-/* $Cambridge: exim/src/src/parse.c,v 1.6 2006/02/03 15:26:54 ph10 Exp $ */
-
/*************************************************
* Exim - an Internet mail transport agent *
*************************************************/
-/* Copyright (c) University of Cambridge 1995 - 2005 */
+/* Copyright (c) University of Cambridge 1995 - 2018 */
/* See the file NOTICE for conditions of use and distribution. */
/* Functions for parsing addresses */
make it possible to ignore comments at the end of compound items.
Argument: current character pointer
-Regurns: new character pointer
+Returns: new character pointer
*/
static uschar *
if (*s == '\"')
{
*t++ = '\"';
- while ((c = *(++s)) != 0 && c != '\"')
+ while ((c = *++s) && c != '\"')
{
*t++ = c;
- if (c == '\\' && s[1] != 0) *t++ = *(++s);
+ if (c == '\\' && s[1]) *t++ = *++s;
}
if (c == '\"')
{
else while (!mac_iscntrl_or_special(*s) || *s == '\\')
{
c = *t++ = *s++;
- if (c == '\\' && *s != 0) *t++ = *s++;
+ if (c == '\\' && *s) *t++ = *s++;
}
/* Terminate the word and skip subsequent comment */
{
s = read_local_part(s, t, errorptr, FALSE);
if (*errorptr == NULL)
- {
if (*s != term)
- {
if (*s != '@')
*errorptr = string_sprintf("\"@\" or \".\" expected after \"%s\"", t);
else
*domainptr = t;
s = read_domain(s, t, errorptr);
}
- }
- }
return s;
}
TRUE and parse_found_group is FALSE when this function is called, an address
which is the start of a group (i.e. preceded by a phrase and a colon) is
recognized; the phrase is ignored and the flag parse_found_group is set. If
-this flag is TRUE at the end of an address, then if an extraneous semicolon is
-found, it is ignored and the flag is cleared. This logic is used only when
-scanning through addresses in headers, either to fulfil the -t option or for
-rewriting or checking header syntax.
+this flag is TRUE at the end of an address, and if an extraneous semicolon is
+found, it is ignored and the flag is cleared.
+
+This logic is used only when scanning through addresses in headers, either to
+fulfil the -t option, or for rewriting, or for checking header syntax. Because
+the group "state" has to be remembered between multiple calls of this function,
+the variables parse_{allow,found}_group are global. It is important to ensure
+that they are reset to FALSE at the end of scanning a header's list of
+addresses.
Arguments:
mailbox points to the RFC822 mailbox
{
uschar *yield = store_get(Ustrlen(mailbox) + 1);
uschar *startptr, *endptr;
-uschar *s = (uschar *)mailbox;
-uschar *t = (uschar *)yield;
+uschar *s = US mailbox;
+uschar *t = US yield;
*domain = 0;
s = skip_comment(s);
startptr = s; /* In case addr-spec */
s = read_local_part(s, t, errorptr, TRUE); /* Dot separated words */
-if (*errorptr != NULL) goto PARSE_FAILED;
+if (*errorptr) goto PARSE_FAILED;
/* If the terminator is neither < nor @ then the format of the address
must either be a bare local-part (we are now at the end), or a phrase
end of string will produce a null local_part and therefore fail. We don't
need to keep updating t, as the phrase isn't to be kept. */
- while (*s != '<' && (!parse_allow_group || *s != ':'))
+ while (*s != '<' && (!f.parse_allow_group || *s != ':'))
{
s = read_local_part(s, t, errorptr, FALSE);
- if (*errorptr != NULL)
+ if (*errorptr)
{
*errorptr = string_sprintf("%s (expected word or \"<\")", *errorptr);
goto PARSE_FAILED;
if (*s == ':')
{
- parse_found_group = TRUE;
- parse_allow_group = FALSE;
+ f.parse_found_group = TRUE;
+ f.parse_allow_group = FALSE;
s++;
goto RESTART;
}
used after reading a preceding phrase.
