xbintsc 0.3.15 → 0.3.18

package/runtime/xt_regex/parser.inc

@@ -0,0 +1,236 @@
+/* -- parser --------------------------------------------------------------- */
+
+typedef struct {
+  const char *s;
+  size_t i;
+  int error;
+  int groupCount;
+} xt_re_parser;
+
+static void xt_re_free_alt(xt_re_alt *alt);
+
+static xt_re_node *xt_re_new_node(int type) {
+  xt_re_node *n = (xt_re_node *)calloc(1, sizeof(xt_re_node));
+  if (n) n->type = type;
+  return n;
+}
+
+static xt_re_node *xt_re_new_char(unsigned char c) {
+  xt_re_node *n = xt_re_new_node(XT_RE_CHAR);
+  if (n) n->ch = c;
+  return n;
+}
+
+static xt_re_node *xt_re_new_class(const unsigned char *bits) {
+  xt_re_node *n = xt_re_new_node(XT_RE_CLASS);
+  if (!n) return NULL;
+  n->classbits = (unsigned char *)malloc(32);
+  if (!n->classbits) {
+    free(n);
+    return NULL;
+  }
+  memcpy(n->classbits, bits, 32);
+  return n;
+}
+
+static void xt_re_push_piece(xt_re_concat *c, xt_re_node *node, int quant) {
+  if (c->count == c->capacity) {
+    int next = c->capacity ? c->capacity * 2 : 4;
+    xt_re_piece *grown = (xt_re_piece *)realloc(c->pieces, (size_t)next * sizeof(xt_re_piece));
+    if (!grown) return;
+    c->pieces = grown;
+    c->capacity = next;
+  }
+  c->pieces[c->count].node = node;
+  c->pieces[c->count].quant = quant;
+  c->count++;
+}
+
+static void xt_re_push_branch(xt_re_alt *alt, xt_re_concat c) {
+  if (alt->count == alt->capacity) {
+    int next = alt->capacity ? alt->capacity * 2 : 4;
+    xt_re_concat *grown = (xt_re_concat *)realloc(alt->branches, (size_t)next * sizeof(xt_re_concat));
+    if (!grown) return;
+    alt->branches = grown;
+    alt->capacity = next;
+  }
+  alt->branches[alt->count++] = c;
+}
+
+static xt_re_alt *xt_re_parse_alt(xt_re_parser *p);
+
+static xt_re_node *xt_re_parse_escape(xt_re_parser *p) {
+  char c;
+  p->i++; /* consume backslash */
+  c = p->s[p->i];
+  if (!c) {
+    p->error = 1;
+    return NULL;
+  }
+  p->i++;
+  if (c == 'd' || c == 'D' || c == 'w' || c == 'W' || c == 's' || c == 'S') {
+    unsigned char bits[32];
+    xt_re_node *n;
+    memset(bits, 0, sizeof(bits));
+    xt_re_add_predefined(bits, c);
+    n = xt_re_new_class(bits);
+    if (!n) p->error = 1;
+    return n;
+  }
+  if (c == 'n') return xt_re_new_char('\n');
+  if (c == 't') return xt_re_new_char('\t');
+  if (c == 'r') return xt_re_new_char('\r');
+  if (c == 'f') return xt_re_new_char('\f');
+  if (c == 'v') return xt_re_new_char('\v');
+  if (c == '0') return xt_re_new_char('\0');
+  return xt_re_new_char((unsigned char)c);
+}
+
+static xt_re_node *xt_re_parse_class(xt_re_parser *p) {
+  unsigned char bits[32];
+  int negate = 0;
+  int first = 1;
+  xt_re_node *n;
+
+  memset(bits, 0, sizeof(bits));
+  p->i++; /* consume '[' */
+  if (p->s[p->i] == '^') {
+    negate = 1;
+    p->i++;
+  }
+  while (p->s[p->i] && (p->s[p->i] != ']' || first)) {
+    unsigned char lo;
+    first = 0;
+    if (p->s[p->i] == '\\' && p->s[p->i + 1]) {
+      char e;
+      p->i++;
+      e = p->s[p->i++];
+      if (xt_re_add_predefined(bits, e)) continue;
+      if (e == 'n') lo = '\n';
+      else if (e == 't') lo = '\t';
+      else if (e == 'r') lo = '\r';
+      else if (e == 'f') lo = '\f';
+      else if (e == 'v') lo = '\v';
+      else lo = (unsigned char)e;
+    } else {
+      lo = (unsigned char)p->s[p->i++];
+    }
+    if (p->s[p->i] == '-' && p->s[p->i + 1] && p->s[p->i + 1] != ']') {
+      unsigned char hi;
+      p->i++; /* consume '-' */
+      hi = (unsigned char)p->s[p->i++];
+      xt_re_add_range(bits, lo, hi);
+    } else {
+      xt_re_set_bit(bits, lo);
+    }
+  }
+  if (p->s[p->i] != ']') {
+    p->error = 1;
+    return NULL;
+  }
+  p->i++;
+  if (negate) {
+    int c;
+    for (c = 0; c < 32; c++) bits[c] = (unsigned char)~bits[c];
+  }
+  n = xt_re_new_class(bits);
+  if (!n) p->error = 1;
+  return n;
+}
+
+static xt_re_node *xt_re_parse_atom(xt_re_parser *p) {
+  char c = p->s[p->i];
+  if (c == '(') {
+    xt_re_node *n;
+    xt_re_alt *group;
+    int index;
+    p->i++;
+    /* Number the group as soon as its '(' is seen so nested groups keep
+     * POSIX/JavaScript numbering (opening-paren order). */
+    index = ++p->groupCount;
+    group = xt_re_parse_alt(p);
+    if (p->error) return NULL;
+    if (p->s[p->i] != ')') {
+      p->error = 1;
+      xt_re_free_alt(group);
+      return NULL;
+    }
+    p->i++;
+    n = xt_re_new_node(XT_RE_GROUP);
+    if (!n) {
+      p->error = 1;
+      return NULL;
+    }
+    n->group = group;
+    n->groupIndex = index;
+    return n;
+  }
+  if (c == '[') return xt_re_parse_class(p);
+  if (c == '.') {
+    p->i++;
+    return xt_re_new_node(XT_RE_ANY);
+  }
+  if (c == '^') {
+    p->i++;
+    return xt_re_new_node(XT_RE_BOL);
+  }
+  if (c == '$') {
+    p->i++;
+    return xt_re_new_node(XT_RE_EOL);
+  }
+  if (c == '\\') return xt_re_parse_escape(p);
+  if (c == '\0') {
+    p->error = 1;
+    return NULL;
+  }
+  p->i++;
+  return xt_re_new_char((unsigned char)c);
+}
+
+static void xt_re_parse_concat(xt_re_parser *p, xt_re_concat *out) {
+  while (p->s[p->i] && p->s[p->i] != '|' && p->s[p->i] != ')') {
+    xt_re_node *node = xt_re_parse_atom(p);
+    int quant = XT_RE_Q_NONE;
+    if (p->error || !node) {
+      p->error = 1;
+      return;
+    }
+    if (p->s[p->i] == '*') {
+      quant = XT_RE_Q_STAR;
+      p->i++;
+    } else if (p->s[p->i] == '+') {
+      quant = XT_RE_Q_PLUS;
+      p->i++;
+    } else if (p->s[p->i] == '?') {
+      quant = XT_RE_Q_QUEST;
+      p->i++;
+    }
+    xt_re_push_piece(out, node, quant);
+  }
+}
+
+static xt_re_alt *xt_re_parse_alt(xt_re_parser *p) {
+  xt_re_alt *alt = (xt_re_alt *)calloc(1, sizeof(xt_re_alt));
+  if (!alt) {
+    p->error = 1;
+    return NULL;
+  }
+  for (;;) {
+    xt_re_concat concat;
+    memset(&concat, 0, sizeof(concat));
+    xt_re_parse_concat(p, &concat);
+    if (p->error) {
+      free(concat.pieces);
+      free(alt->branches);
+      free(alt);
+      return NULL;
+    }
+    xt_re_push_branch(alt, concat);
+    if (p->s[p->i] == '|') {
+      p->i++;
+      continue;
+    }
+    break;
+  }
+  return alt;
+}

