gnu/dist/groff/src/preproc/refer/refer.cpp

   1 /*      $NetBSD$        */
   2
   3 // -*- C++ -*-
   4 /* Copyright (C) 1989-1992, 2000, 2001, 2002, 2004
   5    Free Software Foundation, Inc.
   6      Written by James Clark (jjc@jclark.com)
   7
   8 This file is part of groff.
   9
  10 groff is free software; you can redistribute it and/or modify it under
  11 the terms of the GNU General Public License as published by the Free
  12 Software Foundation; either version 2, or (at your option) any later
  13 version.
  14
  15 groff is distributed in the hope that it will be useful, but WITHOUT ANY
  16 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  17 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  18 for more details.
  19
  20 You should have received a copy of the GNU General Public License along
  21 with groff; see the file COPYING.  If not, write to the Free Software
  22 Foundation, 51 Franklin St - Fifth Floor, Boston, MA 02110-1301, USA. */
  23
  24 #include "refer.h"
  25 #include "refid.h"
  26 #include "ref.h"
  27 #include "token.h"
  28 #include "search.h"
  29 #include "command.h"
  30
  31 extern "C" const char *Version_string;
  32
  33 const char PRE_LABEL_MARKER = '\013';
  34 const char POST_LABEL_MARKER = '\014';
  35 const char LABEL_MARKER = '\015'; // label_type is added on
  36
  37 #define FORCE_LEFT_BRACKET 04
  38 #define FORCE_RIGHT_BRACKET 010
  39
  40 static FILE *outfp = stdout;
  41
  42 string capitalize_fields;
  43 string reverse_fields;
  44 string abbreviate_fields;
  45 string period_before_last_name = ". ";
  46 string period_before_initial = ".";
  47 string period_before_hyphen = "";
  48 string period_before_other = ". ";
  49 string sort_fields;
  50 int annotation_field = -1;
  51 string annotation_macro;
  52 string discard_fields = "XYZ";
  53 string pre_label = "\\*([.";
  54 string post_label = "\\*(.]";
  55 string sep_label = ", ";
  56 int accumulate = 0;
  57 int move_punctuation = 0;
  58 int abbreviate_label_ranges = 0;
  59 string label_range_indicator;
  60 int label_in_text = 1;
  61 int label_in_reference = 1;
  62 int date_as_label = 0;
  63 int sort_adjacent_labels = 0;
  64 // Join exactly two authors with this.
  65 string join_authors_exactly_two = " and ";
  66 // When there are more than two authors join the last two with this.
  67 string join_authors_last_two = ", and ";
  68 // Otherwise join authors with this.
  69 string join_authors_default = ", ";
  70 string separate_label_second_parts = ", ";
  71 // Use this string to represent that there are other authors.
  72 string et_al = " et al";
  73 // Use et al only if it can replace at least this many authors.
  74 int et_al_min_elide = 2;
  75 // Use et al only if the total number of authors is at least this.
  76 int et_al_min_total = 3;
  77
  78
  79 int compatible_flag = 0;
  80
  81 int short_label_flag = 0;
  82
  83 static int recognize_R1_R2 = 1;
  84
  85 search_list database_list;
  86 int search_default = 1;
  87 static int default_database_loaded = 0;
  88
  89 static reference **citation = 0;
  90 static int ncitations = 0;
  91 static int citation_max = 0;
  92
  93 static reference **reference_hash_table = 0;
  94 static int hash_table_size;
  95 static int nreferences = 0;
  96
  97 static int need_syncing = 0;
  98 string pending_line;
  99 string pending_lf_lines;
 100
 101 static void output_pending_line();
 102 static unsigned immediately_handle_reference(const string &);
 103 static void immediately_output_references();
 104 static unsigned store_reference(const string &);
 105 static void divert_to_temporary_file();
 106 static reference *make_reference(const string &, unsigned *);
 107 static void usage(FILE *stream);
 108 static void do_file(const char *);
 109 static void split_punct(string &line, string &punct);
 110 static void output_citation_group(reference **v, int n, label_type, FILE *fp);
 111 static void possibly_load_default_database();
 112
 113 int main(int argc, char **argv)
 114 {
 115   program_name = argv[0];
 116   static char stderr_buf[BUFSIZ];
 117   setbuf(stderr, stderr_buf);
 118   outfp = stdout;
 119   int finished_options = 0;
 120   int bib_flag = 0;
 121   int done_spec = 0;
 122
 123   for (--argc, ++argv;
 124        !finished_options && argc > 0 && argv[0][0] == '-'
 125        && argv[0][1] != '\0';
 126        argv++, argc--) {
 127     const char *opt = argv[0] + 1;
 128     while (opt != 0 && *opt != '\0') {
 129       switch (*opt) {
 130       case 'C':
 131         compatible_flag = 1;
 132         opt++;
 133         break;
 134       case 'B':
 135         bib_flag = 1;
 136         label_in_reference = 0;
 137         label_in_text = 0;
 138         ++opt;
 139         if (*opt == '\0') {
 140           annotation_field = 'X';
 141           annotation_macro = "AP";
 142         }
 143         else if (csalnum(opt[0]) && opt[1] == '.' && opt[2] != '\0') {
 144           annotation_field = opt[0];
 145           annotation_macro = opt + 2;
 146         }
 147         opt = 0;
 148         break;
 149       case 'P':
 150         move_punctuation = 1;
 151         opt++;
 152         break;
 153       case 'R':
 154         recognize_R1_R2 = 0;
 155         opt++;
 156         break;
 157       case 'S':
 158         // Not a very useful spec.