There are a lot of broken sendmails out there that put additional pairs of <>
-round <route-addr>s. If strip_excess_angle_brackets is set, allow any number of
-them, as long as they match. */
+round <route-addr>s. If strip_excess_angle_brackets is set, allow a limited
+number of them, as long as they match. */
if (*s == '<')
{
int bracket_count = 1;
s++;
- if (strip_excess_angle_brackets)
- while (*s == '<') { bracket_count++; s++; }
+ if (strip_excess_angle_brackets) while (*s == '<')
+ {
+ if(bracket_count++ > 5) FAILED(US"angle-brackets nested too deep");
+ s++;
+ }
t = yield;
startptr = s;
if (*s == '@')
{
s = read_route(s, t, errorptr);
- if (*errorptr != NULL) goto PARSE_FAILED;
+ if (*errorptr) goto PARSE_FAILED;
*t = 0; /* Ensure route is ignored - probably overkill */
source_routed = TRUE;
}
else
{
s = read_addr_spec(s, t, '>', errorptr, &domainptr);
- if (*errorptr != NULL) goto PARSE_FAILED;
+ if (*errorptr) goto PARSE_FAILED;
*domain = domainptr - yield;
if (source_routed && *domain == 0)
FAILED(US"domain missing in source-routed address");
if (*errorptr != NULL) goto PARSE_FAILED;
while (bracket_count-- > 0) if (*s++ != '>')
{
- *errorptr = (s[-1] == 0)? US"'>' missing at end of address" :
- string_sprintf("malformed address: %.32s may not follow %.*s",
- s-1, s - (uschar *)mailbox - 1, mailbox);
+ *errorptr = s[-1] == 0
+ ? US"'>' missing at end of address"
+ : string_sprintf("malformed address: %.32s may not follow %.*s",
+ s-1, (int)(s - US mailbox - 1), mailbox);
goto PARSE_FAILED;
}
PARSE_SUCCEEDED:
if (*s != 0)
{
- if (parse_found_group && *s == ';')
+ if (f.parse_found_group && *s == ';')
{
- parse_found_group = FALSE;
- parse_allow_group = TRUE;
+ f.parse_found_group = FALSE;
+ f.parse_allow_group = TRUE;
}
else
{
*errorptr = string_sprintf("malformed address: %.32s may not follow %.*s",
- s, s - (uschar *)mailbox, mailbox);
+ s, (int)(s - US mailbox), mailbox);
goto PARSE_FAILED;
}
}
-*start = startptr - (uschar *)mailbox; /* Return offsets */
+*start = startptr - US mailbox; /* Return offsets */
while (isspace(endptr[-1])) endptr--;
-*end = endptr - (uschar *)mailbox;
+*end = endptr - US mailbox;
/* Although this code has no limitation on the length of address extracted,
other parts of Exim may have limits, and in any case, RFC 2821 limits local
return NULL;
}
-return (uschar *)yield;
+return yield;
/* Use goto (via the macro FAILED) to get to here from a variety of places.
We might have an empty address in a group - the caller can choose to ignore
this. We must, however, keep the flags correct. */
PARSE_FAILED:
-if (parse_found_group && *s == ';')
+if (f.parse_found_group && *s == ';')
{
- parse_found_group = FALSE;
- parse_allow_group = TRUE;
+ f.parse_found_group = FALSE;
+ f.parse_allow_group = TRUE;
}
return NULL;
}
charset the name of the character set; NULL => iso-8859-1
buffer the buffer to put the answer in
buffer_size the size of the buffer
+ fold if TRUE, a newline is inserted before the separating space when
+ more than one encoded-word is generated
Returns: pointer to the original string, if no quoting needed, or
pointer to buffer containing the quoted string, or
the introduction
*/
-uschar *
-parse_quote_2047(uschar *string, int len, uschar *charset, uschar *buffer,
