lexer: Add scanning of identifiers

Signed-off-by: Chloe M. <chloe@mirocom.org>

3 files changed, 87 insertions, 0 deletions

diff --git a/core/lexer.c b/core/lexer.c
index a11a630..51e4d40 100644
--- a/core/lexer.c
+++ b/core/lexer.c
@@ -5,8 +5,35 @@
 
 #include <errno.h>
 #include <stdbool.h>
+#include <stdlib.h>
+#include <ctype.h>
 #include "cescal/lexer.h"
 #include "cescal/log.h"
+#include "cescal/ptrbox.h"
+
+static inline void
+lexer_putback(struct cescal_state *state, char c)
+{
+    if (state == NULL) {
+        return;
+    }
+
+    state->lex_putback = c;
+}
+
+static inline char
+lexer_putback_pop(struct cescal_state *state)
+{
+    char retc;
+
+    if (state == NULL) {
+        return '\0';
+    }
+
+    retc = state->lex_putback;
+    state->lex_putback = '\0';
+    return retc;
+}
 
 /*
  * Returns true if the given character is a whitespace
@@ -44,6 +71,12 @@ lexer_consume_single(struct cescal_state *state, bool skip_ws)
         return '\0';
     }
 
+    if ((c = lexer_putback_pop(state)) != '\0') {
+        if (lexer_is_ws(c) && !skip_ws) {
+            return c;
+        }
+    }
+
     while ((c = readbuf_read(&state->rb, state->in_fd)) != '\0') {
         if (lexer_is_ws(c)) {
             continue;
@@ -55,6 +88,53 @@ lexer_consume_single(struct cescal_state *state, bool skip_ws)
     return '\0';
 }
 
+static int
+lexer_scan_ident(struct cescal_state *state, char lc, struct token *res)
+{
+    char *buf, c;
+    size_t bufsz, bufcap;
+
+    if (state == NULL || res == NULL) {
+        errno = EINVAL;
+        return -1;
+    }
+
+    bufsz = 0;
+    bufcap = 8;
+    if ((buf = malloc(bufcap)) == NULL) {
+        return -1;
+    }
+
+    if (lc != '_' && !isalpha(lc)) {
+        cc_error("bad identifier\n");
+    }
+
+    buf[bufsz++] = lc;
+    for (;;) {
+        c = lexer_consume_single(state, false);
+        if (c != '_' && !isalnum(c)) {
+            lexer_putback(state, c);
+            buf[bufsz] = '\0';
+            break;
+        }
+
+        buf[bufsz++] = c;
+        if (bufsz >= bufcap) {
+            bufcap += 8;
+            buf = realloc(buf, bufcap);
+        }
+
+        if (buf == NULL) {
+            return -1;
+        }
+    }
+
+    res->s = ptrbox_strdup(&state->ptrbox, buf);
+    res->type = TT_IDENT;
+    free(buf);
+    return 0;
+}
+
 int
 lexer_nom(struct cescal_state *state, struct token *res)
 {
@@ -82,6 +162,10 @@ lexer_nom(struct cescal_state *state, struct token *res)
         res->type = TT_COMMA;
         res->c = c;
         return 0;
+    default:
+        if (lexer_scan_ident(state, c, res) == 0) {
+            return 0;
+        }
     }
 
     cc_error("got unknown token '%c'\n", c);
diff --git a/include/cescal/state.h b/include/cescal/state.h
index 566b5a2..3c1ad86 100644
--- a/include/cescal/state.h
+++ b/include/cescal/state.h
@@ -18,12 +18,14 @@
  * @rb:     Read buffer
  * @tokbuf: Token buffer
  * @ptrbox: Global pointer box
+ * @lex_putback: Lexer putback buffer
  */
 struct cescal_state {
     int in_fd;
     struct readbuf rb;
     struct tokbuf tokbuf;
     struct ptrbox ptrbox;
+    char lex_putback;
 };
 
 /*
diff --git a/include/cescal/token.h b/include/cescal/token.h
index 0be8ff0..990df77 100644
--- a/include/cescal/token.h
+++ b/include/cescal/token.h
@@ -32,6 +32,7 @@ struct token {
     tt_t type;
     union {
         char c;
+        char *s;
     };
 };