1 /* PSPP - computes sample statistics.
2 Copyright (C) 1997-2004, 2006 Free Software Foundation, Inc.
3 Written by Ben Pfaff <blp@gnu.org>.
5 This program is free software; you can redistribute it and/or
6 modify it under the terms of the GNU General Public License as
7 published by the Free Software Foundation; either version 2 of the
8 License, or (at your option) any later version.
10 This program is distributed in the hope that it will be useful, but
11 WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 General Public License for more details.
15 You should have received a copy of the GNU General Public License
16 along with this program; if not, write to the Free Software
17 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
22 #include <language/data-io/data-reader.h>
29 #include <data/file-handle-def.h>
30 #include <data/file-name.h>
31 #include <data/procedure.h>
32 #include <language/command.h>
33 #include <language/data-io/file-handle.h>
34 #include <language/lexer/lexer.h>
35 #include <language/line-buffer.h>
36 #include <libpspp/alloc.h>
37 #include <libpspp/assertion.h>
38 #include <libpspp/message.h>
39 #include <libpspp/str.h>
45 #define _(msgid) gettext (msgid)
47 /* Flags for DFM readers. */
50 DFM_ADVANCE = 002, /* Read next line on dfm_get_record() call? */
51 DFM_SAW_BEGIN_DATA = 004, /* For inline_file only, whether we've
52 already read a BEGIN DATA line. */
53 DFM_TABS_EXPANDED = 010, /* Tabs have been expanded. */
56 /* Data file reader. */
59 struct file_handle *fh; /* File handle. */
60 struct msg_locator where; /* Current location in data file. */
61 struct string line; /* Current line. */
62 struct string scratch; /* Extra line buffer. */
63 enum dfm_reader_flags flags; /* Zero or more of DFM_*. */
64 FILE *file; /* Associated file. */
65 size_t pos; /* Offset in line of current character. */
66 unsigned eof_cnt; /* # of attempts to advance past EOF. */
69 /* Closes reader R opened by dfm_open_reader(). */
71 dfm_close_reader (struct dfm_reader *r)
80 is_inline = r->fh == fh_inline_file ();
81 file_name = is_inline ? NULL : xstrdup (fh_get_file_name (r->fh));
82 still_open = fh_close (r->fh, "data file", "rs");
90 fn_close (file_name, r->file);
93 /* Skip any remaining data on the inline file. */
94 if (r->flags & DFM_SAW_BEGIN_DATA)
96 dfm_reread_record (r, 0);
98 dfm_forward_record (r);
102 ds_destroy (&r->line);
103 ds_destroy (&r->scratch);
108 /* Opens the file designated by file handle FH for reading as a
109 data file. Providing fh_inline_file() for FH designates the
110 "inline file", that is, data included inline in the command
111 file between BEGIN FILE and END FILE. Returns a reader if
112 successful, or a null pointer otherwise. */
114 dfm_open_reader (struct file_handle *fh)
116 struct dfm_reader *r;
119 rp = fh_open (fh, FH_REF_FILE | FH_REF_INLINE, "data file", "rs");
125 r = xmalloc (sizeof *r);
127 ds_init_empty (&r->line);
128 ds_init_empty (&r->scratch);
129 r->flags = DFM_ADVANCE;
131 if (fh != fh_inline_file ())
133 r->where.file_name = fh_get_file_name (fh);
134 r->where.line_number = 0;
135 r->file = fn_open (fh_get_file_name (fh), "rb");
138 msg (ME, _("Could not open \"%s\" for reading as a data file: %s."),
139 fh_get_file_name (r->fh), strerror (errno));
140 fh_close (fh,"data file", "rs");
150 /* Returns true if an I/O error occurred on READER, false otherwise. */
152 dfm_reader_error (const struct dfm_reader *r)
154 return fh_get_referent (r->fh) == FH_REF_FILE && ferror (r->file);
157 /* Reads a record from the inline file into R.
