X-Git-Url: https://vcs.fsf.org/?p=exim.git;a=blobdiff_plain;f=src%2Fsrc%2Fexpand.c;h=ec4dd71f94392407f12a78208b6af1c107825871;hp=1517c0625726b102b3fc07b69094613878b32ad1;hb=ac53fcdaf9c772ee8e70ca4f14ed19b39e12eb68;hpb=d45b1de81aa47cef4ca098a4b2725d855b154162 diff --git a/src/src/expand.c b/src/src/expand.c index 1517c0625..ec4dd71f9 100644 --- a/src/src/expand.c +++ b/src/src/expand.c @@ -1,10 +1,8 @@ -/* $Cambridge: exim/src/src/expand.c,v 1.60 2006/09/18 14:49:23 ph10 Exp $ */ - /************************************************* * Exim - an Internet mail transport agent * *************************************************/ -/* Copyright (c) University of Cambridge 1995 - 2006 */ +/* Copyright (c) University of Cambridge 1995 - 2009 */ /* See the file NOTICE for conditions of use and distribution. */ @@ -13,12 +11,20 @@ #include "exim.h" +/* Recursively called function */ + +static uschar *expand_string_internal(uschar *, BOOL, uschar **, BOOL); + #ifdef STAND_ALONE #ifndef SUPPORT_CRYPTEQ #define SUPPORT_CRYPTEQ #endif #endif +#ifdef LOOKUP_LDAP +#include "lookups/ldap.h" +#endif + #ifdef SUPPORT_CRYPTEQ #ifdef CRYPT_H #include @@ -28,15 +34,63 @@ extern char* crypt16(char*, char*); #endif #endif -#ifdef LOOKUP_LDAP -#include "lookups/ldap.h" -#endif - - - -/* Recursively called function */ +/* The handling of crypt16() is a mess. I will record below the analysis of the +mess that was sent to me. We decided, however, to make changing this very low +priority, because in practice people are moving away from the crypt() +algorithms nowadays, so it doesn't seem worth it. + + +There is an algorithm named "crypt16" in Ultrix and Tru64. It crypts +the first 8 characters of the password using a 20-round version of crypt +(standard crypt does 25 rounds). It then crypts the next 8 characters, +or an empty block if the password is less than 9 characters, using a +20-round version of crypt and the same salt as was used for the first +block. Charaters after the first 16 are ignored. It always generates +a 16-byte hash, which is expressed together with the salt as a string +of 24 base 64 digits. Here are some links to peruse: + + http://cvs.pld.org.pl/pam/pamcrypt/crypt16.c?rev=1.2 + http://seclists.org/bugtraq/1999/Mar/0076.html + +There's a different algorithm named "bigcrypt" in HP-UX, Digital Unix, +and OSF/1. This is the same as the standard crypt if given a password +of 8 characters or less. If given more, it first does the same as crypt +using the first 8 characters, then crypts the next 8 (the 9th to 16th) +using as salt the first two base 64 digits from the first hash block. +If the password is more than 16 characters then it crypts the 17th to 24th +characters using as salt the first two base 64 digits from the second hash +block. And so on: I've seen references to it cutting off the password at +40 characters (5 blocks), 80 (10 blocks), or 128 (16 blocks). Some links: + + http://cvs.pld.org.pl/pam/pamcrypt/bigcrypt.c?rev=1.2 + http://seclists.org/bugtraq/1999/Mar/0109.html + http://h30097.www3.hp.com/docs/base_doc/DOCUMENTATION/HTML/AA-Q0R2D- + TET1_html/sec.c222.html#no_id_208 + +Exim has something it calls "crypt16". It will either use a native +crypt16 or its own implementation. A native crypt16 will presumably +be the one that I called "crypt16" above. The internal "crypt16" +function, however, is a two-block-maximum implementation of what I called +"bigcrypt". The documentation matches the internal code. + +I suspect that whoever did the "crypt16" stuff for Exim didn't realise +that crypt16 and bigcrypt were different things. + +Exim uses the LDAP-style scheme identifier "{crypt16}" to refer +to whatever it is using under that name. This unfortunately sets a +precedent for using "{crypt16}" to identify two incompatible algorithms +whose output can't be distinguished. With "{crypt16}" thus rendered +ambiguous, I suggest you deprecate it and invent two new identifiers +for the two algorithms. + +Both crypt16 and bigcrypt are very poor algorithms, btw. Hashing parts +of the password separately means they can be cracked separately, so +the double-length hash only doubles the cracking effort instead of +squaring it. I recommend salted SHA-1 ({SSHA}), or the Blowfish-based +bcrypt ({CRYPT}$2a$). + +*/ -static uschar *expand_string_internal(uschar *, BOOL, uschar **, BOOL); @@ -50,17 +104,20 @@ alphabetical order. */ static uschar *item_table[] = { US"dlfunc", US"extract", + US"filter", US"hash", US"hmac", US"if", US"length", US"lookup", + US"map", US"nhash", US"perl", US"prvs", US"prvscheck", US"readfile", US"readsocket", + US"reduce", US"run", US"sg", US"substr", @@ -69,17 +126,20 @@ static uschar *item_table[] = { enum { EITEM_DLFUNC, EITEM_EXTRACT, + EITEM_FILTER, EITEM_HASH, EITEM_HMAC, EITEM_IF, EITEM_LENGTH, EITEM_LOOKUP, + EITEM_MAP, EITEM_NHASH, EITEM_PERL, EITEM_PRVS, EITEM_PRVSCHECK, EITEM_READFILE, EITEM_READSOCK, + EITEM_REDUCE, EITEM_RUN, EITEM_SG, EITEM_SUBSTR, @@ -94,6 +154,7 @@ static uschar *op_table_underscore[] = { US"from_utf8", US"local_part", US"quote_local_part", + US"reverse_ip", US"time_eval", US"time_interval"}; @@ -101,11 +162,13 @@ enum { EOP_FROM_UTF8, EOP_LOCAL_PART, EOP_QUOTE_LOCAL_PART, + EOP_REVERSE_IP, EOP_TIME_EVAL, EOP_TIME_INTERVAL }; static uschar *op_table_main[] = { US"address", + US"addresses", US"base62", US"base62d", US"domain", @@ -124,7 +187,9 @@ static uschar *op_table_main[] = { US"nh", US"nhash", US"quote", + US"randint", US"rfc2047", + US"rfc2047d", US"rxquote", US"s", US"sha1", @@ -136,6 +201,7 @@ static uschar *op_table_main[] = { enum { EOP_ADDRESS = sizeof(op_table_underscore)/sizeof(uschar *), + EOP_ADDRESSES, EOP_BASE62, EOP_BASE62D, EOP_DOMAIN, @@ -154,7 +220,9 @@ enum { EOP_NH, EOP_NHASH, EOP_QUOTE, + EOP_RANDINT, EOP_RFC2047, + EOP_RFC2047D, EOP_RXQUOTE, EOP_S, EOP_SHA1, @@ -176,12 +244,16 @@ static uschar *cond_table[] = { US">", US">=", US"and", + US"bool", + US"bool_lax", US"crypteq", US"def", US"eq", US"eqi", US"exists", US"first_delivery", + US"forall", + US"forany", US"ge", US"gei", US"gt", @@ -215,12 +287,16 @@ enum { ECOND_NUM_G, ECOND_NUM_GE, ECOND_AND, + ECOND_BOOL, + ECOND_BOOL_LAX, ECOND_CRYPTEQ, ECOND_DEF, ECOND_STR_EQ, ECOND_STR_EQI, ECOND_EXISTS, ECOND_FIRST_DELIVERY, + ECOND_FORALL, + ECOND_FORANY, ECOND_STR_GE, ECOND_STR_GEI, ECOND_STR_GT, @@ -250,9 +326,9 @@ enum { /* Type for main variable table */ typedef struct { - char *name; - int type; - void *value; + const char *name; + int type; + void *value; } var_entry; /* Type for entries pointing to address/length pairs. Not currently @@ -274,11 +350,13 @@ enum { vtype_stringptr, /* value is address of pointer to string */ vtype_msgbody, /* as stringptr, but read when first required */ vtype_msgbody_end, /* ditto, the end of the message */ - vtype_msgheaders, /* the message's headers */ + vtype_msgheaders, /* the message's headers, processed */ + vtype_msgheaders_raw, /* the message's headers, unprocessed */ vtype_localpart, /* extract local part from string */ vtype_domain, /* extract domain from string */ vtype_recipients, /* extract recipients from recipients list */ - /* (enabled only during system filtering */ + /* (available only in system filters, ACLs, and */ + /* local_scan()) */ vtype_todbsdin, /* value not used; generate BSD inbox tod */ vtype_tode, /* value not used; generate tod in epoch format */ vtype_todf, /* value not used; generate full tod */ @@ -292,14 +370,16 @@ enum { vtype_load_avg, /* value not used; result is int from os_getloadavg */ vtype_pspace, /* partition space; value is T/F for spool/log */ vtype_pinodes /* partition inodes; value is T/F for spool/log */ -#ifdef EXPERIMENTAL_DOMAINKEYS - ,vtype_dk_verify /* Serve request out of DomainKeys verification structure */ -#endif + #ifndef DISABLE_DKIM + ,vtype_dkim /* Lookup of value in DKIM signature */ + #endif }; /* This table must be kept in alphabetical order. */ static var_entry var_table[] = { + /* WARNING: Do not invent variables whose names start acl_c or acl_m because + they will be confused with user-creatable ACL variables. */ { "acl_verify_message", vtype_stringptr, &acl_verify_message }, { "address_data", vtype_stringptr, &deliver_address_data }, { "address_file", vtype_stringptr, &address_file }, @@ -307,6 +387,9 @@ static var_entry var_table[] = { { "authenticated_id", vtype_stringptr, &authenticated_id }, { "authenticated_sender",vtype_stringptr, &authenticated_sender }, { "authentication_failed",vtype_int, &authentication_failed }, +#ifdef WITH_CONTENT_SCAN + { "av_failed", vtype_int, &av_failed }, +#endif #ifdef EXPERIMENTAL_BRIGHTMAIL { "bmi_alt_location", vtype_stringptr, &bmi_alt_location }, { "bmi_base64_tracker_verdict", vtype_stringptr, &bmi_base64_tracker_verdict }, @@ -322,24 +405,38 @@ static var_entry var_table[] = { { "compile_date", vtype_stringptr, &version_date }, { "compile_number", vtype_stringptr, &version_cnumber }, { "csa_status", vtype_stringptr, &csa_status }, +#ifdef EXPERIMENTAL_DCC + { "dcc_header", vtype_stringptr, &dcc_header }, + { "dcc_result", vtype_stringptr, &dcc_result }, +#endif #ifdef WITH_OLD_DEMIME { "demime_errorlevel", vtype_int, &demime_errorlevel }, { "demime_reason", vtype_stringptr, &demime_reason }, #endif -#ifdef EXPERIMENTAL_DOMAINKEYS - { "dk_domain", vtype_stringptr, &dk_signing_domain }, - { "dk_is_signed", vtype_dk_verify, NULL }, - { "dk_result", vtype_dk_verify, NULL }, - { "dk_selector", vtype_stringptr, &dk_signing_selector }, - { "dk_sender", vtype_dk_verify, NULL }, - { "dk_sender_domain", vtype_dk_verify, NULL }, - { "dk_sender_local_part",vtype_dk_verify, NULL }, - { "dk_sender_source", vtype_dk_verify, NULL }, - { "dk_signsall", vtype_dk_verify, NULL }, - { "dk_status", vtype_dk_verify, NULL }, - { "dk_testing", vtype_dk_verify, NULL }, +#ifndef DISABLE_DKIM + { "dkim_algo", vtype_dkim, (void *)DKIM_ALGO }, + { "dkim_bodylength", vtype_dkim, (void *)DKIM_BODYLENGTH }, + { "dkim_canon_body", vtype_dkim, (void *)DKIM_CANON_BODY }, + { "dkim_canon_headers", vtype_dkim, (void *)DKIM_CANON_HEADERS }, + { "dkim_copiedheaders", vtype_dkim, (void *)DKIM_COPIEDHEADERS }, + { "dkim_created", vtype_dkim, (void *)DKIM_CREATED }, + { "dkim_cur_signer", vtype_stringptr, &dkim_cur_signer }, + { "dkim_domain", vtype_stringptr, &dkim_signing_domain }, + { "dkim_expires", vtype_dkim, (void *)DKIM_EXPIRES }, + { "dkim_headernames", vtype_dkim, (void *)DKIM_HEADERNAMES }, + { "dkim_identity", vtype_dkim, (void *)DKIM_IDENTITY }, + { "dkim_key_granularity",vtype_dkim, (void *)DKIM_KEY_GRANULARITY }, + { "dkim_key_nosubdomains",vtype_dkim, (void *)DKIM_NOSUBDOMAINS }, + { "dkim_key_notes", vtype_dkim, (void *)DKIM_KEY_NOTES }, + { "dkim_key_srvtype", vtype_dkim, (void *)DKIM_KEY_SRVTYPE }, + { "dkim_key_testing", vtype_dkim, (void *)DKIM_KEY_TESTING }, + { "dkim_selector", vtype_stringptr, &dkim_signing_selector }, + { "dkim_signers", vtype_stringptr, &dkim_signers }, + { "dkim_verify_reason", vtype_dkim, (void *)DKIM_VERIFY_REASON }, + { "dkim_verify_status", vtype_dkim, (void *)DKIM_VERIFY_STATUS}, #endif { "dnslist_domain", vtype_stringptr, &dnslist_domain }, + { "dnslist_matched", vtype_stringptr, &dnslist_matched }, { "dnslist_text", vtype_stringptr, &dnslist_text }, { "dnslist_value", vtype_stringptr, &dnslist_value }, { "domain", vtype_stringptr, &deliver_domain }, @@ -359,6 +456,7 @@ static var_entry var_table[] = { { "inode", vtype_ino, &deliver_inode }, { "interface_address", vtype_stringptr, &interface_address }, { "interface_port", vtype_int, &interface_port }, + { "item", vtype_stringptr, &iterate_item }, #ifdef LOOKUP_LDAP { "ldap_dn", vtype_stringptr, &eldap_dn }, #endif @@ -377,12 +475,14 @@ static var_entry var_table[] = { #ifdef WITH_CONTENT_SCAN { "malware_name", vtype_stringptr, &malware_name }, #endif + { "max_received_linelength", vtype_int, &max_received_linelength }, { "message_age", vtype_int, &message_age }, { "message_body", vtype_msgbody, &message_body }, { "message_body_end", vtype_msgbody_end, &message_body_end }, { "message_body_size", vtype_int, &message_body_size }, { "message_exim_id", vtype_stringptr, &message_id }, { "message_headers", vtype_msgheaders, NULL }, + { "message_headers_raw", vtype_msgheaders_raw, NULL }, { "message_id", vtype_stringptr, &message_id }, { "message_linecount", vtype_int, &message_linecount }, { "message_size", vtype_int, &message_size }, @@ -432,6 +532,8 @@ static var_entry var_table[] = { { "rcpt_fail_count", vtype_int, &rcpt_fail_count }, { "received_count", vtype_int, &received_count }, { "received_for", vtype_stringptr, &received_for }, + { "received_ip_address", vtype_stringptr, &interface_address }, + { "received_port", vtype_int, &interface_port }, { "received_protocol", vtype_stringptr, &received_protocol }, { "received_time", vtype_int, &received_time }, { "recipient_data", vtype_stringptr, &recipient_data }, @@ -463,9 +565,13 @@ static var_entry var_table[] = { { "sender_rate_period", vtype_stringptr, &sender_rate_period }, { "sender_rcvhost", vtype_stringptr, &sender_rcvhost }, { "sender_verify_failure",vtype_stringptr, &sender_verify_failure }, + { "sending_ip_address", vtype_stringptr, &sending_ip_address }, + { "sending_port", vtype_int, &sending_port }, { "smtp_active_hostname", vtype_stringptr, &smtp_active_hostname }, { "smtp_command", vtype_stringptr, &smtp_cmd_buffer }, { "smtp_command_argument", vtype_stringptr, &smtp_cmd_argument }, + { "smtp_count_at_connection_start", vtype_int, &smtp_accept_count }, + { "smtp_notquit_reason", vtype_stringptr, &smtp_notquit_reason }, { "sn0", vtype_filter_int, &filter_sn[0] }, { "sn1", vtype_filter_int, &filter_sn[1] }, { "sn2", vtype_filter_int, &filter_sn[2] }, @@ -483,6 +589,7 @@ static var_entry var_table[] = { { "spam_score_int", vtype_stringptr, &spam_score_int }, #endif #ifdef EXPERIMENTAL_SPF + { "spf_guess", vtype_stringptr, &spf_guess }, { "spf_header_comment", vtype_stringptr, &spf_header_comment }, { "spf_received", vtype_stringptr, &spf_received }, { "spf_result", vtype_stringptr, &spf_result }, @@ -526,9 +633,9 @@ static BOOL malformed_header; /* For textual hashes */ -static char *hashcodes = "abcdefghijklmnopqrtsuvwxyz" - "ABCDEFGHIJKLMNOPQRSTUVWXYZ" - "0123456789"; +static const char *hashcodes = "abcdefghijklmnopqrtsuvwxyz" + "ABCDEFGHIJKLMNOPQRSTUVWXYZ" + "0123456789"; enum { HMAC_MD5, HMAC_SHA1 }; @@ -630,6 +737,8 @@ or "false" value. Failure of the expansion yields FALSE; logged unless it was a forced fail or lookup defer. All store used by the function can be released on exit. +The actual false-value tests should be replicated for ECOND_BOOL_LAX. + Arguments: condition the condition string m1 text to be incorporated in panic error @@ -659,6 +768,75 @@ return rc; +/************************************************* +* Pseudo-random number generation * +*************************************************/ + +/* Pseudo-random number generation. The result is not "expected" to be +cryptographically strong but not so weak that someone will shoot themselves +in the foot using it as a nonce in some email header scheme or whatever +weirdness they'll twist this into. The result should ideally handle fork(). + +However, if we're stuck unable to provide this, then we'll fall back to +appallingly bad randomness. + +If SUPPORT_TLS is defined and OpenSSL is used, then this will not be used. +The GNUTLS randomness functions found do not seem amenable to extracting +random numbers outside of a TLS context. Any volunteers? + +Arguments: + max range maximum +Returns a random number in range [0, max-1] +*/ + +#if !defined(SUPPORT_TLS) || defined(USE_GNUTLS) +int +pseudo_random_number(int max) +{ + static pid_t pid = 0; + pid_t p2; +#if defined(HAVE_SRANDOM) && !defined(HAVE_SRANDOMDEV) + struct timeval tv; +#endif + + p2 = getpid(); + if (p2 != pid) + { + if (pid != 0) + { + +#ifdef HAVE_ARC4RANDOM + /* cryptographically strong randomness, common on *BSD platforms, not + so much elsewhere. Alas. */ + arc4random_stir(); +#elif defined(HAVE_SRANDOM) || defined(HAVE_SRANDOMDEV) +#ifdef HAVE_SRANDOMDEV + /* uses random(4) for seeding */ + srandomdev(); +#else + gettimeofday(&tv, NULL); + srandom(tv.tv_sec | tv.tv_usec | getpid()); +#endif +#else + /* Poor randomness and no seeding here */ +#endif + + } + pid = p2; + } + +#ifdef HAVE_ARC4RANDOM + return arc4random() % max; +#elif defined(HAVE_SRANDOM) || defined(HAVE_SRANDOMDEV) + return random() % max; +#else + /* This one returns a 16-bit number, definitely not crypto-strong */ + return random_number(max); +#endif +} + +#endif + /************************************************* * Pick out a name from a string * *************************************************/ @@ -1076,7 +1254,8 @@ Arguments: newsize return the size of memory block that was obtained; may be NULL if exists_only is TRUE want_raw TRUE if called for $rh_ or $rheader_ variables; no processing, - other than concatenating, will be done on the header + other than concatenating, will be done on the header. Also used + for $message_headers_raw. charset name of charset to translate MIME words to; used only if want_raw is false; if NULL, no translation is done (this is used for $bh_ and $bheader_) @@ -1119,6 +1298,12 @@ for (i = 0; i < 2; i++) while (isspace(*t)) t++; /* remove leading white space */ ilen = h->slen - (t - h->text); /* length to insert */ + /* Unless wanted raw, remove trailing whitespace, including the + newline. */ + + if (!want_raw) + while (ilen > 0 && isspace(t[ilen-1])) ilen--; + /* Set comma = 1 if handling a single header and it's one of those that contains an address list, except when asked for raw headers. Only need to do this once. */ @@ -1130,7 +1315,7 @@ for (i = 0; i < 2; i++) /* First pass - compute total store needed; second pass - compute total store used, including this header. */ - size += ilen + comma; + size += ilen + comma + 1; /* +1 for the newline */ /* Second pass - concatentate the data, up to a maximum. Note that the loop stops when size hits the limit. */ @@ -1139,14 +1324,19 @@ for (i = 0; i < 2; i++) { if (size > header_insert_maxlen) { - ilen -= size - header_insert_maxlen; + ilen -= size - header_insert_maxlen - 1; comma = 0; } Ustrncpy(ptr, t, ilen); ptr += ilen; - if (comma != 0 && ilen > 0) + + /* For a non-raw header, put in the comma if needed, then add + back the newline we removed above, provided there was some text in + the header. */ + + if (!want_raw && ilen > 0) { - ptr[-1] = ','; + if (comma != 0) *ptr++ = ','; *ptr++ = '\n'; } } @@ -1154,8 +1344,9 @@ for (i = 0; i < 2; i++) } } - /* At end of first pass, truncate size if necessary, and get the buffer - to hold the data, returning the buffer size. */ + /* At end of first pass, return NULL if no header found. Then truncate size + if necessary, and get the buffer to hold the data, returning the buffer size. + */ if (i == 0) { @@ -1166,10 +1357,6 @@ for (i = 0; i < 2; i++) } } -/* Remove a redundant added comma if present */ - -if (comma != 0 && ptr > yield) ptr -= 2; - /* That's all we do for raw header expansion. */ if (want_raw) @@ -1177,15 +1364,16 @@ if (want_raw) *ptr = 0; } -/* Otherwise, we remove trailing whitespace, including newlines. Then we do RFC -2047 decoding, translating the charset if requested. The rfc2047_decode2() +/* Otherwise, remove a final newline and a redundant added comma. Then we do +RFC 2047 decoding, translating the charset if requested. The rfc2047_decode2() function can return an error with decoded data if the charset translation fails. If decoding fails, it returns NULL. */ else { uschar *decoded, *error; - while (ptr > yield && isspace(ptr[-1])) ptr--; + if (ptr > yield && ptr[-1] == '\n') ptr--; + if (ptr > yield && comma != 0 && ptr[-1] == ',') ptr--; *ptr = 0; decoded = rfc2047_decode2(yield, check_rfc2047_length, charset, '?', NULL, newsize, &error); @@ -1231,37 +1419,26 @@ find_variable(uschar *name, BOOL exists_only, BOOL skipping, int *newsize) int first = 0; int last = var_table_size; -/* Handle ACL variables, which are not in the table because their number may -vary depending on a build-time setting. If the variable's name is not of the -form acl_mddd or acl_cddd, where the d's are digits, fall through to look for -other names that start with acl_. */ - -if (Ustrncmp(name, "acl_", 4) == 0) - { - uschar *endptr; - int offset = -1; - int max = 0; +/* Handle ACL variables, whose names are of the form acl_cxxx or acl_mxxx. +Originally, xxx had to be a number in the range 0-9 (later 0-19), but from +release 4.64 onwards arbitrary names are permitted, as long as the first 5 +characters are acl_c or acl_m and the sixth is either a digit or an underscore +(this gave backwards compatibility at the changeover). There may be built-in +variables whose names start acl_ but they should never start in this way. This +slightly messy specification is a consequence of the history, needless to say. - if (name[4] == 'm') - { - offset = ACL_CVARS; - max = ACL_MVARS; - } - else if (name[4] == 'c') - { - offset = 0; - max = ACL_CVARS; - } +If an ACL variable does