Contents of /code/trunk/HACKING

Technical Notes about PCRE
--------------------------

Historical note 1
-----------------

Many years ago I implemented some regular expression functions to an algorithm
suggested by Martin Richards. These were not Unix-like in form, and were quite
restricted in what they could do by comparison with Perl. The interesting part
about the algorithm was that the amount of space required to hold the compiled
form of an expression was known in advance. The code to apply an expression did
not operate by backtracking, as the original Henry Spencer code and current
Perl code does, but instead checked all possibilities simultaneously by keeping
a list of current states and checking all of them as it advanced through the
subject string. In the terminology of Jeffrey Friedl's book, it was a "DFA
algorithm". When the pattern was all used up, all remaining states were
possible matches, and the one matching the longest subset of the subject string
was chosen. This did not necessarily maximize the individual wild portions of
the pattern, as is expected in Unix and Perl-style regular expressions.

Historical note 2
-----------------

By contrast, the code originally written by Henry Spencer and subsequently
heavily modified for Perl actually compiles the expression twice: once in a
dummy mode in order to find out how much store will be needed, and then for
real. The execution function operates by backtracking and maximizing (or,
optionally, minimizing in Perl) the amount of the subject that matches
individual wild portions of the pattern. This is an "NFA algorithm" in Friedl's
terminology.

OK, here's the real stuff
-------------------------

For the set of functions that form the "basic" PCRE library (which are
unrelated to those mentioned above), I tried at first to invent an algorithm
that used an amount of store bounded by a multiple of the number of characters
in the pattern, to save on compiling time. However, because of the greater
complexity in Perl regular expressions, I couldn't do this. In any case, a
first pass through the pattern is needed, for a number of reasons. PCRE works
by running a very degenerate first pass to calculate a maximum store size, and
then a second pass to do the real compile - which may use a bit less than the
predicted amount of store. The idea is that this is going to turn out faster
because the first pass is degenerate and the second pass can just store stuff
straight into the vector, which it knows is big enough. It does make the
compiling functions bigger, of course, but they have got quite big anyway to
handle all the Perl stuff.

Traditional matching function
-----------------------------

The "traditional", and original, matching function is called pcre_exec(), and 
it implements an NFA algorithm, similar to the original Henry Spencer algorithm 
and the way that Perl works. Not surprising, since it is intended to be as 
compatible with Perl as possible. This is the function most users of PCRE will 
use most of the time.

Supplementary matching function
-------------------------------

From PCRE 6.0, there is also a supplementary matching function called 
pcre_dfa_exec(). This implements a DFA matching algorithm that searches 
simultaneously for all possible matches that start at one point in the subject 
string. (Going back to my roots: see Historical Note 1 above.) This function 
intreprets the same compiled pattern data as pcre_exec(); however, not all the 
facilities are available, and those that are don't always work in quite the 
same way. See the user documentation for details.

Format of compiled patterns
---------------------------

The compiled form of a pattern is a vector of bytes, containing items of
variable length. The first byte in an item is an opcode, and the length of the
item is either implicit in the opcode or contained in the data bytes that
follow it. 

In many cases below "two-byte" data values are specified. This is in fact just
a default. PCRE can be compiled to use 3-byte or 4-byte values (impairing the
performance). This is necessary only when patterns whose compiled length is
greater than 64K are going to be processed. In this description, we assume the 
"normal" compilation options.

A list of all the opcodes follows:

Opcodes with no following data
------------------------------

These items are all just one byte long

  OP_END                 end of pattern
  OP_ANY                 match any character
  OP_ANYBYTE             match any single byte, even in UTF-8 mode
  OP_SOD                 match start of data: \A
  OP_SOM,                start of match (subject + offset): \G
  OP_CIRC                ^ (start of data, or after \n in multiline)
  OP_NOT_WORD_BOUNDARY   \W
  OP_WORD_BOUNDARY       \w
  OP_NOT_DIGIT           \D
  OP_DIGIT               \d
  OP_NOT_WHITESPACE      \S
  OP_WHITESPACE          \s
  OP_NOT_WORDCHAR        \W
  OP_WORDCHAR            \w
  OP_EODN                match end of data or \n at end: \Z
  OP_EOD                 match end of data: \z
  OP_DOLL                $ (end of data, or before \n in multiline)
  OP_EXTUNI              match an extended Unicode character 
  

Repeating single characters
---------------------------

The common repeats (*, +, ?) when applied to a single character use the
following opcodes:

  OP_STAR
  OP_MINSTAR
  OP_PLUS
  OP_MINPLUS
  OP_QUERY
  OP_MINQUERY

In ASCII mode, these are two-byte items; in UTF-8 mode, the length is variable.
Those with "MIN" in their name are the minimizing versions. Each is followed by
the character that is to be repeated. Other repeats make use of

  OP_UPTO
  OP_MINUPTO
  OP_EXACT

which are followed by a two-byte count (most significant first) and the
repeated character. OP_UPTO matches from 0 to the given number. A repeat with a
non-zero minimum and a fixed maximum is coded as an OP_EXACT followed by an
OP_UPTO (or OP_MINUPTO).