158 Returns true if successful, false on failure. */
160 read_inline_record (struct dfm_reader *r)
162 if ((r->flags & DFM_SAW_BEGIN_DATA) == 0)
164 r->flags |= DFM_SAW_BEGIN_DATA;
168 if (!lex_force_match_id ("BEGIN") || !lex_force_match_id ("DATA"))
170 getl_set_prompt_style (GETL_PROMPT_DATA);
173 if (!getl_read_line (NULL))
175 msg (SE, _("Unexpected end-of-file while reading data in BEGIN "
176 "DATA. This probably indicates "
177 "a missing or misformatted END DATA command. "
178 "END DATA must appear by itself on a single line "
179 "with exactly one space between words."));
183 if (ds_length (&getl_buf) >= 8
184 && !strncasecmp (ds_cstr (&getl_buf), "end data", 8))
186 lex_set_prog (ds_end (&getl_buf));
190 ds_assign_string (&r->line, &getl_buf);
194 /* Reads a record from a disk file into R.
195 Returns true if successful, false on failure. */
197 read_file_record (struct dfm_reader *r)
199 assert (r->fh != fh_inline_file ());
201 if (fh_get_mode (r->fh) == FH_MODE_TEXT)
203 if (!ds_read_line (&r->line, r->file))
205 if (ferror (r->file))
206 msg (ME, _("Error reading file %s: %s."),
207 fh_get_name (r->fh), strerror (errno));
211 else if (fh_get_mode (r->fh) == FH_MODE_BINARY)
213 size_t record_width = fh_get_record_width (r->fh);
214 size_t amt = ds_read_stream (&r->line, 1, record_width, r->file);
215 if (record_width != amt)
217 if (ferror (r->file))
218 msg (ME, _("Error reading file %s: %s."),
219 fh_get_name (r->fh), strerror (errno));
221 msg (ME, _("%s: Partial record at end of file."),
222 fh_get_name (r->fh));
230 r->where.line_number++;
235 /* Reads a record from R, setting the current position to the
236 start of the line. If an error occurs or end-of-file is
237 encountered, the current line is set to null. */
239 read_record (struct dfm_reader *r)
241 return (fh_get_referent (r->fh) == FH_REF_FILE
242 ? read_file_record (r)
243 : read_inline_record (r));
246 /* Returns the number of attempts, thus far, to advance past
247 end-of-file in reader R. Reads forward in HANDLE's file, if
248 necessary, to find out.
250 Normally, the user stops attempting to read from the file the
251 first time EOF is reached (a return value of 1). If the user
252 tries to read past EOF again (a return value of 2 or more),
253 an error message is issued, and the caller should more
254 forcibly abort to avoid an infinite loop. */
256 dfm_eof (struct dfm_reader *r)
258 if (r->flags & DFM_ADVANCE)
260 r->flags &= ~DFM_ADVANCE;
262 if (r->eof_cnt == 0 && read_record (r))
271 if (r->fh != fh_inline_file ())
272 msg (ME, _("Attempt to read beyond end-of-file on file %s."),
273 fh_get_name (r->fh));
275 msg (ME, _("Attempt to read beyond END DATA."));
282 /* Returns the current record in the file corresponding to
283 HANDLE. Aborts if reading from the file is necessary or at
284 end of file, so call dfm_eof() first. */
286 dfm_get_record (struct dfm_reader *r)
288 assert ((r->flags & DFM_ADVANCE) == 0);
289 assert (r->eof_cnt == 0);
291 return ds_substr (&r->line, r->pos, SIZE_MAX);
294 /* Expands tabs in the current line into the equivalent number of
295 spaces, if appropriate for this kind of file. Aborts if
296 reading from the file is necessary or at end of file, so call
299 dfm_expand_tabs (struct dfm_reader *r)
301 size_t ofs, new_pos, tab_width;
303 assert ((r->flags & DFM_ADVANCE) == 0);
304 assert (r->eof_cnt == 0);
306 if (r->flags & DFM_TABS_EXPANDED)
308 r->flags |= DFM_TABS_EXPANDED;