 159         set_label_spec("(A.n|Q)', '(D.y|D)");
 160         done_spec = 1;
 161         pre_label = " (";
 162         post_label = ")";
 163         sep_label = "; ";
 164         opt++;
 165         break;
 166       case 'V':
 167         verify_flag = 1;
 168         opt++;
 169         break;
 170       case 'f':
 171         {
 172           const char *num = 0;
 173           if (*++opt == '\0') {
 174             if (argc > 1) {
 175               num = *++argv;
 176               --argc;
 177             }
 178             else {
 179               error("option `f' requires an argument");
 180               usage(stderr);
 181               exit(1);
 182             }
 183           }
 184           else {
 185             num = opt;
 186             opt = 0;
 187           }
 188           const char *ptr;
 189           for (ptr = num; *ptr; ptr++)
 190             if (!csdigit(*ptr)) {
 191               error("bad character `%1' in argument to -f option", *ptr);
 192               break;
 193             }
 194           if (*ptr == '\0') {
 195             string spec;
 196             spec = '%';
 197             spec += num;
 198             spec += '\0';
 199             set_label_spec(spec.contents());
 200             done_spec = 1;
 201           }
 202           break;
 203         }
 204       case 'b':
 205         label_in_text = 0;
 206         label_in_reference = 0;
 207         opt++;
 208         break;
 209       case 'e':
 210         accumulate = 1;
 211         opt++;
 212         break;
 213       case 'c':
 214         capitalize_fields = ++opt;
 215         opt = 0;
 216         break;
 217       case 'k':
 218         {
 219           char buf[5];
 220           if (csalpha(*++opt))
 221             buf[0] = *opt++;
 222           else {
 223             if (*opt != '\0')
 224               error("bad field name `%1'", *opt++);
 225             buf[0] = 'L';
 226           }
 227           buf[1] = '~';
 228           buf[2] = '%';
 229           buf[3] = 'a';
 230           buf[4] = '\0';
 231           set_label_spec(buf);
 232           done_spec = 1;
 233         }
 234         break;
 235       case 'a':
 236         {
 237           const char *ptr;
 238           for (ptr = ++opt; *ptr; ptr++)
 239             if (!csdigit(*ptr)) {
 240               error("argument to `a' option not a number");
 241               break;
 242             }
 243           if (*ptr == '\0') {
 244             reverse_fields = 'A';
 245             reverse_fields += opt;
 246           }
 247           opt = 0;
 248         }
 249         break;
 250       case 'i':
 251         linear_ignore_fields = ++opt;
 252         opt = 0;
 253         break;
 254       case 'l':
 255         {
 256           char buf[INT_DIGITS*2 + 11]; // A.n+2D.y-3%a
 257           strcpy(buf, "A.n");
 258           if (*++opt != '\0' && *opt != ',') {
 259             char *ptr;
 260             long n = strtol(opt, &ptr, 10);
 261             if (n == 0 && ptr == opt) {
 262               error("bad integer `%1' in `l' option", opt);
 263               opt = 0;
 264               break;
 265             }
 266             if (n < 0)
 267               n = 0;
 268             opt = ptr;
 269             sprintf(strchr(buf, '\0'), "+%ld", n);
 270           }
 271           strcat(buf, "D.y");
 272           if (*opt == ',')
 273             opt++;
 274           if (*opt != '\0') {
 275             char *ptr;
 276             long n = strtol(opt, &ptr, 10);
 277             if (n == 0 && ptr == opt) {
 278               error("bad integer `%1' in `l' option", opt);
 279               opt = 0;
 280               break;
 281             }
 282             if (n < 0)
 283               n = 0;
 284             sprintf(strchr(buf, '\0'), "-%ld", n);
 285             opt = ptr;
 286             if (*opt != '\0')
 287               error("argument to `l' option not of form `m,n'");
 288           }
 289           strcat(buf, "%a");
 290           if (!set_label_spec(buf))
 291             assert(0);
 292           done_spec = 1;
 293         }
 294         break;
 295       case 'n':
 296         search_default = 0;
 297         opt++;
 298         break;
 299       case 'p':
 300         {
 301           const char *filename = 0;
 302           if (*++opt == '\0') {
 303             if (argc > 1) {
 304               filename = *++argv;
 305               argc--;
 306             }
 307             else {