package/runtime/xt_regex/types.inc

@@ -0,0 +1,58 @@
+#define REG_EXTENDED 1
+#define REG_ICASE 2
+#define REG_NOSUB 4
+
+typedef struct {
+  int rm_so;
+  int rm_eo;
+} regmatch_t;
+
+/* Node kinds. */
+enum {
+  XT_RE_CHAR = 0,
+  XT_RE_ANY,
+  XT_RE_CLASS,
+  XT_RE_BOL,
+  XT_RE_EOL,
+  XT_RE_GROUP
+};
+
+/* Quantifiers. */
+enum {
+  XT_RE_Q_NONE = 0,
+  XT_RE_Q_STAR,
+  XT_RE_Q_PLUS,
+  XT_RE_Q_QUEST
+};
+
+typedef struct xt_re_alt xt_re_alt;
+
+typedef struct {
+  int type;
+  unsigned char ch;
+  unsigned char *classbits; /* 256 bits == 32 bytes, for XT_RE_CLASS */
+  xt_re_alt *group;         /* for XT_RE_GROUP */
+  int groupIndex;           /* 1-based capture number, for XT_RE_GROUP */
+} xt_re_node;
+
+typedef struct {
+  xt_re_node *node;
+  int quant;
+} xt_re_piece;
+
+typedef struct {
+  xt_re_piece *pieces;
+  int count;
+  int capacity;
+} xt_re_concat;
+
+struct xt_re_alt {
+  xt_re_concat *branches;
+  int count;
+  int capacity;
+};
+
+typedef struct {
+  xt_re_alt *root;
+  int icase;
+} regex_t;