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/prompt.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. */
67 struct lexer *lexer; /* The lexer reading the file */
70 /* Closes reader R opened by dfm_open_reader(). */
72 dfm_close_reader (struct dfm_reader *r)
81 is_inline = r->fh == fh_inline_file ();
82 file_name = is_inline ? NULL : xstrdup (fh_get_file_name (r->fh));
83 still_open = fh_close (r->fh, "data file", "rs");
91 fn_close (file_name, r->file);
94 /* Skip any remaining data on the inline file. */
95 if (r->flags & DFM_SAW_BEGIN_DATA)
97 dfm_reread_record (r, 0);
99 dfm_forward_record (r);
103 ds_destroy (&r->line);
104 ds_destroy (&r->scratch);
109 /* Opens the file designated by file handle FH for reading as a
110 data file. Providing fh_inline_file() for FH designates the
111 "inline file", that is, data included inline in the command
112 file between BEGIN FILE and END FILE. Returns a reader if
113 successful, or a null pointer otherwise. */
115 dfm_open_reader (struct file_handle *fh, struct lexer *lexer)
117 struct dfm_reader *r;
120 rp = fh_open (fh, FH_REF_FILE | FH_REF_INLINE, "data file", "rs");
126 r = xmalloc (sizeof *r);
129 ds_init_empty (&r->line);
130 ds_init_empty (&r->scratch);
131 r->flags = DFM_ADVANCE;
133 if (fh != fh_inline_file ())
135 r->where.file_name = fh_get_file_name (fh);
136 r->where.line_number = 0;
137 r->file = fn_open (fh_get_file_name (fh), "rb");
140 msg (ME, _("Could not open \"%s\" for reading as a data file: %s."),
141 fh_get_file_name (r->fh), strerror (errno));
142 fh_close (fh,"data file", "rs");
152 /* Returns true if an I/O error occurred on READER, false otherwise. */
154 dfm_reader_error (const struct dfm_reader *r)
156 return fh_get_referent (r->fh) == FH_REF_FILE && ferror (r->file);
159 /* Reads a record from the inline file into R.
160 Returns true if successful, false on failure. */
162 read_inline_record (struct dfm_reader *r)
164 if ((r->flags & DFM_SAW_BEGIN_DATA) == 0)
166 r->flags |= DFM_SAW_BEGIN_DATA;
168 while (lex_token (r->lexer) == '.')
170 if (!lex_force_match_id (r->lexer, "BEGIN") || !lex_force_match_id (r->lexer, "DATA"))
172 prompt_set_style (PROMPT_DATA);
175 if (!lex_get_line_raw (r->lexer, NULL))
177 msg (SE, _("Unexpected end-of-file while reading data in BEGIN "
178 "DATA. This probably indicates "
179 "a missing or misformatted END DATA command. "
180 "END DATA must appear by itself on a single line "
181 "with exactly one space between words."));
185 if (ds_length (lex_entire_line_ds (r->lexer) ) >= 8
186 && !strncasecmp (lex_entire_line (r->lexer), "end data", 8))
188 lex_discard_line (r->lexer);
192 ds_assign_string (&r->line, lex_entire_line_ds (r->lexer) );
197 /* Reads a record from a disk file into R.
198 Returns true if successful, false on failure. */
200 read_file_record (struct dfm_reader *r)
202 assert (r->fh != fh_inline_file ());
204 if (fh_get_mode (r->fh) == FH_MODE_TEXT)
206 if (!ds_read_line (&r->line, r->file))
208 if (ferror (r->file))
209 msg (ME, _("Error reading file %s: %s."),
210 fh_get_name (r->fh), strerror (errno));
213 ds_chomp (&r->line, '\n');
215 else if (fh_get_mode (r->fh) == FH_MODE_BINARY)
217 size_t record_width = fh_get_record_width (r->fh);
218 size_t amt = ds_read_stream (&r->line, 1, record_width, r->file);
219 if (record_width != amt)
221 if (ferror (r->file))
222 msg (ME, _("Error reading file %s: %s."),
223 fh_get_name (r->fh), strerror (errno));
225 msg (ME, _("%s: Partial record at end of file."),
226 fh_get_name (r->fh));
234 r->where.line_number++;
239 /* Reads a record from R, setting the current position to the
240 start of the line. If an error occurs or end-of-file is
241 encountered, the current line is set to null. */
243 read_record (struct dfm_reader *r)
245 return (fh_get_referent (r->fh) == FH_REF_FILE
246 ? read_file_record (r)
247 : read_inline_record (r));
250 /* Returns the number of attempts, thus far, to advance past
251 end-of-file in reader R. Reads forward in HANDLE's file, if
252 necessary, to find out.
