kpdyer / regex2dfa Goto Github PK

This file, when compiled

#include "fst/fstlib.h"
#include "re2/re2.h"
int main () {
}

results in the following error.

$ g++ -o test -Ipath/to/re2 -Ipath/to/openfst test.cc
...
third_party/openfst/src/include/fst/log.h:50:9: note: previous definition is here
#define LOG(type) LogMessage(#type).stream()
        ^
In file included from test.cc:1:
In file included from third_party/re2/re2/regexp.h:89:
In file included from third_party/re2/util/util.h:118:
third_party/re2/util/logging.h:49:7: error: redefinition of 'LogMessage'
class LogMessage {
      ^
third_party/openfst/src/include/fst/log.h:34:7: note: previous definition is here
class LogMessage {
      ^

This is a placeholder issue to raise an issue against re2 and openfst, they should use namespaces.

Create a Python API for regex2dfa.

Currently, regex2dfa depends upon fstcompile, fstprint, and fstminimize being in the system path. This is not okay --- regex2dfa should be a stand-alone executable.

Use emscripten to compile regex2dfa to JS.

We don't yet have Makefile targets for static/dynamic libraries.

The regex2dfa repo needs to expose a C++ API. In order to complete this, we need to:

• Define an API for regex2dfa.
• Introduce regex2dfa.h

Currently dot . is not matching newline, which results in unexpected behavior. Specifically we have that the dfa for ^.*$ and ^\C*$ are not equal:

$ ./bin/regex2dfa -r "^.*$"
0   0   0   0
0   0   1   1
0   0   2   2
0   0   3   3
0   0   4   4
0   0   5   5
0   0   6   6
0   0   7   7
0   0   8   8
0   0   9   9
0   0   11  11

$ ./bin/regex2dfa -r "^\\C*$"
0   0   0   0
0   0   1   1
0   0   2   2
0   0   3   3
0   0   4   4
0   0   5   5
0   0   6   6
0   0   7   7
0   0   8   8
0   0   9   9
0   0   10  10
0   0   11  11

notice the missing line 0 0 10 10 for ^.*$.

Figure out a way to remove our (patched) re2 from the third_party directory.

The following regular expression, the DNS regex from l7, fails to compile to a DFA with regex2dfa 0.1.4.

^.?.?.?.?[\x01\x02].?.?.?.?.?.?[\x01-?][a-z0-9][\x01-?a-z]*[\x02-\x06][a-z][a-z][fglmoprstuvz]?[aeop]?(um)?[\x01-\x10\x1c][\x01\x03\x04\xFF]

Figure out a way to remove openfst from the third_party directory.