- int buffer_size)
+const uschar *
+parse_quote_2047(const uschar *string, int len, uschar *charset, uschar *buffer,
+ int buffer_size, BOOL fold)
{
-uschar *s = string;
+const uschar *s = string;
uschar *p, *t;
int hlen;
BOOL coded = FALSE;
+BOOL first_byte = FALSE;
-if (charset == NULL) charset = US"iso-8859-1";
+if (!charset) charset = US"iso-8859-1";
/* We don't expect this to fail! */
int ch = *s++;
if (t > buffer + buffer_size - hlen - 8) break;
- if (t - p > 70)
+ if ((t - p > 67) && !first_byte)
{
*t++ = '?';
*t++ = '=';
+ if (fold) *t++ = '\n';
*t++ = ' ';
p = t;
Ustrncpy(p, buffer, hlen);
if (ch < 33 || ch > 126 ||
Ustrchr("?=()<>@,;:\\\".[]_", ch) != NULL)
{
- if (ch == ' ') *t++ = '_'; else
+ if (ch == ' ')
+ {
+ *t++ = '_';
+ first_byte = FALSE;
+ }
+ else
{
- sprintf(CS t, "=%02X", ch);
- while (*t != 0) t++;
+ t += sprintf(CS t, "=%02X", ch);
coded = TRUE;
+ first_byte = !first_byte;
}
}
- else *t++ = ch;
+ else { *t++ = ch; first_byte = FALSE; }
}
*t++ = '?';
*t++ = '=';
*t = 0;
-return coded? buffer : string;
+return coded ? buffer : string;
}
Returns: the fixed RFC822 phrase
*/
-uschar *
-parse_fix_phrase(uschar *phrase, int len, uschar *buffer, int buffer_size)
+const uschar *
+parse_fix_phrase(const uschar *phrase, int len, uschar *buffer, int buffer_size)
{
int ch, i;
BOOL quoted = FALSE;
-uschar *s, *t, *end, *yield;
+const uschar *s, *end;
+uschar *t, *yield;
while (len > 0 && isspace(*phrase)) { phrase++; len--; }
if (len > buffer_size/4) return US"Name too long";
if ((*s < 32 && *s != '\t') || *s > 126) break;
if (i < len) return parse_quote_2047(phrase, len, headers_charset, buffer,
- buffer_size);
+ buffer_size, FALSE);
/* No non-printers; use the RFC 822 quoting rules */
else if (ch == '(')
{
- uschar *ss = s; /* uschar after '(' */
+ const uschar *ss = s; /* uschar after '(' */
int level = 1;
while(ss < end)
{
int
parse_forward_list(uschar *s, int options, address_item **anchor,
- uschar **error, uschar *incoming_domain, uschar *directory,
+ uschar **error, const uschar *incoming_domain, uschar *directory,
error_block **syntax_errors)
{
int count = 0;
/* Check file name if required */
- if (directory != NULL)
+ if (directory)
{
int len = Ustrlen(directory);
uschar *p = filename + len;
return FF_ERROR;
}
+#ifdef EXIM_HAVE_OPENAT
+ /* It is necessary to check that every component inside the directory
+ is NOT a symbolic link, in order to keep the file inside the directory.
+ This is mighty tedious. We open the directory and openat every component,
+ with a flag that fails symlinks. */
+
+ {
+ int fd = open(CS directory, O_RDONLY);
+ if (fd < 0)
+ {
+ *error = string_sprintf("failed to open directory %s", directory);
+ return FF_ERROR;
+ }
+ while (*p)
+ {
+ uschar temp;
+ int fd2;
+ uschar * q = p;
+
+ while (*++p && *p != '/') ;
+ temp = *p;
+ *p = '\0';
+
+ fd2 = openat(fd, CS q, O_RDONLY|O_NOFOLLOW);
+ close(fd);
+ *p = temp;
+ if (fd2 < 0)
+ {
+ *error = string_sprintf("failed to open %s (component of included "
+ "file); could be symbolic link", filename);
+ return FF_ERROR;
+ }
+ fd = fd2;
+ }
+ f = fdopen(fd, "rb");
+ }
+#else
/* It is necessary to check that every component inside the directory
is NOT a symbolic link, in order to keep the file inside the directory.
