X-Git-Url: https://git.exim.org/exim.git/blobdiff_plain/f83d4a2b3fedd9a8a0e7367db82a68a719f08e30..a85c067ba6c6940512cf57ec213277a370d87e70:/src/src/parse.c diff --git a/src/src/parse.c b/src/src/parse.c index fa3395206..93b12bc77 100644 --- a/src/src/parse.c +++ b/src/src/parse.c @@ -2,9 +2,10 @@ * Exim - an Internet mail transport agent * *************************************************/ +/* Copyright (c) The Exim Maintainers 2020 - 2022 */ /* Copyright (c) University of Cambridge 1995 - 2018 */ -/* Copyright (c) The Exim Maintainers 2020 */ /* See the file NOTICE for conditions of use and distribution. */ +/* SPDX-License-Identifier: GPL-2.0-only */ /* Functions for parsing addresses */ @@ -22,22 +23,25 @@ redundant apparatus. */ #ifdef STAND_ALONE -address_item *deliver_make_addr(uschar *address, BOOL copy) +address_item * +deliver_make_addr(uschar *address, BOOL copy) { -address_item *addr = store_get(sizeof(address_item), FALSE); +address_item *addr = store_get(sizeof(address_item), GET_UNTAINTED); addr->next = NULL; addr->parent = NULL; addr->address = address; return addr; } -uschar *rewrite_address(uschar *recipient, BOOL dummy1, BOOL dummy2, rewrite_rule +uschar * +rewrite_address(uschar *recipient, BOOL dummy1, BOOL dummy2, rewrite_rule *dummy3, int dummy4) { return recipient; } -uschar *rewrite_address_qualify(uschar *recipient, BOOL dummy1) +uschar * +rewrite_address_qualify(uschar *recipient, BOOL dummy1) { return recipient; } @@ -65,7 +69,7 @@ Returns: pointer past the end of the address */ uschar * -parse_find_address_end(uschar *s, BOOL nl_ends) +parse_find_address_end(const uschar *s, BOOL nl_ends) { BOOL source_routing = *s == '@'; int no_term = source_routing? 1 : 0; @@ -121,7 +125,7 @@ while (*s != 0 && (*s != ',' || no_term > 0) && (*s != '\n' || !nl_ends)) } } -return s; +return US s; } @@ -224,16 +228,20 @@ If allow_domain_literals is TRUE, a "domain" may also be an IP address enclosed in []. Make sure the output is set to the null string if there is a syntax error as well as if there is no domain at all. +Optionally, msg_id domain literals ( printable-ascii enclosed in [] ) +are permitted. + Arguments: s current character pointer t where to put the domain + msg_id_literals flag for relaxed domain-literal processing errorptr put error message here on failure (*t will be 0 on exit) Returns: new character pointer */ static const uschar * -read_domain(const uschar *s, uschar *t, uschar **errorptr) +read_domain(const uschar *s, uschar *t, BOOL msg_id_literals, uschar **errorptr) { uschar *tt = t; s = skip_comment(s); @@ -259,7 +267,11 @@ if (*s == '[') t += 5; s += 5; } - while (*s == '.' || *s == ':' || isxdigit(*s)) *t++ = *s++; + + if (msg_id_literals) + while (*s >= 33 && *s <= 90 || *s >= 94 && *s <= 126) *t++ = *s++; + else + while (*s == '.' || *s == ':' || isxdigit(*s)) *t++ = *s++; if (*s == ']') *t++ = *s++; else { @@ -267,7 +279,7 @@ if (*s == '[') *tt = 0; } - if (!allow_domain_literals) + if (!allow_domain_literals && !msg_id_literals) { *errorptr = US"domain literals not allowed"; *tt = 0; @@ -500,7 +512,7 @@ BOOL commas = FALSE; while (*s == '@') { *t++ = '@'; - s = read_domain(s+1, t, errorptr); + s = read_domain(s+1, t, FALSE, errorptr); if (*t == 0) return s; t += Ustrlen((const uschar *)t); if (*s != ',') break; @@ -559,7 +571,7 @@ if (*errorptr == NULL) t += Ustrlen((const uschar *)t); *t++ = *s++; *domainptr = t; - s = read_domain(s, t, errorptr); + s = read_domain(s, t, FALSE, errorptr); } return s; } @@ -619,7 +631,7 @@ uschar * parse_extract_address(const uschar *mailbox, uschar **errorptr, int *start, int *end, int *domain, BOOL allow_null) { -uschar *yield = store_get(Ustrlen(mailbox) + 1, is_tainted(mailbox)); +uschar * yield = store_get(Ustrlen(mailbox) + 1, mailbox); const uschar *startptr, *endptr; const uschar *s = US mailbox; uschar *t = US yield; @@ -647,9 +659,9 @@ followed by a route-addr (more words must follow). */ if (*s != '@' && *s != '<') { - if (*s == 0 || *s == ';') + if (!