 308               error("option `p' requires an argument");
 309               usage(stderr);
 310               exit(1);
 311             }
 312           }
 313           else {
 314             filename = opt;
 315             opt = 0;
 316           }
 317           database_list.add_file(filename);
 318         }
 319         break;
 320       case 's':
 321         if (*++opt == '\0')
 322           sort_fields = "AD";
 323         else {
 324           sort_fields = opt;
 325           opt = 0;
 326         }
 327         accumulate = 1;
 328         break;
 329       case 't':
 330         {
 331           char *ptr;
 332           long n = strtol(opt, &ptr, 10);
 333           if (n == 0 && ptr == opt) {
 334             error("bad integer `%1' in `t' option", opt);
 335             opt = 0;
 336             break;
 337           }
 338           if (n < 1)
 339             n = 1;
 340           linear_truncate_len = int(n);
 341           opt = ptr;
 342           break;
 343         }
 344       case '-':
 345         if (opt[1] == '\0') {
 346           finished_options = 1;
 347           opt++;
 348           break;
 349         }
 350         if (strcmp(opt,"-version")==0) {
 351       case 'v':
 352           printf("GNU refer (groff) version %s\n", Version_string);
 353           exit(0);
 354           break;
 355         }
 356         if (strcmp(opt,"-help")==0) {
 357           usage(stdout);
 358           exit(0);
 359           break;
 360         }
 361         // fall through
 362       default:
 363         error("unrecognized option `%1'", *opt);
 364         usage(stderr);
 365         exit(1);
 366         break;
 367       }
 368     }
 369   }
 370   if (!done_spec)
 371     set_label_spec("%1");
 372   if (argc <= 0) {
 373     if (bib_flag)
 374       do_bib("-");
 375     else
 376       do_file("-");
 377   }
 378   else {
 379     for (int i = 0; i < argc; i++) {
 380       if (bib_flag)
 381         do_bib(argv[i]);
 382       else
 383         do_file(argv[i]);
 384     }
 385   }
 386   if (accumulate)
 387     output_references();
 388   if (fflush(stdout) < 0)
 389     fatal("output error");
 390   return 0;
 391 }
 392
 393 static void usage(FILE *stream)
 394 {
 395   fprintf(stream,
 396 "usage: %s [-benvCPRS] [-aN] [-cXYZ] [-fN] [-iXYZ] [-kX] [-lM,N] [-p file]\n"
 397 "       [-sXYZ] [-tN] [-BL.M] [files ...]\n",
 398           program_name);
 399 }
 400
 401 static void possibly_load_default_database()
 402 {
 403   if (search_default && !default_database_loaded) {
 404     char *filename = getenv("REFER");
 405     if (filename)
 406       database_list.add_file(filename);
 407     else
 408       database_list.add_file(DEFAULT_INDEX, 1);
 409     default_database_loaded = 1;
 410   }
 411 }
 412
 413 static int is_list(const string &str)
 414 {
 415   const char *start = str.contents();
 416   const char *end = start + str.length();
 417   while (end > start && csspace(end[-1]))
 418     end--;
 419   while (start < end && csspace(*start))
 420     start++;
 421   return end - start == 6 && memcmp(start, "$LIST$", 6) == 0;
 422 }
 423
 424 static void do_file(const char *filename)
 425 {
 426   FILE *fp;
 427   if (strcmp(filename, "-") == 0) {
 428     fp = stdin;
 429   }
 430   else {
 431     errno = 0;
 432     fp = fopen(filename, "r");
 433     if (fp == 0) {
 434       error("can't open `%1': %2", filename, strerror(errno));
 435       return;
 436     }
 437   }
 438   current_filename = filename;
 439   fprintf(outfp, ".lf 1 %s\n", filename);
 440   string line;
 441   current_lineno = 0;
 442   for (;;) {
 443     line.clear();
 444     for (;;) {
 445       int c = getc(fp);
 446       if (c == EOF) {
 447         if (line.length() > 0)
 448           line += '\n';
 449         break;
 450       }
 451       if (invalid_input_char(c))
 452         error("invalid input character code %1", c);
 453       else {
 454         line += c;
 455         if (c == '\n')
 456           break;
 457       }
 458     }
 459     int len = line.length();
 460     if (len == 0)
 461       break;
 462     current_lineno++;
 463     if (len >= 2 && line[0] == '.' && line[1] == '[') {
 464       int start_lineno = current_lineno;
 465       int start_of_line = 1;
 466       string str;
 467       string post;
 468       string pre(line.contents() + 2, line.length() - 3);
 469       for (;;) {
 470         int c = getc(fp);
 471         if (c == EOF) {
