.grbc file serialization/deserialization

src/main.c

@@ -2,7 +2,9 @@
 #include "interpreter.h"
 #include "compiler.h"
 #include "types.h"
+#include "serialize.h"
 #include <stdio.h>
+#include <string.h>
 
 char* getFileContents(const char* filename) {
     // https://stackoverflow.com/questions/3747086/reading-the-whole-text-file-into-a-char-array-in-c
@@ -47,19 +49,42 @@ int main(int argc, char** argv) {
     }
 
     bool compile = false;
+    bool writeBytecode = false;
+    bool readBytecode = false;
     char* fileName = NULL;
+    char* outFileName = NULL;
     List groundArgs = createList();
     for (int i = 1; i < argc; i++) {
         if (strcmp("--compile", argv[i]) == 0 || strcmp("-c", argv[i]) == 0) {
+            if (writeBytecode) {
+                printf("Cannot choose both bytecode and compilation");
+                exit(1);
+            }
             compile = true;
         }
         else if (strcmp("--help", argv[i]) == 0 || strcmp("-h", argv[i]) == 0) {
             printf("GroundVM help\n");
-            printf("Usage: %s <file> [-c] [--compile] [-h] [--help]\n", argv[0]);
+            printf("Usage: %s <file> [-c] [--compile] [-h] [--help] [-w <output>] [--writeBytecode <output>] [-b] [--bytecode]\n", argv[0]);
             printf("Options:\n");
             printf("  -c or --compile: Outputs Linux x86_64 assembly instead of interpreting (WIP)\n");
             printf("  -h or --help: Shows this help message\n");
+            printf("  -w <output> or --writebytecode <output>: Outputs binary Ground bytecode");
+            printf("  -b <output> or --bytecode <output>: Inputs binary Ground bytecode");
             exit(0);
+        } else if (strcmp("--writebytecode", argv[i]) == 0 || strcmp("-w", argv[i]) == 0) {
+            if (compile) {
+                printf("Cannot choose both bytecode and compilation");
+                exit(1);
+            }
+            writeBytecode = true;
+            if (i + 1 >= argc) {
+                printf("Usage: %s %s <output>", argv[0], argv[i]);
+                exit(1);
+            }
+            i++;
+            outFileName = argv[i];
+        } else if (strcmp("--bytecode", argv[i]) == 0 || strcmp("-b", argv[i]) == 0) {
+            readBytecode = true;
         } else {
             if (fileName == NULL) {
                 fileName = argv[i];
@@ -70,18 +95,26 @@ int main(int argc, char** argv) {
     }
 
     if (fileName == NULL) {
-        printf("Usage: %s <file> [-c] [--compile] [-h] [--help]\n", argv[0]);
+        printf("Usage: %s <file> [-c] [--compile] [-h] [--help] [-w <output>] [--writeBytecode <output>] [-b] [--bytecode]\n", argv[0]);
         printf("Error: No file name provided\n");
         exit(1);
     }
 
-    char* file = getFileContents(fileName);
+    GroundProgram program;
-    GroundProgram program = parseFile(file);
+
-    free(file);
+    if (readBytecode) {
+        deserializeProgramFromFile(fileName, &program);
+    } else {
+        char* file = getFileContents(fileName);
+        program = parseFile(file);
+        free(file);
+    }
 
     if (compile) {
         char* compiled = compileGroundProgram(&program);
         printf("%s\n", compiled);
+    } else if (writeBytecode) {
+        serializeProgramToFile(outFileName, &program);
     } else {
         GroundVariable* variables = NULL;
         GroundLabel* labels = NULL;