Repeating character types
-------------------------

Repeats of things like \d are done exactly as for single characters, except
that instead of a character, the opcode for the type is stored in the data
byte. The opcodes are:

  OP_TYPESTAR
  OP_TYPEMINSTAR
  OP_TYPEPLUS
  OP_TYPEMINPLUS
  OP_TYPEQUERY
  OP_TYPEMINQUERY
  OP_TYPEUPTO
  OP_TYPEMINUPTO
  OP_TYPEEXACT


Match by Unicode property
-------------------------

OP_PROP and OP_NOTPROP are used for positive and negative matches of a 
character by testing its Unicode property (the \p and \P escape sequences).
Each is followed by a single byte that encodes the desired property value.

Repeats of these items use the OP_TYPESTAR etc. set of opcodes, followed by two 
bytes: OP_PROP or OP_NOTPROP and then the desired property value.


Matching literal characters
---------------------------

The OP_CHAR opcode is followed by a single character that is to be matched 
casefully. For caseless matching, OP_CHARNC is used. In UTF-8 mode, the 
character may be more than one byte long. (Earlier versions of PCRE used 
multi-character strings, but this was changed to allow some new features to be 
added.)


Character classes
-----------------

If there is only one character, OP_CHAR or OP_CHARNC is used for a positive
class, and OP_NOT for a negative one (that is, for something like [^a]).
However, in UTF-8 mode, the use of OP_NOT applies only to characters with
values < 128, because OP_NOT is confined to single bytes.

Another set of repeating opcodes (OP_NOTSTAR etc.) are used for a repeated,
negated, single-character class. The normal ones (OP_STAR etc.) are used for a
repeated positive single-character class.

When there's more than one character in a class and all the characters are less
than 256, OP_CLASS is used for a positive class, and OP_NCLASS for a negative
one. In either case, the opcode is followed by a 32-byte bit map containing a 1
bit for every character that is acceptable. The bits are counted from the least
significant end of each byte.

The reason for having both OP_CLASS and OP_NCLASS is so that, in UTF-8 mode,
subject characters with values greater than 256 can be handled correctly. For
OP_CLASS they don't match, whereas for OP_NCLASS they do.

For classes containing characters with values > 255, OP_XCLASS is used. It
optionally uses a bit map (if any characters lie within it), followed by a list
of pairs and single characters. There is a flag character than indicates
whether it's a positive or a negative class.


Back references
---------------

OP_REF is followed by two bytes containing the reference number.


Repeating character classes and back references
-----------------------------------------------

Single-character classes are handled specially (see above). This applies to
OP_CLASS and OP_REF. In both cases, the repeat information follows the base
item. The matching code looks at the following opcode to see if it is one of

  OP_CRSTAR
  OP_CRMINSTAR
  OP_CRPLUS
  OP_CRMINPLUS
  OP_CRQUERY
  OP_CRMINQUERY
  OP_CRRANGE
  OP_CRMINRANGE

All but the last two are just single-byte items. The others are followed by
four bytes of data, comprising the minimum and maximum repeat counts.


Brackets and alternation
------------------------

A pair of non-capturing (round) brackets is wrapped round each expression at
compile time, so alternation always happens in the context of brackets.

Non-capturing brackets use the opcode OP_BRA, while capturing brackets use
OP_BRA+1, OP_BRA+2, etc. [Note for North Americans: "bracket" to some English
speakers, including myself, can be round, square, curly, or pointy. Hence this
usage.]

Originally PCRE was limited to 99 capturing brackets (so as not to use up all
the opcodes). From release 3.5, there is no limit. What happens is that the
first ones, up to EXTRACT_BASIC_MAX are handled with separate opcodes, as
above. If there are more, the opcode is set to EXTRACT_BASIC_MAX+1, and the
first operation in the bracket is OP_BRANUMBER, followed by a 2-byte bracket
number. This opcode is ignored while matching, but is fished out when handling
the bracket itself. (They could have all been done like this, but I was making
minimal changes.)

