. */
-static int
-vmsperms (const char *s)
-{
- int perms = 0;
- do
+
+/* Convert the VMS-style directory listing stored in "file" to a
+ linked list of fileinfo (system-independent) entries. The contents
+ of FILE are considered to be produced by the standard VMS
+ "DIRECTORY [/SIZE [= ALL]] /DATE [/OWNER] [/PROTECTION]" command,
+ more or less. (Different VMS FTP servers may have different headers,
+ and may not supply the same data, but all should be subsets of this.)
+
+ VMS normally provides local (server) time and date information.
+ Define the logical name or environment variable
+ "WGET_TIMEZONE_DIFFERENTIAL" (seconds) to adjust the receiving local
+ times if different from the remote local times.
+
+ 2005-02-23 SMS.
+ Added code to eliminate "^" escape characters from ODS5 extended file
+ names. The TCPIP FTP server (V5.4) seems to prefer requests which do
+ not use the escaped names which it provides.
+*/
+
+#define VMS_DEFAULT_PROT_FILE 0644
+#define VMS_DEFAULT_PROT_DIR 0755
+
+/* 2005-02-23 SMS.
+ eat_carets().
+
+ Delete ODS5 extended file name escape characters ("^") in the
+ original buffer.
+ Note that the current scheme does not handle all EFN cases, but it
+ could be made more complicated.
+*/
+
+static void eat_carets( char *str)
+/* char *str; Source pointer. */
+{
+ char *strd; /* Destination pointer. */
+ char hdgt;
+ unsigned char uchr;
+ unsigned char prop;
+
+ /* Skip ahead to the first "^", if any. */
+ while ((*str != '\0') && (*str != '^'))
+ str++;
+
+ /* If no caret was found, quit early. */
+ if (*str != '\0')
+ {
+ /* Shift characters leftward as carets are found. */
+ strd = str;
+ while (*str != '\0')
{
- switch (*s) {
- case ',': perms <<= 3; break;
- case 'R': perms |= 4; break;
- case 'W': perms |= 2; break;
- case 'D': perms |= 2; break;
- case 'E': perms |= 1; break;
- default: DEBUGP(("wrong VMS permissons!\n"));
+ uchr = *str;
+ if (uchr == '^')
+ {
+ /* Found a caret. Skip it, and check the next character. */
+ uchr = *(++str);
+ prop = char_prop[ uchr];
+ if (prop& 64)
+ {
+ /* Hex digit. Get char code from this and next hex digit. */
+ if (uchr <= '9')
+ {
+ hdgt = uchr- '0'; /* '0' - '9' -> 0 - 9. */
+ }
+ else
+ {
+ hdgt = ((uchr- 'A')& 7)+ 10; /* [Aa] - [Ff] -> 10 - 15. */
+ }
+ hdgt <<= 4; /* X16. */
+ uchr = *(++str); /* Next char must be hex digit. */
+ if (uchr <= '9')
+ {
+ uchr = hdgt+ uchr- '0';
+ }
+ else
+ {
+ uchr = hdgt+ ((uchr- 'A')& 15)+ 10;
+ }
+ }
+ else if (uchr == '_')
+ {
+ /* Convert escaped "_" to " ". */
+ uchr = ' ';
+ }
+ else if (uchr == '/')
+ {
+ /* Convert escaped "/" (invalid Zip) to "?" (invalid VMS). */
+ /* Note that this is a left-over from Info-ZIP code, and is
+ probably of little value here, except perhaps to avoid
+ directory confusion which an unconverted slash might cause.
+ */
+ uchr = '?';
+ }
+ /* Else, not a hex digit. Must be a simple escaped character
+ (or Unicode, which is not yet handled here).
