Another combinator, matcher for identifiers.

src/text/prefix_primitives.c

@@ -78,20 +78,63 @@ DEFINE_SINGLE_CTYPE_MATCHER(digit);
 DEFINE_SINGLE_CTYPE_MATCHER(xdigit);
 DEFINE_SINGLE_CTYPE_MATCHER(alnum);
 DEFINE_SINGLE_CTYPE_MATCHER(punct);
+
 DEFINE_CTYPE_SEQUENCE_MATCHER(space);
 DEFINE_CTYPE_SEQUENCE_MATCHER(alpha);
 DEFINE_CTYPE_SEQUENCE_MATCHER(digit);
 DEFINE_CTYPE_SEQUENCE_MATCHER(xdigit);
 DEFINE_CTYPE_SEQUENCE_MATCHER(alnum);
 DEFINE_CTYPE_SEQUENCE_MATCHER(punct);
+
 DEFINE_TO_EOL_MATCHER(line_comment, "//");
 DEFINE_DELIMITED_MATCHER(block_comment, "/*", "*/", 0);
 DEFINE_DELIMITED_MATCHER(double_quoted_string, "\"", "\"", 1);
 DEFINE_DELIMITED_MATCHER(single_quoted_string, "\'", "\'", 1);
 DEFINE_DELIMITED_MATCHER(interpolant, "#{", "}", 0);
 
-// int main() {
-//   char *p = prefix_sequence("hello", prefix_is_alphas);
-//   if (!*p) putchar('0');
-//   return 0;
-// }
+DEFINE_CHAR_MATCHER (lparen,      '(');
+DEFINE_CHAR_MATCHER (rparen,      ')');
+DEFINE_CHAR_MATCHER (lbrack,      '[');
+DEFINE_CHAR_MATCHER (rbrack,      ']');
+DEFINE_CHAR_MATCHER (lbrace,      '{');
\ No newline at end of file
+DEFINE_CHAR_MATCHER (rbrace,      '}');
+
+DEFINE_CHAR_MATCHER (underscore,  '_');
+DEFINE_CHAR_MATCHER (hyphen,      '-');
+DEFINE_CHAR_MATCHER (semicolon,   ';');
+DEFINE_CHAR_MATCHER (colon,       ':');
+DEFINE_CHAR_MATCHER (period,      '.');
+DEFINE_CHAR_MATCHER (question,    '?');
+DEFINE_CHAR_MATCHER (exclamation, '!');
+DEFINE_CHAR_MATCHER (tilde,       '~');
+DEFINE_CHAR_MATCHER (backquote,   '`');
+DEFINE_CHAR_MATCHER (quote,       '\"');
+DEFINE_CHAR_MATCHER (apostrophe,  '\'');
+DEFINE_CHAR_MATCHER (ampersand,   '&');
+DEFINE_CHAR_MATCHER (caret,       '^');
+DEFINE_CHAR_MATCHER (pipe,        '|');
+DEFINE_CHAR_MATCHER (slash,       '/');
+DEFINE_CHAR_MATCHER (backslash,   '\\');
+DEFINE_CHAR_MATCHER (asterisk,    '*');
+DEFINE_CHAR_MATCHER (pound,       '#');
+DEFINE_CHAR_MATCHER (hash,        '#');
+
+DEFINE_CHAR_MATCHER (plus,        '+');
+DEFINE_CHAR_MATCHER (minus,       '-');
+DEFINE_CHAR_MATCHER (times,       '*');
+DEFINE_CHAR_MATCHER (divide,      '/');
+
+DEFINE_CHAR_MATCHER (percent,     '%');
+DEFINE_CHAR_MATCHER (dollar,      '$');
+
+DEFINE_CHAR_MATCHER (gt,          '>');
+DEFINE_CHARS_MATCHER(gte,         ">=");
+DEFINE_CHAR_MATCHER (lt,          '<');
+DEFINE_CHARS_MATCHER(lte,         "<=");
+DEFINE_CHAR_MATCHER (eq,          '=');
+DEFINE_CHAR_MATCHER (assign,      '=');
+DEFINE_CHARS_MATCHER(equal,       "==");
+
+static DEFINE_ALTERNATIVES_MATCHER(identifier_initial, prefix_is_alphas, prefix_is_underscore);
+static DEFINE_ALTERNATIVES_MATCHER(identifier_trailer, prefix_is_alnums, prefix_is_underscore);
+DEFINE_FIRST_REST_MATCHER(identifier, prefix_is_identifier_initial, prefix_is_identifier_trailer);
\ No newline at end of file

src/text/prefix_primitives.h

@@ -3,6 +3,11 @@ typedef char *(*prefix_matcher)(char *);
 #define DECLARE_MATCHER(name) \
 char *prefix_is_ ## name(char *)
 
+#define DEFINE_CHAR_MATCHER(name, prefix) \
+char *prefix_is_ ## name(char *src) { \
+  return prefix_is_char(src, prefix); \
+}
+
 #define DEFINE_CHARS_MATCHER(name, prefix) \
 char *prefix_is_ ## name(char *src) { \
   return prefix_is_chars(src, prefix); \
@@ -32,6 +37,27 @@ char *prefix_is_ ## name(char *src) { \
   return prefix_is_delimited_by(src, begin, end, escapable); \
 }
 
+#define DEFINE_ALTERNATIVES_MATCHER(name, ...) \
+char *prefix_is_ ## name(char *src) { \
+  return prefix_alternatives(src, __VA_ARGS__); \
+}
+
+#define DEFINE_SEQUENCE_MATCHER(name, ...) \
+char *prefix_is_ ## name(char *src) { \
+  return prefix_sequence(src, __VA_ARGS__); \
+}
+
+#define DEFINE_OPTIONAL_MATCHER(name, matcher) \
+char *prefix_is_ ## name(char *src) { \
+  return prefix_optional(src, matcher); \
+}
+
+#define DEFINE_FIRST_REST_MATCHER(name, first_matcher, rest_matcher) \
+char *prefix_is_ ## name(char *src) { \
+  if (src = first_matcher(src)) src = prefix_zero_plus(src, rest_matcher); \
+  return src; \
+}
+
 char *prefix_is_char(char *src, char pre);
 char *prefix_is_chars(char *src, char *pre);
 char *prefix_is_one_of(char *src, char *class);
@@ -64,3 +90,48 @@ DECLARE_MATCHER(block_comment);
 DECLARE_MATCHER(double_quoted_string);
 DECLARE_MATCHER(single_quoted_string);
 DECLARE_MATCHER(interpolant);
+
+DECLARE_MATCHER(lparen);
+DECLARE_MATCHER(rparen);
+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);
+
+DECLARE_MATCHER(identifier);

src/text/test.c

@@ -28,7 +28,7 @@ int main() {
   char *noncomment = "/* blah blah";
   char *interpolant = "#{ this is an interpolant \\} blah blah";
   char *words = "hello my name is aaron";
-  char *id1 = "_identifier123";
+  char *id1 = "_identifier123{blah bloo}";
   char *non_id = "12non_ident_ifier_";
   char *word2 = "-blah-blah_blah";
   char *selector = "#foo > :first-child { color: #abcdef; }";
@@ -74,5 +74,8 @@ int main() {
   testn(prefix_one_plus, words, prefix_is_alphas);
   testn(prefix_one_plus, non_id, prefix_is_alphas);
   
+  test1(prefix_is_identifier, id1);
+  test1(prefix_is_identifier, non_id);
+
   return 0;
 }