1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 2010, 2011 Free Software Foundation, Inc.
4 This program is free software: you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation, either version 3 of the License, or
7 (at your option) any later version.
9 This program is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU General Public License for more details.
14 You should have received a copy of the GNU General Public License
15 along with this program. If not, see <http://www.gnu.org/licenses/>. */
19 #include "u8-istream.h"
33 #include "libpspp/assertion.h"
34 #include "libpspp/cast.h"
35 #include "libpspp/compiler.h"
36 #include "libpspp/encoding-guesser.h"
38 #include "gl/c-strcase.h"
39 #include "gl/localcharset.h"
40 #include "gl/minmax.h"
44 S_AUTO, /* Stream encoding not yet known. */
45 S_UTF8, /* Stream encoding is known to be UTF-8. */
46 S_CONVERT /* Stream encoding is known but not UTF-8. */
53 enum u8_istream_state state;
63 static ssize_t fill_buffer (struct u8_istream *);
65 /* Opens FILENAME, which is encoded in FROMCODE, for reading as an UTF-8
66 stream, passing FLAGS to the open() function. Returns a new u8_istream if
67 successful, otherwise returns NULL and sets errno to an appropriate value.
69 The accepted forms for FROMCODE are listed at the top of
70 encoding-guesser.h. */
72 u8_istream_for_file (const char *fromcode, const char *filename, int flags)
74 struct u8_istream *is;
77 assert (!(flags & O_CREAT));
79 fd = open (filename, flags);
83 is = u8_istream_for_fd (fromcode, fd);
86 int save_errno = errno;
94 /* Creates and returns a new u8_istream that reads its input from FD. Returns
95 a new u8_istream if successful, otherwise returns NULL and sets errno to an
98 The accepted forms for FROMCODE are listed at the top of
99 encoding-guesser.h. */
101 u8_istream_for_fd (const char *fromcode, int fd)
103 struct u8_istream *is;
104 const char *encoding;
106 is = malloc (sizeof *is);
111 is->converter = (iconv_t) -1;
112 is->buffer = malloc (U8_ISTREAM_BUFFER_SIZE);
113 if (is->buffer == NULL)
115 is->head = is->buffer;
119 if (fill_buffer (is) < 0)
122 encoding = encoding_guess_head_encoding (fromcode, is->buffer, is->length);
123 if (!strcmp (encoding, "UTF-8"))
127 if (encoding_guess_encoding_is_auto (fromcode)
128 && !strcmp (encoding, "ASCII"))
131 is->state = S_CONVERT;
133 is->converter = iconv_open ("UTF-8",
134 encoding_guess_parse_encoding (fromcode));
135 if (is->converter == (iconv_t) -1)
142 u8_istream_free (is);
146 /* Closes IS and its underlying file descriptor and frees all associated
147 resources. Returns the return value from close(). */
149 u8_istream_close (struct u8_istream *is)
154 u8_istream_free (is);
160 /* Frees IS and associated resources, but does not close the underlying file
161 descriptor. (Thus, the client must close the file descriptor when it is no
164 u8_istream_free (struct u8_istream *is)
168 if (is->converter != (iconv_t) -1)
169 iconv_close (is->converter);
176 substitute_invalid_input_byte (struct u8_istream *is)
178 assert (is->outlen == 0);
181 is->outlen = u8_uctomb (CHAR_CAST (uint8_t *, is->outbuf),
182 0xfffd, sizeof is->outbuf);
186 fill_buffer (struct u8_istream *is)
190 /* Move any unused bytes to the beginning of the input buffer. */
191 if (is->length > 0 && is->buffer != is->head)
192 memmove (is->buffer, is->head, is->length);
193 is->head = is->buffer;
195 /* Read more input. */
196 n = read (is->fd, is->buffer + is->length,
197 U8_ISTREAM_BUFFER_SIZE - is->length);
204 read_auto (struct u8_istream *is, char *buffer, size_t size)
206 size_t original_size = size;
215 n_ascii = encoding_guess_count_ascii (is->head,
216 MIN (is->length, size));
218 memcpy (buffer, is->head, n_ascii);
223 is->length -= n_ascii;
231 retval = fill_buffer (is);
238 /* is->head points to a byte that isn't a printable ASCII character.
