1This code is structured roughly as follows:
2
3xmalloc.c:
4  - Wrappers for the malloc() functions, for error generation and
5    memory leak checking purposes.
6
7tre-mem.c:
8  - A simple and efficient memory allocator.
9
10tre-stack.c:
11  - Implements a simple stack data structure.
12
13tre-ast.c:
14  - Abstract syntax tree (AST) definitions.
15
16tre-parse.c:
17  - Regexp parser.  Parses a POSIX regexp (with TRE extensions) into
18    an abstract syntax tree (AST).
19
20tre-compile.c:
21  - Compiles ASTs to ready-to-use regex objects.  Comprised of two parts:
22    * Routine to convert an AST to a tagged AST.  A tagged AST has
23      appropriate minimized or maximized tags added to keep track of
24      submatches.
25    * Routine to convert tagged ASTs to tagged nondeterministic state
26      machines (TNFAs) without epsilon transitions (transitions on
27      empty strings).
28
29tre-match-parallel.c:
30  - Parallel TNFA matcher.
31    * The matcher basically takes a string and a TNFA and finds the
32      leftmost longest match and submatches in one pass over the input
33      string.  Only the beginning of the input string is scanned until
34      a leftmost match and longest match is found.
35    * The matcher cannot handle back references, but the worst case
36      time consumption is O(l) where l is the length of the input
37      string.  The space consumption is constant.
38
39tre-match-backtrack.c:
40  - A traditional backtracking matcher.
41    * Like the parallel matcher, takes a string and a TNFA and finds
42      the leftmost longest match and submatches.  Portions of the
43      input string may (and usually are) scanned multiple times.
44    * Can handle back references.  The worst case time consumption,
45      however, is O(k^l) where k is some constant and l is the length
46      of the input string.  The worst case space consumption is O(l).
47
48tre-match-approx.c:
49  - Approximate parallel TNFA matcher.
50    * Finds the leftmost and longest match and submatches in one pass
51      over the input string.  The match may contain errors.  Each
52      missing, substituted, or extra character in the match increases
53      the cost of the match.  A maximum cost for the returned match
54      can be given.  The cost of the found match is returned.
55    * Cannot handle back references.  The space and time consumption
56      bounds are the same as for the parallel exact matcher, but
57      in general this matcher is slower than the exact matcher.
58
59regcomp.c:
60  - Implementation of the regcomp() family of functions as simple
61    wrappers for tre_compile().
62
63regexec.c:
64  - Implementation of the regexec() family of functions.
65    * The appropriate matcher is dispatched according to the
66      features used in the compiled regex object.
67
68regerror.c:
69  - Implements the regerror() function.
70