not exist, treat it as empty, unless strict_acl_vars is +set, in which case give an error. */ - if (offset >= 0) - { - int n = Ustrtoul(name + 5, &endptr, 10); - if (*endptr == 0 && n < max) - return (acl_var[offset + n] == NULL)? US"" : acl_var[offset + n]; - } +if ((Ustrncmp(name, "acl_c", 5) == 0 || Ustrncmp(name, "acl_m", 5) == 0) && + !isalpha(name[5])) + { + tree_node *node = + tree_search((name[4] == 'c')? acl_var_c : acl_var_m, name + 4); + return (node == NULL)? (strict_acl_vars? NULL : US"") : node->data.ptr; } -/* Similarly for $auth variables. */ +/* Handle $auth variables. */ if (Ustrncmp(name, "auth", 4) == 0) { @@ -1290,51 +1467,6 @@ while (last > first) switch (var_table[middle].type) { -#ifdef EXPERIMENTAL_DOMAINKEYS - - case vtype_dk_verify: - if (dk_verify_block == NULL) return US""; - s = NULL; - if (Ustrcmp(var_table[middle].name, "dk_result") == 0) - s = dk_verify_block->result_string; - if (Ustrcmp(var_table[middle].name, "dk_sender") == 0) - s = dk_verify_block->address; - if (Ustrcmp(var_table[middle].name, "dk_sender_domain") == 0) - s = dk_verify_block->domain; - if (Ustrcmp(var_table[middle].name, "dk_sender_local_part") == 0) - s = dk_verify_block->local_part; - - if (Ustrcmp(var_table[middle].name, "dk_sender_source") == 0) - switch(dk_verify_block->address_source) { - case DK_EXIM_ADDRESS_NONE: s = US"0"; break; - case DK_EXIM_ADDRESS_FROM_FROM: s = US"from"; break; - case DK_EXIM_ADDRESS_FROM_SENDER: s = US"sender"; break; - } - - if (Ustrcmp(var_table[middle].name, "dk_status") == 0) - switch(dk_verify_block->result) { - case DK_EXIM_RESULT_ERR: s = US"error"; break; - case DK_EXIM_RESULT_BAD_FORMAT: s = US"bad format"; break; - case DK_EXIM_RESULT_NO_KEY: s = US"no key"; break; - case DK_EXIM_RESULT_NO_SIGNATURE: s = US"no signature"; break; - case DK_EXIM_RESULT_REVOKED: s = US"revoked"; break; - case DK_EXIM_RESULT_NON_PARTICIPANT: s = US"non-participant"; break; - case DK_EXIM_RESULT_GOOD: s = US"good"; break; - case DK_EXIM_RESULT_BAD: s = US"bad"; break; - } - - if (Ustrcmp(var_table[middle].name, "dk_signsall") == 0) - s = (dk_verify_block->signsall)? US"1" : US"0"; - - if (Ustrcmp(var_table[middle].name, "dk_testing") == 0) - s = (dk_verify_block->testing)? US"1" : US"0"; - - if (Ustrcmp(var_table[middle].name, "dk_is_signed") == 0) - s = (dk_verify_block->is_signed)? US"1" : US"0"; - - return (s == NULL)? US"" : s; -#endif - case vtype_filter_int: if (!filter_running) return NULL; /* Fall through */ @@ -1364,7 +1496,7 @@ while (last > first) return var_buffer; case vtype_load_avg: - sprintf(CS var_buffer, "%d", os_getloadavg()); /* load_average */ + sprintf(CS var_buffer, "%d", OS_GETLOADAVG()); /* load_average */ return var_buffer; case vtype_host_lookup: /* Lookup if not done so */ @@ -1394,6 +1526,9 @@ while (last > first) case vtype_msgheaders: return find_header(NULL, exists_only, newsize, FALSE, NULL); + case vtype_msgheaders_raw: + return find_header(NULL, exists_only, newsize, TRUE, NULL); + case vtype_msgbody: /* Pointer to msgbody string */ case vtype_msgbody_end: /* Ditto, the end of the msg */ ss = (uschar **)(var_table[middle].value); @@ -1420,9 +1555,15 @@ while (last > first) if (len > 0) { body[len] = 0; - while (len > 0) + if (message_body_newlines) /* Separate loops for efficiency */ + { + while (len > 0) + { if (body[--len] == 0) body[len] = ' '; } + } + else { - if (body[--len] == '\n' || body[len] == 0) body[len] = ' '; + while (len > 0) + { if (body[--len] == '\n' || body[len] == 0) body[len] = ' '; } } } } @@ -1447,7 +1588,7 @@ while (last > first) return tod_stamp(tod_zulu); case vtype_todlf: /* Log file datestamp tod */ - return tod_stamp(tod_log_datestamp); + return tod_stamp(tod_log_datestamp_daily); case vtype_reply: /* Get reply address */ s = find_header(US"reply-to:", exists_only, newsize, TRUE, @@ -1504,6 +1645,12 @@ while (last > first) sprintf(CS var_buffer, "%d", inodes); } return var_buffer; + + #ifndef DISABLE_DKIM + case vtype_dkim: + return dkim_exim_expand_query((int)(long)var_table[middle].value); + #endif + } } @@ -1575,6 +1722,33 @@ return 0; +/************************************************* +* Elaborate message for bad variable * +*************************************************/ + +/* For the "unknown variable" message, take a look at the variable's name, and +give additional information about possible ACL variables. The extra information +is added on to expand_string_message. + +Argument: the name of the variable +Returns: nothing +*/ + +static void +check_variable_error_message(uschar *name) +{ +if (Ustrncmp(name, "acl_", 4) == 0) + expand_string_message = string_sprintf("%s (%s)", expand_string_message, + (name[4] == 'c' || name[4] == 'm')? + (isalpha(name[5])? + US"6th character of a user-defined ACL variable must be a digit or underscore" : + US"strict_acl_vars is set" /* Syntax is OK, it has to be this */ + ) : + US"user-defined ACL variables must start acl_c or acl_m"); +} + + + /************************************************* * Read and evaluate a condition * *************************************************/ @@ -1656,7 +1830,9 @@ switch(cond_type) s = read_name(name, 256, s+1, US"_"); - /* Test for a header's existence */ + /* Test for a header's existence. If the name contains a closing brace + character, this may be a user error where the terminating colon has been + omitted. Set a flag to adjust a subsequent error message in this case. */ if (Ustrncmp(name, "h_", 2) == 0 || Ustrncmp(name, "rh_", 3) == 0 || @@ -1666,6 +1842,7 @@ switch(cond_type) Ustrncmp(name, "bheader_", 8) == 0) { s = read_header_name(name, 256, s); + if (Ustrchr(name, '}') != NULL) malformed_header = TRUE; if (yield != NULL) *yield = (find_header(name, TRUE, NULL, FALSE, NULL) != NULL) == testfor; } @@ -1681,6 +1858,7 @@ switch(cond_type) expand_string_message = (name[0] == 0)? string_sprintf("variable name omitted after \"def:\"") : string_sprintf("unknown variable \"%s\" after \"def:\"", name); + check_variable_error_message(name); return NULL; } if (yield != NULL) *yield = (value[0] != 0) == testfor; @@ -1891,10 +2069,19 @@ switch(cond_type) conditions that compare numbers do not start with a letter. This just saves checking for them individually. */ - if (!isalpha(name[0])) + if (!isalpha(name[0]) && yield != NULL) { - num[i] = expand_string_integer(sub[i], FALSE); - if (expand_string_message != NULL) return NULL; + if (sub[i][0] == 0) + { + num[i] = 0; + DEBUG(D_expand) + debug_printf("empty string cast to zero for numerical comparison\n"); + } + else + { + num[i] = expand_string_integer(sub[i], FALSE); + if (expand_string_message != NULL) return NULL; + } } } @@ -2240,6 +2427,142 @@ switch(cond_type) return ++s; + /* forall/forany: iterates a condition with different values */ + + case ECOND_FORALL: + case ECOND_FORANY: + { + int sep = 0; + uschar *save_iterate_item = iterate_item; + + while (isspace(*s)) s++; + if (*s++ != '{') goto COND_FAILED_CURLY_START; + sub[0] = expand_string_internal(s, TRUE, &s, (yield == NULL)); + if (sub[0] == NULL) return NULL; + if (*s++ != '}') goto COND_FAILED_CURLY_END; + + while (isspace(*s)) s++; + if (*s++ != '{') goto COND_FAILED_CURLY_START; + + sub[1] = s; + + /* Call eval_condition once, with result discarded (as if scanning a + "false" part). This allows us to find the end of the condition, because if + the list it empty, we won't actually evaluate the condition for real. */ + + s = eval_condition(sub[1], NULL); + if (s == NULL) + { + expand_string_message = string_sprintf("%s inside \"%s\" condition", + expand_string_message, name); + return NULL; + } + while (isspace(*s)) s++; + + if (*s++ != '}') + { + expand_string_message = string_sprintf("missing } at end of condition " + "inside \"%s\"", name); + return NULL; + } + + if (yield != NULL) *yield = !testfor; + while ((iterate_item = string_nextinlist(&sub[0], &sep, NULL, 0)) != NULL) + { + DEBUG(D_expand) debug_printf("%s: $item = \"%s\"\n", name, iterate_item); + if (eval_condition(sub[1], &tempcond) == NULL) + { + expand_string_message = string_sprintf("%s inside \"%s\" condition", + expand_string_message, name); + iterate_item = save_iterate_item; + return NULL; + } + DEBUG(D_expand) debug_printf("%s: condition evaluated to %s\n", name, + tempcond? "true":"false"); + + if (yield != NULL) *yield = (tempcond == testfor); + if (tempcond == (cond_type == ECOND_FORANY)) break; + } + + iterate_item = save_iterate_item; + return s; + } + + + /* The bool{} expansion condition maps a string to boolean. + The values supported should match those supported by the ACL condition + (acl.c, ACLC_CONDITION) so that we keep to a minimum the different ideas + of true/false. Note that Router "condition" rules have a different + interpretation, where general data can be used and only a few values + map to FALSE. + Note that readconf.c boolean matching, for boolean configuration options, + only matches true/yes/false/no. + The bool_lax{} condition matches the Router logic, which is much more + liberal. */ + case ECOND_BOOL: + case ECOND_BOOL_LAX: + { + uschar *sub_arg[1]; + uschar *t, *t2; + uschar *ourname; + size_t len; + BOOL boolvalue = FALSE; + while (isspace(*s)) s++; + if (*s != '{') goto COND_FAILED_CURLY_START; + ourname = cond_type == ECOND_BOOL_LAX ? US"bool_lax" : US"bool"; + switch(read_subs(sub_arg, 1, 1, &s, yield == NULL, FALSE, ourname)) + { + case 1: expand_string_message = string_sprintf( + "too few arguments or bracketing error for %s", + ourname); + /*FALLTHROUGH*/ + case 2: + case 3: return NULL; + } + t = sub_arg[0]; + while (isspace(*t)) t++; + len = Ustrlen(t); + if (len) + { + /* trailing whitespace: seems like a good idea to ignore it too */ + t2 = t + len - 1; + while (isspace(*t2)) t2--; + if (t2 != (t + len)) + { + *++t2 = '\0'; + len = t2 - t; + } + } + DEBUG(D_expand) + debug_printf("considering %s: %s\n", ourname, len ? t : US""); + /* logic for the lax case from expand_check_condition(), which also does + expands, and the logic is both short and stable enough that there should + be no maintenance burden from replicating it. */ + if (len == 0) + boolvalue = FALSE; + else if (Ustrspn(t, "0123456789") == len) + { + boolvalue = (Uatoi(t) == 0) ? FALSE : TRUE; + /* expand_check_condition only does a literal string "0" check */ + if ((cond_type == ECOND_BOOL_LAX) && (len > 1)) + boolvalue = TRUE; + } + else if (strcmpic(t, US"true") == 0 || strcmpic(t, US"yes") == 0) + boolvalue = TRUE; + else if (strcmpic(t, US"false") == 0 || strcmpic(t, US"no") == 0) + boolvalue = FALSE; + else if (cond_type == ECOND_BOOL_LAX) + boolvalue = TRUE; + else + { + expand_string_message = string_sprintf("unrecognised boolean " + "value \"%s\"", t); + return NULL; + } + if (yield != NULL) *yield = (boolvalue == testfor); + return s; + } + /* Unknown condition */ default: @@ -2682,63 +3005,53 @@ return yield; * Evaluate numeric expression * *************************************************/ -/* This is a set of mutually recursive functions that evaluate a simple -arithmetic expression involving only + - * / and parentheses. The only one that -is called from elsewhere is eval_expr, whose interface is: +/* This is a set of mutually recursive functions that evaluate an arithmetic +expression involving + - * / % & | ^ ~ << >> and parentheses. The only one of +these functions that is called from elsewhere is eval_expr, whose interface is: Arguments: - sptr pointer to the pointer to the string - gets updated - decimal TRUE if numbers are to be assumed decimal - error pointer to where to put an error message - must be NULL on input - endket TRUE if ')' must terminate - FALSE for external call - + sptr pointer to the pointer to the string - gets updated + decimal TRUE if numbers are to be assumed decimal + error pointer to where to put an error message - must be NULL on input + endket TRUE if ')' must terminate - FALSE for external call -Returns: on success: the value of the expression, with *error still NULL - on failure: an undefined value, with *error = a message +Returns: on success: the value of the expression, with *error still NULL + on failure: an undefined value, with *error = a message */ -static int eval_sumterm(uschar **, BOOL, uschar **); +static int eval_op_or(uschar **, BOOL, uschar **); + static int eval_expr(uschar **sptr, BOOL decimal, uschar **error, BOOL endket) { uschar *s = *sptr; -int x = eval_sumterm(&s, decimal, error); +int x = eval_op_or(&s, decimal, error); if (*error == NULL) { - while (*s == '+' || *s == '-') + if (endket) { - int op = *s++; - int y = eval_sumterm(&s, decimal, error); - if (*error != NULL) break; - if (op == '+') x += y; else x -= y; - } - if (*error == NULL) - { - if (endket) - { - if (*s != ')') - *error = US"expecting closing parenthesis"; - else - while (isspace(*(++s))); - } - else if (*s != 0) *error = US"expecting + or -"; + if (*s != ')') + *error = US"expecting closing parenthesis"; + else + while (isspace(*(++s))); } + else if (*s != 0) *error = US"expecting operator"; } - *sptr = s; return x; } + static int -eval_term(uschar **sptr, BOOL decimal, uschar **error) +eval_number(uschar **sptr, BOOL decimal, uschar **error) { register int c; int n; uschar *s = *sptr; while (isspace(*s)) s++; c = *s; -if (isdigit(c) || ((c == '-' || c == '+') && isdigit(s[1]))) +if (isdigit(c)) { int count; (void)sscanf(CS s, (decimal? "%d%n" : "%i%n"), &n, &count); @@ -2761,20 +3074,80 @@ else return n; } -static int eval_sumterm(uschar **sptr, BOOL decimal, uschar **error) + +static int eval_op_unary(uschar **sptr, BOOL decimal, uschar **error) { uschar *s = *sptr; -int x = eval_term(&s, decimal, error); +int x; +while (isspace(*s)) s++; +if (*s == '+' || *s == '-' || *s == '~') + { + int op = *s++; + x = eval_op_unary(&s, decimal, error); + if (op == '-') x = -x; + else if (op == '~') x = ~x; + } +else + { + x = eval_number(&s, decimal, error); + } +*sptr = s; +return x; +} + + +static int eval_op_mult(uschar **sptr, BOOL decimal, uschar **error) +{ +uschar *s = *sptr; +int x = eval_op_unary(&s, decimal, error); if (*error == NULL) { while (*s == '*' || *s == '/' || *s == '%') { int op = *s++; - int y = eval_term(&s, decimal, error); + int y = eval_op_unary(&s, decimal, error); if (*error != NULL) break; - if (op == '*') x *= y; - else if (op == '/') x /= y; - else x %= y; + /* SIGFPE both on div/mod by zero and on INT_MIN / -1, which would give + * a value of INT_MAX+1. Note that INT_MIN * -1 gives INT_MIN for me, which + * is a bug somewhere in [gcc 4.2.1, FreeBSD, amd64]. In fact, -N*-M where + * -N*M is INT_MIN will yielf INT_MIN. + * Since we don't support floating point, this is somewhat simpler. + * Ideally, we'd return an error, but since we overflow for all other + * arithmetic, consistency suggests otherwise, but what's the correct value + * to use? There is none. + * The C standard guarantees overflow for unsigned arithmetic but signed + * overflow invokes undefined behaviour; in practice, this is overflow + * except for converting INT_MIN to INT_MAX+1. We also can't guarantee + * that long/longlong larger than int are available, or we could just work + * with larger types. We should consider whether to guarantee 32bit eval + * and 64-bit working variables, with errors returned. For now ... + * So, the only SIGFPEs occur with a non-shrinking div/mod, thus -1; we + * can just let the other invalid results occur otherwise, as they have + * until now. For this one case, we can coerce. + */ + if (y == -1 && x == INT_MIN && op != '*') + { + DEBUG(D_expand) + debug_printf("Integer exception dodging: %d%c-1 coerced to %d\n", + INT_MIN, op, INT_MAX); + x = INT_MAX; + continue; + } + if (op == '*') + x *= y; + else + { + if (y == 0) + { + *error = (op == '/') ? US"divide by zero" : US"modulo by zero"; + x = 0; + break; + } + if (op == '/') + x /= y; + else + x %= y; + } } } *sptr = s; @@ -2782,6 +3155,105 @@ return x; } +static int eval_op_sum(uschar **sptr, BOOL decimal, uschar **error) +{ +uschar *s = *sptr; +int x = eval_op_mult(&s, decimal, error); +if (*error == NULL) + { + while (*s == '+' || *s == '-') + { + int op = *s++; + int y = eval_op_mult(&s, decimal, error); + if (*error != NULL) break; + if (op == '+') x += y; else x -= y; + } + } +*sptr = s; +return x; +} + + +static int eval_op_shift(uschar **sptr, BOOL decimal, uschar **error) +{ +uschar *s = *sptr; +int x = eval_op_sum(&s, decimal, error); +if (*error == NULL) + { + while ((*s == '<' || *s == '>') && s[1] == s[0]) + { + int y; + int op = *s++; + s++; + y = eval_op_sum(&s, decimal, error); + if (*error != NULL) break; + if (op == '<') x <<= y; else x >>= y; + } + } +*sptr = s; +return x; +} + + +static int eval_op_and(uschar **sptr, BOOL decimal, uschar **error) +{ +uschar *s = *sptr; +int x = eval_op_shift(&s, decimal, error); +if (*error == NULL) + { + while (*s == '&') + { + int y; + s++; + y = eval_op_shift(&s, decimal, error); + if (*error != NULL) break; + x &= y; + } + } +*sptr = s; +return x; +} + + +static int eval_op_xor(uschar **sptr, BOOL decimal, uschar **error) +{ +uschar *s = *sptr; +int x = eval_op_and(&s, decimal, error); +if (*error == NULL) + { + while (*s == '^') + { + int y; + s++; + y = eval_op_and(&s, decimal, error); + if (*error != NULL) break; + x ^= y; + } + } +*sptr = s; +return x; +} + + +static int eval_op_or(uschar **sptr, BOOL decimal, uschar **error) +{ +uschar *s = *sptr; +int x = eval_op_xor(&s, decimal, error); +if (*error == NULL) + { + while (*s == '|') + { + int y; + s++; + y = eval_op_xor(&s, decimal, error); + if (*error != NULL) break; + x |= y; + } + } +*sptr = s; +return x; +} + /************************************************* @@ -2822,6 +3294,12 @@ we reset the store before processing it; if the result is in fresh store, we use that without copying. This is helpful for expanding strings like $message_headers which can get very long. +There's a problem if a ${dlfunc item has side-effects that cause allocation, +since resetting the store at the end of the expansion will free store that was +allocated by the plugin code as well as the slop after the expanded string. So +we skip any resets if ${dlfunc has been used. This is an unfortunate +consequence of string expansion becoming too powerful. + Arguments: string the string to be expanded ket_ends true if expansion is to stop at } @@ -2847,6 +3325,7 @@ uschar *yield = store_get(size); uschar *s = string; uschar *save_expand_nstring[EXPAND_MAXN+1]; int save_expand_nlength[EXPAND_MAXN+1]; +BOOL resetok = TRUE; expand_string_forcedfail = FALSE; expand_string_message = US""; @@ -2919,7 +3398,7 @@ while (*s != 0) if (ptr == 0 && yield != NULL) { - store_reset(yield); + if (resetok) store_reset(yield); yield = NULL; size = 0; } @@ -2939,7 +3418,7 @@ while (*s != 0) value = find_header(name, FALSE, &newsize, want_raw, charset); /* If we didn't find the header, and the header contains a closing brace - characters, this may be a user error where the terminating colon + character, this may be a user error where the terminating colon has been omitted. Set a flag to adjust the error message in this case. But there is no error here - nothing gets inserted. */ @@ -2959,6 +3438,7 @@ while (*s != 0) { expand_string_message = string_sprintf("unknown variable name \"%s\"", name); + check_variable_error_message(name); goto EXPAND_FAILED; } } @@ -3234,8 +3714,8 @@ while (*s != 0) if (search_find_defer) { expand_string_message = - string_sprintf("lookup of \"%s\" gave DEFER: %s", key, - search_error_message); + string_sprintf("lookup of \"%s\" gave DEFER: %s", + string_printing2(key, FALSE), search_error_message); goto EXPAND_FAILED; } if (expand_setup > 0) expand_nmax = expand_setup; @@ -3402,11 +3882,11 @@ while (*s != 0) *domain++ = '\0'; yield = string_cat(yield,&size,&ptr,US"prvs=",5); - string_cat(yield,&size,&ptr,sub_arg[0],Ustrlen(sub_arg[0])); - string_cat(yield,&size,&ptr,US"/",1); string_cat(yield,&size,&ptr,(sub_arg[2] != NULL) ? sub_arg[2] : US"0", 1); string_cat(yield,&size,&ptr,prvs_daystamp(7),3); string_cat(yield,&size,&ptr,p,6); + string_cat(yield,&size,&ptr,US"=",1); + string_cat(yield,&size,&ptr,sub_arg[0],Ustrlen(sub_arg[0])); string_cat(yield,&size,&ptr,US"@",1); string_cat(yield,&size,&ptr,domain,Ustrlen(domain)); @@ -3444,15 +3924,15 @@ while (*s != 0) case 3: goto EXPAND_FAILED; } - re = regex_must_compile(US"^prvs\\=(.+)\\/([0-9])([0-9]{3})([A-F0-9]{6})\\@(.+)$", + re = regex_must_compile(US"^prvs\\=([0-9])([0-9]{3})([A-F0-9]{6})\\=(.+)\\@(.