A bracket opcode is followed by LINK_SIZE bytes which give the offset to the
next alternative OP_ALT or, if there aren't any branches, to the matching
OP_KET opcode. Each OP_ALT is followed by LINK_SIZE bytes giving the offset to
the next one, or to the OP_KET opcode.

OP_KET is used for subpatterns that do not repeat indefinitely, while
OP_KETRMIN and OP_KETRMAX are used for indefinite repetitions, minimally or
maximally respectively. All three are followed by LINK_SIZE bytes giving (as a
positive number) the offset back to the matching OP_BRA opcode.

If a subpattern is quantified such that it is permitted to match zero times, it
is preceded by one of OP_BRAZERO or OP_BRAMINZERO. These are single-byte
opcodes which tell the matcher that skipping this subpattern entirely is a
valid branch.

A subpattern with an indefinite maximum repetition is replicated in the
compiled data its minimum number of times (or once with OP_BRAZERO if the
minimum is zero), with the final copy terminating with OP_KETRMIN or OP_KETRMAX
as appropriate.

A subpattern with a bounded maximum repetition is replicated in a nested
fashion up to the maximum number of times, with OP_BRAZERO or OP_BRAMINZERO
before each replication after the minimum, so that, for example, (abc){2,5} is
compiled as (abc)(abc)((abc)((abc)(abc)?)?)?.


Assertions
----------

Forward assertions are just like other subpatterns, but starting with one of
the opcodes OP_ASSERT or OP_ASSERT_NOT. Backward assertions use the opcodes
OP_ASSERTBACK and OP_ASSERTBACK_NOT, and the first opcode inside the assertion
is OP_REVERSE, followed by a two byte count of the number of characters to move
back the pointer in the subject string. When operating in UTF-8 mode, the count
is a character count rather than a byte count. A separate count is present in
each alternative of a lookbehind assertion, allowing them to have different
fixed lengths.


Once-only subpatterns
---------------------

These are also just like other subpatterns, but they start with the opcode
OP_ONCE.


Conditional subpatterns
-----------------------

These are like other subpatterns, but they start with the opcode OP_COND. If
the condition is a back reference, this is stored at the start of the
subpattern using the opcode OP_CREF followed by two bytes containing the
reference number. If the condition is "in recursion" (coded as "(?(R)"), the
same scheme is used, with a "reference number" of 0xffff. Otherwise, a
conditional subpattern always starts with one of the assertions.


Recursion
---------

Recursion either matches the current regex, or some subexpression. The opcode
OP_RECURSE is followed by an value which is the offset to the starting bracket
from the start of the whole pattern. From release 6.5, OP_RECURSE is 
automatically wrapped inside OP_ONCE brackets (because otherwise some patterns 
broke it). OP_RECURSE is also used for "subroutine" calls, even though they 
are not strictly a recursion.


Callout
-------

OP_CALLOUT is followed by one byte of data that holds a callout number in the
range 0 to 254 for manual callouts, or 255 for an automatic callout. In both 
cases there follows a two-byte value giving the offset in the pattern to the
start of the following item, and another two-byte item giving the length of the
next item.


Changing options
----------------

If any of the /i, /m, or /s options are changed within a pattern, an OP_OPT
opcode is compiled, followed by one byte containing the new settings of these
flags. If there are several alternatives, there is an occurrence of OP_OPT at
the start of all those following the first options change, to set appropriate
options for the start of the alternative. Immediately after the end of the
group there is another such item to reset the flags to their previous values. A
change of flag right at the very start of the pattern can be handled entirely
at compile time, and so does not cause anything to be put into the compiled
data.

