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