| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| |
|
|
|
|
| |
Nested captures can be recognized because they start (and end) inside
the character range of the full capture. This optimization can remove
a lot of 'close' captures from the capture logs.
|
| |
|
|
|
| |
That uses 4 bytes (uint) instead of 8 (char*); the size of the
structure 'Capture' reduces from 16 to 8 bytes in 64-bit machines.
|
| | |
|
| |
|
|
|
|
|
| |
Most patterns do not have code, as they are not directly used for
a match; they are created only to compose larger patterns. So, we
shouldn't waste space to store the size of their code, as a NULL
pointer already indicates that the size is zero.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
So that we can get its space for other uses, if needed.
|
| |
|
|
|
| |
Removed '$Id' from all files + updated copyright year + other changes
in comments and documentation
|
| |
|
|
|
| |
New constructor 'lpeg.utfR(from, to)' creates a pattern that matches
UTF-8 byte sequences representing code points in the range [from, to].
|
| |
|
|
|
|
| |
Added a new tree-type node 'TXInfo', which follows 'TRule' nodes,
to store extra information about a node. (In this case, the rule
number, with an 'unsigned short' field.)
|
| |
|
|
|
| |
Version numbers and dates from RCS keyword strings removed from all
source files; keeps only the file name.
|
|
|
LPeg repository is being moved to git. Past versions won't be moved;
they are still available in RCS.
|
