hash table is useful while reading the input, but it |
| 565 | * becomes a liability thereafter. The code below converts |
| 566 | * it to a more easily processed table. |
| 567 | */ |
| 568 | |
| 569 | static void |
| 570 | MakeIdentTable() |
| 571 | { |
| 572 | intish i; |
| 573 | intish j; |
| 574 | struct entry* e; |
| 575 | |
| 576 | nidents = 0; |
| 577 | for (i = 0; i < N_HASH; i++) { |
| 578 | for (e = hashtable[ i ]; e; e = e->next) { |
| 579 | nidents++; |
| 580 | } |
| 581 | } |
| 582 | |
| 583 | identtable = (struct entry**) xmalloc(nidents * sizeof(struct entry*)); |
| 584 | j = 0; |
| 585 | |
| 586 | for (i = 0; i < N_HASH; i++) { |
| 587 | for (e = hashtable[ i ]; e; e = e->next, j++) { |
| 588 | identtable[ j ] = e; |
| 589 | } |
| 590 | } |
| 591 | } |