prism 2.0.0-arm64-darwin

data/include/prism/constant_pool.h

@@ -0,0 +1,106 @@
+/**
+ * @file constant_pool.h
+ *
+ * A data structure that stores a set of strings.
+ *
+ * Each string is assigned a unique id, which can be used to compare strings for
+ * equality. This comparison ends up being much faster than strcmp, since it
+ * only requires a single integer comparison.
+ */
+#ifndef PRISM_CONSTANT_POOL_H
+#define PRISM_CONSTANT_POOL_H
+
+#include "prism/compiler/exported.h"
+#include "prism/compiler/inline.h"
+#include "prism/compiler/nodiscard.h"
+#include "prism/compiler/nonnull.h"
+
+#include "prism/arena.h"
+
+#include <stddef.h>
+#include <stdint.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/**
+ * A constant id is a unique identifier for a constant in the constant pool.
+ */
+typedef uint32_t pm_constant_id_t;
+
+/**
+ * Returns the index of the constant with the given id among the constants of
+ * the parser that produced it. The indices run from 0 to one less than
+ * pm_parser_constants_size, so they can index a cache with one entry per
+ * constant that was filled using pm_parser_constant_at.
+ *
+ * @param constant_id the id of the constant. Optional constant fields on nodes
+ *     hold 0 when they are absent, which is not a valid id, so check for it
+ *     before calling this function.
+ * @returns the index of the constant
+ */
+static PRISM_INLINE size_t
+pm_constant_id_index(pm_constant_id_t constant_id) {
+    return (size_t) constant_id - 1;
+}
+
+/**
+ * A list of constant IDs. Usually used to represent a set of locals.
+ */
+typedef struct {
+    /** The number of constant ids in the list. */
+    size_t size;
+
+    /** The number of constant ids that have been allocated in the list. */
+    size_t capacity;
+
+    /** The constant ids in the list. */
+    pm_constant_id_t *ids;
+} pm_constant_id_list_t;
+
+/** A constant in the pool which effectively stores a string. */
+typedef struct pm_constant_t pm_constant_t;
+
+/**
+ * The overall constant pool, which stores constants found while parsing.
+ */
+typedef struct pm_constant_pool_t pm_constant_pool_t;
+
+/**
+ * Return a raw pointer to the start of a constant.
+ *
+ * @param constant The constant to get the start of.
+ * @returns A raw pointer to the start of the constant.
+ */
+PRISM_EXPORTED_FUNCTION const uint8_t * pm_constant_start(const pm_constant_t *constant) PRISM_NONNULL(1);
+
+/**
+ * Return the length of a constant.
+ *
+ * @param constant The constant to get the length of.
+ * @returns The length of the constant.
+ */
+PRISM_EXPORTED_FUNCTION size_t pm_constant_length(const pm_constant_t *constant) PRISM_NONNULL(1);
+
+/**
+ * Initialize a list of constant ids.
+ *
+ * @param list The list to initialize.
+ */
+PRISM_EXPORTED_FUNCTION void pm_constant_id_list_init(pm_constant_id_list_t *list) PRISM_NONNULL(1);
+
+/**
+ * Append a constant id to a list of constant ids.
+ *
+ * @param arena The arena to use for allocations.
+ * @param list The list to append to.
+ * @param id The constant id to append.
+ */
+PRISM_EXPORTED_FUNCTION void pm_constant_id_list_append(pm_arena_t *arena, pm_constant_id_list_t *list, pm_constant_id_t id) PRISM_NONNULL(1, 2);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif

data/include/prism/diagnostic.h

@@ -0,0 +1,111 @@
+/**
+ * @file diagnostic.h
+ *
+ * A list of diagnostics generated during parsing.
+ */
+#ifndef PRISM_DIAGNOSTIC_H
+#define PRISM_DIAGNOSTIC_H
+
+#include "prism/compiler/exported.h"
+#include "prism/compiler/nodiscard.h"
+#include "prism/compiler/nonnull.h"
+
+#include "prism/ast.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/**
+ * An opaque pointer to a diagnostic generated during parsing.
+ */
+typedef struct pm_diagnostic_t pm_diagnostic_t;
+
+/**
+ * The levels of errors generated during parsing.
+ */
+typedef enum {
+    /** For errors that should raise a syntax error. */
+    PM_ERROR_LEVEL_SYNTAX = 0,
+
+    /** For errors that should raise an argument error. */
+    PM_ERROR_LEVEL_ARGUMENT = 1,
+
+    /** For errors that should raise a load error. */
+    PM_ERROR_LEVEL_LOAD = 2
+} pm_error_level_t;
+
+/**
+ * The levels of warnings generated during parsing.
+ */
+typedef enum {
+    /** For warnings which should be emitted if $VERBOSE != nil. */
+    PM_WARNING_LEVEL_DEFAULT = 0,
+
+    /** For warnings which should be emitted if $VERBOSE == true. */
+    PM_WARNING_LEVEL_VERBOSE = 1
+} pm_warning_level_t;
+
+/**
+ * Get the type of the given diagnostic.
+ *
+ * @param diagnostic The diagnostic to get the type of.
+ * @returns The type of the given diagnostic. Note that this is a string
+ *     representation of an internal ID, and is not meant to be relied upon as a
+ *     stable identifier for the diagnostic. We do not guarantee that these will
+ *     not change in the future. This is meant to be used for debugging and
+ *     error reporting purposes, and not for programmatic checks.
+ */
+PRISM_EXPORTED_FUNCTION const char * pm_diagnostic_type(const pm_diagnostic_t *diagnostic) PRISM_NONNULL(1);
+
+/**
+ * Get the location of the given diagnostic.
+ *
+ * @param diagnostic The diagnostic to get the location of.
+ * @returns The location of the given diagnostic.
+ */
+PRISM_EXPORTED_FUNCTION pm_location_t pm_diagnostic_location(const pm_diagnostic_t *diagnostic) PRISM_NONNULL(1);
+
+/**
+ * Get the message of the given diagnostic.
+ *
+ * @param diagnostic The diagnostic to get the message of.
+ * @returns The message of the given diagnostic.
+ */
+PRISM_EXPORTED_FUNCTION const char * pm_diagnostic_message(const pm_diagnostic_t *diagnostic) PRISM_NONNULL(1);
+
+/**
+ * Get the error level associated with the given diagnostic.
+ *
+ * @param diagnostic The diagnostic to get the error level of.
+ * @returns The error level of the given diagnostic. If the diagnostic was a
+ *     warning, or is in any way not an error, then the return value is
+ *     undefined and should not be relied upon.
+ */
+PRISM_EXPORTED_FUNCTION pm_error_level_t pm_diagnostic_error_level(const pm_diagnostic_t *diagnostic) PRISM_NONNULL(1);
+
+/**
+ * Get the warning level associated with the given diagnostic.
+ *
+ * @param diagnostic The diagnostic to get the warning level of.
+ * @returns The warning level of the given diagnostic. If the diagnostic was an
+ *     error, or is in any way not a warning, then the return value is
+ *     undefined and should not be relied upon.
+ */
+PRISM_EXPORTED_FUNCTION pm_warning_level_t pm_diagnostic_warning_level(const pm_diagnostic_t *diagnostic) PRISM_NONNULL(1);
+
+/**
+ * Get the diagnostic that follows the given diagnostic among the errors or the
+ * warnings of the parser that produced it.
+ *
+ * @param diagnostic The diagnostic to get the successor of.
+ * @returns The next diagnostic of the same kind, or NULL if the given
+ *     diagnostic is the last one.
+ */
+PRISM_EXPORTED_FUNCTION const pm_diagnostic_t * pm_diagnostic_next(const pm_diagnostic_t *diagnostic) PRISM_NONNULL(1);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif

data/include/prism/errors_format.h

@@ -0,0 +1,61 @@
+/**
+ * @file errors_format.h
+ *
+ * A function for formatting the errors in a parser into a buffer.
+ */
+#ifndef PRISM_ERRORS_FORMAT_H
+#define PRISM_ERRORS_FORMAT_H
+
+#include "prism/compiler/exported.h"
+#include "prism/compiler/nodiscard.h"
+#include "prism/compiler/nonnull.h"
+
+#include "prism/buffer.h"
+#include "prism/diagnostic.h"
+#include "prism/parser.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/** The type of formatting to use when formatting errors. */
+typedef enum {
+    /** Format errors in a plain format with no colors or styles. Typically
+     * used when outputting to a file or when the output is not a terminal.
+     */
+    PM_ERRORS_FORMAT_PLAIN = 1,
+
+    /** Format errors in a style format with bold only. Typically used when
+     * outputting to a terminal where you do not want colors.
+     */
+    PM_ERRORS_FORMAT_STYLE = 2,
+
+    /** Format errors in a color format with bold and colors. The default case
+     * when outputting on the terminal.
+     */
+    PM_ERRORS_FORMAT_COLOR = 3
+} pm_errors_format_type_t;
+
+/**
+ * Format the errors in a parser into a buffer. After return, the buffer will
+ * contain the formatted errors and any relevant source snippets if available
+ * and possible.
+ *
+ * @param parser The parser containing the errors to format. Note that the
+ *   parser does not need to be fully initialized, but it will need to have
+ *   start, end, start_line, encoding, line_offsets, filepath, and error_list
+ *   initialized.
+ * @param buffer The buffer to format the errors into.
+ * @param format_type The type of formatting to use when formatting the errors.
+ * @returns The error level of the error with the highest precedence that was
+ *   formatted. In practice this means it will be the argument or load error
+ *   level if any were found, otherwise it will be the syntax error level.
+ *   Callers should raise appropriate error types based on the returned level.
+ */
+PRISM_EXPORTED_FUNCTION pm_error_level_t pm_errors_format(const pm_parser_t *parser, pm_buffer_t *buffer, pm_errors_format_type_t format_type) PRISM_NONNULL(1, 2);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif

data/include/prism/excludes.h

@@ -0,0 +1,29 @@
+/**
+ * @file excludes.h
+ *
+ * A header file that defines macros to exclude certain features of the prism
+ * library. This is useful for reducing the size of the library when certain
+ * features are not needed.
+ */
+#ifndef PRISM_EXCLUDES_H
+#define PRISM_EXCLUDES_H
+
+/**
+ * If PRISM_BUILD_MINIMAL is defined, then we're going to define every possible
+ * switch that will turn off certain features of prism.
+ */
+#ifdef PRISM_BUILD_MINIMAL
+    /** Exclude the serialization API. */
+    #define PRISM_EXCLUDE_SERIALIZATION
+
+    /** Exclude the JSON serialization API. */
+    #define PRISM_EXCLUDE_JSON
+
+    /** Exclude the prettyprint API. */
+    #define PRISM_EXCLUDE_PRETTYPRINT
+
+    /** Exclude the full set of encodings, using the minimal only. */
+    #define PRISM_ENCODING_EXCLUDE_FULL
+#endif
+
+#endif

data/include/prism/integer.h

@@ -0,0 +1,41 @@
+/**
+ * @file integer.h
+ *
+ * This module provides functions for working with arbitrary-sized integers.
+ */
+#ifndef PRISM_INTEGER_H
+#define PRISM_INTEGER_H
+
+#include <stdbool.h>
+#include <stddef.h>
+#include <stdint.h>
+
+/**
+ * A structure represents an arbitrary-sized integer.
+ */
+typedef struct {
+    /**
+     * The number of allocated values. length is set to 0 if the integer fits
+     * into uint32_t.
+     */
+    size_t length;
+
+    /**
+     * List of 32-bit integers. Set to NULL if the integer fits into uint32_t.
+     */
+    uint32_t *values;
+
+    /**
+     * Embedded value for small integer. This value is set to 0 if the value
+     * does not fit into uint32_t.
+     */
+    uint32_t value;
+
+    /**
+     * Whether or not the integer is negative. It is stored this way so that a
+     * zeroed pm_integer_t is always positive zero.
+     */
+    bool negative;
+} pm_integer_t;
+
+#endif

