src/tex2lyx/table.C

   1 /** The .tex to .lyx converter
   2     \author André Pönitz (2003)
   3  */
   4
   5 // {[(
   6
   7 #include <config.h>
   8
   9 #include "tex2lyx.h"
  10
  11 #include <cctype>
  12 #include <fstream>
  13 #include <iostream>
  14 #include <sstream>
  15 #include <vector>
  16 #include <map>
  17
  18 using std::cerr;
  19 using std::endl;
  20 using std::istringstream;
  21 using std::ostream;
  22 using std::ostringstream;
  23 using std::string;
  24 using std::vector;
  25 using std::map;
  26
  27 #include "mathed/math_gridinfo.h"
  28
  29
  30 // filled in preamble.C
  31 std::map<char, int> special_columns;
  32
  33
  34 namespace {
  35
  36 int string2int(string const & s, int deflt = 0)
  37 {
  38         istringstream is(s);
  39         int i = deflt;
  40         is >> i;
  41         return i;
  42 }
  43
  44
  45 string read_hlines(Parser & p)
  46 {
  47         ostringstream os;
  48         p.skip_spaces();
  49         while (p.good()) {
  50                 if (p.next_token().cs() == "hline") {
  51                         p.get_token();
  52                         os << "\\hline";
  53                 } else if (p.next_token().cs() == "cline") {
  54                         p.get_token();
  55                         os << "\\cline{" << p.verbatim_item() << "}";
  56                 } else
  57                         break;
  58                 p.skip_spaces();
  59         };
  60         //cerr << "read_hlines(), read: '" << os.str() << "'\n";
  61         //cerr << "read_hlines(), next token: " << p.next_token() << "\n";
  62         return os.str();
  63 }
  64
  65
  66 /* rather brutish way to code table structure in a string:
  67
  68   \begin{tabular}{ccc}
  69     1 & 2 & 3\\ \hline
  70     \multicolumn{2}{c}{4} & 5 //
  71     6 & 7 \\
  72   \end{tabular}
  73
  74  gets "translated" to:
  75
  76   1 TAB 2 TAB 3 LINE
  77   \hline HLINE  TAB 5 LINE
  78   5 TAB 7 LINE
  79 */
  80
  81 char const TAB   = '\001';
  82 char const LINE  = '\002';
  83 char const HLINE = '\004';
  84
  85
  86 void handle_colalign(Parser & p, vector<ColInfo> & colinfo)
  87 {
  88         if (p.get_token().cat() != catBegin)
  89                 cerr << "wrong syntax for table column alignment. '{' expected\n";
  90
  91         char nextalign = 'b';
  92         bool leftline = false;
  93         for (Token t=p.get_token(); p.good() && t.cat() != catEnd; t = p.get_token()){
  94 #ifdef FILEDEBUG
  95                 cerr << "t: " << t << "  c: '" << t.character() << "'\n";
  96 #endif
  97
  98                 switch (t.character()) {
  99                         case 'c':
 100                         case 'l':
 101                         case 'r': {
 102                                 ColInfo ci;
 103                                 ci.align = t.character();
 104                                 if (colinfo.size() && colinfo.back().rightline > 1) {
 105                                         ci.leftline = true;
 106                                         --colinfo.back().rightline;
 107                                 }
 108                                 colinfo.push_back(ci);
 109                                 break;
 110                         }
 111                         case 'p':
 112                                 colinfo.push_back(ColInfo());
 113                                 colinfo.back().align = nextalign;
 114                                 colinfo.back().width = p.verbatim_item();
 115                                 nextalign = 'b';
 116                                 break;
 117                         case '|':
 118                                 if (colinfo.empty())
 119                                         leftline = true;
 120                                 else
 121                                         ++colinfo.back().rightline;
 122                                 break;
 123                         case '>': {
 124                                 string s = p.verbatim_item();
 125                                 if (s == "\\raggedleft ")
 126                                         nextalign = 'l';
 127                                 else if (s == "\\raggedright ")
 128                                         nextalign = 'r';
 129                                 else
 130                                         cerr << "unknown '>' column '" << s << "'\n";
 131                                 break;
 132                         }
 133                         default:
 134                                 if (special_columns.find(t.character()) != special_columns.end()) {
 135                                         ColInfo ci;
 136                                         ci.align = 'c';
 137                                         ci.special += t.character();
 138                                         int const nargs = special_columns[t.character()];
 139                                         for (int i = 0; i < nargs; ++i)
 140                                                 ci.special += "{" + p.verbatim_item() + "}";
 141                                         //cerr << "handling special column '" << t << "' " << nargs
 142                                         //      << "  '" << ci.special << "'\n";
 143                                         colinfo.push_back(ci);
 144                                 } else {
 145                                         cerr << "ignoring special separator '" << t << "'\n";
 146                                 }
 147                                 break;
 148                         }
 149         }
 150         if (colinfo.size() && leftline)
 151                 colinfo[0].leftline = true;
 152 }
 153
 154
 155 } // anonymous namespace
 156
 157
 158 void parse_table(Parser & p, ostream & os, unsigned flags)
