Start working on modules, including a built-in "boot" module

FossilOrigin-Name: 4829dbeb2071fe0be41ded66a61ba2e156119a47def8a23415b8227a00d53ce7

9 files changed, 188 insertions, 20 deletions

diff --git a/CMakeLists.txt b/CMakeLists.txt
index f59b019..d3450d7 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -48,13 +48,25 @@ target_link_options(common INTERFACE
 )
 
 add_library(aplwc
+	${CMAKE_CURRENT_BINARY_DIR}/generated_headers/boot.h
 	src/lib/eval.c
 	src/lib/main.c
 	src/lib/memory.c
+	src/lib/module.c
 	src/lib/string.c
 	src/lib/syntax.c
 )
 target_link_libraries(aplwc PRIVATE common)
+target_include_directories(aplwc PRIVATE
+	"${CMAKE_CURRENT_BINARY_DIR}/generated_headers/"
+)
+
+add_custom_command(
+	OUTPUT "${CMAKE_CURRENT_BINARY_DIR}/generated_headers/boot.h"
+	COMMAND xxd -i -n boot_src "${CMAKE_CURRENT_SOURCE_DIR}/src/lib/boot.aplwc" "${CMAKE_CURRENT_BINARY_DIR}/generated_headers/boot.h"
+	DEPENDS "${CMAKE_CURRENT_SOURCE_DIR}/src/lib/boot.aplwc"
+	COMMENT "Generate boot.h from boot.aplwc"
+)
 
 add_executable(aplwc-repl
 	src/bin/main.c
diff --git a/src/include/aplwc.h b/src/include/aplwc.h
index 58db819..a22bd8d 100644
--- a/src/include/aplwc.h
+++ b/src/include/aplwc.h
@@ -28,7 +28,8 @@ void aplwc_free_line(struct aplwc_line *);
 struct aplwc_value *aplwc_eval(struct aplwc_line *);
 void aplwc_free_value(struct aplwc_value *);
 const char *aplwc_value_string(struct aplwc_value *);
+enum aplwc_rc aplwc_load_module(struct aplwc *, const char *, const char *, int n);
 
-#define APLWC_MIN_MEM (64)
+#define APLWC_MIN_MEM (2048)
 
 #endif /* APLWC_H */
diff --git a/src/lib/aplwc_internal.h b/src/lib/aplwc_internal.h
index 33feafb..b027073 100644
--- a/src/lib/aplwc_internal.h
+++ b/src/lib/aplwc_internal.h
@@ -36,7 +36,13 @@ enum token_tag {
 	TOKEN_TAG_NEWLINE,
 	TOKEN_TAG_NUMBER,
 	TOKEN_TAG_LEFT_ARROW,
-	TOKEN_TAG_DIAMOND
+	TOKEN_TAG_DIAMOND,
+	TOKEN_TAG_KEYWORD,
+	TOKEN_TAG_TEXT,
+	TOKEN_TAG_DEL,
+	TOKEN_TAG_LEFT_PAREN,
+	TOKEN_TAG_RIGHT_PAREN,
+	TOKEN_TAG_IBEAM
 };
 
 struct token {
@@ -77,11 +83,14 @@ void memory_free_token_list(struct memory *, struct token_list *);
 void memory_free_aplwc_value(struct memory *, struct aplwc_value *);
 
 /* string.c */
-struct string *chars_to_string(struct aplwc *, const char *);
+struct string *chars_to_string(struct aplwc *, const char *, int len);
 uint32_t string_char(struct string *, size_t);
 size_t char_length(struct string *, size_t);
 
 /* syntax.c */
-struct token_list *scan_line(struct aplwc *, struct string *);
+struct token_list *scan_string(struct aplwc *, struct string *, int);
+
+/* module.c */
+enum aplwc_rc load_module(struct aplwc *, struct string *, struct token_list *);
 
 #endif /* APLWC_INTERNAL_H */
diff --git a/src/lib/boot.aplwc b/src/lib/boot.aplwc
new file mode 100644
index 0000000..6248427
--- /dev/null
+++ b/src/lib/boot.aplwc
@@ -0,0 +1,13 @@
+:module Boot
+:symbol plus '+'
+:symbol minus '-'
+:export '+' '-'
+
+∇ r←a plus b
+  r←a (1⌶) b
+∇
+
+∇ r←a minus b
+  r←a (2⌶) b
+∇
+
diff --git a/src/lib/main.c b/src/lib/main.c
index e9d33cb..ae17f34 100644
--- a/src/lib/main.c
+++ b/src/lib/main.c
@@ -1,4 +1,5 @@
 #include "aplwc_internal.h"
+#include "boot.h"
 
 enum aplwc_rc
 aplwc_new(struct aplwc **p, uint8_t *data, size_t size)
@@ -12,7 +13,7 @@ aplwc_new(struct aplwc **p, uint8_t *data, size_t size)
 		return APLWC_OUT_OF_MEMORY;
 	(*p)->memory = m;
 