+ */
}
+ /* Else, not a caret. Use as-is. */
+ *strd = uchr;
+
+ /* Advance destination and source pointers. */
+ strd++;
+ str++;
}
- while (*++s);
- return perms;
+ /* Terminate the destination string. */
+ *strd = '\0';
+ }
}
@@ -586,20 +680,17 @@ static struct fileinfo *
ftp_parse_vms_ls (const char *file)
{
FILE *fp;
- /* #### A third copy of more-or-less the same array ? */
- static const char *months[] = {
- "JAN", "FEB", "MAR", "APR", "MAY", "JUN",
- "JUL", "AUG", "SEP", "OCT", "NOV", "DEC"
- };
- int i;
- int year, month, day; /* for time analysis */
- int hour, min, sec;
- struct tm timestruct;
+ int dt, i, j, len;
+ int perms;
+ size_t bufsize = 0;
+ time_t timenow;
+ struct tm *timestruct;
+ char date_str[ 32];
- char *line, *tok; /* tokenizer */
+ char *line = NULL, *tok; /* tokenizer */
struct fileinfo *dir, *l, cur; /* list creation */
- fp = fopen (file, "rb");
+ fp = fopen (file, "r");
if (!fp)
{
logprintf (LOG_NOTQUIET, "%s: %s\n", file, strerror (errno));
@@ -607,190 +698,311 @@ ftp_parse_vms_ls (const char *file)
}
dir = l = NULL;
- /* Skip empty line. */
- line = read_whole_line (fp);
- xfree_null (line);
-
- /* Skip "Directory PUB$DEVICE[PUB]" */
- line = read_whole_line (fp);
- xfree_null (line);
-
- /* Skip empty line. */
- line = read_whole_line (fp);
- xfree_null (line);
+ /* Skip blank lines, Directory heading, and more blank lines. */
- /* Line loop to end of file: */
- while ((line = read_whole_line (fp)) != NULL)
+ for (j = 0; (i = getline (&line, &bufsize, fp)) > 0; )
{
- char *p;
- i = clean_line (line);
- if (!i)
+ i = clean_line (line, i);
+ if (i <= 0)
+ continue; /* Ignore blank line. */
+
+ if ((j == 0) && (line[i - 1] == ']'))
+ {
+ /* Found Directory heading line. Next non-blank line
+ is significant. */
+ j = 1;
+ }
+ else if (!strncmp (line, "Total of ", 9))
{
- xfree (line);
+ /* Found "Total of ..." footing line. No valid data
+ will follow (empty directory). */
+ i = 0; /* Arrange for early exit. */
break;
}
+ else
+ {
+ break; /* Must be significant data. */
+ }
+ }
+
+ /* Read remainder of file until the next blank line or EOF. */
- /* First column: Name. A bit of black magic again. The name my be
- either ABCD.EXT or ABCD.EXT;NUM and it might be on a separate
- line. Therefore we will first try to get the complete name
- until the first space character; if it fails, we assume that the name
- occupies the whole line. After that we search for the version
- separator ";", we remove it and check the extension of the file;
- extension .DIR denotes directory. */
+ while (i > 0)
+ {
+ char *p;
+
+ /* The first token is the file name. After a long name, other
+ data may be on the following line. A valid directory name ends
+ in ".DIR;1" (any case), although some VMS FTP servers may omit
+ the version number (";1").
+ */
tok = strtok(line, " ");
if (tok == NULL) tok = line;
- DEBUGP(("file name: '%s'\n", tok));
- for (p = tok ; *p && *p != ';' ; p++)
- ;
- if (*p == ';') *p = '\0';
- p = tok + strlen(tok) - 4;
- if (!strcmp(p, ".DIR")) *p = '\0';
- cur.name = xstrdup(tok);
- DEBUGP(("Name: '%s'\n", cur.name));
-
- /* If the name ends on .DIR or .DIR;#, it's a directory. We also set
- the file size to zero as the listing does tell us only the size in
- filesystem blocks - for an integrity check (when mirroring, for
- example) we would need the size in bytes. */
-
- if (! *p)
+ DEBUGP (("file name: '%s'\n", tok));
+
+ /* Stripping the version number on a VMS system would be wrong.
