diff --git a/.gitignore b/.gitignore
index 0d20b64..34f2b04 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1 +1,3 @@
*.pyc
+*.o
+*.a
diff --git a/.hgignore b/.hgignore
index bb6d589..b8428bb 100644
--- a/.hgignore
+++ b/.hgignore
@@ -1,3 +1,5 @@
syntax: glob
*.pyc
+*.o
+*.a
diff --git a/README.markdown b/README.markdown
index f860ef5..53df835 100644
--- a/README.markdown
+++ b/README.markdown
@@ -180,23 +180,21 @@
TODO
----
-* the compiler generates *horrible* string-handling code. fix!
+* libtamsin contains *horrible* string-handling code. fix!
* meta-circular implementation of scanner -- what we have is pretty close
* meta-circular implementation of parser
* meta-circular implementation of interpreter!
-* system library
+* system library in its own Python module
* `bin/tamsin runast astfile.txt` -- for testing the meta-circular parser
+* `bin/tamsin runscan scanfile.txt` -- for testing the meta-circular scanner
+* `bin/tamsin scan file.tamsin` -- to generate a scanfile
### compiler ###
* handle term concatenation
-* literal string escape sequences
* render NUL as EOF. actually, we want 8-bit clean strings eventually
-* implement any
-* implement fail
* implement !
* implement alnum
-* write `-ltamsin` library and link to it
### document ###
diff --git a/c_src/build.sh b/c_src/build.sh
new file mode 100755
index 0000000..197a8d1
--- /dev/null
+++ b/c_src/build.sh
@@ -0,0 +1,10 @@
+#!/bin/sh
+
+FILES="scanner term tamsin"
+
+for FILE in $FILES; do
+ gcc -c $FILE.c -o $FILE.o
+done
+
+ar -r libtamsin.a scanner.o term.o tamsin.o
+
diff --git a/c_src/scanner.c b/c_src/scanner.c
new file mode 100644
index 0000000..060c340
--- /dev/null
+++ b/c_src/scanner.c
@@ -0,0 +1,19 @@
+#include "tamsin.h"
+
+char scan(struct scanner *s) {
+ char c = s->buffer[s->position];
+ if (c == '\0') {
+ return '\0';
+ } else {
+ s->position++;
+ return c;
+ }
+};
+
+void unscan(struct scanner *s) {
+ s->position = s->reset_position;
+}
+
+void commit(struct scanner *s) {
+ s->reset_position = s->position;
+}
diff --git a/c_src/tamsin.c b/c_src/tamsin.c
new file mode 100644
index 0000000..c354e80
--- /dev/null
+++ b/c_src/tamsin.c
@@ -0,0 +1,48 @@
+#include "tamsin.h"
+
+void tamsin_eof(struct scanner *s) {
+ char c = scan(s);
+ unscan(s);
+ if (c == '\0') {
+ result = new_term("EOF");
+ ok = 1;
+ } else {
+ char t[100];
+ sprintf(t, "expected EOF found '%c'", c);
+ result = new_term(t);
+ ok = 0;
+ }
+}
+
+void tamsin_any(struct scanner *s) {
+ char c = scan(s);
+ if (c == '\0') {
+ unscan(s);
+ result = new_term("expected any token, found EOF");
+ ok = 0;
+ } else {
+ char t[2];
+ commit(s);
+ sprintf(t, "%c", c);
+ result = new_term(t);
+ ok = 1;
+ }
+}
+
+void tamsin_expect(struct scanner *s, char *token) {
+ char c = scan(s);
+ if (c == token[0]) {
+ commit(s);
+ char s[10];
+ strcpy(s, "a");
+ s[0] = c;
+ result = new_term(s);
+ ok = 1;
+ } else {
+ unscan(s);
+ char s[100];
+ sprintf(s, "expected '%c' found '%c'", token[0], c);
+ result = new_term(s);
+ ok = 0;
+ }
+};
diff --git a/c_src/tamsin.h b/c_src/tamsin.h
new file mode 100644
index 0000000..8036dc0
--- /dev/null
+++ b/c_src/tamsin.h
@@ -0,0 +1,43 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+/* terms */
+
+struct term {
+ const char *atom;
+ struct term_list *subterms;
+};
+
+struct term_list {
+ struct term *term;
+ struct term_list *next;
+};
+
+struct term *new_term(const char *);
+void add_subterm(struct term *, struct term *);
+void term_format_r(struct term *t);
+char *term_format(struct term *t);
+
+/* scanner */
+
+struct scanner {
+ char *buffer;
+ int position;
+ int reset_position;
+};
+
+char scan(struct scanner *);
+void unscan(struct scanner *);
+void commit(struct scanner *);
+
+/* tamsin */
+
+void tamsin_eof(struct scanner *);
+void tamsin_any(struct scanner *);
+void tamsin_expect(struct scanner *, char *);
+
+/* global state: result of last action */
+