-	return APLWC_OK;
+	return aplwc_load_module(*p, "Boot", (const char *)boot_src, boot_src_len);
 }
 
 void
@@ -70,11 +71,11 @@ aplwc_read(struct aplwc *aplwc, const char *line)
 	struct token_list *tokens;
 	struct aplwc_line *ret;
 
-	text = chars_to_string(aplwc, line);
+	text = chars_to_string(aplwc, line, -1);
 	if(text == NULL)
 		return NULL;
 
-	tokens = scan_line(aplwc, text);
+	tokens = scan_string(aplwc, text, 1);
 	if(tokens == NULL)
 		return NULL;
 
@@ -124,3 +125,39 @@ aplwc_value_string(struct aplwc_value *value)
 		return "???";
 	}
 }
+
+enum aplwc_rc
+aplwc_load_module(struct aplwc *aplwc, const char *name, const char *src, int srclen)
+{
+	struct string *text, *namestr;
+	struct token_list *tokens;
+
+	namestr = NULL;
+	text = NULL;
+	tokens = NULL;
+
+	namestr = chars_to_string(aplwc, name, -1);
+	if(namestr == NULL)
+		goto error;
+
+	text = chars_to_string(aplwc, src, srclen);
+	if(text == NULL)
+		goto error;
+	
+	tokens = scan_string(aplwc, text, 0);
+	if(tokens == NULL)
+		goto error;
+
+	if(load_module(aplwc, namestr, tokens) == APLWC_OK)
+		return APLWC_OK;
+	else{
+error:
+		if(tokens != NULL)
+			memory_free_token_list(&aplwc->memory, tokens);
+		if(text != NULL)
+			memory_free_string(&aplwc->memory, text);
+		if(namestr != NULL)
+			memory_free_string(&aplwc->memory, namestr);
+		return aplwc_last_error(aplwc);
+	}
+}
diff --git a/src/lib/module.c b/src/lib/module.c
new file mode 100644
index 0000000..4ce9f8c
--- /dev/null
+++ b/src/lib/module.c
@@ -0,0 +1,65 @@
+#include "aplwc_internal.h"
+
+static int
+read_keyword_decl(struct token_list *tokens, size_t offset)
+{
+	int len;
+
+	for(len = 1; (offset + len) < tokens->length; len++)
+		if(tokens->tokens[offset+len].tag == TOKEN_TAG_NEWLINE)
+			break;
+	return len;
+}
+
+static int
+read_function_decl(struct token_list *tokens, size_t offset)
+{
+	int len;
+
+	for(len = 1; (offset + len) < tokens->length; len++)
+		if(tokens->tokens[offset+len].tag == TOKEN_TAG_DEL)
+			break;
+	return len;
+}
+
+static int
+read_decl(struct token_list *tokens, size_t offset)
+{
+	UNUSED_PARAMETER(tokens);
+	UNUSED_PARAMETER(offset);
+
+	switch(tokens->tokens[offset].tag){
+	case TOKEN_TAG_NEWLINE:
+		return 0;
+	case TOKEN_TAG_KEYWORD:
+		return read_keyword_decl(tokens, offset);
+	case TOKEN_TAG_DEL:
+		return read_function_decl(tokens, offset);
+	default:
+		return -1;
+	}
+}
+
+static enum aplwc_rc
+read_decls(struct token_list *tokens)
+{
+	size_t offset;
+	int skip;
+
+	for(offset = 0; offset < tokens->length; offset++){
+		skip = read_decl(tokens, offset);
+		if(skip == -1)
+			return APLWC_INVALID_SYNTAX;
+		offset += skip;
+	}
+	return APLWC_OK;
+}
+
+enum aplwc_rc
+load_module(struct aplwc *aplwc, struct string *name, struct token_list *tokens)
+{
+	UNUSED_PARAMETER(aplwc);
+	UNUSED_PARAMETER(name);
+
+	return read_decls(tokens);
+}
diff --git a/src/lib/string.c b/src/lib/string.c
index 123e81b..ecac9d7 100644
--- a/src/lib/string.c
+++ b/src/lib/string.c
@@ -1,7 +1,7 @@
 #include "aplwc_internal.h"
 
