String

String/c_BoyerMoore.c

@@ -0,0 +1,94 @@
+#include <c_BoyerMoore.h>
+
+C_STATIC_FORCE_INLINE
+c_int_t max(c_int_t a, c_int_t b) {
+    return a > b ? a : b;
+}
+
+c_err_t c_BoyerMoore_Init(c_BoyerMoore_t* self, const char* pattern, c_size_t pattern_size, int R, c_Allocator_t* allocator) {
+    if (!self || !pattern || pattern_size==0 || R==0) return C_ERR_PARAM;
+
+    // 前置逆向除法算术溢出防御审计
+    if (((c_size_t)-1) / sizeof(c_int_t) < (c_size_t)R) return C_ERR_PARAM;
+    if (((c_size_t)-1) / sizeof(char) < pattern_size) return C_ERR_PARAM;
+
+    self->R = R;
+    self->allocator = allocator?*allocator:c_DefaultAllocator;
+
+    self->pattern = c_Allocator_Alloc(&self->allocator, sizeof(char) * pattern_size);
+    self->right = c_Allocator_Alloc(&self->allocator, sizeof(*self->right) * R);
+
+    if (!self->pattern || !self->right) {
+        if (self->pattern) c_Allocator_Free(&self->allocator, self->pattern);
+        if (self->right)   c_Allocator_Free(&self->allocator, self->right);
+        return C_ERR_NOMEM;
+    }
+
+    memcpy(self->pattern, pattern, pattern_size);
+    self->pattern_size = pattern_size;
+
+    for (int c=0; c<R; c++) {
+        self->right[c] = -1;
+    }
+
+    // 若在当前模式串内部出现过，则记录其最右侧出现的绝对物理位置下标
+    for (c_size_t i = 0; i < pattern_size; i++) {
+        unsigned char c = (unsigned char)self->pattern[i];
+
+        // 防御性高位拦截：若模式串内包含超出指定字母表 R 上限的越界字符，安全抛出异常拦截
+        if ((int)c >= R) {
+            c_Allocator_Free(&self->allocator, self->pattern);
+            c_Allocator_Free(&self->allocator, self->right);
+            return C_ERR_OUTOFBOUND;
+        }
+        self->right[c] = (c_int_t)i;
+    }
+
+    return C_ERR_OK;
+}
+
+c_err_t c_BoyerMoore_Search(c_BoyerMoore_t* self, const char* text, c_size_t text_size, c_size_t* index) {
+    if (!self || !text || text_size==0 || !self->pattern || self->pattern_size==0) return C_ERR_PARAM;
+
+    c_size_t m = self->pattern_size;
+    c_size_t n = text_size;
+
+    // 区间饱和过滤：主文本字节宽度短于子串宽度，绝无命中可能，平滑拦截
+    if (n < m) {
+        return C_ERR_NOTFOUND;
+    }
+
+    c_size_t skip = 0;
+    for (c_size_t i=0; i<=n-m; i+=skip) {
+        skip = 0;
+        for (c_size_t j=m-1; j!=-1; j--) {
+            if (self->pattern[j]!=text[i+j]) {
+                skip = max(1, j-self->right[text[i+j]]);
+                break;
+            }
+        }
+        if (skip==0) {
+            if (index) {
+                *index = i;
+                return C_ERR_OK;
+            }
+        }
+    }
+    if (index) {
+        *index = n;
+    }
+    return C_ERR_NOTFOUND;
+}
+
+void c_BoyerMoore_Destroy(c_BoyerMoore_t* self) {
+    if (!self) return;
+    if (self->pattern) {
+        c_Allocator_Free(&self->allocator, self->pattern);
+        self->pattern = NULL;
+        self->pattern_size = 0;
+    }
+    if (self->right) {
+        c_Allocator_Free(&self->allocator, self->right);
+        self->right = NULL;
+    }
+}