*s || *s == ';') { - if (*t == 0) FAILED(US"empty address"); + if (!*t) FAILED(US"empty address"); endptr = last_comment_position; goto PARSE_SUCCEEDED; /* Bare local part */ } @@ -740,7 +752,7 @@ if (*s == '<') } endptr = s; - if (*errorptr != NULL) goto PARSE_FAILED; + if (*errorptr) goto PARSE_FAILED; while (bracket_count-- > 0) if (*s++ != '>') { *errorptr = s[-1] == 0 @@ -759,14 +771,14 @@ should be the domain. However, for flexibility we allow for a route-address not enclosed in <> as well, which is indicated by an empty first local part preceding '@'. The source routing is, however, ignored. */ -else if (*t == 0) +else if (!*t) { uschar *domainptr = yield; 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 */ s = read_addr_spec(s, t, 0, errorptr, &domainptr); - if (*errorptr != NULL) goto PARSE_FAILED; + if (*errorptr) goto PARSE_FAILED; *domain = domainptr - yield; endptr = last_comment_position; if (*domain == 0) FAILED(US"domain missing in source-routed address"); @@ -779,8 +791,8 @@ else t += Ustrlen((const uschar *)t); *t++ = *s++; *domain = t - yield; - s = read_domain(s, t, errorptr); - if (*t == 0) goto PARSE_FAILED; + s = read_domain(s, t, TRUE, errorptr); + if (!*t) goto PARSE_FAILED; endptr = last_comment_position; } @@ -789,7 +801,7 @@ through for other cases. Endptr may have been moved over whitespace, so move it back past white space if necessary. */ PARSE_SUCCEEDED: -if (*s != 0) +if (*s) { if (f.parse_found_group && *s == ';') { @@ -863,7 +875,8 @@ Returns: pointer to the original string, if no quoting needed, or */ const uschar * -parse_quote_2047(const uschar *string, int len, uschar *charset, BOOL fold) +parse_quote_2047(const uschar *string, int len, const uschar *charset, + BOOL fold) { const uschar * s = string; int hlen, l; @@ -985,11 +998,9 @@ if (i < len) /* No non-printers; use the RFC 822 quoting rules */ if (len <= 0 || len >= INT_MAX/4) - { - return string_copy_taint(CUS"", is_tainted(phrase)); - } + return string_copy_taint(CUS"", phrase); -buffer = store_get((len+1)*4, is_tainted(phrase)); +buffer = store_get((len+1)*4, phrase); s = phrase; end = s + len; @@ -1233,8 +1244,8 @@ Returns: FF_DELIVERED addresses extracted */ int -parse_forward_list(uschar *s, int options, address_item **anchor, - uschar **error, const uschar *incoming_domain, uschar *directory, +parse_forward_list(const uschar *s, int options, address_item **anchor, + uschar **error, const uschar *incoming_domain, const uschar *directory, error_block **syntax_errors) { int count = 0; @@ -1243,18 +1254,15 @@ DEBUG(D_route) debug_printf("parse_forward_list: %s\n", s); for (;;) { - int len; - int special = 0; - int specopt = 0; - int specbit = 0; - uschar *ss, *nexts; - address_item *addr; + int len, special = 0, specopt = 0, specbit = 0; + const uschar * ss, * nexts; + address_item * addr; BOOL inquote = FALSE; for (;;) { while (isspace(*s) || *s == ',') s++; - if (*s == '#') { while (*s != 0 && *s != '\n') s++; } else break; + if (*s == '#') { while (*s && *s != '\n') s++; } else break; } /* When we reach the end of the list, we return FF_DELIVERED if any child @@ -1275,18 +1283,17 @@ for (;;) syntax error has been skipped. I now think it is the wrong approach, but have left this here just in case, and for the record. */ - #ifdef NEVER +#ifdef NEVER if (count > 0) return FF_DELIVERED; /* Something was generated */ - if (syntax_errors == NULL || /* Not skipping syntax errors, or */ - *syntax_errors == NULL) /* we didn't actually skip any */ + if (!syntax_errors || /* Not skipping syntax errors, or */ + !