 159 {
 160         string hlines;
 161
 162         while (p.good()) {
 163                 Token const & t = p.get_token();
 164
 165 #ifdef FILEDEBUG
 166                 cerr << "t: " << t << " flags: " << flags << "\n";
 167 #endif
 168
 169                 //
 170                 // cat codes
 171                 //
 172                 if (t.cat() == catMath) {
 173                                 // we are inside some text mode thingy, so opening new math is allowed
 174                         Token const & n = p.get_token();
 175                         if (n.cat() == catMath) {
 176                                 // TeX's $$...$$ syntax for displayed math
 177                                 os << "\\[";
 178                                 parse_math(p, os, FLAG_SIMPLE, MATH_MODE);
 179                                 os << "\\]";
 180                                 p.get_token(); // skip the second '$' token
 181                         } else {
 182                                 // simple $...$  stuff
 183                                 p.putback();
 184                                 os << '$';
 185                                 parse_math(p, os, FLAG_SIMPLE, MATH_MODE);
 186                                 os << '$';
 187                         }
 188                 }
 189
 190                 else if (t.cat() == catLetter ||
 191                                t.cat() == catSpace ||
 192                                t.cat() == catSuper ||
 193                                t.cat() == catSub ||
 194                                t.cat() == catOther ||
 195                                t.cat() == catActive ||
 196                                t.cat() == catNewline ||
 197                                t.cat() == catParameter)
 198                         os << t.character();
 199
 200                 else if (t.cat() == catBegin) {
 201                         os << '{';
 202                         parse_table(p, os, FLAG_BRACE_LAST);
 203                         os << '}';
 204                 }
 205
 206                 else if (t.cat() == catEnd) {
 207                         if (flags & FLAG_BRACE_LAST)
 208                                 return;
 209                         cerr << "unexpected '}'\n";
 210                 }
 211
 212                 else if (t.cat() == catAlign) {
 213                         os << TAB;
 214                 }
 215
 216                 else if (t.cs() == "tabularnewline" || t.cs() == "\\") {
 217                         // stuff before the line break
 218                         // and look ahead for stuff after the line break
 219                         os << HLINE << hlines << HLINE << LINE << read_hlines(p) << HLINE;
 220                         hlines.erase();
 221                 }
 222
 223                 else if (t.cs() == "hline")
 224                         hlines += "\\hline";
 225
 226                 else if (t.cs() == "cline")
 227                         hlines += "\\cline{" + p.verbatim_item() + '}';
 228
 229                 else if (t.cat() == catComment)
 230                         handle_comment(p);
 231
 232                 else if (t.cs() == "(") {
 233                         os << "\\(";
 234                         parse_math(p, os, FLAG_SIMPLE2, MATH_MODE);
 235                         os << "\\)";
 236                 }
 237
 238                 else if (t.cs() == "[") {
 239                         os << "\\[";
 240                         parse_math(p, os, FLAG_EQUATION, MATH_MODE);
 241                         os << "\\]";
 242                 }
 243
 244                 else if (t.cs() == "begin") {
 245                         string const name = p.getArg('{', '}');
 246                         active_environments.push_back(name);
 247                         parse_table(p, os, FLAG_END);
 248                 }
 249
 250                 else if (t.cs() == "end") {
 251                         if (flags & FLAG_END) {
 252                                 // eat environment name
 253                                 string const name = p.getArg('{', '}');
 254                                 if (name != active_environment())
 255                                         p.error("\\end{" + name + "} does not match \\begin{"
 256                                                 + active_environment() + "}");
 257                                 active_environments.pop_back();
 258                                 return;
 259                         }
 260                         p.error("found 'end' unexpectedly");
 261                 }
 262
 263                 else
 264                         os << t.asInput();
 265         }
 266 }
 267
 268
 269 void handle_hline_above(RowInfo & ri, vector<CellInfo> & ci)
 270 {
 271         ri.topline = true;
 272         for (size_t col = 0; col < ci.size(); ++col)
 273                 ci[col].topline = true;
 274 }
 275
 276
 277 void handle_hline_below(RowInfo & ri, vector<CellInfo> & ci)
 278 {
 279         ri.bottomline = true;
 280         for (size_t col = 0; col < ci.size(); ++col)
 281                 ci[col].bottomline = true;
 282 }
 283
 284
 285 void handle_tabular(Parser & p, ostream & os,
 286                     LyXTextClass const & textclass)
 287 {
 288         string posopts = p.getOpt();
 289         if (posopts.size())
 290                 cerr << "vertical tabular positioning '" << posopts << "' ignored\n";
 291
 292         vector<ColInfo>            colinfo;
 293
 294         // handle column formatting
 295         handle_colalign(p, colinfo);
 296
 297         // handle initial hlines
 298
 299         // first scan of cells
 300         // use table mode to keep it minimal-invasive
 301         // not exactly what's TeX doing...