310 if (r->fh != fh_inline_file ()
311 && (fh_get_mode (r->fh) == FH_MODE_BINARY
312 || fh_get_tab_width (r->fh) == 0
313 || ds_find_char (&r->line, '\t') == SIZE_MAX))
316 /* Expand tabs from r->line into r->scratch, and figure out
317 new value for r->pos. */
318 tab_width = fh_get_tab_width (r->fh);
319 ds_clear (&r->scratch);
321 for (ofs = 0; ofs < ds_length (&r->line); ofs++)
326 new_pos = ds_length (&r->scratch);
328 c = ds_data (&r->line)[ofs];
330 ds_put_char (&r->scratch, c);
334 ds_put_char (&r->scratch, ' ');
335 while (ds_length (&r->scratch) % tab_width != 0);
338 if (new_pos == SIZE_MAX)
340 /* Maintain the same relationship between position and line
341 length that we had before. DATA LIST uses a
342 beyond-the-end position to deal with an empty field at
343 the end of the line. */
344 assert (r->pos >= ds_length (&r->line));
345 new_pos = (r->pos - ds_length (&r->line)) + ds_length (&r->scratch);
348 /* Swap r->line and r->scratch and set new r->pos. */
349 ds_swap (&r->line, &r->scratch);
353 /* Causes dfm_get_record() or dfm_get_whole_record() to read in
354 the next record the next time it is executed on file
357 dfm_forward_record (struct dfm_reader *r)
359 r->flags |= DFM_ADVANCE;
362 /* Cancels the effect of any previous dfm_fwd_record() executed
363 on file HANDLE. Sets the current line to begin in the 1-based
366 dfm_reread_record (struct dfm_reader *r, size_t column)
368 r->flags &= ~DFM_ADVANCE;
369 r->pos = MAX (column, 1) - 1;
372 /* Sets the current line to begin COLUMNS characters following
373 the current start. */
375 dfm_forward_columns (struct dfm_reader *r, size_t columns)
377 dfm_reread_record (r, (r->pos + 1) + columns);
380 /* Returns the 1-based column to which the line pointer in HANDLE
381 is set. Unless dfm_reread_record() or dfm_forward_columns()
382 have been called, this is 1. */
384 dfm_column_start (const struct dfm_reader *r)
389 /* Returns the number of columns we are currently beyond the end
390 of the line. At or before end-of-line, this is 0; one column
391 after end-of-line, this is 1; and so on. */
393 dfm_columns_past_end (const struct dfm_reader *r)
395 return r->pos < ds_length (&r->line) ? 0 : ds_length (&r->line) - r->pos;
398 /* Returns the 1-based column within the current line that P
401 dfm_get_column (const struct dfm_reader *r, const char *p)
403 return ds_pointer_to_position (&r->line, p) + 1;
406 /* Pushes the file name and line number on the fn/ln stack. */
408 dfm_push (struct dfm_reader *r)
410 if (r->fh != fh_inline_file ())
411 msg_push_msg_locator (&r->where);
414 /* Pops the file name and line number from the fn/ln stack. */
416 dfm_pop (struct dfm_reader *r)
418 if (r->fh != fh_inline_file ())
419 msg_pop_msg_locator (&r->where);
422 /* BEGIN DATA...END DATA procedure. */
424 /* Perform BEGIN DATA...END DATA as a procedure in itself. */
426 cmd_begin_data (struct dataset *ds)
428 struct dfm_reader *r;
431 if (!fh_is_open (fh_inline_file ()))
433 msg (SE, _("This command is not valid here since the current "
434 "input program does not access the inline file."));
435 return CMD_CASCADING_FAILURE;
438 /* Open inline file. */
439 r = dfm_open_reader (fh_inline_file ());
440 r->flags |= DFM_SAW_BEGIN_DATA;
442 /* Input procedure reads from inline file. */
443 getl_set_prompt_style (GETL_PROMPT_DATA);
444 ok = procedure (ds, NULL, NULL);
446 dfm_close_reader (r);
448 return ok ? CMD_SUCCESS : CMD_CASCADING_FAILURE;