+extern int ok;
+extern struct term *result;
diff --git a/c_src/term.c b/c_src/term.c
new file mode 100644
index 0000000..f344ea9
--- /dev/null
+++ b/c_src/term.c
@@ -0,0 +1,42 @@
+#include "tamsin.h"
+
+struct term *new_term(const char *atom) {
+ struct term *t;
+ t = malloc(sizeof(struct term));
+ t->atom = strdup(atom);
+ t->subterms = NULL;
+}
+
+void add_subterm(struct term *term, struct term *subterm) {
+ struct term_list *tl;
+ tl = malloc(sizeof(struct term_list));
+ tl->term = subterm;
+ tl->next = term->subterms;
+ term->subterms = tl;
+}
+
+char fmtbuf[1000]; /* yeesh */
+
+void term_format_r(struct term *t) {
+ struct term_list *tl;
+
+ strcat(fmtbuf, t->atom);
+ if (t->subterms != NULL) {
+ strcat(fmtbuf, "(");
+
+ for (tl = t->subterms; tl != NULL; tl = tl->next) {
+ term_format_r(tl->term);
+ if (tl->next != NULL) {
+ strcat(fmtbuf, ", ");
+ }
+ }
+
+ strcat(fmtbuf, ")");
+ }
+}
+
+char *term_format(struct term *t) {
+ strcpy(fmtbuf, "");
+ term_format_r(t);
+ return fmtbuf;
+}
diff --git a/loadngo.sh b/loadngo.sh
index 1745b05..9c51761 100755
--- a/loadngo.sh
+++ b/loadngo.sh
@@ -1,6 +1,8 @@
#!/bin/sh
-INPUT=`cat`
bin/tamsin compile $1 > foo.c && \
- gcc foo.c -o foo && \
- ./foo "$INPUT"
+ gcc -Ic_src -Lc_src foo.c -o foo -ltamsin && \
+ ./foo `cat`
+R=$?
+rm -f foo.c foo
+exit $R
diff --git a/src/tamsin/compiler.py b/src/tamsin/compiler.py
index 5ac3799..83faed5 100644
--- a/src/tamsin/compiler.py
+++ b/src/tamsin/compiler.py
@@ -10,140 +10,12 @@
/*
* Generated code! Edit at your own risk!
*/
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h>
+#include <tamsin.h>
/* global state: result of last action */
int ok;
struct term *result;
-
-/* terms */
-
-struct term {
- const char *atom;
- struct term_list *subterms;
-};
-
-struct term_list {
- struct term *term;
- struct term_list *next;
-};
-
-struct term *new_term(const char *atom) {
- struct term *t;
- t = malloc(sizeof(struct term));
- t->atom = strdup(atom);
- t->subterms = NULL;
-}
-
-void add_subterm(struct term *term, struct term *subterm) {
- struct term_list *tl;
- tl = malloc(sizeof(struct term_list));
- tl->term = subterm;
- tl->next = term->subterms;
- term->subterms = tl;
-}
-
-char fmtbuf[1000]; /* yeesh */
-
-void term_format_r(struct term *t) {
- struct term_list *tl;
-
- strcat(fmtbuf, t->atom);
- if (t->subterms != NULL) {
- strcat(fmtbuf, "(");
-
- for (tl = t->subterms; tl != NULL; tl = tl->next) {
- term_format_r(tl->term);
- if (tl->next != NULL) {
- strcat(fmtbuf, ", ");
- }
- }
-
- strcat(fmtbuf, ")");
- }
-}
-
-char *term_format(struct term *t) {
- strcpy(fmtbuf, "");
- term_format_r(t);
- return fmtbuf;
-}
-
-/* scanner */
-
-struct scanner {
- char *buffer;
- int position;
- int reset_position;
-};
-
-char scan(struct scanner *s) {
- char c = s->buffer[s->position];
- if (c == '\0') {
- return '\0';
- } else {
- s->position++;
- return c;
- }
-};
-
-void unscan(struct scanner *s) {
- s->position = s->reset_position;
-}
-
-void commit(struct scanner *s) {
- s->reset_position = s->position;
-}
-
-void tamsin_eof(struct scanner *s) {
- char c = scan(s);
- unscan(s);
- if (c == '\0') {
- result = new_term("EOF");
- ok = 1;
- } else {
- char t[100];
- sprintf(t, "expected EOF found '%c'", c);
- result = new_term(t);
- ok = 0;
- }
-}
-
-void tamsin_any(struct scanner *s) {
- char c = scan(s);
- if (c == '\0') {
- unscan(s);
- result = new_term("expected any token, found EOF");
- ok = 0;
- } else {
- char t[2];
- commit(s);
- sprintf(t, "%c", c);
- result = new_term(t);
- ok = 1;
- }
-}
-
-void tamsin_expect(struct scanner *s, char *token) {
- char c = scan(s);
- if (c == token[0]) {
- commit(s);
- char s[10];
- strcpy(s, "a");
- s[0] = c;
- result = new_term(s);
- ok = 1;
- } else {
- unscan(s);
- char s[100];
- sprintf(s, "expected '%c' found '%c'", token[0], c);
- result = new_term(s);
- ok = 0;
- }
-};
struct scanner * scanner;
'''