1PCRE_TABLE(5) File Formats Manual PCRE_TABLE(5)
2
6 pcre_table - format of Postfix PCRE tables
7
9 postmap -q "string" pcre:/etc/postfix/filename
10 postmap -q - pcre:/etc/postfix/filename <inputfile
12 postmap -hmq - pcre:/etc/postfix/filename <inputfile
14 postmap -bmq - pcre:/etc/postfix/filename <inputfile
16
18 The Postfix mail system uses optional tables for address rewriting,
19 mail routing, or access control. These tables are usually in dbm or db
20 format.
21 Alternatively, lookup tables can be specified in Perl Compatible Regu‐
23 lar Expression form. In this case, each input is compared against a
24 list of patterns. When a match is found, the corresponding result is
25 returned and the search is terminated.
26 To find out what types of lookup tables your Postfix system supports
28 use the "postconf -m" command.
29 To test lookup tables, use the "postmap -q" command as described in the
31 SYNOPSIS above. Use "postmap -hmq - <file" for header_checks(5) pat‐
32 terns, and "postmap -bmq - <file" for body_checks(5) (Postfix 2.6 and
33 later).
34 This driver can be built with the pcre2 library (Postfix 3.7 and
36 later), or with the legacy pcre library (all Postfix versions).
37
39 With Postfix version 2.2 and earlier specify "postmap -fq" to query a
40 table that contains case sensitive patterns. Patterns are case insensi‐
41 tive by default.
42
44 The general form of a PCRE table is:
45 /pattern/flags result
47 When pattern matches the input string, use the corresponding re‐
48 sult value.
49 !/pattern/flags result
51 When pattern does not match the input string, use the corre‐
52 sponding result value.
53 if /pattern/flags
55 endif If the input string matches /pattern/, then match that input
57 string against the patterns between if and endif. The if..endif
58 can nest.
59 Note: do not prepend whitespace to patterns inside if..endif.
61 This feature is available in Postfix 2.1 and later.
63 if !/pattern/flags
65 endif If the input string does not match /pattern/, then match that
67 input string against the patterns between if and endif. The
68 if..endif can nest.
69 Note: do not prepend whitespace to patterns inside if..endif.
71 This feature is available in Postfix 2.1 and later.
73 blank lines and comments
75 Empty lines and whitespace-only lines are ignored, as are lines
76 whose first non-whitespace character is a `#'.
77 multi-line text
79 A logical line starts with non-whitespace text. A line that
80 starts with whitespace continues a logical line.
81 Each pattern is a perl-like regular expression. The expression delim‐
83 iter can be any non-alphanumeric character, except whitespace or char‐
84 acters that have special meaning (traditionally the forward slash is
85 used). The regular expression can contain whitespace.
86 By default, matching is case-insensitive, and newlines are not treated
88 as special characters. The behavior is controlled by flags, which are
89 toggled by appending one or more of the following characters after the
90 pattern:
91 i (default: on)
93 Toggles the case sensitivity flag. By default, matching is case
94 insensitive.
95 m (default: off)
97 Toggles the pcre MULTILINE flag. When this flag is on, the ^ and
98 $ metacharacters match immediately after and immediately before
99 a newline character, respectively, in addition to matching at
100 the start and end of the subject string.
101 s (default: on)
103 Toggles the pcre DOTALL flag. When this flag is on, the .
104 metacharacter matches the newline character. With Postfix ver‐
105 sions prior to 2.0, the flag is off by default, which is incon‐
106 venient for multi-line message header matching.
107 x (default: off)
109 Toggles the pcre extended flag. When this flag is on, whitespace
110 characters in the pattern (other than in a character class) are
111 ignored. To include a whitespace character as part of the pat‐
112 tern, escape it with backslash.
113 Note: do not use #comment after patterns.
115 A (default: off)
117 Toggles the pcre ANCHORED flag. When this flag is on, the pat‐
118 tern is forced to be "anchored", that is, it is constrained to
119 match only at the start of the string which is being searched
120 (the "subject string"). This effect can also be achieved by ap‐
121 propriate constructs in the pattern itself.
