More tokenizing stuff. May have to re-invent closures to make this cleaner.

src/parser/lexical_patterns.c

+#include "prefix_primitives.h"
+
 int prefix_starts_with_identifier(char *src) {
   int p = prefix_is_alphas(src) || prefix_is_exactly(src, "_");
   if (!p++) return 0;
@@ -18,8 +20,7 @@ DEFINE_EXACT_MATCHER(hash, "#");
 DEFINE_EXACT_MATCHER(adjacent_to, "+");
 DEFINE_EXACT_MATCHER(precedes, "~");
 DEFINE_EXACT_MATCHER(parent_of, ">");
-int (*prefix_starts_with_ancestor_of)(char *src) = prefix_starts_with_spaces;
-
+prefix_matcher prefix_starts_with_ancestor_of = prefix_starts_with_spaces;
 DEFINE_EXACT_MATCHER(exclamation, "!");
 DEFINE_EXACT_MATCHER(dollar, "$");
 DEFINE_EXACT_MATCHER(percent, "%");

src/parser/prefix_primitives.c

@@ -34,15 +34,31 @@ int prefix_is_delimited_by(char *src, char *beg, char *end, int esc) {
   }
 }
 
-int prefix_try_alternatives(char *src, ...) {
+int prefix_alternatives(char *src, ...) {
   int p = 0;
   va_list ap;
   va_start(ap, src);
   prefix_matcher m = va_arg(ap, prefix_matcher);  
   while (m && !(p = (*m)(src))) m = va_arg(ap, prefix_matcher);
+  va_end(ap);
   return p;
 }
 
+int prefix_sequence(char *src, ...) {
+  int p = 0, p_sum = 0;
+  va_list ap;
+  va_start(ap, src);
+  prefix_matcher m = va_arg(ap, prefix_matcher);
+  while (m && (p = (*m)(src))) p_sum += p, m = va_arg(ap, prefix_matcher);
+  va_end(ap);
+  return p ? p_sum : 0;
+}
+
+int prefix_optional(char *src, prefix_matcher m) {
+  int p = m(src);
+  return p ? p : -1;
+}
+
 DEFINE_SINGLE_CTYPE_MATCHER(space);
 DEFINE_SINGLE_CTYPE_MATCHER(alpha);
 DEFINE_SINGLE_CTYPE_MATCHER(digit);
@@ -62,14 +78,58 @@ DEFINE_DELIMITED_MATCHER(single_quoted_string, "'", "'", 1);
 DEFINE_DELIMITED_MATCHER(interpolant, "#{", "}", 0);
 
 int prefix_is_string(char *src) {
-  return prefix_try_alternatives(src,
-                                 prefix_is_double_quoted_string,
-                                 prefix_is_single_quoted_string);
+  return prefix_alternatives(src, prefix_is_double_quoted_string,
+                                      prefix_is_single_quoted_string);
 }
 
-DEFINE_EXACT_MATCHER(lparen, "(");
-DEFINE_EXACT_MATCHER(rparen, ")");
-DEFINE_EXACT_MATCHER(lbrack, "[");
-DEFINE_EXACT_MATCHER(rbrack, "]");
-DEFINE_EXACT_MATCHER(lbrace, "{");
-DEFINE_EXACT_MATCHER(rbrace, "}");
+DEFINE_EXACT_MATCHER(lparen,      "(");
+DEFINE_EXACT_MATCHER(rparen,      ")");
+DEFINE_EXACT_MATCHER(lbrack,      "[");
+DEFINE_EXACT_MATCHER(rbrack,      "]");
+DEFINE_EXACT_MATCHER(lbrace,      "{");
+DEFINE_EXACT_MATCHER(rbrace,      "}");
+
+DEFINE_EXACT_MATCHER(underscore,  "_");
+DEFINE_EXACT_MATCHER(hyphen,      "-");
+DEFINE_EXACT_MATCHER(semicolon,   ";");
+DEFINE_EXACT_MATCHER(colon,       ":");
+DEFINE_EXACT_MATCHER(period,      ".");
+DEFINE_EXACT_MATCHER(question,    "?");
+DEFINE_EXACT_MATCHER(exclamation, "!");
+DEFINE_EXACT_MATCHER(tilde,       "~");
+DEFINE_EXACT_MATCHER(backquote,   "`");
+DEFINE_EXACT_MATCHER(quote,       "\"");
+DEFINE_EXACT_MATCHER(apostrophe,  "'");
+DEFINE_EXACT_MATCHER(ampersand,   "&");
+DEFINE_EXACT_MATCHER(caret,       "^");
+DEFINE_EXACT_MATCHER(pipe,        "|");
+DEFINE_EXACT_MATCHER(slash,       "/");
+DEFINE_EXACT_MATCHER(backslash,   "\\");
+DEFINE_EXACT_MATCHER(asterisk,    "*");
+DEFINE_EXACT_MATCHER(pound,       "#");
+DEFINE_EXACT_MATCHER(hash,        "#");
+
+DEFINE_EXACT_MATCHER(plus,        "+");
+DEFINE_EXACT_MATCHER(minus,       "-");
+DEFINE_EXACT_MATCHER(times,       "*");
+DEFINE_EXACT_MATCHER(divide,      "/");
+
+DEFINE_EXACT_MATCHER(percent,     "%");
+DEFINE_EXACT_MATCHER(dollar,      "$");
+
+DEFINE_EXACT_MATCHER(gt,          ">");
+DEFINE_EXACT_MATCHER(gte,         ">=");
+DEFINE_EXACT_MATCHER(lt,          "<");
+DEFINE_EXACT_MATCHER(lte,         "<=");
+DEFINE_EXACT_MATCHER(eq,          "=");
+DEFINE_EXACT_MATCHER(assign,      "=");
+DEFINE_EXACT_MATCHER(equal,       "==");
+
+DEFINE_ALTERNATIVES_MATCHER(identifier_initial, prefix_is_alphas, prefix_is_underscore);
+DEFINE_ALTERNATIVES_MATCHER(identifier_trailing, prefix_is_alnums, prefix_is_underscore);
+DEFINE_FIRST_REST_MATCHER(identifier, prefix_is_identifier_initial, prefix_is_identifier_trailing);
+
+// DEFINE_ALTERNATIVES_MATCHER(
+// DEFINE_ALTERNATIVES_MATCHER(word_initial, prefix_is_identifier, prefix_is_hyphen_);
+DEFINE_ALTERNATIVES_MATCHER(word_trailing, prefix_is_alnums, prefix_is_underscore, prefix_is_hyphen);
+DEFINE_FIRST_REST_MATCHER(word, prefix_is_word_initial, prefix_is_word_trailing);

