lily/source-file.cc

   1 /*
   2   source-file.cc -- implement Source_file
   3
   4   source file of the GNU LilyPond music typesetter
   5
   6   (c) 1997--2009 Jan Nieuwenhuizen <janneke@gnu.org>
   7   Han-Wen Nienhuys <hanwen@xs4all.nl>
   8 */
   9
  10 #if GCC_MAJOR < 4
  11 #define _GLIBCXX_HAVE_MBSTATE_T
  12 #include <wchar.h>
  13 #endif /* GCC_MAJOR < 4 */
  14
  15 #include "source-file.hh"
  16
  17 #include "config.hh"
  18
  19 #include <cstdio>
  20
  21 #if HAVE_SSTREAM
  22 #include <sstream>
  23 #else
  24 #include <strstream>
  25 #define istringstream(x) istrstream (x, length ())
  26 #endif
  27 using namespace std;
  28
  29 #include "file-name-map.hh"
  30 #include "international.hh"
  31 #include "misc.hh"
  32 #include "warn.hh"
  33
  34 void
  35 Source_file::load_stdin ()
  36 {
  37   characters_.clear ();
  38   int c;
  39   while ((c = fgetc (stdin)) != EOF)
  40     characters_.push_back (c);
  41 }
  42
  43 /*
  44   return contents of FILENAME. *Not 0-terminated!*
  45  */
  46 vector<char>
  47 gulp_file (string filename, int desired_size)
  48 {
  49   /* "b" must ensure to open literally, avoiding text (CR/LF)
  50      conversions.  */
  51   FILE *f = fopen (filename.c_str (), "rb");
  52   if (!f)
  53     {
  54       warning (_f ("cannot open file: `%s'", filename.c_str ()));
  55
  56       vector<char> cxx_arr;
  57       return cxx_arr;
  58     }
  59
  60   fseek (f, 0, SEEK_END);
  61   int real_size = ftell (f);
  62   int read_count = real_size;
  63
  64   if (desired_size > 0)
  65     read_count = min (read_count, desired_size);
  66
  67   rewind (f);
  68
  69   char *str = new char[read_count + 1];
  70   str[read_count] = 0;
  71
  72   int bytes_read = fread (str, sizeof (char), read_count, f);
  73   if (bytes_read != read_count)
  74     warning (_f ("expected to read %d characters, got %d", bytes_read,
  75                  read_count));
  76   fclose (f);
  77   int filesize = bytes_read;
  78
  79   vector<char> cxx_arr;
  80   cxx_arr.resize (filesize);
  81
  82   copy (str, str + filesize, cxx_arr.begin ());
  83
  84   delete[] str;
  85   return cxx_arr;
  86 }
  87
  88 void
  89 Source_file::init ()
  90 {
  91   istream_ = 0;
  92   line_offset_ = 0;
  93   str_port_ = SCM_EOL;
  94   self_scm_ = SCM_EOL;
  95   smobify_self ();
  96 }
  97
  98 Source_file::Source_file (string filename, string data)
  99 {
 100   init ();
 101
 102   name_ = filename;
 103
 104   characters_.resize (data.length ());
 105   copy (data.begin (), data.end (), characters_.begin ());
 106
 107   characters_.push_back (0);
 108
 109   init_port ();
 110
 111   for (vsize i = 0; i < characters_.size (); i++)
 112     if (characters_[i] == '\n')
 113       newline_locations_.push_back (&characters_[0] + i);
 114 }
 115
 116 Source_file::Source_file (string filename_string)
 117 {
 118   init ();
 119
 120   name_ = filename_string;
 121
 122   if (filename_string == "-")
 123     load_stdin ();
 124   else
 125     {
 126       characters_ = gulp_file (filename_string, -1);
 127     }
 128
 129   characters_.push_back (0);
 130
 131   init_port ();
 132
 133   for (vsize i = 0; i < characters_.size (); i++)
 134     if (characters_[i] == '\n')
 135       newline_locations_.push_back (&characters_[0] + i);
 136 }
 137
 138 void
 139 Source_file::init_port ()
 140 {
 141   SCM str = scm_from_locale_string (c_str ());
 142   str_port_ = scm_mkstrport (SCM_INUM0, str, SCM_OPN | SCM_RDNG, __FUNCTION__);
 143   scm_set_port_filename_x (str_port_, ly_string2scm (name_));
 144 }
 145
 146
 147 istream *
 148 Source_file::get_istream ()
 149 {
 150   if (!istream_)
 151     {
 152       if (length ()) // can-t this be done without such a hack?