122 E (default: off)
124 Toggles the pcre DOLLAR_ENDONLY flag. When this flag is on, a $
125 metacharacter in the pattern matches only at the end of the sub‐
126 ject string. Without this flag, a dollar also matches immedi‐
127 ately before the final character if it is a newline character
128 (but not before any other newline characters). This flag is ig‐
129 nored if the pcre MULTILINE flag is set.
130 U (default: off)
132 Toggles the pcre UNGREEDY flag. When this flag is on, the pat‐
133 tern matching engine inverts the "greediness" of the quantifiers
134 so that they are not greedy by default, but become greedy if
135 followed by "?". This flag can also set by a (?U) modifier
136 within the pattern.
137 X (default: off)
139 Toggles the pcre EXTRA flag. When this flag is on, any back‐
140 slash in a pattern that is followed by a letter that has no spe‐
141 cial meaning causes an error, thus reserving these combinations
142 for future expansion.
143 This feature is not supported with PCRE2.
145
147 Patterns are applied in the order as specified in the table, until a
148 pattern is found that matches the input string.
149 Each pattern is applied to the entire input string. Depending on the
151 application, that string is an entire client hostname, an entire client
152 IP address, or an entire mail address. Thus, no parent domain or par‐
153 ent network search is done, and user@domain mail addresses are not bro‐
154 ken up into their user and domain constituent parts, nor is user+foo
155 broken up into user and foo.
156
158 Substitution of substrings (text that matches patterns inside "()")
159 from the matched expression into the result string is requested with
160 $1, $2, etc.; specify $$ to produce a $ character as output. The
161 macros in the result string may need to be written as ${n} or $(n) if
162 they aren't followed by whitespace. This feature does not support
163 pcre2 substring names.
164 Note: since negated patterns (those preceded by !) return a result when
166 the expression does not match, substitutions are not available for
167 negated patterns.
168
170 The contents of a table may be specified in the table name (Postfix 3.7
171 and later). The basic syntax is:
172 main.cf:
174 parameter = .. pcre:{ { rule-1 }, { rule-2 } .. } ..
175 master.cf:
177 .. -o { parameter = .. pcre:{ { rule-1 }, { rule-2 } .. } .. } ..
178 Postfix ignores whitespace after '{' and before '}', and writes each
180 rule as one text line to an in-memory file:
181 in-memory file:
183 rule-1
184 rule-2
185 ..
186 Postfix parses the result as if it is a file in /etc/postfix.
188 Note: if a rule contains $, specify $$ to keep Postfix from trying to
190 do $name expansion as it evaluates a parameter value.
191
193 # Protect your outgoing majordomo exploders
194 /^(?!owner-)(.*)-outgoing@(.*)/ 550 Use ${1}@${2} instead
195 # Bounce friend@whatever, except when whatever is our domain (you would
197 # be better just bouncing all friend@ mail - this is just an example).
198 /^(friend@(?!my\.domain$).*)$/ 550 Stick this in your pipe $1
199 # A multi-line entry. The text is sent as one line.
201 #
202 /^noddy@my\.domain$/
203 550 This user is a funny one. You really don't want to send mail to
204 them as it only makes their head spin.
205
207 /^Subject: make money fast/ REJECT
208 /^To: friend@public\.com/ REJECT
209
211 # First skip over base 64 encoded text to save CPU cycles.
212 # Requires PCRE version 3.
213 ~^[[:alnum:]+/]{60,}$~ OK
214 # Put your own body patterns here.
216
218 postmap(1), Postfix lookup table manager
219 postconf(5), configuration parameters
220 regexp_table(5), format of POSIX regular expression tables
221
223 Use "postconf readme_directory" or "postconf html_directory" to locate
224 this information.
225 DATABASE_README, Postfix lookup table overview
226
228 The PCRE table lookup code was originally written by:
229 Andrew McNamara
230 andrewm@connect.com.au
231 connect.com.au Pty. Ltd.
232 Level 3, 213 Miller St
233 North Sydney, NSW, Australia
234 Adopted and adapted by:
236 Wietse Venema
237 IBM T.J. Watson Research
238 P.O. Box 704
239 Yorktown Heights, NY 10598, USA
240 Wietse Venema
242 Google, Inc.
243 111 8th Avenue
244 New York, NY 10011, USA
245 PCRE_TABLE(5)