src/tex2lyx/table.C

   1 /**
   2  * \file table.C
   3  * This file is part of LyX, the document processor.
   4  * Licence details can be found in the file COPYING.
   5  *
   6  * \author André Pönitz
   7  * \author Jean-Marc Lasgouttes
   8  *
   9  * Full author contact details are available in file CREDITS.
  10  */
  11
  12 // {[(
  13
  14 #include <config.h>
  15
  16 #include "tex2lyx.h"
  17
  18 #include <cctype>
  19 #include <fstream>
  20 #include <iostream>
  21 #include <sstream>
  22 #include <vector>
  23 #include <map>
  24
  25 using std::cerr;
  26 using std::endl;
  27 using std::istringstream;
  28 using std::ostream;
  29 using std::ostringstream;
  30 using std::string;
  31 using std::vector;
  32
  33 #include "mathed/math_gridinfo.h"
  34
  35 // filled in preamble.C
  36 std::map<char, int> special_columns;
  37
  38
  39 namespace {
  40
  41 int string2int(string const & s, int deflt = 0)
  42 {
  43         istringstream is(s);
  44         int i = deflt;
  45         is >> i;
  46         return i;
  47 }
  48
  49
  50 string read_hlines(Parser & p)
  51 {
  52         ostringstream os;
  53         p.skip_spaces();
  54         while (p.good()) {
  55                 if (p.next_token().cs() == "hline") {
  56                         p.get_token();
  57                         os << "\\hline";
  58                 } else if (p.next_token().cs() == "cline") {
  59                         p.get_token();
  60                         os << "\\cline{" << p.verbatim_item() << "}";
  61                 } else
  62                         break;
  63                 p.skip_spaces();
  64         };
  65         //cerr << "read_hlines(), read: '" << os.str() << "'\n";
  66         //cerr << "read_hlines(), next token: " << p.next_token() << "\n";
  67         return os.str();
  68 }
  69
  70
  71 /* rather brutish way to code table structure in a string:
  72
  73   \begin{tabular}{ccc}
  74     1 & 2 & 3\\ \hline
  75     \multicolumn{2}{c}{4} & 5 //
  76     6 & 7 \\
  77   \end{tabular}
  78
  79  gets "translated" to:
  80
  81   1 TAB 2 TAB 3 LINE
  82   \hline HLINE  TAB 5 LINE
  83   5 TAB 7 LINE
  84 */
  85
  86 char const TAB   = '\001';
  87 char const LINE  = '\002';
  88 char const HLINE = '\004';
  89
  90
  91 void handle_colalign(Parser & p, vector<ColInfo> & colinfo)
  92 {
  93         if (p.get_token().cat() != catBegin)
  94                 cerr << "wrong syntax for table column alignment. '{' expected\n";
  95
  96         char nextalign = 'b';
  97         bool leftline = false;
  98         for (Token t=p.get_token(); p.good() && t.cat() != catEnd; t = p.get_token()){
  99 #ifdef FILEDEBUG
 100                 cerr << "t: " << t << "  c: '" << t.character() << "'\n";
 101 #endif
 102
 103                 switch (t.character()) {
 104                         case 'c':
 105                         case 'l':
 106                         case 'r': {
 107                                 ColInfo ci;
 108                                 ci.align = t.character();
 109                                 if (colinfo.size() && colinfo.back().rightline > 1) {
 110                                         ci.leftline = true;
 111                                         --colinfo.back().rightline;
 112                                 }
 113                                 colinfo.push_back(ci);
 114                                 break;
 115                         }
 116                         case 'p':
 117                                 colinfo.push_back(ColInfo());
 118                                 colinfo.back().align = nextalign;
 119                                 colinfo.back().width = p.verbatim_item();
 120                                 nextalign = 'b';
 121                                 break;
 122                         case '|':
 123                                 if (colinfo.empty())
 124                                         leftline = true;
 125                                 else
 126                                         ++colinfo.back().rightline;
 127                                 break;
 128                         case '>': {
 129                                 string s = p.verbatim_item();
 130                                 if (s == "\\raggedleft ")
 131                                         nextalign = 'l';
 132                                 else if (s == "\\raggedright ")
 133                                         nextalign = 'r';
 134                                 else
 135                                         cerr << "unknown '>' column '" << s << "'\n";
 136                                 break;
 137                         }
 138                         default:
 139                                 if (special_columns.find(t.character()) != special_columns.end()) {
 140                                         ColInfo ci;
 141                                         ci.align = 'c';