 472           error_with_file_and_line(current_filename, start_lineno,
 473                                    "missing `.]' line");
 474           break;
 475         }
 476         if (start_of_line)
 477           current_lineno++;
 478         if (start_of_line && c == '.') {
 479           int d = getc(fp);
 480           if (d == ']') {
 481             while ((d = getc(fp)) != '\n' && d != EOF) {
 482               if (invalid_input_char(d))
 483                 error("invalid input character code %1", d);
 484               else
 485                 post += d;
 486             }
 487             break;
 488           }
 489           if (d != EOF)
 490             ungetc(d, fp);
 491         }
 492         if (invalid_input_char(c))
 493           error("invalid input character code %1", c);
 494         else
 495           str += c;
 496         start_of_line = (c == '\n');
 497       }
 498       if (is_list(str)) {
 499         output_pending_line();
 500         if (accumulate)
 501           output_references();
 502         else
 503           error("found `$LIST$' but not accumulating references");
 504       }
 505       else {
 506         unsigned flags = (accumulate
 507                           ? store_reference(str)
 508                           : immediately_handle_reference(str));
 509         if (label_in_text) {
 510           if (accumulate && outfp == stdout)
 511             divert_to_temporary_file();
 512           if (pending_line.length() == 0) {
 513             warning("can't attach citation to previous line");
 514           }
 515           else
 516             pending_line.set_length(pending_line.length() - 1);
 517           string punct;
 518           if (move_punctuation)
 519             split_punct(pending_line, punct);
 520           int have_text = pre.length() > 0 || post.length() > 0;
 521           label_type lt = label_type(flags & ~(FORCE_LEFT_BRACKET
 522                                                |FORCE_RIGHT_BRACKET));
 523           if ((flags & FORCE_LEFT_BRACKET) || !have_text)
 524             pending_line += PRE_LABEL_MARKER;
 525           pending_line += pre;
 526           char lm = LABEL_MARKER + (int)lt;
 527           pending_line += lm;
 528           pending_line += post;
 529           if ((flags & FORCE_RIGHT_BRACKET) || !have_text)
 530             pending_line += POST_LABEL_MARKER;
 531           pending_line += punct;
 532           pending_line += '\n';
 533         }
 534       }
 535       need_syncing = 1;
 536     }
 537     else if (len >= 4
 538              && line[0] == '.' && line[1] == 'l' && line[2] == 'f'
 539              && (compatible_flag || line[3] == '\n' || line[3] == ' ')) {
 540       pending_lf_lines += line;
 541       line += '\0';
 542       if (interpret_lf_args(line.contents() + 3))
 543         current_lineno--;
 544     }
 545     else if (recognize_R1_R2
 546              && len >= 4
 547              && line[0] == '.' && line[1] == 'R' && line[2] == '1'
 548              && (compatible_flag || line[3] == '\n' || line[3] == ' ')) {
 549       line.clear();
 550       int start_of_line = 1;
 551       int start_lineno = current_lineno;
 552       for (;;) {
 553         int c = getc(fp);
 554         if (c != EOF && start_of_line)
 555           current_lineno++;
 556         if (start_of_line && c == '.') {
 557           c = getc(fp);
 558           if (c == 'R') {
 559             c = getc(fp);
 560             if (c == '2') {
 561               c = getc(fp);
 562               if (compatible_flag || c == ' ' || c == '\n' || c == EOF) {
 563                 while (c != EOF && c != '\n')
 564                   c = getc(fp);
 565                 break;
 566               }
 567               else {
 568                 line += '.';
 569                 line += 'R';
 570                 line += '2';
 571               }
 572             }
 573             else {
 574               line += '.';
 575               line += 'R';
 576             }
 577           }
 578           else
 579             line += '.';
 580         }
 581         if (c == EOF) {
 582           error_with_file_and_line(current_filename, start_lineno,
 583                                    "missing `.R2' line");
 584           break;
 585         }
 586         if (invalid_input_char(c))
 587           error("invalid input character code %1", int(c));
 588         else {
 589           line += c;
 590           start_of_line = c == '\n';