data/include/prism/internal/allocator.h

@@ -0,0 +1,68 @@
+#ifndef PRISM_INTERNAL_ALLOCATOR_H
+#define PRISM_INTERNAL_ALLOCATOR_H
+
+/* If you build Prism with a custom allocator, configure it with
+ * "-D PRISM_XALLOCATOR" to use your own allocator that defines xmalloc,
+ * xrealloc, xcalloc, and xfree.
+ *
+ * For example, your `prism_xallocator.h` file could look like this:
+ *
+ * ```
+ * #ifndef PRISM_XALLOCATOR_H
+ * #define PRISM_XALLOCATOR_H
+ * #define xmalloc          my_malloc
+ * #define xrealloc         my_realloc
+ * #define xcalloc          my_calloc
+ * #define xfree            my_free
+ * #define xrealloc_sized   my_realloc_sized // (optional)
+ * #define xfree_sized      my_free_sized    // (optional)
+ * #endif
+ * ```
+ */
+#ifdef PRISM_XALLOCATOR
+    #include "prism_xallocator.h"
+#else
+    #ifndef xmalloc
+        /* The malloc function that should be used. This can be overridden with
+         * the PRISM_XALLOCATOR define. */
+        #define xmalloc malloc
+    #endif
+
+    #ifndef xrealloc
+        /* The realloc function that should be used. This can be overridden with
+         * the PRISM_XALLOCATOR define. */
+        #define xrealloc realloc
+    #endif
+
+    #ifndef xcalloc
+        /* The calloc function that should be used. This can be overridden with
+         * the PRISM_XALLOCATOR define. */
+        #define xcalloc calloc
+    #endif
+
+    #ifndef xfree
+        /* The free function that should be used. This can be overridden with
+         * the PRISM_XALLOCATOR define. */
+        #define xfree free
+    #endif
+#endif
+
+#ifndef xfree_sized
+    /* The free_sized function that should be used. This can be overridden with
+     * the PRISM_XALLOCATOR define. If not defined, defaults to calling xfree.
+     */
+    #define xfree_sized(p, s) xfree(((void)(s), (p)))
+#endif
+
+#ifndef xrealloc_sized
+    /* The xrealloc_sized function that should be used. This can be overridden
+     * with the PRISM_XALLOCATOR define. If not defined, defaults to calling
+     * xrealloc. */
+    #define xrealloc_sized(p, ns, os) xrealloc((p), ((void)(os), (ns)))
+#endif
+
+#ifdef PRISM_BUILD_DEBUG
+    #include "prism/internal/allocator_debug.h"
+#endif
+
+#endif