*syntax_errors) /* we didn't actually skip any */ return FF_NOTDELIVERED; *error = string_sprintf("no addresses generated: syntax error in %s: %s", (*syntax_errors)->text2, (*syntax_errors)->text1); return FF_ERROR; - #endif - +#endif } /* Find the end of the next address. Quoted strings in addresses may contain @@ -1303,7 +1310,7 @@ for (;;) /* Remove any trailing spaces; we know there's at least one non-space. */ - while (isspace((ss[-1]))) ss--; + while (isspace(ss[-1])) ss--; /* We now have s->start and ss->end of the next address. Remove quotes if they completely enclose, remembering the address started with a quote @@ -1316,20 +1323,14 @@ for (;;) ss--; inquote = TRUE; while (s < ss && isspace(*s)) s++; - while (ss > s && isspace((ss[-1]))) ss--; + while (ss > s && isspace(ss[-1])) ss--; } /* Set up the length of the address. */ len = ss - s; - DEBUG(D_route) - { - int save = s[len]; - s[len] = 0; - debug_printf("extract item: %s\n", s); - s[len] = save; - } + DEBUG(D_route) debug_printf("extract item: %.*s\n", len, s); /* Handle special addresses if permitted. If the address is :unknown: ignore it - this is for backward compatibility with old alias files. You @@ -1350,18 +1351,18 @@ for (;;) else if (Ustrncmp(s, ":fail:", 6) == 0) { special = FF_FAIL; specopt = RDO_FAIL; } /* specbit is 0 */ - if (special != 0) + if (special) { - uschar *ss = Ustrchr(s+1, ':') + 1; + uschar * ss = Ustrchr(s+1, ':') + 1; /* line after the special... */ if ((options & specopt) == specbit) { *error = string_sprintf("\"%.*s\" is not permitted", len, s); return FF_ERROR; } - while (*ss != 0 && isspace(*ss)) ss++; - while (s[len] != 0 && s[len] != '\n') len++; - s[len] = 0; - *error = string_copy(ss); + while (*ss && isspace(*ss)) ss++; /* skip leading whitespace */ + if ((len = Ustrlen(ss)) > 0) /* ignore trailing newlines */ + for (const uschar * t = ss + len - 1; t >= ss && *t == '\n'; t--) len--; + *error = string_copyn(ss, len); /* becomes the error */ return special; } @@ -1372,14 +1373,14 @@ for (;;) if (Ustrncmp(s, ":include:", 9) == 0) { - uschar *filebuf; + uschar * filebuf; uschar filename[256]; - uschar *t = s+9; + const uschar * t = s+9; int flen = len - 9; int frc; struct stat statbuf; - address_item *last; - FILE *f; + address_item * last; + FILE * f; while (flen > 0 && isspace(*t)) { t++; flen--; } @@ -1389,7 +1390,7 @@ for (;;) return FF_ERROR; } - if (flen > 255) + if (flen > sizeof(filename)-1) { *error = string_sprintf("included file name \"%s\" is too long", t); return FF_ERROR; @@ -1427,8 +1428,10 @@ for (;;) if (directory) { int len = Ustrlen(directory); - uschar *p = filename + len; + uschar * p; + while (len > 0 && directory[len-1] == '/') len--; /* ignore trailing '/' */ + p = filename + len; if (Ustrncmp(filename, directory, len) != 0 || *p != '/') { *error = string_sprintf("included file %s is not in directory %s", @@ -1443,7 +1446,7 @@ for (;;) with a flag that fails symlinks. */ { - int fd = exim_open2(CS directory, O_RDONLY); + int fd = exim_open2(CCS directory, O_RDONLY); if (fd < 0) { *error = string_sprintf("failed to open directory %s", directory); @@ -1453,9 +1456,10 @@ for (;;) { uschar temp; int fd2; - uschar * q = p; + uschar * q = p + 1; /* skip dividing '/' */ - while (*++p && *p != '/') ; + while (*q == '/') q++; /* skip extra '/' */ + while (*++p && *p != '/') ; /* end of component */ temp = *p; *p = '\0'; @@ -1543,7 +1547,7 @@ for (;;) return FF_ERROR; } - filebuf = store_get(statbuf.st_size + 1, is_tainted(filename)); + filebuf = store_get(statbuf.st_size + 1, filename); if (fread(filebuf, 1, statbuf.st_size, f) != statbuf.st_size) { *error = string_sprintf("error while reading included file %s: %s", @@ -1591,18 +1595,17 @@ for (;;) { int start, end, domain; const uschar *recipient = NULL; - int save = s[len]; - s[len] = 0; + uschar * s_ltd = string_copyn(s, len); /* If