 142                                         ci.special += t.character();
 143                                         int const nargs = special_columns[t.character()];
 144                                         for (int i = 0; i < nargs; ++i)
 145                                                 ci.special += "{" + p.verbatim_item() + "}";
 146                                         //cerr << "handling special column '" << t << "' " << nargs
 147                                         //      << "  '" << ci.special << "'\n";
 148                                         colinfo.push_back(ci);
 149                                 } else {
 150                                         cerr << "ignoring special separator '" << t << "'\n";
 151                                 }
 152                                 break;
 153                         }
 154         }
 155         if (colinfo.size() && leftline)
 156                 colinfo[0].leftline = true;
 157 }
 158
 159
 160 } // anonymous namespace
 161
 162
 163 void parse_table(Parser & p, ostream & os, unsigned flags)
 164 {
 165         string hlines;
 166
 167         while (p.good()) {
 168                 Token const & t = p.get_token();
 169
 170 #ifdef FILEDEBUG
 171                 cerr << "t: " << t << " flags: " << flags << "\n";
 172 #endif
 173
 174                 //
 175                 // cat codes
 176                 //
 177                 if (t.cat() == catMath) {
 178                         // we are inside some text mode thingy, so opening new math is allowed
 179                         Token const & n = p.get_token();
 180                         if (n.cat() == catMath) {
 181                                 // TeX's $$...$$ syntax for displayed math
 182                                 os << "\\[";
 183                                 parse_math(p, os, FLAG_SIMPLE, MATH_MODE);
 184                                 os << "\\]";
 185                                 p.get_token(); // skip the second '$' token
 186                         } else {
 187                                 // simple $...$  stuff
 188                                 p.putback();
 189                                 os << '$';
 190                                 parse_math(p, os, FLAG_SIMPLE, MATH_MODE);
 191                                 os << '$';
 192                         }
 193                 }
 194
 195                 else if (t.cat() == catSpace || t.cat() == catNewline)
 196                                 os << t.cs();
 197
 198                 else if (t.cat() == catLetter ||
 199                                t.cat() == catSuper ||
 200                                t.cat() == catSub ||
 201                                t.cat() == catOther ||
 202                                t.cat() == catActive ||
 203                                t.cat() == catParameter)
 204                         os << t.character();
 205
 206                 else if (t.cat() == catBegin) {
 207                         os << '{';
 208                         parse_table(p, os, FLAG_BRACE_LAST);
 209                         os << '}';
 210                 }
 211
 212                 else if (t.cat() == catEnd) {
 213                         if (flags & FLAG_BRACE_LAST)
 214                                 return;
 215                         cerr << "unexpected '}'\n";
 216                 }
 217
 218                 else if (t.cat() == catAlign) {
 219                         os << TAB;
 220                         p.skip_spaces();
 221                 }
 222
 223                 else if (t.cs() == "tabularnewline" || t.cs() == "\\") {
 224                         // stuff before the line break
 225                         // and look ahead for stuff after the line break
 226                         os << HLINE << hlines << HLINE << LINE << read_hlines(p) << HLINE;
 227                         hlines.erase();
 228                 }
 229
 230                 else if (t.cs() == "hline")
 231                         hlines += "\\hline";
 232
 233                 else if (t.cs() == "cline")
 234                         hlines += "\\cline{" + p.verbatim_item() + '}';
 235
 236                 else if (t.cat() == catComment)
 237                         os << t.asInput();
 238
 239                 else if (t.cs() == "(") {
 240                         os << "\\(";
 241                         parse_math(p, os, FLAG_SIMPLE2, MATH_MODE);
 242                         os << "\\)";
 243                 }
 244
 245                 else if (t.cs() == "[") {
 246                         os << "\\[";
 247                         parse_math(p, os, FLAG_EQUATION, MATH_MODE);
 248                         os << "\\]";
 249                 }
 250
 251                 else if (t.cs() == "begin") {
 252                         string const name = p.getArg('{', '}');
 253                         active_environments.push_back(name);
 254                         os << "\\begin{" << name << '}';