+ It may be foolish on a non-VMS system, too, but that's someone
+ else's problem. (Define PRESERVE_VMS_VERSIONS for proper
+ operation on other operating systems.)
+
+ 2005-02-23 SMS.
+ ODS5 extended file names may contain escaped semi-colons, so
+ the version number is identified as right-side decimal digits
+ led by a non-escaped semi-colon. It may be absent.
+ */
+
+#if (!defined( __VMS) && !defined( PRESERVE_VMS_VERSIONS))
+ for (p = tok + strlen (tok); (--p > tok) && c_isdigit( *p); );
+ if ((*p == ';') && (*(p- 1) != '^'))
+ {
+ *p = '\0';
+ }
+#endif /* (!defined( __VMS) && !defined( PRESERVE_VMS_VERSIONS)) */
+
+ /* 2005-02-23 SMS.
+ Eliminate "^" escape characters from ODS5 extended file name.
+ (A caret is invalid in an ODS2 name, so this is always safe.)
+ */
+ eat_carets (tok);
+ DEBUGP (("file name-^: '%s'\n", tok));
+
+ /* Differentiate between a directory and any other file. A VMS
+ listing may not include file protections (permissions). Set a
+ default permissions value (according to the file type), which
+ may be overwritten later. Store directory names without the
+ ".DIR;1" file type and version number, as the plain name is
+ what will work in a CWD command.
+ */
+ len = strlen (tok);
+ if (!strncasecmp((tok + (len - 4)), ".DIR", 4))
{
+ *(tok+ (len - 4)) = '\0'; /* Discard ".DIR". */
cur.type = FT_DIRECTORY;
- cur.size = 0;
- DEBUGP(("Directory\n"));
+ cur.perms = VMS_DEFAULT_PROT_DIR;
+ DEBUGP (("Directory (nv)\n"));
+ }
+ else if (!strncasecmp ((tok + (len - 6)), ".DIR;1", 6))
+ {
+ *(tok+ (len - 6)) = '\0'; /* Discard ".DIR;1". */
+ cur.type = FT_DIRECTORY;
+ cur.perms = VMS_DEFAULT_PROT_DIR;
+ DEBUGP (("Directory (v)\n"));
}
else
{
cur.type = FT_PLAINFILE;
- DEBUGP(("File\n"));
+ cur.perms = VMS_DEFAULT_PROT_FILE;
+ DEBUGP (("File\n"));
}
+ cur.name = xstrdup (tok);
+ DEBUGP (("Name: '%s'\n", cur.name));
- cur.size = 0;
+ /* Null the date and time string. */
+ *date_str = '\0';
- /* Second column, if exists, or the first column of the next line
- contain file size in blocks. We will skip it. */
+ /* VMS lacks symbolic links. */
+ cur.linkto = NULL;
- tok = strtok(NULL, " ");
- if (tok == NULL)
- {
- DEBUGP(("Getting additional line\n"));
- xfree (line);
- line = read_whole_line (fp);
- if (!line)
- {
- DEBUGP(("empty line read, leaving listing parser\n"));
- break;
- }
- i = clean_line (line);
- if (!i)
- {
- DEBUGP(("confusing VMS listing item, leaving listing parser\n"));
- xfree (line);
- break;
- }
- tok = strtok(line, " ");
- }
- DEBUGP(("second token: '%s'\n", tok));
+ /* VMS reports file sizes in (512-byte) disk blocks, not bytes,
+ hence useless for an integrity check based on byte-count.
+ Set size to unknown.
+ */
+ cur.size = 0;
- /* Third/Second column: Date DD-MMM-YYYY. */
+ /* Get token 2, if any. A long name may force all other data onto
+ a second line. If needed, read the second line.
