]> git.cameronkatri.com Git - mandoc.git/blobdiff - preconv.c
Make sure that the configuration file is always read, even when
[mandoc.git] / preconv.c
index a53504ece9e47783ac4c233014e631e06cd14d59..9ed627d4467d5a320f39c456f8b21c3c9280f0e7 100644 (file)
--- a/preconv.c
+++ b/preconv.c
@@ -1,6 +1,7 @@
-/*     $Id: preconv.c,v 1.4 2011/05/26 21:13:07 kristaps Exp $ */
+/*     $Id: preconv.c,v 1.17 2018/12/13 11:55:47 schwarze Exp $ */
 /*
  * Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
+ * Copyright (c) 2014 Ingo Schwarze <schwarze@openbsd.org>
  *
  * Permission to use, copy, modify, and distribute this software for any
  * purpose with or without fee is hereby granted, provided that the above
  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  */
-#ifdef HAVE_CONFIG_H
 #include "config.h"
-#endif
 
-#include <sys/stat.h>
-#include <sys/mman.h>
+#include <sys/types.h>
 
 #include <assert.h>
-#include <fcntl.h>
 #include <stdio.h>
-#include <stdlib.h>
 #include <string.h>
-#include <unistd.h>
 
-/* 
- * The read_whole_file() and resize_buf() functions are copied from
- * read.c, including all dependency code (MAP_FILE, etc.).
- */
-
-#ifndef MAP_FILE
-#define        MAP_FILE        0
-#endif
-
-enum   enc {
-       ENC_UTF_8, /* UTF-8 */
-       ENC_US_ASCII, /* US-ASCII */
-       ENC_LATIN_1, /* Latin-1 */
-       ENC__MAX
-};
-
-struct buf {
-       char             *buf; /* binary input buffer */
-       size_t            sz; /* size of binary buffer */
-       size_t            offs; /* starting buffer offset */
-};
-
-struct encode {
-       const char       *name;
-       int             (*conv)(const struct buf *);
-};
-
-static int      cue_enc(const struct buf *, size_t *, enum enc *);
-static int      conv_latin_1(const struct buf *);
-static int      conv_us_ascii(const struct buf *);
-static int      conv_utf_8(const struct buf *);
-static int      read_whole_file(const char *, int, 
-                       struct buf *, int *);
-static void     resize_buf(struct buf *, size_t);
-static void     usage(void);
-
-static const struct encode encs[ENC__MAX] = {
-       { "utf-8", conv_utf_8 }, /* ENC_UTF_8 */
-       { "us-ascii", conv_us_ascii }, /* ENC_US_ASCII */
-       { "latin-1", conv_latin_1 }, /* ENC_LATIN_1 */
-};
-
-static const char       *progname;
-
-static void
-usage(void)
-{
-
-       fprintf(stderr, "usage: %s "
-                       "[-D enc] "
-                       "[-e ENC] "
-                       "[file]\n", progname);
-}
-
-static int
-conv_latin_1(const struct buf *b)
-{
-       size_t           i;
-       unsigned char    cu;
-       const char      *cp;
-
-       cp = b->buf + (int)b->offs;
-
-       /*
-        * Latin-1 falls into the first 256 code-points of Unicode, so
-        * there's no need for any sort of translation.  Just make the
-        * 8-bit characters use the Unicode escape.
-        * Note that binary values 128 < v < 160 are passed through
-        * unmodified to mandoc.
-        */
-
-       for (i = b->offs; i < b->sz; i++) {
-               cu = (unsigned char)*cp++;
-               cu < 160U ? putchar(cu) : printf("\\[u%.4X]", cu);
-       }
+#include "mandoc.h"
+#include "roff.h"
+#include "mandoc_parse.h"
+#include "libmandoc.h"
 