Philip Hazel
January 2006
1	Technical Notes about PCRE
2	--------------------------
3
4	Historical note 1
5	-----------------
6
7	Many years ago I implemented some regular expression functions to an algorithm
8	suggested by Martin Richards. These were not Unix-like in form, and were quite
9	restricted in what they could do by comparison with Perl. The interesting part
10	about the algorithm was that the amount of space required to hold the compiled
11	form of an expression was known in advance. The code to apply an expression did
12	not operate by backtracking, as the original Henry Spencer code and current
13	Perl code does, but instead checked all possibilities simultaneously by keeping
14	a list of current states and checking all of them as it advanced through the
15	subject string. In the terminology of Jeffrey Friedl's book, it was a "DFA
16	algorithm". When the pattern was all used up, all remaining states were
17	possible matches, and the one matching the longest subset of the subject string
18	was chosen. This did not necessarily maximize the individual wild portions of
19	the pattern, as is expected in Unix and Perl-style regular expressions.
20
21	Historical note 2
22	-----------------
23
24	By contrast, the code originally written by Henry Spencer and subsequently
25	heavily modified for Perl actually compiles the expression twice: once in a
26	dummy mode in order to find out how much store will be needed, and then for
27	real. The execution function operates by backtracking and maximizing (or,
28	optionally, minimizing in Perl) the amount of the subject that matches
29	individual wild portions of the pattern. This is an "NFA algorithm" in Friedl's
30	terminology.
31
32	OK, here's the real stuff
33	-------------------------
34
35	For the set of functions that form the "basic" PCRE library (which are
36	unrelated to those mentioned above), I tried at first to invent an algorithm
37	that used an amount of store bounded by a multiple of the number of characters
38	in the pattern, to save on compiling time. However, because of the greater
39	complexity in Perl regular expressions, I couldn't do this. In any case, a
40	first pass through the pattern is needed, for a number of reasons. PCRE works
41	by running a very degenerate first pass to calculate a maximum store size, and
42	then a second pass to do the real compile - which may use a bit less than the
43	predicted amount of store. The idea is that this is going to turn out faster
44	because the first pass is degenerate and the second pass can just store stuff
45	straight into the vector, which it knows is big enough. It does make the
46	compiling functions bigger, of course, but they have got quite big anyway to
47	handle all the Perl stuff.
48
49	Traditional matching function
50	-----------------------------
51
52	The "traditional", and original, matching function is called pcre_exec(), and
53	it implements an NFA algorithm, similar to the original Henry Spencer algorithm
54	and the way that Perl works. Not surprising, since it is intended to be as
55	compatible with Perl as possible. This is the function most users of PCRE will
56	use most of the time.
57
58	Supplementary matching function
59	-------------------------------
60
61	From PCRE 6.0, there is also a supplementary matching function called
62	pcre_dfa_exec(). This implements a DFA matching algorithm that searches
63	simultaneously for all possible matches that start at one point in the subject
64	string. (Going back to my roots: see Historical Note 1 above.) This function
65	intreprets the same compiled pattern data as pcre_exec(); however, not all the
66	facilities are available, and those that are don't always work in quite the
67	same way. See the user documentation for details.
68
69	Format of compiled patterns
70	---------------------------
71
72	The compiled form of a pattern is a vector of bytes, containing items of
73	variable length. The first byte in an item is an opcode, and the length of the
74	item is either implicit in the opcode or contained in the data bytes that
75	follow it.
76
77	In many cases below "two-byte" data values are specified. This is in fact just
78	a default. PCRE can be compiled to use 3-byte or 4-byte values (impairing the
79	performance). This is necessary only when patterns whose compiled length is
80	greater than 64K are going to be processed. In this description, we assume the
81	"normal" compilation options.
82
83	A list of all the opcodes follows:
84
85	Opcodes with no following data
86	------------------------------
87
88	These items are all just one byte long
89
90	OP_END end of pattern
91	OP_ANY match any character
92	OP_ANYBYTE match any single byte, even in UTF-8 mode
93	OP_SOD match start of data: \A
94	OP_SOM, start of match (subject + offset): \G
95	OP_CIRC ^ (start of data, or after \n in multiline)
96	OP_NOT_WORD_BOUNDARY \W
97	OP_WORD_BOUNDARY \w
98	OP_NOT_DIGIT \D
99	OP_DIGIT \d
100	OP_NOT_WHITESPACE \S
101	OP_WHITESPACE \s