This is mighty tedious. It is also not totally foolproof in that it
leaves the possibility of a race attack, but I don't know how to do
any better. */
- while (*p != 0)
+ while (*p)
{
int temp;
- while (*(++p) != 0 && *p != '/');
+ while (*++p && *p != '/');
temp = *p;
*p = 0;
if (Ulstat(filename, &statbuf) != 0)
return FF_ERROR;
}
}
+#endif
}
- /* Open and stat the file */
+#ifdef EXIM_HAVE_OPENAT
+ else
+#endif
+ /* Open and stat the file */
+ f = Ufopen(filename, "rb");
- if ((f = Ufopen(filename, "rb")) == NULL)
+ if (!f)
{
*error = string_open_failed(errno, "included file %s", filename);
return FF_INCLUDEFAIL;
/* If directory was checked, double check that we opened a regular file */
- if (directory != NULL && (statbuf.st_mode & S_IFMT) != S_IFREG)
+ if (directory && (statbuf.st_mode & S_IFMT) != S_IFREG)
{
*error = string_sprintf("included file %s is not a regular file in "
"the %s directory", filename, directory);
error, incoming_domain, directory, syntax_errors);
if (frc != FF_DELIVERED && frc != FF_NOTDELIVERED) return frc;
- if (addr != NULL)
+ if (addr)
{
- last = addr;
- while (last->next != NULL) { count++; last = last->next; }
+ for (last = addr; last->next; last = last->next) count++;
last->next = *anchor;
*anchor = addr;
count++;
}
+/*************************************************
+* Extract a Message-ID *
+*************************************************/
+
+/* This function is used to extract message ids from In-Reply-To: and
+References: header lines.
+
+Arguments:
+ str pointer to the start of the message-id
+ yield put pointer to the message id (in dynamic memory) here
+ error put error message here on failure
+
+Returns: points after the processed message-id or NULL on error
+*/
+
+uschar *
+parse_message_id(uschar *str, uschar **yield, uschar **error)
+{
+uschar *domain = NULL;
+uschar *id;
+
+str = skip_comment(str);
+if (*str != '<')
+ {
+ *error = US"Missing '<' before message-id";
+ return NULL;
+ }
+
+/* Getting a block the size of the input string will definitely be sufficient
+for the answer, but it may also be very long if we are processing a header
+line. Therefore, take care to release unwanted store afterwards. */
+
+id = *yield = store_get(Ustrlen(str) + 1);
+*id++ = *str++;
+
+str = read_addr_spec(str, id, '>', error, &domain);
+
+if (*error == NULL)
+ {
+ if (*str != '>') *error = US"Missing '>' after message-id";
+ else if (domain == NULL) *error = US"domain missing in message-id";
+ }
+
+if (*error != NULL)
+ {
+ store_reset(*yield);
+ return NULL;
+ }
+
+while (*id != 0) id++;
+*id++ = *str++;
+*id++ = 0;
+store_reset(id);
+
+str = skip_comment(str);
+return str;
+}
+
+
+/*************************************************
+* Parse a fixed digit number *
+*************************************************/
+
+/* Parse a string containing an ASCII encoded fixed digits number
+
+Arguments:
+ str pointer to the start of the ASCII encoded number
+ n pointer to the resulting value
+ digits number of required digits
+
+Returns: points after the processed date or NULL on error
+*/
+
+static uschar *
+parse_number(uschar *str, int *n, int digits)
+{
+ *n=0;
+ while (digits--)
+ {
+ if (*str<'0' || *str>'9') return NULL;
+ *n=10*(*n)+(*str++-'0');
+ }
+ return str;
+}
+
+
+/*************************************************
+* Parse a RFC 2822 day of week *
+*************************************************/
+
+/* Parse the day of the week from a RFC 2822 date, but do not
+ decode it, because it is only for humans.
+
+Arguments:
+ str pointer to the start of the day of the week
+
+Returns: points after the parsed day or NULL on error
+*/
+
+static uschar *
+parse_day_of_week(uschar *str)
+{
+/*
+day-of-week = ([FWS] day-name) / obs-day-of-week
+
+day-name = "Mon" / "Tue" / "Wed" / "Thu" /
+ "Fri" / "Sat" / "Sun"
+
+obs-day-of-week = [CFWS] day-name [CFWS]
+*/
+
+static const uschar *day_name[7]={ US"mon", US"tue", US"wed", US"thu", US"fri", US"sat", US"sun" };
+int i;
+uschar day[4];
+
+str=skip_comment(str);
+for (i=0; i<3; ++i)
+ {
+ if ((day[i]=tolower(*str))=='\0') return NULL;
+ ++str;
+ }
+day[3]='\0';
+for (i=0; i<7; ++i) if (Ustrcmp(day,day_name[i])==0) break;
+if (i==7) return NULL;
+str=skip_comment(str);
+return str;
+}
+
+
+/*************************************************
+* Parse a RFC 2822 date *
+*************************************************/
+
+/* Parse the date part of a RFC 2822 date-time, extracting the
+ day, month and year.
