Split out number parsing

4 files changed, 95 insertions, 72 deletions

diff --git a/src/parse/numbers.c b/src/parse/numbers.c
new file mode 100644
index 00000000..a0db1273
--- /dev/null
+++ b/src/parse/numbers.c
@@ -0,0 +1,83 @@
+
+#include <ctype.h>
+#include <gc.h>
+#include <stdarg.h>
+#include <stdbool.h>
+#include <string.h>
+
+#include <unictype.h>
+#include <uniname.h>
+
+#include "../ast.h"
+#include "../stdlib/util.h"
+#include "context.h"
+#include "errors.h"
+#include "utils.h"
+
+static const double RADIANS_PER_DEGREE = 0.0174532925199432957692369076848861271344287188854172545609719144;
+
+public
+ast_t *parse_int(parse_ctx_t *ctx, const char *pos) {
+    const char *start = pos;
+    (void)match(&pos, "-");
+    if (!isdigit(*pos)) return NULL;
+    if (match(&pos, "0x")) { // Hex
+        pos += strspn(pos, "0123456789abcdefABCDEF_");
+    } else if (match(&pos, "0b")) { // Binary
+        pos += strspn(pos, "01_");
+    } else if (match(&pos, "0o")) { // Octal
+        pos += strspn(pos, "01234567_");
+    } else { // Decimal
+        pos += strspn(pos, "0123456789_");
+    }
+    char *str = GC_MALLOC_ATOMIC((size_t)(pos - start) + 1);
+    memset(str, 0, (size_t)(pos - start) + 1);
+    for (char *src = (char *)start, *dest = str; src < pos; ++src) {
+        if (*src != '_') *(dest++) = *src;
+    }
+
+    if (match(&pos, "e") || match(&pos, "f")) // floating point literal
+        return NULL;
+
+    if (match(&pos, "%")) {
+        double n = strtod(str, NULL) / 100.;
+        return NewAST(ctx->file, start, pos, Num, .n = n);
+    } else if (match(&pos, "deg")) {
+        double n = strtod(str, NULL) * RADIANS_PER_DEGREE;
+        return NewAST(ctx->file, start, pos, Num, .n = n);
+    }
+
+    return NewAST(ctx->file, start, pos, Int, .str = str);
+}
+
+public
+ast_t *parse_num(parse_ctx_t *ctx, const char *pos) {
+    const char *start = pos;
+    bool negative = match(&pos, "-");
+    if (!isdigit(*pos) && *pos != '.') return NULL;
+    else if (*pos == '.' && !isdigit(pos[1])) return NULL;
+
+    size_t len = strspn(pos, "0123456789_");
+    if (strncmp(pos + len, "..", 2) == 0) return NULL;
+    else if (pos[len] == '.') len += 1 + strspn(pos + len + 1, "0123456789");
+    else if (pos[len] != 'e' && pos[len] != 'f' && pos[len] != '%') return NULL;
+    if (pos[len] == 'e') {
+        len += 1;
+        if (pos[len] == '-') len += 1;
+        len += strspn(pos + len, "0123456789_");
+    }
+    char *buf = GC_MALLOC_ATOMIC(len + 1);
+    memset(buf, 0, len + 1);
+    for (char *src = (char *)pos, *dest = buf; src < pos + len; ++src) {
+        if (*src != '_') *(dest++) = *src;
+    }
+    double d = strtod(buf, NULL);
+    pos += len;
+
+    if (negative) d *= -1;
+
+    if (match(&pos, "%")) d /= 100.;
+    else if (match(&pos, "deg")) d *= RADIANS_PER_DEGREE;
+
+    return NewAST(ctx->file, start, pos, Num, .n = d);
+}
diff --git a/src/parse/numbers.h b/src/parse/numbers.h
new file mode 100644
index 00000000..89f5e28b
--- /dev/null
+++ b/src/parse/numbers.h
@@ -0,0 +1,10 @@
+// Logic for parsing numbers
+#pragma once
+
+// Parsing logic
+
+#include "../ast.h"
+#include "context.h"
+
+ast_t *parse_int(parse_ctx_t *ctx, const char *pos);
+ast_t *parse_num(parse_ctx_t *ctx, const char *pos);
diff --git a/src/parse/parse.c b/src/parse/parse.c
index 48c92ef9..53365ee7 100644
--- a/src/parse/parse.c
+++ b/src/parse/parse.c
@@ -1,5 +1,5 @@
 // Recursive descent parser for parsing code
-#include <ctype.h>
+
 #include <gc.h>
 #include <stdarg.h>
 #include <stdbool.h>
@@ -16,10 +16,10 @@
 #include "context.h"
 #include "errors.h"
 #include "files.h"
+#include "numbers.h"
 #include "parse.h"
 #include "utils.h"
 
-static const double RADIANS_PER_DEGREE = 0.0174532925199432957692369076848861271344287188854172545609719144;
 static const char closing[128] = {['('] = ')', ['['] = ']', ['<'] = '>', ['{'] = '}'};
 
 int op_tightness[] = {
@@ -49,10 +49,6 @@ int op_tightness[] = {
     [Xor] = 1,
 };
 