 struct string *
-chars_to_string(struct aplwc *aplwc, const char *chars)
+chars_to_string(struct aplwc *aplwc, const char *chars, int len)
 {
 	const uint8_t *utf8;
 	size_t bytecount, charcount;
@@ -12,7 +12,7 @@ chars_to_string(struct aplwc *aplwc, const char *chars)
 
 	utf8 = (const uint8_t *)chars;
 	seqlen = extra = 0;
-	for(bytecount = 0, charcount = 0; utf8[bytecount]; bytecount++){
+	for(bytecount = 0, charcount = 0; ((bytecount < (unsigned int)len) || (len == -1)) && utf8[bytecount]; bytecount++){
 		b = utf8[bytecount];
 		if(extra == 0){
 			c = 0;
diff --git a/src/lib/syntax.c b/src/lib/syntax.c
index c3452f2..f930973 100644
--- a/src/lib/syntax.c
+++ b/src/lib/syntax.c
@@ -59,6 +59,11 @@ single_sym(struct scan_state *state, struct token *token)
 	switch(peek(state)){
 	case 0x2190: token->tag = TOKEN_TAG_LEFT_ARROW; break;
 	case 0x22C4: token->tag = TOKEN_TAG_DIAMOND; break;
+	case 0x2207: token->tag = TOKEN_TAG_DEL; break;
+	case 0x2336: token->tag = TOKEN_TAG_IBEAM; break;
+	case '\n':   token->tag = TOKEN_TAG_NEWLINE; break;
+	case '(':    token->tag = TOKEN_TAG_LEFT_PAREN; break;
+	case ')':    token->tag = TOKEN_TAG_RIGHT_PAREN; break;
 	default:
 		return 0;
 	}
@@ -83,6 +88,21 @@ identifier(struct scan_state *state, struct token *token)
 	return 1;
 }
 
+static int
+keyword(struct scan_state *state, struct token *token)
+{
+	uint32_t c;
+	if(peek(state) != ':')
+		return 0;
+
+	do
+		c = advance(state);
+	while(is_letter(c));
+
+	token->tag = TOKEN_TAG_KEYWORD;
+	return 1;
+}
+
 static int 
 number(struct scan_state *state, struct token *token)
 {
@@ -125,14 +145,25 @@ number(struct scan_state *state, struct token *token)
 	return 1;
 }
 
-static int 
-newline(struct scan_state *state, struct token *token)
+static int
+text(struct scan_state *state, struct token *token)
 {
-	if(peek(state) != '\n')
+	uint32_t c;
+
+	c = peek(state);
+	if(c != '\'')
 		return 0;
 
-	advance(state);
-	token->tag = TOKEN_TAG_NEWLINE;
+	do
+		c = advance(state);
+	while((c != '\n') && (c != 0) && (c != '\''));
+
+	if(c == '\'')
+		advance(state);
+	else
+		return 0;
+
+	token->tag = TOKEN_TAG_TEXT;
 	return 1;
 }
 
@@ -143,8 +174,9 @@ next_token(struct scan_state *state, struct token *token)
 	int (*scans[])(struct scan_state *, struct token *token) = {
 		single_sym,
 		identifier,
+		keyword,
 		number,
-		newline
+		text,
 	};
 
 	skip_space(state);
@@ -198,7 +230,7 @@ scan(struct string *text, struct token_list *tokens)
 }
 
 struct token_list *
-scan_line(struct aplwc *aplwc, struct string *text)
+scan_string(struct aplwc *aplwc, struct string *text, int oneline)
 {
 	enum aplwc_rc rc;
 	struct token_list *tokens;
@@ -218,7 +250,7 @@ scan_line(struct aplwc *aplwc, struct string *text)
 	tokens->length = n;
 	scan(text, tokens);
 
-	for(i = 0; i < n; i++){
+	for(i = 0; oneline && (i < n); i++){
 		if(tokens->tokens[i].tag != TOKEN_TAG_NEWLINE)
 			continue;
 
diff --git a/tests/aplwc_new.c b/tests/aplwc_new.c
index 55e742f..8d57ecd 100644
--- a/tests/aplwc_new.c
+++ b/tests/aplwc_new.c
@@ -11,15 +11,14 @@ main(void)
 	for(i = 0; i < 32; i++){
 		for(j = 0; j <= sizeof(data)-i; j++){
 			ret = aplwc_new(&aplwc, data+i, j);
+			aplwc_destroy(aplwc);
 			if(ret){
 				if(ret != APLWC_OUT_OF_MEMORY)
 					return 1;
-				if(aplwc != NULL)
-					return 1;
 				if(j >= APLWC_MIN_MEM)
 					return 1;
+				break;
 			}
-			aplwc_destroy(aplwc);
 		}
 	}