src/serialize.c

@@ -0,0 +1,302 @@
+#include "serialize.h"
+#include "types.h"
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <stdbool.h>
+#include <string.h>
+#include <inttypes.h>
+
+static bool writeBytes(FILE* f, const void* data, size_t n) {
+    return fwrite(data, 1, n, f) == n;
+}
+
+static bool readBytes(FILE* f, void* data, size_t n) {
+    return fread(data, 1, n, f) == n;
+}
+
+/* Convenience macros for writing/reading a single typed value. */
+#define WRITE(f, val)  writeBytes((f), &(val), sizeof(val))
+#define READ(f, val)   readBytes((f),  &(val), sizeof(val))
+
+
+bool serializeWriteString(FILE* f, const char* s) {
+    if (s == NULL) {
+        uint32_t sentinel = UINT32_MAX;
+        return WRITE(f, sentinel);
+    }
+    uint32_t len = (uint32_t)strlen(s);
+    if (!WRITE(f, len)) return false;
+    if (len > 0 && !writeBytes(f, s, len)) return false;
+    return true;
+}
+
+char* serializeReadString(FILE* f) {
+    uint32_t len;
+    if (!READ(f, len)) return NULL;
+    if (len == UINT32_MAX) return NULL;
+    char* s = malloc(len + 1);
+    if (!s) return NULL;
+    if (len > 0 && !readBytes(f, s, len)) {
+        free(s);
+        return NULL;
+    }
+    s[len] = '\0';
+    return s;
+}
+
+/* -----------------------------------------------------------------------
+ * GroundValue
+ *
+ * Format:
+ *   uint32_t  type
+ *   <payload depending on type>
+ *
+ * Only INT, DOUBLE, CHAR, BOOL, STRING, NONE are expected here.
+ * Any other type is treated as a serialization error.
+ * ----------------------------------------------------------------------- 
+*/
+
+bool serializeWriteValue(FILE* f, const GroundValue* gv) {
+    uint32_t type = (uint32_t)gv->type;
+    if (!WRITE(f, type)) return false;
+
+    switch (gv->type) {
+        case INT:
+            return WRITE(f, gv->data.intVal);
+        case DOUBLE:
+            return WRITE(f, gv->data.doubleVal);
+        case CHAR:
+            return WRITE(f, gv->data.charVal);
+        case BOOL:
+            return WRITE(f, gv->data.boolVal);
+        case STRING:
+            return serializeWriteString(f, gv->data.stringVal);
+        case NONE:
+            return true;
+        default:
+            /* LIST, FUNCTION, STRUCTVAL, CUSTOM, ERROR:
+             * These don't exist at serialization time. If you're hitting
+             * this, something has gone wrong well before we got here. */
+            fprintf(stderr, "serializeWriteValue: unexpected type %d\n", gv->type);
+            return false;
+    }
+}
+
+bool serializeReadValue(FILE* f, GroundValue* out) {
+    memset(out, 0, sizeof(*out));
+
+    uint32_t type;
+    if (!READ(f, type)) return false;
+    out->type = (GroundValueType)type;
+
+    switch (out->type) {
+        case INT:
+            return READ(f, out->data.intVal);
+        case DOUBLE:
+            return READ(f, out->data.doubleVal);
+        case CHAR:
+            return READ(f, out->data.charVal);
+        case BOOL:
+            return READ(f, out->data.boolVal);
+        case STRING: {
+            char* s = serializeReadString(f);
+            /* NULL is a valid encoded value (the sentinel case), but
+             * we only wrote non-NULL strings, so treat NULL-read as error. */
+            if (!s) return false;
+            out->data.stringVal = s;
+            return true;
+        }
+        case NONE:
+            return true;
+        default:
+            fprintf(stderr, "serializeReadValue: unexpected type %d\n", out->type);
+            return false;
+    }
+}
+
+/* -----------------------------------------------------------------------
+ * GroundArg
+ *
+ * Format:
+ *   uint32_t  argType
+ *   <if VALUE>  serialized GroundValue
+ *   <otherwise> length-prefixed refName string
+ * ----------------------------------------------------------------------- 
+*/
+
+bool serializeWriteArg(FILE* f, const GroundArg* ga) {
+    uint32_t type = (uint32_t)ga->type;
+    if (!WRITE(f, type)) return false;
+
+    if (ga->type == VALUE) {
+        return serializeWriteValue(f, &ga->value.value);
+    } else {
+        /* VALREF, DIRREF, LINEREF, LABEL, FNREF, TYPEREF — all carry a refName */
+        return serializeWriteString(f, ga->value.refName);
+    }
+}
+
+bool serializeReadArg(FILE* f, GroundArg* out) {
+    memset(out, 0, sizeof(*out));
+
+    uint32_t type;
+    if (!READ(f, type)) return false;
+    out->type = (GroundArgType)type;
+
+    if (out->type == VALUE) {
+        return serializeReadValue(f, &out->value.value);
+    } else {
+        char* ref = serializeReadString(f);
+        if (!ref) return false;
+        out->value.refName = ref;
+        return true;
+    }
+}
+
+/* -----------------------------------------------------------------------
+ * GroundInstruction
+ *
+ * Format:
+ *   uint32_t  instType
+ *   uint64_t  argCount
+ *   <argCount x serialized GroundArg>
+ * ----------------------------------------------------------------------- 
+*/
+
+bool serializeWriteInstruction(FILE* f, const GroundInstruction* gi) {
+    uint32_t type = (uint32_t)gi->type;
+    if (!WRITE(f, type)) return false;
+
+    uint64_t argc = (uint64_t)gi->args.length;
+    if (!WRITE(f, argc)) return false;
+
+    for (size_t i = 0; i < gi->args.length; i++) {
+        if (!serializeWriteArg(f, &gi->args.args[i])) return false;
+    }
+    return true;
+}
+
+bool serializeReadInstruction(FILE* f, GroundInstruction* out) {
+    uint32_t type;
+    if (!READ(f, type)) return false;
+
+    *out = createGroundInstruction((GroundInstType)type);
+
+    uint64_t argc;
+    if (!READ(f, argc)) return false;
+
+    for (uint64_t i = 0; i < argc; i++) {
+        GroundArg arg;
+        if (!serializeReadArg(f, &arg)) {
+            /* Free whatever args we've already read before bailing. */
+            freeGroundInstruction(out);
+            return false;
+        }
+        addArgToInstruction(out, arg);
+    }
+    return true;
+}
+
+/* -----------------------------------------------------------------------
+ * GroundProgram  —  top-level
+ *
+ * File layout:
+ *   uint32_t  magic    (GROUND_MAGIC)
+ *   uint32_t  version  (GROUND_VERSION)
+ *   uint64_t  instrCount
+ *   <instrCount x serialized GroundInstruction>
+ * ----------------------------------------------------------------------- 
+*/
+
+bool serializeProgramToFile(const char* path, const GroundProgram* prog) {
+    FILE* f = fopen(path, "wb");
+    if (!f) {
+        perror("serializeProgramToFile: fopen");
+        return false;
+    }
+
+    bool ok = true;
+
+    /* Header */
+    uint32_t magic   = GROUND_MAGIC;
+    uint32_t version = GROUND_VERSION;
+    ok = ok && WRITE(f, magic);
+    ok = ok && WRITE(f, version);
+
+    /* Instruction count, then instructions */
+    uint64_t count = (uint64_t)prog->size;
+    ok = ok && WRITE(f, count);
+
+    for (size_t i = 0; i < prog->size && ok; i++) {
+        ok = serializeWriteInstruction(f, &prog->instructions[i]);
+    }
+
+    if (!ok) {
+        fprintf(stderr, "serializeProgramToFile: write error\n");
+    }
+
+    fclose(f);
+    return ok;
+}
+
+bool deserializeProgramFromFile(const char* path, GroundProgram* out) {
+    memset(out, 0, sizeof(*out));
+
+    FILE* f = fopen(path, "rb");
+    if (!f) {
+        perror("deserializeProgramFromFile: fopen");
+        return false;
+    }
+
+    bool ok = true;
+
+    /* Validate header */
+    uint32_t magic, version;
+    ok = ok && READ(f, magic);
+    ok = ok && READ(f, version);
+
+    if (!ok || magic != GROUND_MAGIC) {
+        fprintf(stderr, "deserializeProgramFromFile: bad magic (got 0x%08X)\n", magic);
+        fclose(f);
+        return false;
+    }
+    if (version != GROUND_VERSION) {
+        fprintf(stderr, "deserializeProgramFromFile: unsupported version %u\n", version);
+        fclose(f);
+        return false;
+    }
+
+    /* Read instruction count */
+    uint64_t count;
+    if (!READ(f, count)) {
+        fclose(f);
+        return false;
+    }
+
+    out->instructions = malloc(sizeof(GroundInstruction) * count);
+    if (!out->instructions && count > 0) {
+        fprintf(stderr, "deserializeProgramFromFile: malloc failed\n");
+        fclose(f);
+        return false;
+    }
+    out->size = 0;  /* incremented as we go so partial frees are safe */
+
+    for (uint64_t i = 0; i < count; i++) {
+        if (!serializeReadInstruction(f, &out->instructions[i])) {
+            fprintf(stderr, "deserializeProgramFromFile: failed at instruction %llu\n",
+                    (unsigned long long)i);
+            /* Free everything successfully read so far. */
+            for (size_t j = 0; j < out->size; j++)
+                freeGroundInstruction(&out->instructions[j]);
+            free(out->instructions);
+            memset(out, 0, sizeof(*out));
+            fclose(f);
+            return false;
+        }
+        out->size++;
+    }
+
+    fclose(f);
+    return true;
+}