 302         vector<string> lines;
 303         ostringstream ss;
 304         ss << read_hlines(p) << HLINE; // handle initial hlines
 305         parse_table(p, ss, FLAG_END);
 306         split(ss.str(), lines, LINE);
 307
 308         vector< vector<CellInfo> > cellinfo(lines.size());
 309         vector<RowInfo> rowinfo(lines.size());
 310
 311         // split into rows
 312         //cerr << "// split into rows\n";
 313         for (size_t row = 0; row < rowinfo.size(); ++row) {
 314
 315                 // init row
 316                 cellinfo[row].resize(colinfo.size());
 317
 318                 // split row
 319                 vector<string> dummy;
 320                 //cerr << "\n########### LINE: " << lines[row] << "########\n";
 321                 split(lines[row], dummy, HLINE);
 322
 323                 // handle horizontal line fragments
 324                 if (dummy.size() != 3) {
 325                         if (dummy.size() != 1)
 326                                 cerr << "unexpected dummy size: " << dummy.size()
 327                                         << " content: " << lines[row] << "\n";
 328                         dummy.resize(3);
 329                 }
 330                 lines[row] = dummy[1];
 331
 332                 //cerr << "line: " << row << " above 0: " << dummy[0] << "\n";
 333                 //cerr << "line: " << row << " below 2: " << dummy[2] <<  "\n";
 334                 //cerr << "line: " << row << " cells 1: " << dummy[1] <<  "\n";
 335
 336                 for (int i = 0; i <= 2; i += 2) {
 337                         //cerr << "   reading from line string '" << dummy[i] << "'\n";
 338                         Parser p1(dummy[i]);
 339                         while (p1.good()) {
 340                                 Token t = p1.get_token();
 341                                 //cerr << "read token: " << t << "\n";
 342                                 if (t.cs() == "hline") {
 343                                         if (i == 0) {
 344                                                 if (rowinfo[row].topline) {
 345                                                         if (row > 0) // extra bottomline above
 346                                                                 handle_hline_below(rowinfo[row - 1], cellinfo[row - 1]);
 347                                                         else
 348                                                                 cerr << "dropping extra hline\n";
 349                                                         //cerr << "below row: " << row-1 << endl;
 350                                                 } else {
 351                                                         handle_hline_above(rowinfo[row], cellinfo[row]);
 352                                                         //cerr << "above row: " << row << endl;
 353                                                 }
 354                                         } else {
 355                                                 //cerr << "below row: " << row << endl;
 356                                                 handle_hline_below(rowinfo[row], cellinfo[row]);
 357                                         }
 358                                 } else if (t.cs() == "cline") {
 359                                         string arg = p1.verbatim_item();
 360                                         //cerr << "read cline arg: '" << arg << "'\n";
 361                                         vector<string> t;
 362                                         split(arg, t, '-');
 363                                         t.resize(2);
 364                                         size_t from = string2int(t[0]) - 1;
 365                                         size_t to = string2int(t[1]);
 366                                         for (size_t col = from; col < to; ++col) {
 367                                                 //cerr << "row: " << row << " col: " << col << " i: " << i << endl;
 368                                                 if (i == 0) {
 369                                                         rowinfo[row].topline = true;
 370                                                         cellinfo[row][col].topline = true;
 371                                                 } else {
 372                                                         rowinfo[row].bottomline = true;
 373                                                         cellinfo[row][col].bottomline = true;
 374                                                 }
 375                                         }
 376                                 } else {
 377                                         cerr << "unexpected line token: " << t << endl;
 378                                 }
 379                         }
 380                 }
 381
 382                 // split into cells
 383                 vector<string> cells;
 384                 split(lines[row], cells, TAB);
 385                 for (size_t col = 0, cell = 0;
 386                                 cell < cells.size() && col < colinfo.size(); ++col, ++cell) {
 387                         //cerr << "cell content: '" << cells[cell] << "'\n";
 388                         Parser p(cells[cell]);
 389                         p.skip_spaces();
 390                         //cells[cell] << "'\n";
 391                         if (p.next_token().cs() == "multicolumn") {
 392                                 // how many cells?