 591         }
 592       }
 593       output_pending_line();
 594       if (accumulate)
 595         output_references();
 596       else
 597         nreferences = 0;
 598       process_commands(line, current_filename, start_lineno + 1);
 599       need_syncing = 1;
 600     }
 601     else {
 602       output_pending_line();
 603       pending_line = line;
 604     }
 605   }
 606   need_syncing = 0;
 607   output_pending_line();
 608   if (fp != stdin)
 609     fclose(fp);
 610 }
 611
 612 class label_processing_state {
 613   enum {
 614     NORMAL,
 615     PENDING_LABEL,
 616     PENDING_LABEL_POST,
 617     PENDING_LABEL_POST_PRE,
 618     PENDING_POST
 619     } state;
 620   label_type type;              // type of pending labels
 621   int count;                    // number of pending labels
 622   reference **rptr;             // pointer to next reference
 623   int rcount;                   // number of references left
 624   FILE *fp;
 625   int handle_pending(int c);
 626 public:
 627   label_processing_state(reference **, int, FILE *);
 628   ~label_processing_state();
 629   void process(int c);
 630 };
 631
 632 static void output_pending_line()
 633 {
 634   if (label_in_text && !accumulate && ncitations > 0) {
 635     label_processing_state state(citation, ncitations, outfp);
 636     int len = pending_line.length();
 637     for (int i = 0; i < len; i++)
 638       state.process((unsigned char)(pending_line[i]));
 639   }
 640   else
 641     put_string(pending_line, outfp);
 642   pending_line.clear();
 643   if (pending_lf_lines.length() > 0) {
 644     put_string(pending_lf_lines, outfp);
 645     pending_lf_lines.clear();
 646   }
 647   if (!accumulate)
 648     immediately_output_references();
 649   if (need_syncing) {
 650     fprintf(outfp, ".lf %d %s\n", current_lineno, current_filename);
 651     need_syncing = 0;
 652   }
 653 }
 654
 655 static void split_punct(string &line, string &punct)
 656 {
 657   const char *start = line.contents();
 658   const char *end = start + line.length();
 659   const char *ptr = start;
 660   const char *last_token_start = 0;
 661   for (;;) {
 662     if (ptr >= end)
 663       break;
 664     last_token_start = ptr;
 665     if (*ptr == PRE_LABEL_MARKER || *ptr == POST_LABEL_MARKER
 666         || (*ptr >= LABEL_MARKER && *ptr < LABEL_MARKER + N_LABEL_TYPES))
 667       ptr++;
 668     else if (!get_token(&ptr, end))
 669       break;
 670   }
 671   if (last_token_start) {
 672     const token_info *ti = lookup_token(last_token_start, end);
 673     if (ti->is_punct()) {
 674       punct.append(last_token_start, end - last_token_start);
 675       line.set_length(last_token_start - start);
 676     }
 677   }
 678 }
 679
 680 static void divert_to_temporary_file()
 681 {
 682   outfp = xtmpfile();
 683 }
 684
 685 static void store_citation(reference *ref)
 686 {
 687   if (ncitations >= citation_max) {
 688     if (citation == 0)
 689       citation = new reference*[citation_max = 100];
 690     else {
 691       reference **old_citation = citation;
 692       citation_max *= 2;
 693       citation = new reference *[citation_max];
 694       memcpy(citation, old_citation, ncitations*sizeof(reference *));
 695       a_delete old_citation;
 696     }
 697   }
 698   citation[ncitations++] = ref;
 699 }
 700
 701 static unsigned store_reference(const string &str)
 702 {
 703   if (reference_hash_table == 0) {
 704     reference_hash_table = new reference *[17];
 705     hash_table_size = 17;
 706     for (int i = 0; i < hash_table_size; i++)
 707       reference_hash_table[i] = 0;
 708   }
 709   unsigned flags;
 710   reference *ref = make_reference(str, &flags);
 711   ref->compute_hash_code();
 712   unsigned h = ref->hash();
 713   reference **ptr;
 714   for (ptr = reference_hash_table + (h % hash_table_size);
 715        *ptr != 0;
 716        ((ptr == reference_hash_table)
 717         ? (ptr = reference_hash_table + hash_table_size - 1)
 718         : --ptr))
 719     if (same_reference(**ptr, *ref))
 720       break;
 721   if (*ptr != 0) {
 722     if (ref->is_merged())
 723       warning("fields ignored because reference already used");
 724     delete ref;
 725     ref = *ptr;
 726   }
 727   else {
 728     *ptr = ref;
 729     ref->set_number(nreferences);
 730     nreferences++;
 731     ref->pre_compute_label();
 732     ref->compute_sort_key();
 733     if (nreferences*2 >= hash_table_size) {
 734       // Rehash it.