 255                         if (is_math_env(name)) {
 256                                 parse_math(p, os, FLAG_END, MATH_MODE);
 257                         } else {
 258                                 parse_table(p, os, FLAG_END);
 259                         }
 260                         os << "\\end{" << name << '}';
 261                         active_environments.pop_back();
 262                 }
 263
 264                 else if (t.cs() == "end") {
 265                         if (flags & FLAG_END) {
 266                                 // eat environment name
 267                                 string const name = p.getArg('{', '}');
 268                                 if (name != active_environment())
 269                                         p.error("\\end{" + name + "} does not match \\begin{"
 270                                                 + active_environment() + "}");
 271                                 return;
 272                         }
 273                         p.error("found 'end' unexpectedly");
 274                 }
 275
 276                 else
 277                         os << t.asInput();
 278         }
 279 }
 280
 281
 282 void handle_hline_above(RowInfo & ri, vector<CellInfo> & ci)
 283 {
 284         ri.topline = true;
 285         for (size_t col = 0; col < ci.size(); ++col)
 286                 ci[col].topline = true;
 287 }
 288
 289
 290 void handle_hline_below(RowInfo & ri, vector<CellInfo> & ci)
 291 {
 292         ri.bottomline = true;
 293         for (size_t col = 0; col < ci.size(); ++col)
 294                 ci[col].bottomline = true;
 295 }
 296
 297
 298 void handle_tabular(Parser & p, ostream & os,
 299                     Context & context)
 300 {
 301         string posopts = p.getOpt();
 302         if (posopts.size())
 303                 cerr << "vertical tabular positioning '" << posopts << "' ignored\n";
 304
 305         vector<ColInfo>            colinfo;
 306
 307         // handle column formatting
 308         handle_colalign(p, colinfo);
 309
 310         // handle initial hlines
 311
 312         // first scan of cells
 313         // use table mode to keep it minimal-invasive
 314         // not exactly what's TeX doing...
 315         vector<string> lines;
 316         ostringstream ss;
 317         ss << read_hlines(p) << HLINE; // handle initial hlines
 318         parse_table(p, ss, FLAG_END);
 319         split(ss.str(), lines, LINE);
 320
 321         vector< vector<CellInfo> > cellinfo(lines.size());
 322         vector<RowInfo> rowinfo(lines.size());
 323
 324         // split into rows
 325         //cerr << "// split into rows\n";
 326         for (size_t row = 0; row < rowinfo.size(); ++row) {
 327
 328                 // init row
 329                 cellinfo[row].resize(colinfo.size());
 330
 331                 // split row
 332                 vector<string> dummy;
 333                 //cerr << "\n########### LINE: " << lines[row] << "########\n";
 334                 split(lines[row], dummy, HLINE);
 335
 336                 // handle horizontal line fragments
 337                 if (dummy.size() != 3) {
 338                         if (dummy.size() != 1)
 339                                 cerr << "unexpected dummy size: " << dummy.size()
 340                                         << " content: " << lines[row] << "\n";
 341                         dummy.resize(3);
 342                 }
 343                 lines[row] = dummy[1];
 344
 345                 //cerr << "line: " << row << " above 0: " << dummy[0] << "\n";
 346                 //cerr << "line: " << row << " below 2: " << dummy[2] <<  "\n";
 347                 //cerr << "line: " << row << " cells 1: " << dummy[1] <<  "\n";
 348
 349                 for (int i = 0; i <= 2; i += 2) {
 350                         //cerr << "   reading from line string '" << dummy[i] << "'\n";
 351                         Parser p1(dummy[i]);
 352                         while (p1.good()) {
 353                                 Token t = p1.get_token();
 354                                 //cerr << "read token: " << t << "\n";
 355                                 if (t.cs() == "hline") {
 356                                         if (i == 0) {
 357                                                 if (rowinfo[row].topline) {
 358                                                         if (row > 0) // extra bottomline above
 359                                                                 handle_hline_below(rowinfo[row - 1], cellinfo[row - 1]);
 360                                                         else
 361                                                                 cerr << "dropping extra hline\n";
 362                                                         //cerr << "below row: " << row-1 << endl;
 363                                                 } else {
 364                                                         handle_hline_above(rowinfo[row], cellinfo[row]);