+)$", TRUE,FALSE); if (regex_match_and_setup(re,sub_arg[0],0,-1)) { - uschar *local_part = string_copyn(expand_nstring[1],expand_nlength[1]); - uschar *key_num = string_copyn(expand_nstring[2],expand_nlength[2]); - uschar *daystamp = string_copyn(expand_nstring[3],expand_nlength[3]); - uschar *hash = string_copyn(expand_nstring[4],expand_nlength[4]); + uschar *local_part = string_copyn(expand_nstring[4],expand_nlength[4]); + uschar *key_num = string_copyn(expand_nstring[1],expand_nlength[1]); + uschar *daystamp = string_copyn(expand_nstring[2],expand_nlength[2]); + uschar *hash = string_copyn(expand_nstring[3],expand_nlength[3]); uschar *domain = string_copyn(expand_nstring[5],expand_nlength[5]); DEBUG(D_expand) debug_printf("prvscheck localpart: %s\n", local_part); @@ -3503,7 +3983,7 @@ while (*s != 0) Adjust "inow" accordingly. */ if ( (iexpire < 7) && (inow >= 993) ) inow = 0; - if (iexpire > inow) + if (iexpire >= inow) { prvscheck_result = US"1"; DEBUG(D_expand) debug_printf("prvscheck: success, $pvrs_result set to 1\n"); @@ -3722,7 +4202,8 @@ while (*s != 0) else { shost.name = server_name; - if (host_find_byname(&shost, NULL, NULL, FALSE) != HOST_FOUND) + if (host_find_byname(&shost, NULL, HOST_FIND_QUALIFY_SINGLE, NULL, + FALSE) != HOST_FOUND) { expand_string_message = string_sprintf("no IP address found for host %s", shost.name); @@ -3762,6 +4243,7 @@ while (*s != 0) else { + int rc; if ((fd = socket(PF_UNIX, SOCK_STREAM, 0)) == -1) { expand_string_message = string_sprintf("failed to create socket: %s", @@ -3772,7 +4254,17 @@ while (*s != 0) sockun.sun_family = AF_UNIX; sprintf(sockun.sun_path, "%.*s", (int)(sizeof(sockun.sun_path)-1), sub_arg[0]); - if(connect(fd, (struct sockaddr *)(&sockun), sizeof(sockun)) == -1) + + sigalrm_seen = FALSE; + alarm(timeout); + rc = connect(fd, (struct sockaddr *)(&sockun), sizeof(sockun)); + alarm(0); + if (sigalrm_seen) + { + expand_string_message = US "socket connect timed out"; + goto SOCK_FAIL; + } + if (rc < 0) { expand_string_message = string_sprintf("failed to connect to socket " "%s: %s", sub_arg[0], strerror(errno)); @@ -3797,6 +4289,14 @@ while (*s != 0) } } + /* Shut down the sending side of the socket. This helps some servers to + recognise that it is their turn to do some work. Just in case some + system doesn't have this function, make it conditional. */ + + #ifdef SHUT_WR + shutdown(fd, SHUT_WR); + #endif + /* Now we need to read from the socket, under a timeout. The function that reads a file can be used. */ @@ -3904,13 +4404,24 @@ while (*s != 0) (void)close(fd_in); + /* Read the pipe to get the command's output into $value (which is kept + in lookup_value). Read during execution, so that if the output exceeds + the OS pipe buffer limit, we don't block forever. */ + + f = fdopen(fd_out, "rb"); + sigalrm_seen = FALSE; + alarm(60); + lookup_value = cat_file(f, lookup_value, &lsize, &lptr, NULL); + alarm(0); + (void)fclose(f); + /* Wait for the process to finish, applying the timeout, and inspect its return code for serious disasters. Simple non-zero returns are passed on. */ - if ((runrc = child_close(pid, 60)) < 0) + if (sigalrm_seen == TRUE || (runrc = child_close(pid, 30)) < 0) { - if (runrc == -256) + if (sigalrm_seen == TRUE || runrc == -256) { expand_string_message = string_sprintf("command timed out"); killpg(pid, SIGKILL); /* Kill the whole process group */ @@ -3926,14 +4437,6 @@ while (*s != 0) goto EXPAND_FAILED; } - - /* Read the pipe to get the command's output into $value (which is kept - in lookup_value). */ - - f = fdopen(fd_out, "rb"); - lookup_value = NULL; - lookup_value = cat_file(f, lookup_value, &lsize, &lptr, NULL); - (void)fclose(f); } /* Process the yes/no strings; $value may be useful in both cases */ @@ -4317,7 +4820,7 @@ while (*s != 0) while (len > 0 && isspace(p[len-1])) len--; p[len] = 0; - if (*p == 0) + if (*p == 0 && !skipping) { expand_string_message = US"first argument of \"extract\" must " "not be empty"; @@ -4374,6 +4877,184 @@ while (*s != 0) } + /* Handle list operations */ + + case EITEM_FILTER: + case EITEM_MAP: + case EITEM_REDUCE: + { + int sep = 0; + int save_ptr = ptr; + uschar outsep[2] = { '\0', '\0' }; + uschar *list, *expr, *temp; + uschar *save_iterate_item = iterate_item; + uschar *save_lookup_value = lookup_value; + + while (isspace(*s)) s++; + if (*s++ != '{') goto EXPAND_FAILED_CURLY; + + list = expand_string_internal(s, TRUE, &s, skipping); + if (list == NULL) goto EXPAND_FAILED; + if (*s++ != '}') goto EXPAND_FAILED_CURLY; + + if (item_type == EITEM_REDUCE) + { + while (isspace(*s)) s++; + if (*s++ != '{') goto EXPAND_FAILED_CURLY; + temp = expand_string_internal(s, TRUE, &s, skipping); + if (temp == NULL) goto EXPAND_FAILED; + lookup_value = temp; + if (*s++ != '}') goto EXPAND_FAILED_CURLY; + } + + while (isspace(*s)) s++; + if (*s++ != '{') goto EXPAND_FAILED_CURLY; + + expr = s; + + /* For EITEM_FILTER, call eval_condition once, with result discarded (as + if scanning a "false" part). This allows us to find the end of the + condition, because if the list is empty, we won't actually evaluate the + condition for real. For EITEM_MAP and EITEM_REDUCE, do the same, using + the normal internal expansion function. */ + + if (item_type == EITEM_FILTER) + { + temp = eval_condition(expr, NULL); + if (temp != NULL) s = temp; + } + else + { + temp = expand_string_internal(s, TRUE, &s, TRUE); + } + + if (temp == NULL) + { + expand_string_message = string_sprintf("%s inside \"%s\" item", + expand_string_message, name); + goto EXPAND_FAILED; + } + + while (isspace(*s)) s++; + if (*s++ != '}') + { + expand_string_message = string_sprintf("missing } at end of condition " + "or expression inside \"%s\"", name); + goto EXPAND_FAILED; + } + + while (isspace(*s)) s++; + if (*s++ != '}') + { + expand_string_message = string_sprintf("missing } at end of \"%s\"", + name); + goto EXPAND_FAILED; + } + + /* If we are skipping, we can now just move on to the next item. When + processing for real, we perform the iteration. */ + + if (skipping) continue; + while ((iterate_item = string_nextinlist(&list, &sep, NULL, 0)) != NULL) + { + *outsep = (uschar)sep; /* Separator as a string */ + + DEBUG(D_expand) debug_printf("%s: $item = \"%s\"\n", name, iterate_item); + + if (item_type == EITEM_FILTER) + { + BOOL condresult; + if (eval_condition(expr, &condresult) == NULL) + { + iterate_item = save_iterate_item; + lookup_value = save_lookup_value; + expand_string_message = string_sprintf("%s inside \"%s\" condition", + expand_string_message, name); + goto EXPAND_FAILED; + } + DEBUG(D_expand) debug_printf("%s: condition is %s\n", name, + condresult? "true":"false"); + if (condresult) + temp = iterate_item; /* TRUE => include this item */ + else + continue; /* FALSE => skip this item */ + } + + /* EITEM_MAP and EITEM_REDUCE */ + + else + { + temp = expand_string_internal(expr, TRUE, NULL, skipping); + if (temp == NULL) + { + iterate_item = save_iterate_item; + expand_string_message = string_sprintf("%s inside \"%s\" item", + expand_string_message, name); + goto EXPAND_FAILED; + } + if (item_type == EITEM_REDUCE) + { + lookup_value = temp; /* Update the value of $value */ + continue; /* and continue the iteration */ + } + } + + /* We reach here for FILTER if the condition is true, always for MAP, + and never for REDUCE. The value in "temp" is to be added to the output + list that is being created, ensuring that any occurrences of the + separator character are doubled. Unless we are dealing with the first + item of the output list, add in a space if the new item begins with the + separator character, or is an empty string. */ + + if (ptr != save_ptr && (temp[0] == *outsep || temp[0] == 0)) + yield = string_cat(yield, &size, &ptr, US" ", 1); + + /* Add the string in "temp" to the output list that we are building, + This is done in chunks by searching for the separator character. */ + + for (;;) + { + size_t seglen = Ustrcspn(temp, outsep); + yield = string_cat(yield, &size, &ptr, temp, seglen + 1); + + /* If we got to the end of the string we output one character + too many; backup and end the loop. Otherwise arrange to double the + separator. */ + + if (temp[seglen] == '\0') { ptr--; break; } + yield = string_cat(yield, &size, &ptr, outsep, 1); + temp += seglen + 1; + } + + /* Output a separator after the string: we will remove the redundant + final one at the end. */ + + yield = string_cat(yield, &size, &ptr, outsep, 1); + } /* End of iteration over the list loop */ + + /* REDUCE has generated no output above: output the final value of + $value. */ + + if (item_type == EITEM_REDUCE) + { + yield = string_cat(yield, &size, &ptr, lookup_value, + Ustrlen(lookup_value)); + lookup_value = save_lookup_value; /* Restore $value */ + } + + /* FILTER and MAP generate lists: if they have generated anything, remove + the redundant final separator. Even though an empty item at the end of a + list does not count, this is tidier. */ + + else if (ptr != save_ptr) ptr--; + + /* Restore preserved $item */ + + iterate_item = save_iterate_item; + continue; + } + + /* If ${dlfunc support is configured, handle calling dynamically-loaded functions, unless locked out at this time. Syntax is ${dlfunc{file}{func}} or ${dlfunc{file}{func}{arg}} or ${dlfunc{file}{func}{arg1}{arg2}} or up to @@ -4450,8 +5131,10 @@ while (*s != 0) returns OK, we have a replacement string; if it returns DEFER then expansion has failed in a non-forced manner; if it returns FAIL then failure was forced; if it returns ERROR or any other value there's a - problem, so panic slightly. */ + problem, so panic slightly. In any case, assume that the function has + side-effects on the store that must be preserved. */ + resetok = FALSE; result = NULL; for (argc = 0; argv[argc] != NULL; argc++); status = func(&result, argc - 2, &argv[2]); @@ -4734,6 +5417,69 @@ while (*s != 0) continue; } + case EOP_ADDRESSES: + { + uschar outsep[2] = { ':', '\0' }; + uschar *address, *error; + int save_ptr = ptr; + int start, end, domain; /* Not really used */ + + while (isspace(*sub)) sub++; + if (*sub == '>') { *outsep = *++sub; ++sub; } + parse_allow_group = TRUE; + + for (;;) + { + uschar *p = parse_find_address_end(sub, FALSE); + uschar saveend = *p; + *p = '\0'; + address = parse_extract_address(sub, &error, &start, &end, &domain, + FALSE); + *p = saveend; + + /* Add the address to the output list that we are building. This is + done in chunks by searching for the separator character. At the + start, unless we are dealing with the first address of the output + list, add in a space if the new address begins with the separator + character, or is an empty string. */ + + if (address != NULL) + { + if (ptr != save_ptr && address[0] == *outsep) + yield = string_cat(yield, &size, &ptr, US" ", 1); + + for (;;) + { + size_t seglen = Ustrcspn(address, outsep); + yield = string_cat(yield, &size, &ptr, address, seglen + 1); + + /* If we got to the end of the string we output one character + too many. */ + + if (address[seglen] == '\0') { ptr--; break; } + yield = string_cat(yield, &size, &ptr, outsep, 1); + address += seglen + 1; + } + + /* Output a separator after the string: we will remove the + redundant final one at the end. */ + + yield = string_cat(yield, &size, &ptr, outsep, 1); + } + + if (saveend == '\0') break; + sub = p + 1; + } + + /* If we have generated anything, remove the redundant final + separator. */ + + if (ptr != save_ptr) ptr--; + parse_allow_group = FALSE; + continue; + } + + /* quote puts a string in quotes if it is empty or contains anything other than alphamerics, underscore, dot, or hyphen. @@ -4802,8 +5548,8 @@ while (*s != 0) goto EXPAND_FAILED; } - if (lookup_list[n].quote != NULL) - sub = (lookup_list[n].quote)(sub, opt); + if (lookup_list[n]->quote != NULL) + sub = (lookup_list[n]->quote)(sub, opt); else if (opt != NULL) sub = NULL; if (sub == NULL) @@ -4845,6 +5591,23 @@ while (*s != 0) continue; } + /* RFC 2047 decode */ + + case EOP_RFC2047D: + { + int len; + uschar *error; + uschar *decoded = rfc2047_decode(sub, check_rfc2047_length, + headers_charset, '?', &len, &error); + if (error != NULL) + { + expand_string_message = error; + goto EXPAND_FAILED; + } + yield = string_cat(yield, &size, &ptr, decoded, len); + continue; + } + /* from_utf8 converts UTF-8 to 8859-1, turning non-existent chars into underscores */ @@ -5087,6 +5850,40 @@ while (*s != 0) continue; } + /* pseudo-random number less than N */ + + case EOP_RANDINT: + { + int max; + uschar *s; + + max = expand_string_integer(sub, TRUE); + if (expand_string_message != NULL) + goto EXPAND_FAILED; + s = string_sprintf("%d", pseudo_random_number(max)); + yield = string_cat(yield, &size, &ptr, s, Ustrlen(s)); + continue; + } + + /* Reverse IP, including IPv6 to dotted-nibble */ + + case EOP_REVERSE_IP: + { + int family, maskptr; + uschar reversed[128]; + + family = string_is_ip_address(sub, &maskptr); + if (family == 0) + { + expand_string_message = string_sprintf( + "reverse_ip() not given an IP address [%s]", sub); + goto EXPAND_FAILED; + } + invert_address(reversed, sub); + yield = string_cat(yield, &size, &ptr, reversed, Ustrlen(reversed)); + continue; + } + /* Unknown operator */ default: @@ -5109,7 +5906,7 @@ while (*s != 0) int newsize = 0; if (ptr == 0) { - store_reset(yield); + if (resetok) store_reset(yield); yield = NULL; size = 0; } @@ -5118,6 +5915,7 @@ while (*s != 0) { expand_string_message = string_sprintf("unknown variable in \"${%s}\"", name); + check_variable_error_message(name); goto EXPAND_FAILED; } len = Ustrlen(value); @@ -5163,7 +5961,7 @@ if (left != NULL) *left = s; In many cases the final string will be the first one that was got and so there will be optimal store usage. */ -store_reset(yield + ptr + 1); +if (resetok) store_reset(yield + ptr + 1); DEBUG(D_expand) { debug_printf("expanding: %.*s\n result: %s\n", (int)(s - string), string, @@ -5271,7 +6069,24 @@ systems, so we set it zero ourselves. */ errno = 0; expand_string_message = NULL; /* Indicates no error */ -value = strtol(CS s, CSS &endptr, 0); + +/* Before Exim 4.64, strings consisting entirely of whitespace compared +equal to 0. Unfortunately, people actually relied upon that, so preserve +the behaviour explicitly. Stripping leading whitespace is a harmless +noop change since strtol skips it anyway (provided that there is a number +to find at all). */ +if (isspace(*s)) + { + while (isspace(*s)) ++s; + if (*s == '\0') + { + DEBUG(D_expand) + debug_printf("treating blank string as number 0\n"); + return 0; + } + } + +value = strtol(CS s, CSS &endptr, 10); if (endptr == s) {