+ */
- tok = strtok(NULL, "-");
- if (tok == NULL) continue;
- DEBUGP(("day: '%s'\n",tok));
- day = atoi(tok);
- tok = strtok(NULL, "-");
- if (!tok)
- {
- /* If the server produces garbage like
- 'EA95_0PS.GZ;1 No privilege for attempted operation'
- the first strtok(NULL, "-") will return everything until the end
- of the line and only the next strtok() call will return NULL. */
- DEBUGP(("nonsense in VMS listing, skipping this line\n"));
- xfree (line);
- break;
- }
- for (i=0; i<12; i++) if (!strcmp(tok,months[i])) break;
- /* Uknown months are mapped to January */
- month = i % 12 ;
tok = strtok (NULL, " ");
- if (tok == NULL) continue;
- year = atoi (tok) - 1900;
- DEBUGP(("date parsed\n"));
+ if (tok == NULL)
+ {
+ DEBUGP (("Getting additional line.\n"));
+ i = getline (&line, &bufsize, fp);
+ if (i <= 0)
+ {
+ DEBUGP (("EOF. Leaving listing parser.\n"));
+ break;
+ }
- /* Fourth/Third column: Time hh:mm[:ss] */
- tok = strtok (NULL, " ");
- if (tok == NULL) continue;
- min = sec = 0;
- p = tok;
- hour = atoi (p);
- for (; *p && *p != ':'; ++p)
- ;
- if (*p)
- min = atoi (++p);
- for (; *p && *p != ':'; ++p)
- ;
- if (*p)
- sec = atoi (++p);
-
- DEBUGP(("YYYY/MM/DD HH:MM:SS - %d/%02d/%02d %02d:%02d:%02d\n",
- year+1900, month, day, hour, min, sec));
-
- /* Build the time-stamp (copy & paste from above) */
- timestruct.tm_sec = sec;
- timestruct.tm_min = min;
- timestruct.tm_hour = hour;
- timestruct.tm_mday = day;
- timestruct.tm_mon = month;
- timestruct.tm_year = year;
- timestruct.tm_wday = 0;
- timestruct.tm_yday = 0;
- timestruct.tm_isdst = -1;
- cur.tstamp = mktime (×truct); /* store the time-stamp */
+ /* Second line must begin with " ". Otherwise, it's a first
+ line (and we may be confused).
+ */
+ i = clean_line (line, i);
+ if (i <= 0)
+ {
+ /* Blank line. End of significant file listing. */
+ DEBUGP (("Blank line. Leaving listing parser.\n"));
+ break;
+ }
+ else if (line[0] != ' ')
+ {
+ DEBUGP (("Non-blank in column 1. Must be a new file name?\n"));
+ continue;
+ }
+ else
+ {
+ tok = strtok (line, " ");
+ if (tok == NULL)
+ {
+ /* Unexpected non-empty but apparently blank line. */
+ DEBUGP (("Null token. Leaving listing parser.\n"));
+ break;
+ }
+ }
+ }
- DEBUGP(("Timestamp: %ld\n", cur.tstamp));
+ /* Analyze tokens. (Order is not significant, except date must
+ precede time.)
- /* Skip the fifth column */
+ Size: ddd or ddd/ddd (where "ddd" is a decimal number)
+ Date: DD-MMM-YYYY
+ Time: HH:MM or HH:MM:SS or HH:MM:SS.CC
+ Owner: [user] or [user,group]
+ Protection: (ppp,ppp,ppp,ppp) (where "ppp" is "RWED" or some
+ subset thereof, for System, Owner, Group, World.
- tok = strtok(NULL, " ");
- if (tok == NULL) continue;
+ If permission is lacking, info may be replaced by the string:
+ "No privilege for attempted operation".