-       return(1);
-}
-
-static int
-conv_us_ascii(const struct buf *b)
-{
-
-       /*
-        * US-ASCII has no conversion since it falls into the first 128
-        * bytes of Unicode.
-        */
-
-       fwrite(b->buf, 1, b->sz, stdout);
-       return(1);
-}
-
-static int
-conv_utf_8(const struct buf *b)
+int
+preconv_encode(const struct buf *ib, size_t *ii, struct buf *ob, size_t *oi,
+    int *filenc)
 {
-       int              state, be;
-       unsigned int     accum;
-       size_t           i;
-       unsigned char    cu;
-       const char      *cp;
-       const long       one = 1L;
-
-       cp = b->buf + (int)b->offs;
-       state = 0;
-       accum = 0U;
-       be = 0;
-
-       /* Quick test for big-endian value. */
-
-       if ( ! (*((const char *)(&one))))
-               be = 1;
-
-       for (i = b->offs; i < b->sz; i++) {
-               cu = (unsigned char)*cp++;
-               if (state) {
-                       if ( ! (cu & 128) || (cu & 64)) {
-                               /* Bad sequence header. */
-                               return(0);
-                       }
-
-                       /* Accept only legitimate bit patterns. */
-
-                       if (cu > 191 || cu < 128) {
-                               /* Bad in-sequence bits. */
-                               return(0);
-                       }
-
-                       accum |= (cu & 63) << --state * 6;
-
-                       /*
-                        * Accum is held in little-endian order as
-                        * stipulated by the UTF-8 sequence coding.  We
-                        * need to convert to a native big-endian if our
-                        * architecture requires it.
-                        */
-
-                       if (0 == state && be) 
-                               accum = (accum >> 24) | 
-                                       ((accum << 8) & 0x00FF0000) |
-                                       ((accum >> 8) & 0x0000FF00) |
-                                       (accum << 24);
-
-                       if (0 == state) {
-                               accum < 128U ? putchar(accum) : 
-                                       printf("\\[u%.4X]", accum);
-                               accum = 0U;
-                       }
-               } else if (cu & (1 << 7)) {
-                       /*
-                        * Entering a UTF-8 state:  if we encounter a
-                        * UTF-8 bitmask, calculate the expected UTF-8
-                        * state from it.
-                        */
-                       for (state = 0; state < 7; state++) 
-                               if ( ! (cu & (1 << (7 - state))))
-                                       break;
-
-                       /* Accept only legitimate bit patterns. */
-
-                       switch (state) {
-                       case (4):
-                               if (cu <= 244 && cu >= 240) {
-                                       accum = (cu & 7) << 18;
-                                       break;
-                               }
-                               /* Bad 4-sequence start bits. */
-                               return(0);
-                       case (3):
-                               if (cu <= 239 && cu >= 224) {
-                                       accum = (cu & 15) << 12;
-                                       break;
-                               }
-                               /* Bad 3-sequence start bits. */
-                               return(0);
-                       case (2):
-                               if (cu <= 223 && cu >= 194) {
-                                       accum = (cu & 31) << 6;
-                                       break;
-                               }
-                               /* Bad 2-sequence start bits. */
-                               return(0);
-                       default:
-                               /* Bad sequence bit mask. */
-                               return(0);
-                       }
-                       state--;
-               } else
-                       putchar(cu);
+       const unsigned char     *cu;
+       int                      nby;
+       unsigned int             accum;
+
+       cu = (const unsigned char *)ib->buf + *ii;
+       assert(*cu & 0x80);
+
+       if ( ! (*filenc & MPARSE_UTF8))
+               goto latin;
+
+       nby = 1;
+       while (nby < 5 && *cu & (1 << (7 - nby)))
+               nby++;
+
+       switch (nby) {
+       case 2:
+               accum = *cu & 0x1f;
+               if (accum < 0x02)  /* Obfuscated ASCII. */
+                       goto latin;
+               break;
+       case 3:
+               accum = *cu & 0x0f;
+               break;
+       case 4:
+               accum = *cu & 0x07;
+               if (accum > 0x04) /* Beyond Unicode. */
+                       goto latin;
+               break;
+       default:  /* Bad sequence header. */
+               goto latin;
        }
 
-       if (0 != state) {
-               /* Bad trailing bits. */
-               return(0);
+       cu++;
+       switch (nby) {
+       case 3:
+               if ((accum == 0x00 && ! (*cu & 0x20)) ||  /* Use 2-byte. */
+                   (accum == 0x0d && *cu & 0x20))  /* Surrogates. */
+                       goto latin;
+               break;
+       case 4:
+               if ((accum == 0x00 && ! (*cu & 0x30)) ||  /* Use 3-byte. */
+                   (accum == 0x04 && *cu & 0x30))  /* Beyond Unicode. */
+                       goto latin;
+               break;
+       default:
+               break;
        }
 
-       return(1);
-}
-
-static void
-resize_buf(struct buf *buf, size_t initial)
-{
-
-       buf->sz = buf->sz > initial / 2 ? 
-               2 * buf->sz : initial;
-
-       buf->buf = realloc(buf->buf, buf->sz);
-       if (NULL == buf->buf) {
-               perror(NULL);
-               exit(EXIT_FAILURE);
+       while (--nby) {
+               if ((*cu & 0xc0) != 0x80)  /* Invalid continuation. */
+                       goto latin;
+               accum <<= 6;
+               accum += *cu & 0x3f;
+               cu++;
        }
-}
 