 365                                                         //cerr << "above row: " << row << endl;
 366                                                 }
 367                                         } else {
 368                                                 //cerr << "below row: " << row << endl;
 369                                                 handle_hline_below(rowinfo[row], cellinfo[row]);
 370                                         }
 371                                 } else if (t.cs() == "cline") {
 372                                         string arg = p1.verbatim_item();
 373                                         //cerr << "read cline arg: '" << arg << "'\n";
 374                                         vector<string> t;
 375                                         split(arg, t, '-');
 376                                         t.resize(2);
 377                                         size_t from = string2int(t[0]) - 1;
 378                                         size_t to = string2int(t[1]);
 379                                         for (size_t col = from; col < to; ++col) {
 380                                                 //cerr << "row: " << row << " col: " << col << " i: " << i << endl;
 381                                                 if (i == 0) {
 382                                                         rowinfo[row].topline = true;
 383                                                         cellinfo[row][col].topline = true;
 384                                                 } else {
 385                                                         rowinfo[row].bottomline = true;
 386                                                         cellinfo[row][col].bottomline = true;
 387                                                 }
 388                                         }
 389                                 } else {
 390                                         cerr << "unexpected line token: " << t << endl;
 391                                 }
 392                         }
 393                 }
 394
 395                 // split into cells
 396                 vector<string> cells;
 397                 split(lines[row], cells, TAB);
 398                 // Has the last multicolumn cell a rightline?
 399                 bool last_rightline = false;
 400                 for (size_t col = 0, cell = 0;
 401                                 cell < cells.size() && col < colinfo.size(); ++col, ++cell) {
 402                         //cerr << "cell content: '" << cells[cell] << "'\n";
 403                         Parser p(cells[cell]);
 404                         p.skip_spaces();
 405                         //cells[cell] << "'\n";
 406                         if (p.next_token().cs() == "multicolumn") {
 407                                 // how many cells?
 408                                 p.get_token();
 409                                 size_t const ncells = string2int(p.verbatim_item());
 410
 411                                 // special cell properties alignment
 412                                 vector<ColInfo> t;
 413                                 handle_colalign(p, t);
 414                                 cellinfo[row][col].multi     = 1;
 415                                 cellinfo[row][col].align     = t.front().align;
 416                                 ostringstream os;
 417                                 parse_text_in_inset(p, os, FLAG_ITEM, false, context);
 418                                 cellinfo[row][col].content   = os.str();
 419
 420                                 // multicolumn cells are tricky: This
 421                                 // \multicolumn{2}{|c|}{col1-2}&
 422                                 // \multicolumn{2}{|c|}{col3-4} "\\"
 423                                 // gives | col1-2 | col3-4 | and not
 424                                 //       | col1-2 || col3-4 |
 425                                 // So:
 426                                 if (last_rightline && t.front().leftline) {
 427                                         t.front().leftline = false;
 428                                 }
 429                                 last_rightline = t.front().rightline;
 430
 431                                 // multicolumn lines override normal cell lines
 432                                 cellinfo[row][col].leftline  = t.front().leftline;
 433                                 cellinfo[row][col].rightline = t.front().rightline;
 434
 435                                 // add dummy cells for multicol
 436                                 for (size_t i = 0; i < ncells - 1 && col < colinfo.size(); ++i) {
 437                                         ++col;
 438                                         cellinfo[row][col].multi = 2;
 439                                         cellinfo[row][col].align = 'c';
 440                                 }
 441
 442                                 // more than one line on the right?