+ */
+ while (tok != NULL)
+ {
+ DEBUGP (("Token: >%s<: ", tok));
- /* Sixth column: Permissions */
+ if ((strlen (tok) < 12) && (strchr( tok, '-') != NULL))
+ {
+ /* Date. */
+ DEBUGP (("Date.\n"));
+ strcpy( date_str, tok);
+ strcat( date_str, " ");
+ }
+ else if ((strlen (tok) < 12) && (strchr( tok, ':') != NULL))
+ {
+ /* Time. */
+ DEBUGP (("Time. "));
+ strncat( date_str,
+ tok,
+ (sizeof( date_str)- strlen (date_str) - 1));
+ DEBUGP (("Date time: >%s<\n", date_str));
+ }
+ else if (strchr (tok, '[') != NULL)
+ {
+ /* Owner. (Ignore.) */
+ DEBUGP (("Owner.\n"));
+ }
+ else if (strchr (tok, '(') != NULL)
+ {
+ /* Protections (permissions). */
+ perms = 0;
+ j = 0;
+ /*FIXME: Should not be using the variable like this. */
+ for (i = 0; i < (int) strlen(tok); i++)
+ {
+ switch (tok[ i])
+ {
+ case '(':
+ break;
+ case ')':
+ break;
+ case ',':
+ if (j == 0)
+ {
+ perms = 0;
+ j = 1;
+ }
+ else
+ {
+ perms <<= 3;
+ }
+ break;
+ case 'R':
+ perms |= 4;
+ break;
+ case 'W':
+ perms |= 2;
+ break;
+ case 'E':
+ perms |= 1;
+ break;
+ case 'D':
+ perms |= 2;
+ break;
+ }
+ }
+ cur.perms = perms;
+ DEBUGP (("Prot. perms = %0o.\n", cur.perms));
+ }
+ else
+ {
+ /* Nondescript. Probably size(s), probably in blocks.
+ Could be "No privilege ..." message. (Ignore.)
+ */
+ DEBUGP (("Ignored (size?).\n"));
+ }
- tok = strtok(NULL, ","); /* Skip the VMS-specific SYSTEM permissons */
- if (tok == NULL) continue;
- tok = strtok(NULL, ")");
- if (tok == NULL)
+ tok = strtok (NULL, " ");
+ }
+
+ /* Tokens exhausted. Interpret the data, and fill in the
+ structure.
+ */
+ /* Fill tm timestruct according to date-time string. Fractional
+ seconds are ignored. Default to current time, if conversion
+ fails.
+ */
+ timenow = time( NULL);
+ timestruct = localtime( &timenow );
+ strptime( date_str, "%d-%b-%Y %H:%M:%S", timestruct);
+
+ /* Convert struct tm local time to time_t local time. */
+ timenow = mktime (timestruct);
+ /* Offset local time according to environment variable (seconds). */
+ if ((tok = getenv ( "WGET_TIMEZONE_DIFFERENTIAL")) != NULL)
{
- DEBUGP(("confusing VMS permissions, skipping line\n"));
- xfree (line);
- continue;
+ dt = atoi (tok);
+ DEBUGP (("Time differential = %d.\n", dt));
}
- /* Permissons have the format "RWED,RWED,RE" */
- cur.perms = vmsperms(tok);
- DEBUGP(("permissions: %s -> 0%o\n", tok, cur.perms));
+ else
+ dt = 0;
- cur.linkto = NULL;
+ if (dt >= 0)
+ timenow += dt;
+ else
+ timenow -= (-dt);
- /* And put everything into the linked list */
+ cur.tstamp = timenow; /* Store the time-stamp. */
+ DEBUGP (("Timestamp: %ld\n", cur.tstamp));
+ cur.ptype = TT_HOUR_MIN;
+
+ /* Add the data for this item to the linked list, */
if (!dir)
{
- l = dir = xnew (struct fileinfo);
+ l = dir = (struct fileinfo *)xmalloc (sizeof (struct fileinfo));
memcpy (l, &cur, sizeof (cur));