 735       reference **old_table = reference_hash_table;
 736       int old_size = hash_table_size;
 737       hash_table_size = next_size(hash_table_size);
 738       reference_hash_table = new reference*[hash_table_size];
 739       int i;
 740       for (i = 0; i < hash_table_size; i++)
 741         reference_hash_table[i] = 0;
 742       for (i = 0; i < old_size; i++)
 743         if (old_table[i]) {
 744           reference **p;
 745           for (p = (reference_hash_table
 746                                 + (old_table[i]->hash() % hash_table_size));
 747                *p;
 748                ((p == reference_hash_table)
 749                 ? (p = reference_hash_table + hash_table_size - 1)
 750                 : --p))
 751             ;
 752           *p = old_table[i];
 753         }
 754       a_delete old_table;
 755     }
 756   }
 757   if (label_in_text)
 758     store_citation(ref);
 759   return flags;
 760 }
 761
 762 unsigned immediately_handle_reference(const string &str)
 763 {
 764   unsigned flags;
 765   reference *ref = make_reference(str, &flags);
 766   ref->set_number(nreferences);
 767   if (label_in_text || label_in_reference) {
 768     ref->pre_compute_label();
 769     ref->immediate_compute_label();
 770   }
 771   nreferences++;
 772   store_citation(ref);
 773   return flags;
 774 }
 775
 776 static void immediately_output_references()
 777 {
 778   for (int i = 0; i < ncitations; i++) {
 779     reference *ref = citation[i];
 780     if (label_in_reference) {
 781       fputs(".ds [F ", outfp);
 782       const string &label = ref->get_label(NORMAL_LABEL);
 783       if (label.length() > 0
 784           && (label[0] == ' ' || label[0] == '\\' || label[0] == '"'))
 785         putc('"', outfp);
 786       put_string(label, outfp);
 787       putc('\n', outfp);
 788     }
 789     ref->output(outfp);
 790     delete ref;
 791   }
 792   ncitations = 0;
 793 }
 794
 795 static void output_citation_group(reference **v, int n, label_type type,
 796                                   FILE *fp)
 797 {
 798   if (sort_adjacent_labels) {
 799     // Do an insertion sort.  Usually n will be very small.
 800     for (int i = 1; i < n; i++) {
 801       int num = v[i]->get_number();
 802       reference *temp = v[i];
 803       int j;
 804       for (j = i - 1; j >= 0 && v[j]->get_number() > num; j--)
 805         v[j + 1] = v[j];
 806       v[j + 1] = temp;
 807     }
 808   }
 809   // This messes up if !accumulate.
 810   if (accumulate && n > 1) {
 811     // remove duplicates
 812     int j = 1;
 813     for (int i = 1; i < n; i++)
 814       if (v[i]->get_label(type) != v[i - 1]->get_label(type))
 815         v[j++] = v[i];
 816     n = j;
 817   }
 818   string merged_label;
 819   for (int i = 0; i < n; i++) {
 820     int nmerged = v[i]->merge_labels(v + i + 1, n - i - 1, type, merged_label);
 821     if (nmerged > 0) {
 822       put_string(merged_label, fp);
 823       i += nmerged;
 824     }
 825     else
 826       put_string(v[i]->get_label(type), fp);
 827     if (i < n - 1)
 828       put_string(sep_label, fp);
 829   }
 830 }
 831
 832
 833 label_processing_state::label_processing_state(reference **p, int n, FILE *f)
 834 : state(NORMAL), count(0), rptr(p), rcount(n), fp(f)
 835 {
 836 }
 837
 838 label_processing_state::~label_processing_state()
 839 {
 840   int handled = handle_pending(EOF);
 841   assert(!handled);
 842   assert(rcount == 0);
 843 }
 844
 845 int label_processing_state::handle_pending(int c)
 846 {
 847   switch (state) {
 848   case NORMAL:
 849     break;
 850   case PENDING_LABEL:
 851     if (c == POST_LABEL_MARKER) {
 852       state = PENDING_LABEL_POST;
 853       return 1;
 854     }
 855     else {
 856       output_citation_group(rptr, count, type, fp);
 857       rptr += count ;
 858       rcount -= count;
 859       state = NORMAL;
 860     }
 861     break;
 862   case PENDING_LABEL_POST:
 863     if (c == PRE_LABEL_MARKER) {
 864       state = PENDING_LABEL_POST_PRE;
 865       return 1;
 866     }
 867     else {
 868       output_citation_group(rptr, count, type, fp);
 869       rptr += count;
 870       rcount -= count;
 871       put_string(post_label, fp);
 872       state = NORMAL;
 873     }
 874     break;
 875   case PENDING_LABEL_POST_PRE:
 876     if (c >= LABEL_MARKER
 877         && c < LABEL_MARKER + N_LABEL_TYPES
 878         && c - LABEL_MARKER == type) {
 879       count += 1;
 880       state = PENDING_LABEL;
 881       return 1;
 882     }
 883     else {
 884       output_citation_group(rptr, count, type, fp);
 885       rptr += count;
 886       rcount -= count;
 887       put_string(sep_label, fp);
 888       state = NORMAL;
 889     }
 890     break;
 891   case PENDING_POST:
 892     if (c == PRE_LABEL_MARKER) {