-static int
-read_whole_file(const char *f, int fd, 
-               struct buf *fb, int *with_mmap)
-{
-       struct stat      st;
-       size_t           off;
-       ssize_t          ssz;
+       assert(accum > 0x7f);
+       assert(accum < 0x110000);
+       assert(accum < 0xd800 || accum > 0xdfff);
 
-       if (-1 == fstat(fd, &st)) {
-               perror(f);
-               return(0);
-       }
+       *oi += snprintf(ob->buf + *oi, 11, "\\[u%.4X]", accum);
+       *ii = (const char *)cu - ib->buf;
+       *filenc &= ~MPARSE_LATIN1;
+       return 1;
 
-       /*
-        * If we're a regular file, try just reading in the whole entry
-        * via mmap().  This is faster than reading it into blocks, and
-        * since each file is only a few bytes to begin with, I'm not
-        * concerned that this is going to tank any machines.
-        */
-
-       if (S_ISREG(st.st_mode) && st.st_size >= (1U << 31)) {
-               fprintf(stderr, "%s: input too large\n", f);
-               return(0);
-       } 
-       
-       if (S_ISREG(st.st_mode)) {
-               *with_mmap = 1;
-               fb->sz = (size_t)st.st_size;
-               fb->buf = mmap(NULL, fb->sz, PROT_READ, 
-                               MAP_FILE|MAP_SHARED, fd, 0);
-               if (fb->buf != MAP_FAILED)
-                       return(1);
-       }
+latin:
+       if ( ! (*filenc & MPARSE_LATIN1))
+               return 0;
 
-       /*
-        * If this isn't a regular file (like, say, stdin), then we must
-        * go the old way and just read things in bit by bit.
-        */
-
-       *with_mmap = 0;
-       off = 0;
-       fb->sz = 0;
-       fb->buf = NULL;
-       for (;;) {
-               if (off == fb->sz && fb->sz == (1U << 31)) {
-                       fprintf(stderr, "%s: input too large\n", f);
-                       break;
-               } 
-               
-               if (off == fb->sz)
-                       resize_buf(fb, 65536);
-
-               ssz = read(fd, fb->buf + (int)off, fb->sz - off);
-               if (ssz == 0) {
-                       fb->sz = off;
-                       return(1);
-               }
-               if (ssz == -1) {
-                       perror(f);
-                       break;
-               }
-               off += (size_t)ssz;
-       }
+       *oi += snprintf(ob->buf + *oi, 11,
+           "\\[u%.4X]", (unsigned char)ib->buf[(*ii)++]);
 
-       free(fb->buf);
-       fb->buf = NULL;
-       return(0);
+       *filenc &= ~MPARSE_UTF8;
+       return 1;
 }
 
-static int
-cue_enc(const struct buf *b, size_t *offs, enum enc *enc)
+int
+preconv_cue(const struct buf *b, size_t offset)
 {
        const char      *ln, *eoln, *eoph;
-       size_t           sz, phsz, nsz;
-       int              i;
+       size_t           sz, phsz;
 
-       ln = b->buf + (int)*offs;
-       sz = b->sz - *offs;
+       ln = b->buf + offset;
+       sz = b->sz - offset;
 
        /* Look for the end-of-line. */
 
        if (NULL == (eoln = memchr(ln, '\n', sz)))
-               return(-1);
-
-       /* Set next-line marker. */
-
-       *offs = (size_t)((eoln + 1) - b->buf);
+               eoln = ln + sz;
 
        /* Check if we have the correct header/trailer. */
 
-       if ((sz = (size_t)(eoln - ln)) < 10 || 
-                       memcmp(ln, ".\\\" -*-", 7) ||
-                       memcmp(eoln - 3, "-*-", 3))
-               return(0);
+       if ((sz = (size_t)(eoln - ln)) < 10 ||
+           memcmp(ln, ".\\\" -*-", 7) || memcmp(eoln - 3, "-*-", 3))
+               return MPARSE_UTF8 | MPARSE_LATIN1;
 
        /* Move after the header and adjust for the trailer. */
 
@@ -357,12 +150,12 @@ cue_enc(const struct buf *b, size_t *offs, enum enc *enc)
 
                /* Only account for the "coding" phrase. */
 
-               if ((phsz = (size_t)(eoph - ln)) < 7 ||
-                               strncasecmp(ln, "coding:", 7)) {
+               if ((phsz = eoph - ln) < 7 ||
+                   strncasecmp(ln, "coding:", 7)) {
                        sz -= phsz;
                        ln += phsz;
                        continue;
-               } 
+               }
 
                sz -= 7;
                ln += 7;
@@ -372,151 +165,15 @@ cue_enc(const struct buf *b, size_t *offs, enum enc *enc)
                        sz--;
                }
                if (0 == sz)
-                       break;
+                       return 0;
 