l->prev = l->next = NULL;
}
else
{
cur.prev = l;
- l->next = xnew (struct fileinfo);
+ l->next = (struct fileinfo *)xmalloc (sizeof (struct fileinfo));
l = l->next;
memcpy (l, &cur, sizeof (cur));
l->next = NULL;
}
- xfree (line);
+ i = getline (&line, &bufsize, fp);
+ if (i > 0)
+ {
+ i = clean_line (line, i);
+ if (i <= 0)
+ {
+ /* Blank line. End of significant file listing. */
+ break;
+ }
+ }
}
+ xfree (line);
fclose (fp);
return dir;
}
@@ -850,7 +1062,9 @@ ftp_index (const char *file, struct url *u, struct fileinfo *f)
{
FILE *fp;
char *upwd;
+ char *htcldir; /* HTML-clean dir name */
char *htclfile; /* HTML-clean file name */
+ char *urlclfile; /* URL-clean file name */
if (!output_stream)
{
@@ -878,12 +1092,16 @@ ftp_index (const char *file, struct url *u, struct fileinfo *f)
}
else
upwd = xstrdup ("");
+
+ htcldir = html_quote_string (u->dir);
+
fprintf (fp, "\n");
fprintf (fp, "\n\n");
- fprintf (fp, _("Index of /%s on %s:%d"), u->dir, u->host, u->port);
+ fprintf (fp, _("Index of /%s on %s:%d"), htcldir, u->host, u->port);
fprintf (fp, "\n\n\n");
- fprintf (fp, _("Index of /%s on %s:%d"), u->dir, u->host, u->port);
+ fprintf (fp, _("Index of /%s on %s:%d"), htcldir, u->host, u->port);
fprintf (fp, "
\n
\n\n");
+
while (f)
{
fprintf (fp, " ");
@@ -895,11 +1113,12 @@ ftp_index (const char *file, struct url *u, struct fileinfo *f)
"Jan", "Feb", "Mar", "Apr", "May", "Jun",
"Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
};
- struct tm *ptm = localtime ((time_t *)&f->tstamp);
+ time_t tstamp = f->tstamp;
+ struct tm *ptm = localtime (&tstamp);
fprintf (fp, "%d %s %02d ", ptm->tm_year + 1900, months[ptm->tm_mon],
ptm->tm_mday);
- if (ptm->tm_hour)
+ if (f->ptype == TT_HOUR_MIN)
fprintf (fp, "%02d:%02d ", ptm->tm_hour, ptm->tm_min);
else
fprintf (fp, " ");
@@ -922,13 +1141,18 @@ ftp_index (const char *file, struct url *u, struct fileinfo *f)
break;
}
htclfile = html_quote_string (f->name);
+ urlclfile = url_escape_unsafe_and_reserved (f->name);
fprintf (fp, "host, u->port);
if (*u->dir != '/')
putc ('/', fp);
- fprintf (fp, "%s", u->dir);
+ /* XXX: Should probably URL-escape dir components here, rather
+ * than just HTML-escape, for consistency with the next bit where
+ * we use urlclfile for the file component. Anyway, this is safer
+ * than what we had... */
+ fprintf (fp, "%s", htcldir);
if (*u->dir)
putc ('/', fp);
- fprintf (fp, "%s", htclfile);
+ fprintf (fp, "%s", urlclfile);
if (f->type == FT_DIRECTORY)
putc ('/', fp);
fprintf (fp, "\">%s", htclfile);
@@ -941,9 +1165,11 @@ ftp_index (const char *file, struct url *u, struct fileinfo *f)
fprintf (fp, "-> %s", f->linkto ? f->linkto : "(nil)");
putc ('\n', fp);
xfree (htclfile);
+ xfree (urlclfile);
f = f->next;
}
fprintf (fp, "\n\n\n");
+ xfree (htcldir);
xfree (upwd);
if (!output_stream)
fclose (fp);