102	OP_NOT_WORDCHAR \W
103	OP_WORDCHAR \w
104	OP_EODN match end of data or \n at end: \Z
105	OP_EOD match end of data: \z
106	OP_DOLL $ (end of data, or before \n in multiline)
107	OP_EXTUNI match an extended Unicode character
108
109
110	Repeating single characters
111	---------------------------
112
113	The common repeats (*, +, ?) when applied to a single character use the
114	following opcodes:
115
116	OP_STAR
117	OP_MINSTAR
118	OP_PLUS
119	OP_MINPLUS
120	OP_QUERY
121	OP_MINQUERY
122
123	In ASCII mode, these are two-byte items; in UTF-8 mode, the length is variable.
124	Those with "MIN" in their name are the minimizing versions. Each is followed by
125	the character that is to be repeated. Other repeats make use of
126
127	OP_UPTO
128	OP_MINUPTO
129	OP_EXACT
130
131	which are followed by a two-byte count (most significant first) and the
132	repeated character. OP_UPTO matches from 0 to the given number. A repeat with a
133	non-zero minimum and a fixed maximum is coded as an OP_EXACT followed by an
134	OP_UPTO (or OP_MINUPTO).
135
136
137	Repeating character types
138	-------------------------
139
140	Repeats of things like \d are done exactly as for single characters, except
141	that instead of a character, the opcode for the type is stored in the data
142	byte. The opcodes are:
143
144	OP_TYPESTAR
145	OP_TYPEMINSTAR
146	OP_TYPEPLUS
147	OP_TYPEMINPLUS
148	OP_TYPEQUERY
149	OP_TYPEMINQUERY
150	OP_TYPEUPTO
151	OP_TYPEMINUPTO
152	OP_TYPEEXACT
153
154
155	Match by Unicode property
156	-------------------------
157
158	OP_PROP and OP_NOTPROP are used for positive and negative matches of a
159	character by testing its Unicode property (the \p and \P escape sequences).
160	Each is followed by a single byte that encodes the desired property value.
161
162	Repeats of these items use the OP_TYPESTAR etc. set of opcodes, followed by two
163	bytes: OP_PROP or OP_NOTPROP and then the desired property value.
164
165
166	Matching literal characters
167	---------------------------
168
169	The OP_CHAR opcode is followed by a single character that is to be matched
170	casefully. For caseless matching, OP_CHARNC is used. In UTF-8 mode, the
171	character may be more than one byte long. (Earlier versions of PCRE used
172	multi-character strings, but this was changed to allow some new features to be
173	added.)
174
175
176	Character classes
177	-----------------
178
179	If there is only one character, OP_CHAR or OP_CHARNC is used for a positive
180	class, and OP_NOT for a negative one (that is, for something like [^a]).
181	However, in UTF-8 mode, the use of OP_NOT applies only to characters with
182	values < 128, because OP_NOT is confined to single bytes.
183
184	Another set of repeating opcodes (OP_NOTSTAR etc.) are used for a repeated,
185	negated, single-character class. The normal ones (OP_STAR etc.) are used for a
186	repeated positive single-character class.
187
188	When there's more than one character in a class and all the characters are less
189	than 256, OP_CLASS is used for a positive class, and OP_NCLASS for a negative
190	one. In either case, the opcode is followed by a 32-byte bit map containing a 1
191	bit for every character that is acceptable. The bits are counted from the least
192	significant end of each byte.
193
194	The reason for having both OP_CLASS and OP_NCLASS is so that, in UTF-8 mode,
195	subject characters with values greater than 256 can be handled correctly. For
196	OP_CLASS they don't match, whereas for OP_NCLASS they do.
197
198	For classes containing characters with values > 255, OP_XCLASS is used. It
199	optionally uses a bit map (if any characters lie within it), followed by a list
200	of pairs and single characters. There is a flag character than indicates
201	whether it's a positive or a negative class.
202
203
204	Back references
205	---------------
206
207	OP_REF is followed by two bytes containing the reference number.
208
209
210	Repeating character classes and back references
211	-----------------------------------------------
212
213	Single-character classes are handled specially (see above). This applies to
214	OP_CLASS and OP_REF. In both cases, the repeat information follows the base
215	item. The matching code looks at the following opcode to see if it is one of
216
217	OP_CRSTAR
218	OP_CRMINSTAR
219	OP_CRPLUS
220	OP_CRMINPLUS
221	OP_CRQUERY
222	OP_CRMINQUERY
223	OP_CRRANGE
224	OP_CRMINRANGE
225
226	All but the last two are just single-byte items. The others are followed by
227	four bytes of data, comprising the minimum and maximum repeat counts.
228
229
230	Brackets and alternation
231	------------------------
232
233	A pair of non-capturing (round) brackets is wrapped round each expression at
234	compile time, so alternation always happens in the context of brackets.
235
236	Non-capturing brackets use the opcode OP_BRA, while capturing brackets use
237	OP_BRA+1, OP_BRA+2, etc. [Note for North Americans: "bracket" to some English