 893       put_string(sep_label, fp);
 894       state = NORMAL;
 895       return 1;
 896     }
 897     else {
 898       put_string(post_label, fp);
 899       state = NORMAL;
 900     }
 901     break;
 902   }
 903   return 0;
 904 }
 905
 906 void label_processing_state::process(int c)
 907 {
 908   if (handle_pending(c))
 909     return;
 910   assert(state == NORMAL);
 911   switch (c) {
 912   case PRE_LABEL_MARKER:
 913     put_string(pre_label, fp);
 914     state = NORMAL;
 915     break;
 916   case POST_LABEL_MARKER:
 917     state = PENDING_POST;
 918     break;
 919   case LABEL_MARKER:
 920   case LABEL_MARKER + 1:
 921     count = 1;
 922     state = PENDING_LABEL;
 923     type = label_type(c - LABEL_MARKER);
 924     break;
 925   default:
 926     state = NORMAL;
 927     putc(c, fp);
 928     break;
 929   }
 930 }
 931
 932 extern "C" {
 933
 934 int rcompare(const void *p1, const void *p2)
 935 {
 936   return compare_reference(**(reference **)p1, **(reference **)p2);
 937 }
 938
 939 }
 940
 941 void output_references()
 942 {
 943   assert(accumulate);
 944   if (!hash_table_size) {
 945     error("nothing to reference (probably `bibliography' before `sort')");
 946     accumulate = 0;
 947     nreferences = 0;
 948     return;
 949   }
 950   if (nreferences > 0) {
 951     int j = 0;
 952     int i;
 953     for (i = 0; i < hash_table_size; i++)
 954       if (reference_hash_table[i] != 0)
 955         reference_hash_table[j++] = reference_hash_table[i];
 956     assert(j == nreferences);
 957     for (; j < hash_table_size; j++)
 958       reference_hash_table[j] = 0;
 959     qsort(reference_hash_table, nreferences, sizeof(reference*), rcompare);
 960     for (i = 0; i < nreferences; i++)
 961       reference_hash_table[i]->set_number(i);
 962     compute_labels(reference_hash_table, nreferences);
 963   }
 964   if (outfp != stdout) {
 965     rewind(outfp);
 966     {
 967       label_processing_state state(citation, ncitations, stdout);
 968       int c;
 969       while ((c = getc(outfp)) != EOF)
 970         state.process(c);
 971     }
 972     ncitations = 0;
 973     fclose(outfp);
 974     outfp = stdout;
 975   }
 976   if (nreferences > 0) {
 977     fputs(".]<\n", outfp);
 978     for (int i = 0; i < nreferences; i++) {
 979       if (sort_fields.length() > 0)
 980         reference_hash_table[i]->print_sort_key_comment(outfp);
 981       if (label_in_reference) {
 982         fputs(".ds [F ", outfp);
 983         const string &label = reference_hash_table[i]->get_label(NORMAL_LABEL);
 984         if (label.length() > 0
 985             && (label[0] == ' ' || label[0] == '\\' || label[0] == '"'))
 986           putc('"', outfp);
 987         put_string(label, outfp);
 988         putc('\n', outfp);
 989       }
 990       reference_hash_table[i]->output(outfp);
 991       delete reference_hash_table[i];
 992       reference_hash_table[i] = 0;
 993     }
 994     fputs(".]>\n", outfp);
 995     nreferences = 0;
 996   }
 997   clear_labels();
 998 }
 999
1000 static reference *find_reference(const char *query, int query_len)
1001 {
1002   // This is so that error messages look better.
1003   while (query_len > 0 && csspace(query[query_len - 1]))
1004     query_len--;
1005   string str;
1006   for (int i = 0; i < query_len; i++)
1007     str += query[i] == '\n' ? ' ' : query[i];
1008   str += '\0';
1009   possibly_load_default_database();
1010   search_list_iterator iter(&database_list, str.contents());
1011   reference_id rid;
1012   const char *start;
1013   int len;
1014   if (!iter.next(&start, &len, &rid)) {
1015     error("no matches for `%1'", str.contents());
1016     return 0;
1017   }
1018   const char *end = start + len;
1019   while (start < end) {
1020     if (*start == '%')
1021       break;
1022     while (start < end && *start++ != '\n')
1023       ;
1024   }
1025   if (start >= end) {
1026     error("found a reference for `%1' but it didn't contain any fields",
1027           str.contents());
1028     return 0;
1029   }
1030   reference *result = new reference(start, end - start, &rid);
1031   if (iter.next(&start, &len, &rid))
1032     warning("multiple matches for `%1'", str.contents());
1033   return result;
1034 }
1035
1036 static reference *make_reference(const string &str, unsigned *flagsp)
1037 {
1038   const char *start = str.contents();
1039   const char *end = start + str.length();
1040   const char *ptr = start;
1041   while (ptr < end) {
1042     if (*ptr == '%')
1043       break;
1044     while (ptr < end && *ptr++ != '\n')
1045       ;
1046   }
1047   *flagsp = 0;
1048   for (; start < ptr; start++) {
1049     if (*start == '#')
1050       *flagsp = (SHORT_LABEL | (*flagsp & (FORCE_RIGHT_BRACKET