data/include/prism/internal/allocator_debug.h

@@ -0,0 +1,88 @@
+#ifndef PRISM_INTERNAL_ALLOCATOR_DEBUG_H
+#define PRISM_INTERNAL_ALLOCATOR_DEBUG_H
+
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+static inline void *
+pm_allocator_debug_malloc(size_t size) {
+    size_t *memory = xmalloc(size + sizeof(size_t));
+    memory[0] = size;
+    return memory + 1;
+}
+
+static inline void *
+pm_allocator_debug_calloc(size_t nmemb, size_t size) {
+    size_t total_size = nmemb * size;
+    void *ptr = pm_allocator_debug_malloc(total_size);
+    memset(ptr, 0, total_size);
+    return ptr;
+}
+
+static inline void *
+pm_allocator_debug_realloc(void *ptr, size_t size) {
+    if (ptr == NULL) {
+        return pm_allocator_debug_malloc(size);
+    }
+
+    size_t *memory = (size_t *)ptr;
+    void *raw_memory = memory - 1;
+    memory = (size_t *)xrealloc(raw_memory, size + sizeof(size_t));
+    memory[0] = size;
+    return memory + 1;
+}
+
+static inline void
+pm_allocator_debug_free(void *ptr) {
+    if (ptr != NULL) {
+        size_t *memory = (size_t *)ptr;
+        xfree(memory - 1);
+    }
+}
+
+static inline void
+pm_allocator_debug_free_sized(void *ptr, size_t old_size) {
+    if (ptr != NULL) {
+        size_t *memory = (size_t *)ptr;
+        if (old_size != memory[-1]) {
+            fprintf(stderr, "[BUG] buffer %p was allocated with size %lu but freed with size %lu\n", ptr, memory[-1], old_size);
+            abort();
+        }
+        xfree_sized(memory - 1, old_size + sizeof(size_t));
+    }
+}
+
+static inline void *
+pm_allocator_debug_realloc_sized(void *ptr, size_t size, size_t old_size) {
+    if (ptr == NULL) {
+        if (old_size != 0) {
+            fprintf(stderr, "[BUG] realloc_sized called with NULL pointer and old size %lu\n", old_size);
+            abort();
+        }
+        return pm_allocator_debug_malloc(size);
+    }
+
+    size_t *memory = (size_t *)ptr;
+    if (old_size != memory[-1]) {
+        fprintf(stderr, "[BUG] buffer %p was allocated with size %lu but realloced with size %lu\n", ptr, memory[-1], old_size);
+        abort();
+    }
+    return pm_allocator_debug_realloc(ptr, size);
+}
+
+#undef xmalloc
+#undef xrealloc
+#undef xcalloc
+#undef xfree
+#undef xrealloc_sized
+#undef xfree_sized
+
+#define xmalloc          pm_allocator_debug_malloc
+#define xrealloc         pm_allocator_debug_realloc
+#define xcalloc          pm_allocator_debug_calloc
+#define xfree            pm_allocator_debug_free
+#define xrealloc_sized   pm_allocator_debug_realloc_sized
+#define xfree_sized      pm_allocator_debug_free_sized
+
+#endif