+
+Arguments:
+ str pointer to the start of the date
+ d pointer to the resulting day
+ m pointer to the resulting month
+ y pointer to the resulting year
+
+Returns: points after the processed date or NULL on error
+*/
+
+static uschar *
+parse_date(uschar *str, int *d, int *m, int *y)
+{
+/*
+date = day month year
+
+year = 4*DIGIT / obs-year
+
+obs-year = [CFWS] 2*DIGIT [CFWS]
+
+month = (FWS month-name FWS) / obs-month
+
+month-name = "Jan" / "Feb" / "Mar" / "Apr" /
+ "May" / "Jun" / "Jul" / "Aug" /
+ "Sep" / "Oct" / "Nov" / "Dec"
+
+obs-month = CFWS month-name CFWS
+
+day = ([FWS] 1*2DIGIT) / obs-day
+
+obs-day = [CFWS] 1*2DIGIT [CFWS]
+*/
+
+uschar *c,*n;
+static const uschar *month_name[]={ US"jan", US"feb", US"mar", US"apr", US"may", US"jun", US"jul", US"aug", US"sep", US"oct", US"nov", US"dec" };
+int i;
+uschar month[4];
+
+str=skip_comment(str);
+if ((str=parse_number(str,d,1))==NULL) return NULL;
+if (*str>='0' && *str<='9') *d=10*(*d)+(*str++-'0');
+c=skip_comment(str);
+if (c==str) return NULL;
+else str=c;
+for (i=0; i<3; ++i) if ((month[i]=tolower(*(str+i)))=='\0') return NULL;
+month[3]='\0';
+for (i=0; i<12; ++i) if (Ustrcmp(month,month_name[i])==0) break;
+if (i==12) return NULL;
+str+=3;
+*m=i;
+c=skip_comment(str);
+if (c==str) return NULL;
+else str=c;
+if ((n=parse_number(str,y,4)))
+ {
+ str=n;
+ if (*y<1900) return NULL;
+ *y=*y-1900;
+ }
+else if ((n=parse_number(str,y,2)))
+ {
+ str=skip_comment(n);
+ while (*(str-1)==' ' || *(str-1)=='\t') --str; /* match last FWS later */
+ if (*y<50) *y+=100;
+ }
+else return NULL;
+return str;
+}
+
+
+/*************************************************
+* Parse a RFC 2822 Time *
+*************************************************/
+
+/* Parse the time part of a RFC 2822 date-time, extracting the
+ hour, minute, second and timezone.
+
+Arguments:
+ str pointer to the start of the time
+ h pointer to the resulting hour
+ m pointer to the resulting minute
+ s pointer to the resulting second
+ z pointer to the resulting timezone (offset in seconds)
+
+Returns: points after the processed time or NULL on error
+*/
+
+static uschar *
+parse_time(uschar *str, int *h, int *m, int *s, int *z)
+{
+/*
+time = time-of-day FWS zone
+
+time-of-day = hour ":" minute [ ":" second ]
+
+hour = 2DIGIT / obs-hour
+
+obs-hour = [CFWS] 2DIGIT [CFWS]
+
+minute = 2DIGIT / obs-minute
+
+obs-minute = [CFWS] 2DIGIT [CFWS]
+
+second = 2DIGIT / obs-second
+
+obs-second = [CFWS] 2DIGIT [CFWS]
+
+zone = (( "+" / "-" ) 4DIGIT) / obs-zone
+
+obs-zone = "UT" / "GMT" / ; Universal Time
+ ; North American UT
+ ; offsets
+ "EST" / "EDT" / ; Eastern: - 5/ - 4
+ "CST" / "CDT" / ; Central: - 6/ - 5
+ "MST" / "MDT" / ; Mountain: - 7/ - 6
+ "PST" / "PDT" / ; Pacific: - 8/ - 7
+
+ %d65-73 / ; Military zones - "A"
+ %d75-90 / ; through "I" and "K"
+ %d97-105 / ; through "Z", both
+ %d107-122 ; upper and lower case
+*/
+
+uschar *c;
+
+str=skip_comment(str);
+if ((str=parse_number(str,h,2))==NULL) return NULL;
+str=skip_comment(str);
+if (*str!=':') return NULL;
+++str;
+str=skip_comment(str);
+if ((str=parse_number(str,m,2))==NULL) return NULL;