 443                                 if (t.front().rightline > 1)
 444                                         cellinfo[row][col + 1].leftline = true;
 445
 446                         } else {
 447                                 // FLAG_END is a hack, we need to read all of it
 448                                 cellinfo[row][col].leftline = colinfo[col].leftline;
 449                                 cellinfo[row][col].rightline = colinfo[col].rightline;
 450                                 cellinfo[row][col].align = colinfo[col].align;
 451                                 ostringstream os;
 452                                 parse_text_in_inset(p, os, FLAG_CELL, false, context);
 453                                 cellinfo[row][col].content   = os.str();
 454                                 last_rightline = false;
 455                         }
 456                 }
 457
 458                 //cerr << "//  handle almost empty last row what we have\n";
 459                 // handle almost empty last row
 460                 if (row && lines[row].empty() && row + 1 == rowinfo.size()) {
 461                         //cerr << "remove empty last line\n";
 462                         if (rowinfo[row].topline)
 463                                 rowinfo[row - 1].bottomline = true;
 464                         for (size_t col = 0; col < colinfo.size(); ++col)
 465                                 if (cellinfo[row][col].topline)
 466                                         cellinfo[row - 1][col].bottomline = true;
 467                         rowinfo.pop_back();
 468                 }
 469
 470         }
 471
 472         //cerr << "// output what we have\n";
 473         // output what we have
 474         os << "\n<lyxtabular version=\"3\" rows=\"" << rowinfo.size()
 475                  << "\" columns=\"" << colinfo.size() << "\">\n"
 476                  << "<features>\n";
 477
 478         //cerr << "// after header\n";
 479         for (size_t col = 0; col < colinfo.size(); ++col) {
 480                 os << "<column alignment=\""
 481                    << verbose_align(colinfo[col].align) << "\"";
 482                 os << " valignment=\"top\"";
 483                 if (colinfo[col].leftline)
 484                         os << " leftline=\"true\"";
 485                 if (colinfo[col].rightline)
 486                         os << " rightline=\"true\"";
 487                 if (colinfo[col].width.size())
 488                         os << " width=\"" << colinfo[col].width << "\"";
 489                 if (colinfo[col].special.size())
 490                         os << " special=\"" << colinfo[col].special << "\"";
 491                 os << ">\n";
 492         }
 493         //cerr << "// after cols\n";
 494
 495         for (size_t row = 0; row < rowinfo.size(); ++row) {
 496                 os << "<row";
 497                 if (rowinfo[row].topline)
 498                         os << " topline=\"true\"";
 499                 if (rowinfo[row].bottomline)
 500                         os << " bottomline=\"true\"";
 501                 os << ">\n";
 502                 for (size_t col = 0; col < colinfo.size(); ++col) {
 503                         CellInfo const & cell = cellinfo[row][col];
 504                         os << "<cell";
 505                         if (cell.multi)
 506                                 os << " multicolumn=\"" << cell.multi << "\"";
 507                         os << " alignment=\"" << verbose_align(cell.align)
 508                            << "\""
 509                            << " valignment=\"top\"";
 510                         if (cell.topline)
 511                                 os << " topline=\"true\"";
 512                         if (cell.bottomline)
 513                                 os << " bottomline=\"true\"";
 514                         if (cell.leftline)
 515                                 os << " leftline=\"true\"";
 516                         if (cell.rightline)
 517                                 os << " rightline=\"true\"";
 518                         //cerr << "\nrow: " << row << " col: " << col;
 519                         //if (cell.topline)
 520                         //      cerr << " topline=\"true\"";
 521                         //if (cell.bottomline)
 522                         //      cerr << " bottomline=\"true\"";
 523                         os << " usebox=\"none\""
 524                            << ">"
 525                            << "\n\\begin_inset Text\n"
 526                            << cell.content
 527                            << "\n\\end_inset \n"
 528                            << "</cell>\n";
 529                 }
 530                 os << "</row>\n";
 531         }
 532
 533         os << "</lyxtabular>\n";
 534 }
 535
 536
 537
 538
 539 // }])