254 Normally, the user stops attempting to read from the file the
255 first time EOF is reached (a return value of 1). If the user
256 tries to read past EOF again (a return value of 2 or more),
257 an error message is issued, and the caller should more
258 forcibly abort to avoid an infinite loop. */
260 dfm_eof (struct dfm_reader *r)
262 if (r->flags & DFM_ADVANCE)
264 r->flags &= ~DFM_ADVANCE;
266 if (r->eof_cnt == 0 && read_record (r) )
275 if (r->fh != fh_inline_file ())
276 msg (ME, _("Attempt to read beyond end-of-file on file %s."),
277 fh_get_name (r->fh));
279 msg (ME, _("Attempt to read beyond END DATA."));
286 /* Returns the current record in the file corresponding to
287 HANDLE. Aborts if reading from the file is necessary or at
288 end of file, so call dfm_eof() first. */
290 dfm_get_record (struct dfm_reader *r)
292 assert ((r->flags & DFM_ADVANCE) == 0);
293 assert (r->eof_cnt == 0);
295 return ds_substr (&r->line, r->pos, SIZE_MAX);
298 /* Expands tabs in the current line into the equivalent number of
299 spaces, if appropriate for this kind of file. Aborts if
300 reading from the file is necessary or at end of file, so call
303 dfm_expand_tabs (struct dfm_reader *r)
305 size_t ofs, new_pos, tab_width;
307 assert ((r->flags & DFM_ADVANCE) == 0);
308 assert (r->eof_cnt == 0);
310 if (r->flags & DFM_TABS_EXPANDED)
312 r->flags |= DFM_TABS_EXPANDED;
314 if (r->fh != fh_inline_file ()
315 && (fh_get_mode (r->fh) == FH_MODE_BINARY
316 || fh_get_tab_width (r->fh) == 0
317 || ds_find_char (&r->line, '\t') == SIZE_MAX))
320 /* Expand tabs from r->line into r->scratch, and figure out
321 new value for r->pos. */
322 tab_width = fh_get_tab_width (r->fh);
323 ds_clear (&r->scratch);
325 for (ofs = 0; ofs < ds_length (&r->line); ofs++)
330 new_pos = ds_length (&r->scratch);
332 c = ds_data (&r->line)[ofs];
334 ds_put_char (&r->scratch, c);
338 ds_put_char (&r->scratch, ' ');
339 while (ds_length (&r->scratch) % tab_width != 0);
342 if (new_pos == SIZE_MAX)
344 /* Maintain the same relationship between position and line
345 length that we had before. DATA LIST uses a
346 beyond-the-end position to deal with an empty field at
347 the end of the line. */
348 assert (r->pos >= ds_length (&r->line));
349 new_pos = (r->pos - ds_length (&r->line)) + ds_length (&r->scratch);
352 /* Swap r->line and r->scratch and set new r->pos. */
353 ds_swap (&r->line, &r->scratch);
357 /* Causes dfm_get_record() or dfm_get_whole_record() to read in
358 the next record the next time it is executed on file
361 dfm_forward_record (struct dfm_reader *r)
363 r->flags |= DFM_ADVANCE;
366 /* Cancels the effect of any previous dfm_fwd_record() executed
367 on file HANDLE. Sets the current line to begin in the 1-based
370 dfm_reread_record (struct dfm_reader *r, size_t column)
372 r->flags &= ~DFM_ADVANCE;
373 r->pos = MAX (column, 1) - 1;
376 /* Sets the current line to begin COLUMNS characters following
377 the current start. */
379 dfm_forward_columns (struct dfm_reader *r, size_t columns)
381 dfm_reread_record (r, (r->pos + 1) + columns);
384 /* Returns the 1-based column to which the line pointer in HANDLE
385 is set. Unless dfm_reread_record() or dfm_forward_columns()
386 have been called, this is 1. */
388 dfm_column_start (const struct dfm_reader *r)
393 /* Returns the number of columns we are currently beyond the end
394 of the line. At or before end-of-line, this is 0; one column
395 after end-of-line, this is 1; and so on. */
397 dfm_columns_past_end (const struct dfm_reader *r)
399 return r->pos < ds_length (&r->line) ? 0 : ds_length (&r->line) - r->pos;
402 /* Returns the 1-based column within the current line that P
405 dfm_get_column (const struct dfm_reader *r, const char *p)
407 return ds_pointer_to_position (&r->line, p) + 1;
410 /* Pushes the file name and line number on the fn/ln stack. */
412 dfm_push (struct dfm_reader *r)
414 if (r->fh != fh_inline_file ())
415 msg_push_msg_locator (&r->where);
418 /* Pops the file name and line number from the fn/ln stack. */
420 dfm_pop (struct dfm_reader *r)
422 if (r->fh != fh_inline_file ())
423 msg_pop_msg_locator (&r->where);
426 /* BEGIN DATA...END DATA procedure. */
428 /* Perform BEGIN DATA...END DATA as a procedure in itself. */
430 cmd_begin_data (struct lexer *lexer, struct dataset *ds)
432 struct dfm_reader *r;
435 if (!fh_is_open (fh_inline_file ()))
437 msg (SE, _("This command is not valid here since the current "
438 "input program does not access the inline file."));
439 return CMD_CASCADING_FAILURE;
442 /* Open inline file. */
443 r = dfm_open_reader (fh_inline_file (), lexer);
444 r->flags |= DFM_SAW_BEGIN_DATA;
446 /* Input procedure reads from inline file. */
447 prompt_set_style (PROMPT_DATA);
448 ok = procedure (ds, NULL, NULL);
450 dfm_close_reader (r);
452 return ok ? CMD_SUCCESS : CMD_CASCADING_FAILURE;