+c=skip_comment(str);
+if (*str==':')
+ {
+ ++str;
+ str=skip_comment(str);
+ if ((str=parse_number(str,s,2))==NULL) return NULL;
+ c=skip_comment(str);
+ }
+if (c==str) return NULL;
+else str=c;
+if (*str=='+' || *str=='-')
+ {
+ int neg;
+
+ neg=(*str=='-');
+ ++str;
+ if ((str=parse_number(str,z,4))==NULL) return NULL;
+ *z=(*z/100)*3600+(*z%100)*60;
+ if (neg) *z=-*z;
+ }
+else
+ {
+ char zone[5];
+ struct { const char *name; int off; } zone_name[10]=
+ { {"gmt",0}, {"ut",0}, {"est",-5}, {"edt",-4}, {"cst",-6}, {"cdt",-5}, {"mst",-7}, {"mdt",-6}, {"pst",-8}, {"pdt",-7}};
+ int i,j;
+
+ for (i=0; i<4; ++i)
+ {
+ zone[i]=tolower(*(str+i));
+ if (zone[i]<'a' || zone[i]>'z') break;
+ }
+ zone[i]='\0';
+ for (j=0; j<10 && strcmp(zone,zone_name[j].name); ++j);
+ /* Besides zones named in the grammar, RFC 2822 says other alphabetic */
+ /* time zones should be treated as unknown offsets. */
+ if (j<10)
+ {
+ *z=zone_name[j].off*3600;
+ str+=i;
+ }
+ else if (zone[0]<'a' || zone[1]>'z') return 0;
+ else
+ {
+ while ((*str>='a' && *str<='z') || (*str>='A' && *str<='Z')) ++str;
+ *z=0;
+ }
+ }
+return str;
+}
+
+
+/*************************************************
+* Parse a RFC 2822 date-time *
+*************************************************/
+
+/* Parse a RFC 2822 date-time and return it in seconds since the epoch.
+
+Arguments:
+ str pointer to the start of the date-time
+ t pointer to the parsed time
+
+Returns: points after the processed date-time or NULL on error
+*/
+
+uschar *
+parse_date_time(uschar *str, time_t *t)
+{
+/*
+date-time = [ day-of-week "," ] date FWS time [CFWS]
+*/
+
+struct tm tm;
+int zone;
+extern char **environ;
+char **old_environ;
+static char gmt0[]="TZ=GMT0";
+static char *gmt_env[]={ gmt0, (char*)0 };
+uschar *try;
+
+if ((try=parse_day_of_week(str)))
+ {
+ str=try;
+ if (*str!=',') return 0;
+ ++str;
+ }
+if ((str=parse_date(str,&tm.tm_mday,&tm.tm_mon,&tm.tm_year))==NULL) return NULL;
+if (*str!=' ' && *str!='\t') return NULL;
+while (*str==' ' || *str=='\t') ++str;
+if ((str=parse_time(str,&tm.tm_hour,&tm.tm_min,&tm.tm_sec,&zone))==NULL) return NULL;
+tm.tm_isdst=0;
+old_environ=environ;
+environ=gmt_env;
+*t=mktime(&tm);
+environ=old_environ;
+if (*t==-1) return NULL;
+*t-=zone;
+str=skip_comment(str);
+return str;
+}
+
+
+
+
/*************************************************
**************************************************
* Stand-alone test program *
printf("Testing parse_extract_address with group syntax\n");
-parse_allow_group = TRUE;
+f.parse_allow_group = TRUE;
while (Ufgets(buffer, sizeof(buffer), stdin) != NULL)
{
uschar *out;
else printf("Failed: %d %s\n", extracted, errmess);
}
+printf("Testing parse_message_id\n");
+
+while (Ufgets(buffer, sizeof(buffer), stdin) != NULL)
+ {
+ uschar *s, *t, *errmess;
+ buffer[Ustrlen(buffer) - 1] = 0;
+ if (buffer[0] == 0) break;
+ s = buffer;
+ while (*s != 0)
+ {
+ s = parse_message_id(s, &t, &errmess);
+ if (errmess != NULL)
+ {
+ printf("Failed: %s\n", errmess);
+ break;
+ }
+ printf("%s\n", t);
+ }
+ }
+
return 0;
}
git://git.exim.org / users / jgh / exim.git / blobdiff