                /* Check us against known encodings. */
 
-               for (i = 0; i < (int)ENC__MAX; i++) {
-                       nsz = strlen(encs[i].name);
-                       if (phsz < nsz)
-                               continue;
-                       if (strncasecmp(ln, encs[i].name, nsz))
-                               continue;
-
-                       *enc = (enum enc)i;
-                       return(1);
-               }
-
-               /* Unknown encoding. */
-
-               *enc = ENC__MAX;
-               return(1);
+               if (phsz > 4 && !strncasecmp(ln, "utf-8", 5))
+                       return MPARSE_UTF8;
+               if (phsz > 10 && !strncasecmp(ln, "iso-latin-1", 11))
+                       return MPARSE_LATIN1;
+               return 0;
        }
-
-       return(0);
-}
-
-int
-main(int argc, char *argv[])
-{
-       int              i, ch, map, fd, rc;
-       struct buf       b;
-       const char      *fn;
-       enum enc         enc, def;
-       unsigned char    bom[3] = { 0xEF, 0xBB, 0xBF };
-       size_t           offs;
-       extern int       optind;
-       extern char     *optarg;
-
-       progname = strrchr(argv[0], '/');
-       if (progname == NULL)
-               progname = argv[0];
-       else
-               ++progname;
-
-       fn = "<stdin>";
-       fd = STDIN_FILENO;
-       rc = EXIT_FAILURE;
-       enc = def = ENC__MAX;
-       map = 0;
-
-       memset(&b, 0, sizeof(struct buf));
-
-       while (-1 != (ch = getopt(argc, argv, "D:e:rdvh")))
-               switch (ch) {
-               case ('D'):
-                       /* FALLTHROUGH */
-               case ('e'):
-                       for (i = 0; i < (int)ENC__MAX; i++) {
-                               if (strcasecmp(optarg, encs[i].name))
-                                       continue;
-                               break;
-                       }
-                       if (i < (int)ENC__MAX) {
-                               if ('D' == ch)
-                                       def = (enum enc)i;
-                               else
-                                       enc = (enum enc)i;
-                               break;
-                       }
-
-                       fprintf(stderr, "%s: Bad encoding\n", optarg);
-                       return(EXIT_FAILURE);
-               case ('r'):
-                       /* FALLTHROUGH */
-               case ('d'):
-                       /* FALLTHROUGH */
-               case ('v'):
-                       /* Compatibility with GNU preconv. */
-                       break;
-               case ('h'):
-                       /* Compatibility with GNU preconv. */
-                       /* FALLTHROUGH */
-               default:
-                       usage();
-                       return(EXIT_FAILURE);
-               }
-
-       argc -= optind;
-       argv += optind;
-       
-       /* 
-        * Open and read the first argument on the command-line.
-        * If we don't have one, we default to stdin.
-        */
-
-       if (argc > 0) {
-               fn = *argv;
-               fd = open(fn, O_RDONLY, 0);
-               if (-1 == fd) {
-                       perror(fn);
-                       return(EXIT_FAILURE);
-               }
-       }
-
-       if ( ! read_whole_file(fn, fd, &b, &map))
-               goto out;
-
-       /* Try to read the UTF-8 BOM. */
-
-       if (ENC__MAX == enc)
-               if (b.sz > 3 && 0 == memcmp(b.buf, bom, 3)) {
-                       b.offs = 3;
-                       enc = ENC_UTF_8;
-               }
-
-       /* Try reading from the "-*-" cue. */
-
-       if (ENC__MAX == enc) {
-               offs = b.offs;
-               ch = cue_enc(&b, &offs, &enc);
-               if (0 == ch)
-                       ch = cue_enc(&b, &offs, &enc);
-       }
-
-       /*
-        * No encoding has been detected.
-        * Thus, we either fall into our default encoder, if specified,
-        * or use Latin-1 if all else fails.
-        */
-
-       if (ENC__MAX == enc) 
-               enc = ENC__MAX == def ? ENC_LATIN_1 : def;
-
-       if ( ! (*encs[(int)enc].conv)(&b)) {
-               fprintf(stderr, "%s: Bad encoding\n", fn);
-               goto out;
-       }
-
-       rc = EXIT_SUCCESS;
-out:
-       if (map)
-               munmap(b.buf, b.sz);
-       else 
-               free(b.buf);
-
-       if (fd > STDIN_FILENO)
-               close(fd);
-
-       return(rc);
+       return MPARSE_UTF8 | MPARSE_LATIN1;
 }