239 Fill up the buffer and check for UTF-8. */
241 is->state = (encoding_guess_tail_is_utf8 (is->head, is->length)
242 ? S_UTF8 : S_CONVERT);
243 if (size == original_size)
244 return u8_istream_read (is, buffer, size);
248 return original_size - size;
252 convert_iconv (iconv_t converter,
253 char **inbufp, size_t *inbytesleft,
254 char **outbufp, size_t *outbytesleft)
256 size_t n = iconv (converter, inbufp, inbytesleft, outbufp, outbytesleft);
257 return n == SIZE_MAX ? errno : 0;
261 convert_utf8 (iconv_t converter UNUSED,
262 char **inbufp, size_t *inbytesleft,
263 char **outbufp, size_t *outbytesleft)
265 const uint8_t *in = CHAR_CAST (const uint8_t *, *inbufp);
266 size_t n = MIN (*inbytesleft, *outbytesleft);
277 error = ofs < *inbytesleft ? E2BIG : 0;
281 mblen = u8_mbtouc (&uc, in + ofs, n - ofs);
284 int retval = u8_mbtoucr (&uc, in + ofs, *inbytesleft - ofs);
287 /* There's an actual U+FFFD in the input stream. Carry on. */
291 error = (retval == -1 ? EILSEQ
292 : retval == -2 ? EINVAL
303 memcpy (*outbufp, *inbufp, ofs);
307 *outbytesleft -= ofs;
314 read_convert (struct u8_istream *is,
315 int (*convert) (iconv_t converter,
316 char **inbufp, size_t *inbytesleft,
317 char **outbufp, size_t *outbytesleft),
318 char *buffer, size_t size)
320 size_t original_size = size;
328 size_t n = MIN (size, is->outlen);
330 memcpy (buffer, is->outbuf, n);
333 memmove (is->outbuf, is->outbuf + n, is->outlen);
344 int error = convert (is->converter,
345 &is->head, &is->length,
353 /* Converted all of the input into output, possibly with space
354 for output left over.
360 substitute_invalid_input_byte (is);
364 /* Incomplete byte sequence at end of input. Read more
369 /* A real error of some kind (ENOMEM?). */
373 /* Ran out of room for output.
374 Convert into outbuf and copy from there instead. */
376 char *outptr = is->outbuf;
377 size_t outleft = sizeof is->outbuf;
379 error = convert (is->converter,
380 &is->head, &is->length,
382 is->outlen = outptr - is->outbuf;
389 substitute_invalid_input_byte (is);
397 /* A real error of some kind (ENOMEM?). */
404 assert (is->length <= MB_LEN_MAX);
405 n_read = fill_buffer (is);
408 if (original_size != size)
410 /* We produced some output so don't report EOF or error yet. */
413 else if (n_read == 0 && is->length != 0)
415 /* Incomplete byte sequence at end of file. */
416 substitute_invalid_input_byte (is);
420 /* Propagate end-of-file or error to caller. */
426 return original_size - size;
429 /* Reads up to SIZE bytes of UTF-8 text from IS into BUFFER. Returns the
430 number of bytes read if successful, 0 at end of file, or -1 if an error
431 occurred before any data could be read. Upon error, sets errno to an
432 appropriate value. */
434 u8_istream_read (struct u8_istream *is, char *buffer, size_t size)
439 return read_convert (is, convert_iconv, buffer, size);
442 return read_auto (is, buffer, size);
445 return read_convert (is, convert_utf8, buffer, size);
451 /* Returns the file descriptor underlying IS. */
453 u8_istream_fileno (const struct u8_istream *is)
460 These functions are probably useful only for white-box testing. */
462 /* Returns true if the encoding of the file being read by IS is not yet
465 u8_istream_is_auto (const struct u8_istream *is)
467 return is->state == S_AUTO;
470 /* Returns true if the encoding of the file being read by IS has been
471 determined to be UTF-8. */
473 u8_istream_is_utf8 (const struct u8_istream *is)
475 return is->state == S_UTF8;