-///////////////////////////////////////////////////////////////////////////////////////////////////////////
-////////////////////////////////////     AST-based parsers    /////////////////////////////////////////////
-///////////////////////////////////////////////////////////////////////////////////////////////////////////
-
 ast_t *parse_parens(parse_ctx_t *ctx, const char *pos) {
     const char *start = pos;
     spaces(&pos);
@@ -75,39 +71,6 @@ ast_t *parse_parens(parse_ctx_t *ctx, const char *pos) {
     return new (ast_t, .file = (ctx)->file, .start = start, .end = pos, .tag = expr->tag, .__data = expr->__data);
 }
 
-ast_t *parse_int(parse_ctx_t *ctx, const char *pos) {
-    const char *start = pos;
-    (void)match(&pos, "-");
-    if (!isdigit(*pos)) return NULL;
-    if (match(&pos, "0x")) { // Hex
-        pos += strspn(pos, "0123456789abcdefABCDEF_");
-    } else if (match(&pos, "0b")) { // Binary
-        pos += strspn(pos, "01_");
-    } else if (match(&pos, "0o")) { // Octal
-        pos += strspn(pos, "01234567_");
-    } else { // Decimal
-        pos += strspn(pos, "0123456789_");
-    }
-    char *str = GC_MALLOC_ATOMIC((size_t)(pos - start) + 1);
-    memset(str, 0, (size_t)(pos - start) + 1);
-    for (char *src = (char *)start, *dest = str; src < pos; ++src) {
-        if (*src != '_') *(dest++) = *src;
-    }
-
-    if (match(&pos, "e") || match(&pos, "f")) // floating point literal
-        return NULL;
-
-    if (match(&pos, "%")) {
-        double n = strtod(str, NULL) / 100.;
-        return NewAST(ctx->file, start, pos, Num, .n = n);
-    } else if (match(&pos, "deg")) {
-        double n = strtod(str, NULL) * RADIANS_PER_DEGREE;
-        return NewAST(ctx->file, start, pos, Num, .n = n);
-    }
-
-    return NewAST(ctx->file, start, pos, Int, .str = str);
-}
-
 type_ast_t *parse_table_type(parse_ctx_t *ctx, const char *pos) {
     const char *start = pos;
     if (!match(&pos, "{")) return NULL;
@@ -230,37 +193,6 @@ type_ast_t *parse_type(parse_ctx_t *ctx, const char *pos) {
     return type;
 }
 
-ast_t *parse_num(parse_ctx_t *ctx, const char *pos) {
-    const char *start = pos;
-    bool negative = match(&pos, "-");
-    if (!isdigit(*pos) && *pos != '.') return NULL;
-    else if (*pos == '.' && !isdigit(pos[1])) return NULL;
-
-    size_t len = strspn(pos, "0123456789_");
-    if (strncmp(pos + len, "..", 2) == 0) return NULL;
-    else if (pos[len] == '.') len += 1 + strspn(pos + len + 1, "0123456789");
-    else if (pos[len] != 'e' && pos[len] != 'f' && pos[len] != '%') return NULL;
-    if (pos[len] == 'e') {
-        len += 1;
-        if (pos[len] == '-') len += 1;
-        len += strspn(pos + len, "0123456789_");
-    }
-    char *buf = GC_MALLOC_ATOMIC(len + 1);
-    memset(buf, 0, len + 1);
-    for (char *src = (char *)pos, *dest = buf; src < pos + len; ++src) {
-        if (*src != '_') *(dest++) = *src;
-    }
-    double d = strtod(buf, NULL);
-    pos += len;
-
-    if (negative) d *= -1;
-
-    if (match(&pos, "%")) d /= 100.;
-    else if (match(&pos, "deg")) d *= RADIANS_PER_DEGREE;
-
-    return NewAST(ctx->file, start, pos, Num, .n = d);
-}
-
 ast_t *parse_list(parse_ctx_t *ctx, const char *pos) {
     const char *start = pos;
     if (!match(&pos, "[")) return NULL;
diff --git a/src/parse/parse.h b/src/parse/parse.h
index 1e8c979f..fed78032 100644
--- a/src/parse/parse.h
+++ b/src/parse/parse.h
@@ -45,7 +45,6 @@ ast_t *parse_func_def(parse_ctx_t *ctx, const char *pos);
 ast_t *parse_heap_alloc(parse_ctx_t *ctx, const char *pos);
 ast_t *parse_if(parse_ctx_t *ctx, const char *pos);
 ast_t *parse_inline_c(parse_ctx_t *ctx, const char *pos);
-ast_t *parse_int(parse_ctx_t *ctx, const char *pos);
 ast_t *parse_lambda(parse_ctx_t *ctx, const char *pos);
 ast_t *parse_lang_def(parse_ctx_t *ctx, const char *pos);
 ast_t *parse_extend(parse_ctx_t *ctx, const char *pos);
@@ -53,7 +52,6 @@ ast_t *parse_namespace(parse_ctx_t *ctx, const char *pos);
 ast_t *parse_negative(parse_ctx_t *ctx, const char *pos);
 ast_t *parse_not(parse_ctx_t *ctx, const char *pos);
 ast_t *parse_none(parse_ctx_t *ctx, const char *pos);
-ast_t *parse_num(parse_ctx_t *ctx, const char *pos);
 ast_t *parse_parens(parse_ctx_t *ctx, const char *pos);
 ast_t *parse_pass(parse_ctx_t *ctx, const char *pos);
 ast_t *parse_path(parse_ctx_t *ctx, const char *pos);