1051                                            | FORCE_LEFT_BRACKET)));
1052     else if (*start == '[')
1053       *flagsp |= FORCE_LEFT_BRACKET;
1054     else if (*start == ']')
1055       *flagsp |= FORCE_RIGHT_BRACKET;
1056     else if (!csspace(*start))
1057       break;
1058   }
1059   if (start >= end) {
1060     error("empty reference");
1061     return new reference;
1062   }
1063   reference *database_ref = 0;
1064   if (start < ptr)
1065     database_ref = find_reference(start, ptr - start);
1066   reference *inline_ref = 0;
1067   if (ptr < end)
1068     inline_ref = new reference(ptr, end - ptr);
1069   if (inline_ref) {
1070     if (database_ref) {
1071       database_ref->merge(*inline_ref);
1072       delete inline_ref;
1073       return database_ref;
1074     }
1075     else
1076       return inline_ref;
1077   }
1078   else if (database_ref)
1079     return database_ref;
1080   else
1081     return new reference;
1082 }
1083
1084 static void do_ref(const string &str)
1085 {
1086   if (accumulate)
1087     (void)store_reference(str);
1088   else {
1089     (void)immediately_handle_reference(str);
1090     immediately_output_references();
1091   }
1092 }
1093
1094 static void trim_blanks(string &str)
1095 {
1096   const char *start = str.contents();
1097   const char *end = start + str.length();
1098   while (end > start && end[-1] != '\n' && csspace(end[-1]))
1099     --end;
1100   str.set_length(end - start);
1101 }
1102
1103 void do_bib(const char *filename)
1104 {
1105   FILE *fp;
1106   if (strcmp(filename, "-") == 0)
1107     fp = stdin;
1108   else {
1109     errno = 0;
1110     fp = fopen(filename, "r");
1111     if (fp == 0) {
1112       error("can't open `%1': %2", filename, strerror(errno));
1113       return;
1114     }
1115     current_filename = filename;
1116   }
1117   enum {
1118     START, MIDDLE, BODY, BODY_START, BODY_BLANK, BODY_DOT
1119     } state = START;
1120   string body;
1121   for (;;) {
1122     int c = getc(fp);
1123     if (c == EOF)
1124       break;
1125     if (invalid_input_char(c)) {
1126       error("invalid input character code %1", c);
1127       continue;
1128     }
1129     switch (state) {
1130     case START:
1131       if (c == '%') {
1132         body = c;
1133         state = BODY;
1134       }
1135       else if (c != '\n')
1136         state = MIDDLE;
1137       break;
1138     case MIDDLE:
1139       if (c == '\n')
1140         state = START;
1141       break;
1142     case BODY:
1143       body += c;
1144       if (c == '\n')
1145         state = BODY_START;
1146       break;
1147     case BODY_START:
1148       if (c == '\n') {
1149         do_ref(body);
1150         state = START;
1151       }
1152       else if (c == '.')
1153         state = BODY_DOT;
1154       else if (csspace(c)) {
1155         state = BODY_BLANK;
1156         body += c;
1157       }
1158       else {
1159         body += c;
1160         state = BODY;
1161       }
1162       break;
1163     case BODY_BLANK:
1164       if (c == '\n') {
1165         trim_blanks(body);
1166         do_ref(body);
1167         state = START;
1168       }
1169       else if (csspace(c))
1170         body += c;
1171       else {
1172         body += c;
1173         state = BODY;
1174       }
1175       break;
1176     case BODY_DOT:
1177       if (c == ']') {
1178         do_ref(body);
1179         state = MIDDLE;
1180       }
1181       else {
1182         body += '.';
1183         body += c;
1184         state = c == '\n' ? BODY_START : BODY;
1185       }
1186       break;
1187     default:
1188       assert(0);
1189     }
1190     if (c == '\n')
1191       current_lineno++;
1192   }
1193   switch (state) {
1194   case START:
1195   case MIDDLE:
1196     break;
1197   case BODY:
1198     body += '\n';
1199     do_ref(body);
1200     break;
1201   case BODY_DOT:
1202   case BODY_START:
1203     do_ref(body);
1204     break;
1205   case BODY_BLANK:
1206     trim_blanks(body);
1207     do_ref(body);
1208     break;
1209   }
1210   fclose(fp);
1211 }
1212
1213 // from the Dragon Book
1214
1215 unsigned hash_string(const char *s, int len)
1216 {
1217   const char *end = s + len;
1218   unsigned h = 0, g;
1219   while (s < end) {
1220     h <<= 4;
1221     h += *s++;
1222     if ((g = h & 0xf0000000) != 0) {
1223       h ^= g >> 24;
1224       h ^= g;
1225     }
1226   }
1227   return h;
1228 }
1229
1230 int next_size(int n)
1231 {
1232   static const int table_sizes[] = {
1233     101, 503, 1009, 2003, 3001, 4001, 5003, 10007, 20011, 40009,
1234     80021, 160001, 500009, 1000003, 2000003, 4000037, 8000009,
1235     16000057, 32000011, 64000031, 128000003, 0
1236   };
1237
1238   const int *p;
1239   for (p = table_sizes; *p <= n && *p != 0; p++)
1240     ;
1241   assert(*p != 0);
1242   return *p;
1243 }
1244