data/include/prism/internal/arena.h

@@ -0,0 +1,108 @@
+#ifndef PRISM_INTERNAL_ARENA_H
+#define PRISM_INTERNAL_ARENA_H
+
+#include "prism/compiler/exported.h"
+#include "prism/compiler/flex_array.h"
+#include "prism/compiler/force_inline.h"
+#include "prism/compiler/inline.h"
+
+#include "prism/arena.h"
+
+#include <stddef.h>
+#include <string.h>
+
+/*
+ * A single block of memory in the arena. Blocks are linked via prev pointers so
+ * they can be freed by walking the chain.
+ */
+typedef struct pm_arena_block {
+    /* The previous block in the chain (for freeing). */
+    struct pm_arena_block *prev;
+
+    /* The total usable bytes in data[]. */
+    size_t capacity;
+
+    /* The number of bytes consumed so far. */
+    size_t used;
+
+    /* The block's data. */
+    char data[PM_FLEX_ARRAY_LENGTH];
+} pm_arena_block_t;
+
+/*
+ * A bump allocator. Allocations are made by bumping a pointer within the
+ * current block. When a block is full, a new block is allocated and linked to
+ * the previous one. All blocks are freed at once by walking the chain.
+ */
+struct pm_arena_t {
+    /* The active block (allocate from here). */
+    pm_arena_block_t *current;
+
+    /* The number of blocks allocated. */
+    size_t block_count;
+};
+
+/*
+ * Free all blocks in the arena. After this call, all pointers returned by
+ * pm_arena_alloc and pm_arena_zalloc are invalid.
+ */
+void pm_arena_cleanup(pm_arena_t *arena);
+
+/*
+ * Ensure the arena has at least `capacity` bytes available in its current
+ * block, allocating a new block if necessary. This allows callers to
+ * pre-size the arena to avoid repeated small block allocations.
+ */
+void pm_arena_reserve(pm_arena_t *arena, size_t capacity);
+
+/*
+ * Slow path for pm_arena_alloc: allocate a new block and return a pointer to
+ * the first `size` bytes. Do not call directly — use pm_arena_alloc instead.
+ */
+void * pm_arena_alloc_slow(pm_arena_t *arena, size_t size);
+
+/*
+ * Allocate memory from the arena. The returned memory is NOT zeroed. This
+ * function is infallible — it aborts on allocation failure.
+ *
+ * The fast path (bump pointer within the current block) is inlined at each
+ * call site. The slow path (new block allocation) is out-of-line.
+ */
+static PRISM_FORCE_INLINE void *
+pm_arena_alloc(pm_arena_t *arena, size_t size, size_t alignment) {
+    if (arena->current != NULL) {
+        size_t used_aligned = (arena->current->used + alignment - 1) & ~(alignment - 1);
+        size_t needed = used_aligned + size;
+
+        if (used_aligned >= arena->current->used && needed >= used_aligned && needed <= arena->current->capacity) {
+            arena->current->used = needed;
+            return arena->current->data + used_aligned;
+        }
+    }
+
+    return pm_arena_alloc_slow(arena, size);
+}
+
+/*
+ * Allocate zero-initialized memory from the arena. This function is infallible
+ * — it aborts on allocation failure.
+ */
+static PRISM_INLINE void *
+pm_arena_zalloc(pm_arena_t *arena, size_t size, size_t alignment) {
+    void *ptr = pm_arena_alloc(arena, size, alignment);
+    memset(ptr, 0, size);
+    return ptr;
+}
+
+/*
+ * Allocate memory from the arena and copy the given data into it. This is a
+ * convenience wrapper around pm_arena_alloc + memcpy.
+ */
+static PRISM_INLINE void *
+pm_arena_memdup(pm_arena_t *arena, const void *src, size_t size, size_t alignment) {
+    void *dst = pm_arena_alloc(arena, size, alignment);
+    memcpy(dst, src, size);
+    return dst;
+}
+
+#endif

data/include/prism/internal/bit.h

@@ -0,0 +1,42 @@
+#ifndef PRISM_INTERNAL_BIT_H
+#define PRISM_INTERNAL_BIT_H
+
+#include "prism/compiler/inline.h"
+
+/*
+ * Count trailing zero bits in a 64-bit value. Used by SWAR identifier scanning
+ * to find the first non-matching byte in a word.
+ *
+ * Precondition: v must be nonzero. The result is undefined when v == 0
+ * (matching the behavior of __builtin_ctzll and _BitScanForward64).
+ */
+#if defined(__GNUC__) || defined(__clang__)
+#define pm_ctzll(v) ((unsigned) __builtin_ctzll(v))
+#elif defined(_MSC_VER) && defined(_WIN64)
+#include <intrin.h>
+#include <stdint.h>
+
+static PRISM_INLINE unsigned
+pm_ctzll(uint64_t v) {
+    unsigned long index;
+    _BitScanForward64(&index, v);
+    return (unsigned) index;
+}
+#else
+#include <stdint.h>
+
+static PRISM_INLINE unsigned
+pm_ctzll(uint64_t v) {
+    unsigned c = 0;
+    v &= (uint64_t) (-(int64_t) v);
+    if (v & 0x00000000FFFFFFFFULL) c += 0;  else c += 32;
+    if (v & 0x0000FFFF0000FFFFULL) c += 0;  else c += 16;
+    if (v & 0x00FF00FF00FF00FFULL) c += 0;  else c += 8;
+    if (v & 0x0F0F0F0F0F0F0F0FULL) c += 0;  else c += 4;
+    if (v & 0x3333333333333333ULL) c += 0;  else c += 2;
+    if (v & 0x5555555555555555ULL) c += 0;  else c += 1;
+    return c;
+}
+#endif
+
+#endif