238	speakers, including myself, can be round, square, curly, or pointy. Hence this
239	usage.]
240
241	Originally PCRE was limited to 99 capturing brackets (so as not to use up all
242	the opcodes). From release 3.5, there is no limit. What happens is that the
243	first ones, up to EXTRACT_BASIC_MAX are handled with separate opcodes, as
244	above. If there are more, the opcode is set to EXTRACT_BASIC_MAX+1, and the
245	first operation in the bracket is OP_BRANUMBER, followed by a 2-byte bracket
246	number. This opcode is ignored while matching, but is fished out when handling
247	the bracket itself. (They could have all been done like this, but I was making
248	minimal changes.)
249
250	A bracket opcode is followed by LINK_SIZE bytes which give the offset to the
251	next alternative OP_ALT or, if there aren't any branches, to the matching
252	OP_KET opcode. Each OP_ALT is followed by LINK_SIZE bytes giving the offset to
253	the next one, or to the OP_KET opcode.
254
255	OP_KET is used for subpatterns that do not repeat indefinitely, while
256	OP_KETRMIN and OP_KETRMAX are used for indefinite repetitions, minimally or
257	maximally respectively. All three are followed by LINK_SIZE bytes giving (as a
258	positive number) the offset back to the matching OP_BRA opcode.
259
260	If a subpattern is quantified such that it is permitted to match zero times, it
261	is preceded by one of OP_BRAZERO or OP_BRAMINZERO. These are single-byte
262	opcodes which tell the matcher that skipping this subpattern entirely is a
263	valid branch.
264
265	A subpattern with an indefinite maximum repetition is replicated in the
266	compiled data its minimum number of times (or once with OP_BRAZERO if the
267	minimum is zero), with the final copy terminating with OP_KETRMIN or OP_KETRMAX
268	as appropriate.
269
270	A subpattern with a bounded maximum repetition is replicated in a nested
271	fashion up to the maximum number of times, with OP_BRAZERO or OP_BRAMINZERO
272	before each replication after the minimum, so that, for example, (abc){2,5} is
273	compiled as (abc)(abc)((abc)((abc)(abc)?)?)?.
274
275
276	Assertions
277	----------
278
279	Forward assertions are just like other subpatterns, but starting with one of
280	the opcodes OP_ASSERT or OP_ASSERT_NOT. Backward assertions use the opcodes
281	OP_ASSERTBACK and OP_ASSERTBACK_NOT, and the first opcode inside the assertion
282	is OP_REVERSE, followed by a two byte count of the number of characters to move
283	back the pointer in the subject string. When operating in UTF-8 mode, the count
284	is a character count rather than a byte count. A separate count is present in
285	each alternative of a lookbehind assertion, allowing them to have different
286	fixed lengths.
287
288
289	Once-only subpatterns
290	---------------------
291
292	These are also just like other subpatterns, but they start with the opcode
293	OP_ONCE.
294
295
296	Conditional subpatterns
297	-----------------------
298
299	These are like other subpatterns, but they start with the opcode OP_COND. If
300	the condition is a back reference, this is stored at the start of the
301	subpattern using the opcode OP_CREF followed by two bytes containing the
302	reference number. If the condition is "in recursion" (coded as "(?(R)"), the
303	same scheme is used, with a "reference number" of 0xffff. Otherwise, a
304	conditional subpattern always starts with one of the assertions.
305
306
307	Recursion
308	---------
309
310	Recursion either matches the current regex, or some subexpression. The opcode
311	OP_RECURSE is followed by an value which is the offset to the starting bracket
312	from the start of the whole pattern. From release 6.5, OP_RECURSE is
313	automatically wrapped inside OP_ONCE brackets (because otherwise some patterns
314	broke it). OP_RECURSE is also used for "subroutine" calls, even though they
315	are not strictly a recursion.
316
317
318	Callout
319	-------
320
321	OP_CALLOUT is followed by one byte of data that holds a callout number in the
322	range 0 to 254 for manual callouts, or 255 for an automatic callout. In both
323	cases there follows a two-byte value giving the offset in the pattern to the
324	start of the following item, and another two-byte item giving the length of the
325	next item.
326
327
328	Changing options
329	----------------
330
331	If any of the /i, /m, or /s options are changed within a pattern, an OP_OPT
332	opcode is compiled, followed by one byte containing the new settings of these
333	flags. If there are several alternatives, there is an occurrence of OP_OPT at
334	the start of all those following the first options change, to set appropriate
335	options for the start of the alternative. Immediately after the end of the
336	group there is another such item to reset the flags to their previous values. A
337	change of flag right at the very start of the pattern can be handled entirely
338	at compile time, and so does not cause anything to be put into the compiled
339	data.
340
341	Philip Hazel
342	January 2006