it starts with \ and the rest of it parses as a valid mail address without a domain, carry on with that address, but qualify it with the incoming domain. Otherwise arrange for the address to fall through, causing an error message on the re-parse. */ - if (*s == '\\') + if (*s_ltd == '\\') { recipient = - parse_extract_address(s+1, error, &start, &end, &domain, FALSE); + parse_extract_address(s_ltd+1, error, &start, &end, &domain, FALSE); if (recipient) recipient = domain != 0 ? NULL : string_sprintf("%s@%s", recipient, incoming_domain); @@ -1611,22 +1614,22 @@ for (;;) /* Try parsing the item as an address. */ if (!recipient) recipient = - parse_extract_address(s, error, &start, &end, &domain, FALSE); + parse_extract_address(s_ltd, error, &start, &end, &domain, FALSE); /* If item starts with / or | and is not a valid address, or there is no domain, treat it as a file or pipe. If it was a quoted item, remove the quoting occurrences of \ within it. */ - if ((*s == '|' || *s == '/') && (recipient == NULL || domain == 0)) + if ((*s_ltd == '|' || *s_ltd == '/') && (!recipient || domain == 0)) { - uschar *t = store_get(Ustrlen(s) + 1, is_tainted(s)); - uschar *p = t; - uschar *q = s; - while (*q != 0) + uschar * t = store_get(Ustrlen(s_ltd) + 1, s_ltd); + uschar * p = t, * q = s_ltd; + + while (*q) { if (inquote) { - *p++ = (*q == '\\')? *(++q) : *q; + *p++ = *q == '\\' ? *++q : *q; q++; } else *p++ = *q++; @@ -1634,7 +1637,7 @@ for (;;) *p = 0; addr = deliver_make_addr(t, TRUE); setflag(addr, af_pfr); /* indicates pipe/file/reply */ - if (*s != '|') setflag(addr, af_file); /* indicates file */ + if (*s_ltd != '|') setflag(addr, af_file); /* indicates file */ } /* Item must be an address. Complain if not, else qualify, rewrite and set @@ -1646,36 +1649,35 @@ for (;;) else { - if (recipient == NULL) + if (!recipient) { if (Ustrcmp(*error, "empty address") == 0) { *error = NULL; - s[len] = save; s = nexts; continue; } - if (syntax_errors != NULL) + if (syntax_errors) { - error_block *e = store_get(sizeof(error_block), FALSE); - error_block *last = *syntax_errors; - if (last == NULL) *syntax_errors = e; else + error_block * e = store_get(sizeof(error_block), GET_UNTAINTED); + error_block * last = *syntax_errors; + if (last) { - while (last->next != NULL) last = last->next; + while (last->next) last = last->next; last->next = e; } + else + *syntax_errors = e; e->next = NULL; e->text1 = *error; - e->text2 = string_copy(s); - s[len] = save; + e->text2 = s_ltd; s = nexts; continue; } else { - *error = string_sprintf("%s in \"%s\"", *error, s); - s[len] = save; /* _after_ using it for *error */ + *error = string_sprintf("%s in \"%s\"", *error, s_ltd); return FF_ERROR; } } @@ -1683,17 +1685,15 @@ for (;;) /* Address was successfully parsed. Rewrite, and then make an address block. */ - recipient = ((options & RDO_REWRITE) != 0)? - rewrite_address(recipient, TRUE, FALSE, global_rewrite_rules, - rewrite_existflags) : - rewrite_address_qualify(recipient, TRUE); /*XXX loses track of const */ + recipient = options & RDO_REWRITE + ? rewrite_address(recipient, TRUE, FALSE, global_rewrite_rules, + rewrite_existflags) + : rewrite_address_qualify(recipient, TRUE); /*XXX loses track of const */ addr = deliver_make_addr(US recipient, TRUE); /* TRUE => copy recipient, so deconst ok */ } - /* Restore the final character in the original data, and add to the - output chain. */ + /* Add the original data to the output chain. */ - s[len] = save; addr->next = *anchor; *anchor = addr; count++; @@ -1740,7 +1740,7 @@ 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. */ reset_point = store_mark(); -id = *yield = store_get(Ustrlen(str) + 1, is_tainted(str)); +id = *yield = store_get(Ustrlen(str) + 1, str); *id++ = *str++; str = read_addr_spec(str, id, '>', error, &domain);