src/parser/prefix_primitives.h

@@ -4,6 +4,9 @@ int prefix_is_exactly(char *, char*);
 int prefix_is_one_of(char *, char *);
 int prefix_is_some_of(char *, char *);
 int prefix_is_delimited_by(char *, char *, char *, int);
+int prefix_alternatives(char *, ...);
+int prefix_sequence(char *, ...);
+int prefix_optional(char *, prefix_matcher);
 
 #define DECLARE_MATCHER(name) \
 int prefix_is_ ## name(char *)
@@ -14,7 +17,7 @@ int prefix_is_ ## name(char *src) { \
 }
 
 #define DEFINE_SINGLE_CTYPE_MATCHER(type) \
-int prefix_is_one_ ## type(char *src) { \
+int prefix_is_ ## type(char *src) { \
   return is ## type(src[0]) ? 1 : 0; \
 }
 
@@ -37,15 +40,38 @@ int prefix_is_ ## name(char *src) { \
   return p; \
 }
 
-#define prefix_list(...) { __VA_ARGS__, NULL }
-  
+#define DEFINE_ALTERNATIVES_MATCHER(name, ...) \
+int prefix_is_ ## name(char *src) { \
+  return prefix_alternatives(src, __VA_ARGS__); \
+}
+
+#define DEFINE_SEQUENCE_MATCHER(name, ...) \
+int prefix_is_ ## name(char *src) { \
+  return prefix_sequence(src, __VA_ARGS__); \
+}
+
+#define DEFINE_OPTIONAL_MATCHER(name, matcher) \
+int prefix_is_ ## name(char *src) { \
+  return prefix_optional(src, matcher); \
+}
+
+#define DEFINE_FIRST_REST_MATCHER(name, first_matcher, rest_matcher) \
+int prefix_is_ ## name(char *src) { \
+  int p = first_matcher(src); \
+  int p_sum = p; \
+  while (p) { \
+    p = rest_matcher(src+p); \
+    p_sum += p; \
+  } \
+  return p_sum; \
+}
 
-DECLARE_MATCHER(one_space);
-DECLARE_MATCHER(one_alpha);
-DECLARE_MATCHER(one_digit);
-DECLARE_MATCHER(one_xdigit);
-DECLARE_MATCHER(one_alnum);
-DECLARE_MATCHER(one_punct);
+DECLARE_MATCHER(space);
+DECLARE_MATCHER(alpha);
+DECLARE_MATCHER(digit);
+DECLARE_MATCHER(xdigit);
+DECLARE_MATCHER(alnum);
+DECLARE_MATCHER(punct);
 DECLARE_MATCHER(spaces);
 DECLARE_MATCHER(alphas);
 DECLARE_MATCHER(digits);
@@ -64,3 +90,39 @@ DECLARE_MATCHER(lbrack);
 DECLARE_MATCHER(rbrack);
 DECLARE_MATCHER(lbrace);
 DECLARE_MATCHER(rbrace);
+
+DECLARE_MATCHER(underscore);
+DECLARE_MATCHER(hyphen);
+DECLARE_MATCHER(semicolon);
+DECLARE_MATCHER(colon);
+DECLARE_MATCHER(period);
+DECLARE_MATCHER(question);
+DECLARE_MATCHER(exclamation);
+DECLARE_MATCHER(tilde);
+DECLARE_MATCHER(backquote);
+DECLARE_MATCHER(quote);
+DECLARE_MATCHER(apostrophe);
+DECLARE_MATCHER(ampersand);
+DECLARE_MATCHER(caret);
+DECLARE_MATCHER(pipe);
+DECLARE_MATCHER(slash);
+DECLARE_MATCHER(backslash);
+DECLARE_MATCHER(asterisk);
+DECLARE_MATCHER(pound);
+DECLARE_MATCHER(hash);
+                                  
+DECLARE_MATCHER(plus);
+DECLARE_MATCHER(minus);
+DECLARE_MATCHER(times);
+DECLARE_MATCHER(divide);
+                                  
+DECLARE_MATCHER(percent);
+DECLARE_MATCHER(dollar);
+                                  
+DECLARE_MATCHER(gt);
+DECLARE_MATCHER(gte);
+DECLARE_MATCHER(lt);
+DECLARE_MATCHER(lte);
+DECLARE_MATCHER(eq);
+DECLARE_MATCHER(assign);
+DECLARE_MATCHER(equal);

src/parser/test.c

@@ -19,6 +19,9 @@ int main() {
   char *t = "/* this is a c comment \\*/ blah blah";
   char *u = "#{ this is an interpolant \\} blah blah";
   char *v = "hello my name is aaron";
+  char *w = "_identifier123";
+  char *x = "12nonidentifier_";
+  char *y = "-blah-blah_blah";
   
   int (*funcs[])(char *) = { prefix_is_string, prefix_is_block_comment, NULL };