Initial commit

api/noose.hpp

@@ -0,0 +1,87 @@
+#ifndef NOOSE_HPP
+#define NOOSE_HPP
+#include "../headers/rope.hpp"
+
+#include <string>
+
+struct Exp;
+
+enum class ExpKind { RANGE, OR, SEQ, ANY, NONE };
+
+struct ExRange {
+  bool negate = false;
+  char start;
+  char end;
+};
+
+struct OpOr {
+  Exp *left;
+  Exp *right;
+};
+
+struct OpSeq {
+  Exp *left;
+  Exp *right;
+};
+
+struct Exp {
+  bool capture;
+  ExpKind kind;
+  union {
+    OpOr *opor;
+    OpSeq *opseq;
+    std::vector<ExRange> ranges;
+  };
+  Exp() {
+    capture = false;
+    kind = ExpKind::NONE;
+  }
+};
+
+struct Parser {
+  std::string s;
+  size_t i;
+  Parser(std::string str) : s(str), i(0) {}
+};
+
+enum Op {
+  // These jump around
+  JMP = 0, // Jump to j.x
+  FRK = 1, // Fork to j.x and j.y (with priority to x)
+  // These consume 1 char from the input, if not then fail thread
+  // (failuire of main thread is not successfull match)
+  MCH = 2, // match with range object
+  NMC = 3, // not match with range object
+  ANY = 4, // Anything
+  // Used to save offsets
+  SVS = 5, // Start save for i cap group
+  SVE = 6, // End save for i cap group
+  // Match is successful if main thread reaches the end
+  END = 7
+};
+
+struct Range { // use start == end to match a particular char
+  char start;
+  char end;
+};
+
+struct Inst {
+  Op op;
+  union {
+    struct {
+      Range *ranges;
+      int len;
+    } r;
+    struct {
+      int x, y;
+    } j;
+  };
+  int idx;
+};
+
+Exp *regex_to_ast(std::string pattern);
+Inst *compile_ast(Exp *root);
+Inst *compile_regex(std::string pattern);
+int next_match(Inst *prog, ByteIterator *it, uint32_t *saved);
+
+#endif