Add integer parsing to pegs.

CHANGELOG.md

@@ -1,6 +1,9 @@
 # Changelog
 All notable changes to this project will be documented in this file.
 
+## Unreleased - ???
+- Add integer parsing forms to pegs. This makes parsing many binary protocols easier.
+
 ## 1.12.2 - 2020-09-20
 - Add janet\_try and janet\_restore to C API.
 - Fix `os/execute` regression on windows.

meson.build

@@ -20,7 +20,7 @@
 
 project('janet', 'c',
   default_options : ['c_std=c99', 'b_lundef=false', 'default_library=both'],
-  version : '1.12.2')
+  version : '1.13.0')
 
 # Global settings
 janet_path = join_paths(get_option('prefix'), get_option('libdir'), 'janet')

src/conf/janetconf.h

@@ -27,10 +27,10 @@
 #define JANETCONF_H
 
 #define JANET_VERSION_MAJOR 1
-#define JANET_VERSION_MINOR 12
-#define JANET_VERSION_PATCH 2
-#define JANET_VERSION_EXTRA ""
-#define JANET_VERSION "1.12.2"
+#define JANET_VERSION_MINOR 13
+#define JANET_VERSION_PATCH 0
+#define JANET_VERSION_EXTRA "-dev"
+#define JANET_VERSION "1.13.0-dev"
 
 /* #define JANET_BUILD "local" */
 

src/core/peg.c

@@ -87,6 +87,12 @@ static void pushcap(PegState *s, Janet capture, uint32_t tag) {
     }
 }
 
+/* Convert a uint64_t to a int64_t by wrapping to a maximum number of bytes */
+static int64_t peg_convert_u64_s64(uint64_t from, int width) {
+    int shift = 8 * (8 - width);
+    return ((int64_t)(from << shift)) >> shift;
+}
+
 /* Prevent stack overflow */
 #define down1(s) do { \
     if (0 == --((s)->depth)) janet_panic("peg/match recursed too deeply"); \
@@ -469,6 +475,47 @@ tail:
             return next_text;
         }
 
+        case RULE_READINT: {
+            uint32_t tag = rule[2];
+            uint32_t signedness = rule[1] & 0x10;
+            uint32_t endianess = rule[1] & 0x20;
+            int width = (int)(rule[1] & 0xF);
+            if (text + width > s->text_end) return NULL;
+            uint64_t accum = 0;
+            if (endianess) {
+                /* BE */
+                for (int i = 0; i < width; i++) accum = (accum << 8) | text[i];
+            } else {
+                /* LE */
+                for (int i = width - 1; i >= 0; i--) accum = (accum << 8) | text[i];
+            }
+
+            Janet capture_value;
+            /* We can only parse integeres of greater than 6 bytes reliable if int-types are enabled.
+             * Otherwise, we may lose precision, so 6 is the maximum size when int-types are disabled. */
+#ifdef JANET_INT_TYPES
+            if (width > 6) {
+                if (signedness) {
+                    capture_value = janet_wrap_s64(peg_convert_u64_s64(accum, width));
+                } else {
+                    capture_value = janet_wrap_u64(accum);
+                }
+            } else
+#endif
+            {
+                double double_value;
+                if (signedness) {
+                    double_value = (double)(peg_convert_u64_s64(accum, width));
+                } else {
+                    double_value = (double)accum;
+                }
+                capture_value = janet_wrap_number(double_value);
+            }
+
+            pushcap(s, capture_value, tag);
+            return text + width;
+        }
+
     }
 }
 
@@ -876,6 +923,36 @@ static void spec_matchtime(Builder *b, int32_t argc, const Janet *argv) {
     emit_3(r, RULE_MATCHTIME, subrule, cindex, tag);
 }
 
