prism 2.0.0.rc1-arm64-darwin

data/src/line_offset_list.c

@@ -0,0 +1,100 @@
+#include "prism/compiler/align.h"
+#include "prism/internal/line_offset_list.h"
+#include "prism/internal/arena.h"
+
+#include <assert.h>
+#include <string.h>
+
+/**
+ * Initialize a new line offset list with the given capacity.
+ */
+void
+pm_line_offset_list_init(pm_arena_t *arena, pm_line_offset_list_t *list, size_t capacity) {
+    list->offsets = (uint32_t *) pm_arena_alloc(arena, capacity * sizeof(uint32_t), PRISM_ALIGNOF(uint32_t));
+
+    // The first line always has offset 0.
+    list->offsets[0] = 0;
+    list->size = 1;
+    list->capacity = capacity;
+}
+
+/**
+ * Clear out the newlines that have been appended to the list.
+ */
+void
+pm_line_offset_list_clear(pm_line_offset_list_t *list) {
+    list->size = 1;
+}
+
+/**
+ * Append a new offset to the newline list (slow path: resize and store).
+ */
+void
+pm_line_offset_list_append_slow(pm_arena_t *arena, pm_line_offset_list_t *list, uint32_t cursor) {
+    size_t new_capacity = (list->capacity * 3) / 2;
+    uint32_t *new_offsets = (uint32_t *) pm_arena_alloc(arena, new_capacity * sizeof(uint32_t), PRISM_ALIGNOF(uint32_t));
+
+    memcpy(new_offsets, list->offsets, list->size * sizeof(uint32_t));
+
+    list->offsets = new_offsets;
+    list->capacity = new_capacity;
+
+    assert(list->size == 0 || cursor > list->offsets[list->size - 1]);
+    list->offsets[list->size++] = cursor;
+}
+
+/**
+ * Returns the line of the given offset. If the offset is not in the list, the
+ * line of the closest offset less than the given offset is returned.
+ */
+int32_t
+pm_line_offset_list_line(const pm_line_offset_list_t *list, uint32_t cursor, int32_t start_line) {
+    size_t left = 0;
+    size_t right = list->size - 1;
+
+    while (left <= right) {
+        size_t mid = left + (right - left) / 2;
+
+        if (list->offsets[mid] == cursor) {
+            return ((int32_t) mid) + start_line;
+        }
+
+        if (list->offsets[mid] < cursor) {
+            left = mid + 1;
+        } else {
+            right = mid - 1;
+        }
+    }
+
+    return ((int32_t) left) + start_line - 1;
+}
+
+/**
+ * Returns the line and column of the given offset. If the offset is not in the
+ * list, the line and column of the closest offset less than the given offset
+ * are returned.
+ */
+pm_line_column_t
+pm_line_offset_list_line_column(const pm_line_offset_list_t *list, uint32_t cursor, int32_t start_line) {
+    size_t left = 0;
+    size_t right = list->size - 1;
+
+    while (left <= right) {
+        size_t mid = left + (right - left) / 2;
+
+        if (list->offsets[mid] == cursor) {
+            return ((pm_line_column_t) { ((int32_t) mid) + start_line, 0 });
+        }
+
+        if (list->offsets[mid] < cursor) {
+            left = mid + 1;
+        } else {
+            right = mid - 1;
+        }
+    }
+
+    return ((pm_line_column_t) {
+        .line = ((int32_t) left) + start_line - 1,
+        .column = cursor - list->offsets[left - 1]
+    });
+}

data/src/list.c

@@ -0,0 +1,24 @@
+#include "prism/internal/list.h"
+
+/**
+ * Returns the size of the list.
+ */
+size_t
+pm_list_size(pm_list_t *list) {
+    return list->size;
+}
+
+/**
+ * Append a node to the given list.
+ */
+void
+pm_list_append(pm_list_t *list, pm_list_node_t *node) {
+    if (list->head == NULL) {
+        list->head = node;
+    } else {
+        list->tail->next = node;
+    }
+
+    list->tail = node;
+    list->size++;
+}

data/src/memchr.c

@@ -0,0 +1,37 @@
+#include "prism/internal/memchr.h"
+
+#include <stdbool.h>
+#include <stdint.h>
+#include <string.h>
+
+#define TRAILING_BYTE_MINIMUM 0x40
+
+/**
+ * We need to roll our own memchr to handle cases where the encoding changes and
+ * we need to search for a character in a buffer that could be the trailing byte
+ * of a multibyte character.
+ */
+const void *
+pm_memchr(const void *memory, int character, size_t number, bool encoding_changed, const pm_encoding_t *encoding) {
+    if (encoding_changed && encoding->multibyte && character >= TRAILING_BYTE_MINIMUM) {
+        const uint8_t *source = (const uint8_t *) memory;
+        size_t index = 0;
+
+        while (index < number) {
+            if (source[index] == character) {
+                return (void *) (source + index);
+            }
+
+            size_t width = encoding->char_width(source + index, (ptrdiff_t) (number - index));
+            if (width == 0) {
+                return NULL;
+            }
+
+            index += width;
+        }
+
+        return NULL;
+    } else {
+        return memchr(memory, character, number);
+    }
+}