 393                                 p.get_token();
 394                                 size_t const ncells = string2int(p.verbatim_item());
 395
 396                                 // special cell properties alignment
 397                                 vector<ColInfo> t;
 398                                 handle_colalign(p, t);
 399                                 cellinfo[row][col].multi     = 1;
 400                                 cellinfo[row][col].align     = t.front().align;
 401                                 cellinfo[row][col].content   = parse_text(p, FLAG_ITEM, false, textclass);
 402                                 cellinfo[row][col].leftline  |= t.front().leftline;
 403                                 cellinfo[row][col].rightline |= t.front().rightline;
 404
 405                                 // add dummy cells for multicol
 406                                 for (size_t i = 0; i < ncells - 1 && col < colinfo.size(); ++i) {
 407                                         ++col;
 408                                         cellinfo[row][col].multi = 2;
 409                                         cellinfo[row][col].align = 'c';
 410                                 }
 411
 412                                 // more than one line on the right?
 413                                 if (t.front().rightline > 1)
 414                                         cellinfo[row][col + 1].leftline = true;
 415
 416                         } else {
 417                                 // FLAG_END is a hack, we need to read all of it
 418                                 cellinfo[row][col].leftline = colinfo[col].leftline;
 419                                 cellinfo[row][col].rightline = colinfo[col].rightline;
 420                                 cellinfo[row][col].align = colinfo[col].align;
 421                                 cellinfo[row][col].content = parse_text(p, FLAG_END, false, textclass);
 422                         }
 423                 }
 424
 425                 //cerr << "//  handle almost empty last row what we have\n";
 426                 // handle almost empty last row
 427                 if (row && lines[row].empty() && row + 1 == rowinfo.size()) {
 428                         //cerr << "remove empty last line\n";
 429                         if (rowinfo[row].topline)
 430                                 rowinfo[row - 1].bottomline = true;
 431                         for (size_t col = 0; col < colinfo.size(); ++col)
 432                                 if (cellinfo[row][col].topline)
 433                                         cellinfo[row - 1][col].bottomline = true;
 434                         rowinfo.pop_back();
 435                 }
 436
 437         }
 438
 439         //cerr << "// output what we have\n";
 440         // output what we have
 441         os << "\n<lyxtabular version=\"3\" rows=\"" << rowinfo.size()
 442                  << "\" columns=\"" << colinfo.size() << "\">\n"
 443                  << "<features>\n";
 444
 445         //cerr << "// after header\n";
 446         for (size_t col = 0; col < colinfo.size(); ++col) {
 447                 os << "<column alignment=\""
 448                    << verbose_align(colinfo[col].align) << "\"";
 449                 os << " valignment=\"top\"";
 450                 if (colinfo[col].leftline)
 451                         os << " leftline=\"true\"";
 452                 if (colinfo[col].rightline)
 453                         os << " rightline=\"true\"";
 454                 if (colinfo[col].width.size())
 455                         os << " width=\"" << colinfo[col].width << "\"";
 456                 if (colinfo[col].special.size())
 457                         os << " special=\"" << colinfo[col].special << "\"";
 458                 os << ">\n";
 459         }
 460         //cerr << "// after cols\n";
 461
 462         for (size_t row = 0; row < rowinfo.size(); ++row) {
 463                 os << "<row";
 464                 if (rowinfo[row].topline)
 465                         os << " topline=\"true\"";
 466                 if (rowinfo[row].bottomline)
 467                         os << " bottomline=\"true\"";
 468                 os << ">\n";
 469                 for (size_t col = 0; col < colinfo.size(); ++col) {
 470                         CellInfo const & cell = cellinfo[row][col];
 471                         os << "<cell";
 472                         if (cell.multi)
 473                                 os << " multicolumn=\"" << cell.multi << "\"";
 474                         os << " alignment=\"" << verbose_align(cell.align)
 475                            << "\""
 476                            << " valignment=\"top\"";
 477                         if (cell.topline)
 478                                 os << " topline=\"true\"";
 479                         if (cell.bottomline)
 480                                 os << " bottomline=\"true\"";
 481                         if (cell.leftline)
 482                                 os << " leftline=\"true\"";
 483                         if (cell.rightline)
 484                                 os << " rightline=\"true\"";
 485                         //cerr << "\nrow: " << row << " col: " << col;
 486                         //if (cell.topline)
 487                         //      cerr << " topline=\"true\"";
 488                         //if (cell.bottomline)
 489                         //      cerr << " bottomline=\"true\"";
 490                         os << " usebox=\"none\""
 491                            << ">"
 492                            << "\n\\begin_inset Text"
 493                            << "\n\n\\layout Standard\n\n"
 494                            << cell.content
 495                            << "\n\\end_inset \n"
 496                            << "</cell>\n";
 497                 }
 498                 os << "</row>\n";
 499         }
 500
 501         os << "</lyxtabular>\n";
 502 }
 503
 504
 505
 506
 507 // }])