+#ifdef JANET_INT_TYPES
+#define JANET_MAX_READINT_WIDTH 8
+#else
+#define JANET_MAX_READINT_WIDTH 6
+#endif
+
+static void spec_readint(Builder *b, int32_t argc, const Janet *argv, uint32_t mask) {
+    peg_arity(b, argc, 1, 2);
+    Reserve r = reserve(b, 3);
+    uint32_t tag = (argc == 2) ? emit_tag(b, argv[3]) : 0;
+    int32_t width = peg_getnat(b, argv[0]);
+    if ((width < 0) || (width > JANET_MAX_READINT_WIDTH)) {
+        peg_panicf(b, "width must be between 0 and %d, got %d", JANET_MAX_READINT_WIDTH, width);
+    }
+    emit_2(r, RULE_READINT, mask | ((uint32_t) width), tag);
+}
+
+static void spec_uint_le(Builder *b, int32_t argc, const Janet *argv) {
+    spec_readint(b, argc, argv, 0x0u);
+}
+static void spec_int_le(Builder *b, int32_t argc, const Janet *argv) {
+    spec_readint(b, argc, argv, 0x10u);
+}
+static void spec_uint_be(Builder *b, int32_t argc, const Janet *argv) {
+    spec_readint(b, argc, argv, 0x20u);
+}
+static void spec_int_be(Builder *b, int32_t argc, const Janet *argv) {
+    spec_readint(b, argc, argv, 0x30u);
+}
+
 /* Special compiler form */
 typedef void (*Special)(Builder *b, int32_t argc, const Janet *argv);
 typedef struct {
@@ -912,6 +989,8 @@ static const SpecialPair peg_specials[] = {
     {"group", spec_group},
     {"if", spec_if},
     {"if-not", spec_ifnot},
+    {"int", spec_int_le},
+    {"int-be", spec_int_be},
     {"lenprefix", spec_lenprefix},
     {"look", spec_look},
     {"not", spec_not},
@@ -926,6 +1005,8 @@ static const SpecialPair peg_specials[] = {
     {"some", spec_some},
     {"thru", spec_thru},
     {"to", spec_to},
+    {"uint", spec_uint_le},
+    {"uint-be", spec_uint_be},
 };
 
 /* Compile a janet value into a rule and return the rule index. */
@@ -1226,6 +1307,11 @@ static void *peg_unmarshal(JanetMarshalContext *ctx) {
                 op_flags[rule[1]] |= 0x01;
                 i += 2;
                 break;
+            case RULE_READINT:
+                /* [ width | (endianess << 5) | (signedness << 6), tag ] */
+                if (rule[1] > JANET_MAX_READINT_WIDTH) goto bad;
+                i += 3;
+                break;
             default:
                 goto bad;
         }

src/include/janet.h

@@ -1652,6 +1652,7 @@ typedef enum {
     RULE_TO,           /* [rule] */
     RULE_THRU,         /* [rule] */
     RULE_LENPREFIX,    /* [rule_a, rule_b (repeat rule_b rule_a times)] */
+    RULE_READINT,      /* [(signedness << 4) | (endianess << 5) | bytewidth, tag] */
 } JanetPegOpcode;
 
 typedef struct {

test/helper.janet

@@ -11,9 +11,12 @@
   (default e "assert error")
   (++ num-tests-run)
   (when x (++ num-tests-passed))
+  (def str (string e))
+  (def truncated
+    (if (> (length e) 40) (string (string/slice e 0 35) "...") (string e)))
   (if x
-    (xprintf stdout "\e[32m✔\e[0m %s: %v" (string e) x)
-    (xprintf stdout "\n\e[31m✘\e[0m %s: %v" (string e) x))
+    (xprintf stdout "\e[32m✔\e[0m %s: %v" truncated x)
+    (xprintf stdout "\n\e[31m✘\e[0m %s: %v" truncated x))
   x)
 
 (defmacro assert-error

test/suite0003.janet

@@ -443,4 +443,26 @@
 (check-match redef-b "aabeef" false)
 (check-match redef-b "aaaaaa" false)
 
+# Integer parsing
+
+(check-deep '(int 1) "a" @[(chr "a")])
+(check-deep '(uint 1) "a" @[(chr "a")])
+(check-deep '(int-be 1) "a" @[(chr "a")])
+(check-deep '(uint-be 1) "a" @[(chr "a")])
+(check-deep '(int 1) "\xFF" @[-1])
+(check-deep '(uint 1) "\xFF" @[255])
+(check-deep '(int-be 1) "\xFF" @[-1])
+(check-deep '(uint-be 1) "\xFF" @[255])
+(check-deep '(int 2) "\xFF\x7f" @[0x7fff])
+(check-deep '(int-be 2) "\x7f\xff" @[0x7fff])
+(check-deep '(uint 2) "\xff\x7f" @[0x7fff])
+(check-deep '(uint-be 2) "\x7f\xff" @[0x7fff])
+(check-deep '(uint-be 2) "\x7f\xff" @[0x7fff])
+(check-deep '(uint 8) "\xff\x7f\x00\x00\x00\x00\x00\x00" @[(int/u64 0x7fff)])
+(check-deep '(int 8) "\xff\x7f\x00\x00\x00\x00\x00\x00" @[(int/s64 0x7fff)])
+(check-deep '(uint 7) "\xff\x7f\x00\x00\x00\x00\x00" @[(int/u64 0x7fff)])
+(check-deep '(int 7) "\xff\x7f\x00\x00\x00\x00\x00" @[(int/s64 0x7fff)])
+
+(check-deep '(* (int 2) -1) "123" nil)
+
 (end-suite)