 153         istream_ = new istringstream (c_str ());
 154       else
 155         {
 156           istream_ = new istringstream ("");
 157           istream_->setstate (ios::eofbit);
 158           //      istream_->set (ios::eofbit);
 159         }
 160     }
 161   return istream_;
 162 }
 163
 164 string
 165 Source_file::file_line_column_string (char const *context_str0) const
 166 {
 167   if (!c_str ())
 168     return " (" + _ ("position unknown") + ")";
 169   else
 170     {
 171       int l, ch, col, offset;
 172       get_counts (context_str0, &l, &ch, &col, &offset);
 173
 174       return name_string () + ":" + to_string (l)
 175         + ":" + to_string (col);
 176     }
 177 }
 178
 179 string
 180 Source_file::quote_input (char const *pos_str0) const
 181 {
 182   if (!contains (pos_str0))
 183     return " (" + _ ("position unknown") + ")";
 184
 185   int l, ch, col, offset;
 186   get_counts (pos_str0, &l, &ch, &col, &offset);
 187   string line = line_string (pos_str0);
 188   string context = line.substr (0, offset)
 189     + to_string ('\n')
 190     + to_string (' ', col)
 191     + line.substr (offset, line.length () - offset);
 192   return context;
 193 }
 194
 195 string
 196 Source_file::name_string () const
 197 {
 198   return map_file_name (name_);
 199 }
 200
 201 Source_file::~Source_file ()
 202 {
 203   delete istream_;
 204 }
 205
 206 Slice
 207 Source_file::line_slice (char const *pos_str0) const
 208 {
 209   if (!contains (pos_str0))
 210     return Slice (0, 0);
 211
 212   char const *data_str0 = c_str ();
 213   char const *eof_C_ = data_str0 + length ();
 214
 215   if (pos_str0 == eof_C_)
 216     pos_str0--;
 217   char const *begin_str0 = pos_str0;
 218   while (begin_str0 > data_str0)
 219     if (*--begin_str0 == '\n')
 220       {
 221         begin_str0++;
 222         break;
 223       }
 224
 225   char const *end_str0 = pos_str0;
 226   while (end_str0 < eof_C_)
 227     if (*end_str0++ == '\n')
 228       {
 229         end_str0--;
 230         break;
 231       }
 232
 233   return Slice (begin_str0 - data_str0, end_str0 - data_str0);
 234 }
 235
 236 string
 237 Source_file::line_string (char const *pos_str0) const
 238 {
 239   if (!contains (pos_str0))
 240     return "";
 241
 242   Slice line = line_slice (pos_str0);
 243   char const *data_str0 = c_str ();
 244   return string (data_str0 + line[LEFT], line.length ());
 245 }
 246
 247 void
 248 Source_file::get_counts (char const *pos_str0,
 249                          int *line_number,
 250                          int *line_char,
 251                          int *column,
 252                          int *byte_offset) const
 253 {
 254   *line_number = 0;
 255
 256   if (!contains (pos_str0))
 257     return;
 258
 259   *line_number = get_line (pos_str0);
 260
 261   Slice line = line_slice (pos_str0);
 262   char const *data = c_str ();
 263   char const *line_start = (char const *)data + line[LEFT];
 264
 265   ssize left = (char const *) pos_str0 - line_start;
 266   string line_begin (line_start, left);
 267   char const *line_chars = line_begin.c_str ();
 268
 269   *line_char = 0;
 270   *column = 0;
 271   *byte_offset = 0;
 272
 273   while (left > 0)
 274     {
 275       size_t thislen = utf8_char_len (*line_chars);
 276
 277       if (thislen == 1 && line_chars[0] == '\t')
 278         (*column) = (*column / 8 + 1) * 8;
 279       else
 280         (*column)++;
 281
 282       (*line_char)++;
 283
 284       /*
 285         To have decent output in UTF-8 aware terminals,
 286         we must keep track of the number of bytes from
 287         the left edge of the terminal.
 288       */
 289       *byte_offset += thislen;
 290
 291       /* Advance past this character. */
 292       line_chars += thislen;
 293       left -= thislen;
 294     }
 295 }
 296
 297 bool
 298 Source_file::contains (char const *pos_str0) const
 299 {
 300   return (pos_str0 && (pos_str0 >= c_str ()) && (pos_str0 <= c_str () + length ()));
 301 }
 302
 303 int
 304 Source_file::get_line (char const *pos_str0) const
 305 {
 306   if (!contains (pos_str0))
 307     return 0;
 308
 309   if (!newline_locations_.size ())
 310     return 1;
 311
 312   /* this will find the '\n' character at the end of our line */
 313   vsize lo = lower_bound (newline_locations_,
 314                           pos_str0,
 315                           less<char const*> ());
 316
 317   /* the return value will be indexed from 1 */
 318   return lo + 1 + line_offset_;
 319 }
 320
 321 void
 322 Source_file::set_line (char const *pos_str0, int line)
 323 {
 324   int current_line = get_line (pos_str0);
 325   line_offset_ += line - current_line;
 326
 327   assert (line == get_line (pos_str0));
 328 }
 329
 330 int
 331 Source_file::length () const
 332 {
 333   return characters_.size ();
 334 }
 335
 336 char const *
 337 Source_file::c_str () const
 338 {
 339   return &characters_[0];
 340 }
 341
 342 SCM
 343 Source_file::get_port () const
 344 {
 345   return str_port_;
 346 }
 347
 348 /****************************************************************/
 349
 350 #include "ly-smobs.icc"
 351
 352 IMPLEMENT_SMOBS (Source_file);
 353 IMPLEMENT_DEFAULT_EQUAL_P (Source_file);
 354 IMPLEMENT_TYPE_P (Source_file, "ly:source-file?");
 355
 356 SCM
 357 Source_file::mark_smob (SCM smob)
 358 {
 359   Source_file *sc = (Source_file *) SCM_CELL_WORD_1 (smob);
 360
 361   return sc->str_port_;
 362 }
 363
 364
 365 int
 366 Source_file::print_smob (SCM smob, SCM port, scm_print_state *)
 367 {
 368   Source_file *sc = (Source_file *) SCM_CELL_WORD_1 (smob);
 369
 370   scm_puts ("#<Source_file ", port);
 371   scm_puts (sc->name_.c_str (), port);
 372
 373   /* Do not print properties, that is too much hassle.  */
 374   scm_puts (" >", port);
 375   return 1;
 376 }
 377