src/serialize.h

@@ -0,0 +1,48 @@
+#ifndef SERIALIZE_H
+#define SERIALIZE_H
+
+#include <stdio.h>
+#include <stdbool.h>
+#include "types.h"
+
+/*
+ * Magic number and version for Ground bytecode files.
+ * The magic bytes spell 'GRND'
+ */
+#define GROUND_MAGIC   0x47524E44u
+#define GROUND_VERSION 1u
+
+/*
+ * File header written at the start of every .grbc file.
+ */
+typedef struct GroundBytecodeHeader {
+    uint32_t magic;
+    uint32_t version;
+} GroundBytecodeHeader;
+
+/*
+ * Writes a length-prefixed UTF-8 string to (f).
+ * NULL is encoded as a sentinel length (UINT32_MAX).
+ * Returns true on success.
+ */
+bool serializeWriteString(FILE* f, const char* s);
+
+char* serializeReadString(FILE* f);
+
+
+bool serializeWriteValue(FILE* f, const GroundValue* gv);
+bool serializeReadValue(FILE* f, GroundValue* out);
+
+
+bool serializeWriteArg(FILE* f, const GroundArg* ga);
+bool serializeReadArg(FILE* f, GroundArg* out);
+
+
+bool serializeWriteInstruction(FILE* f, const GroundInstruction* gi);
+bool serializeReadInstruction(FILE* f, GroundInstruction* out);
+
+bool serializeProgramToFile(const char* path, const GroundProgram* prog);
+
+bool deserializeProgramFromFile(const char* path, GroundProgram* out);
+
+#endif