tranfi 0.1.2 → 0.2.0

package/csrc/op_acf.c

@@ -1,6 +1,6 @@
 /*
  * op_acf.c — Autocorrelation function.
- * Aggregate op: buffers all values, computes ACF for lags 0..N.
+ * Aggregate op: buffers numeric values, computes ACF for lags 0..N.
  * Output: 2-column table (lag, acf).
  *
  * Config: {"column": "price", "lags": 20}
@@ -12,6 +12,18 @@
 #include <string.h>
 #include <stdio.h>
 #include <math.h>
+#include <limits.h>
+
+typedef enum {
+    ACF_MISSING_ERROR,
+    ACF_MISSING_NULL,
+    ACF_MISSING_IGNORE
+} acf_missing_policy;
+
+typedef enum {
+    ACF_TYPE_FAIL,
+    ACF_TYPE_NULL
+} acf_type_policy;
 
 typedef struct {
     char   *column;
@@ -19,30 +31,112 @@ typedef struct {
     double *values;
     size_t  n_values;
     size_t  cap_values;
+    int     checked_schema;
+    int     col_idx;
+    int     ignore;
+    int     force_null;
+    acf_missing_policy missing;
+    acf_type_policy on_type_error;
 } acf_state;
 
+static int acf_is_numeric(tf_type type) {
+    return type == TF_TYPE_INT64 || type == TF_TYPE_FLOAT64;
+}
+
 static double get_numeric(const tf_batch *b, size_t r, int ci) {
     if (b->col_types[ci] == TF_TYPE_INT64) return (double)tf_batch_get_int64(b, r, ci);
     if (b->col_types[ci] == TF_TYPE_FLOAT64) return tf_batch_get_float64(b, r, ci);
     return 0;
 }
 
+static int acf_parse_missing_policy(const cJSON *args, acf_missing_policy *out) {
+    *out = ACF_MISSING_ERROR;
+    const cJSON *j = cJSON_GetObjectItemCaseSensitive(args, "missing");
+    if (!j) return TF_OK;
+    if (!cJSON_IsString(j)) {
+        tf_set_last_error("acf: missing must be error, null, or ignore");
+        return TF_ERROR;
+    }
+    if (strcmp(j->valuestring, "error") == 0) *out = ACF_MISSING_ERROR;
+    else if (strcmp(j->valuestring, "null") == 0) *out = ACF_MISSING_NULL;
+    else if (strcmp(j->valuestring, "ignore") == 0) *out = ACF_MISSING_IGNORE;
+    else {
+        tf_set_last_error("acf: missing must be error, null, or ignore");
+        return TF_ERROR;
+    }
+    return TF_OK;
+}
+
+static int acf_parse_type_policy(const cJSON *args, acf_type_policy *out) {
+    *out = ACF_TYPE_FAIL;
+    const cJSON *j = cJSON_GetObjectItemCaseSensitive(args, "on_type_error");
+    if (!j) return TF_OK;
+    if (!cJSON_IsString(j)) {
+        tf_set_last_error("acf: on_type_error must be fail or null");
+        return TF_ERROR;
+    }
+    if (strcmp(j->valuestring, "fail") == 0) *out = ACF_TYPE_FAIL;
+    else if (strcmp(j->valuestring, "null") == 0) *out = ACF_TYPE_NULL;
+    else {
+        tf_set_last_error("acf: on_type_error must be fail or null");
+        return TF_ERROR;
+    }
+    return TF_OK;
+}
+
+static int acf_set_col_error(acf_state *st, const tf_batch *in) {
+    if (st->col_idx < 0) {
+        char msg[256];
+        snprintf(msg, sizeof(msg), "acf: column '%s' not found", st->column);
+        tf_set_last_error(msg);
+        return TF_ERROR;
+    }
+    if (!acf_is_numeric(in->col_types[st->col_idx])) {
+        char msg[256];
+        snprintf(msg, sizeof(msg), "acf: column '%s' must be numeric", st->column);
+        tf_set_last_error(msg);
+        return TF_ERROR;
+    }
+    return TF_OK;
+}
+
+static int acf_check_schema(acf_state *st, const tf_batch *in) {
+    if (st->checked_schema) return TF_OK;
+    st->checked_schema = 1;
+    st->col_idx = tf_batch_col_index(in, st->column);
+    if (st->col_idx < 0) {
+        if (st->missing == ACF_MISSING_ERROR) return acf_set_col_error(st, in);
+        if (st->missing == ACF_MISSING_IGNORE) st->ignore = 1;
+        else st->force_null = 1;
+        return TF_OK;
+    }
+    if (!acf_is_numeric(in->col_types[st->col_idx])) {
+        if (st->on_type_error == ACF_TYPE_FAIL) return acf_set_col_error(st, in);
+        st->force_null = 1;
+    }
+    return TF_OK;
+}
+
 static int acf_process(tf_step *self, tf_batch *in, tf_batch **out,
                        tf_side_channels *side) {
     (void)side;
     acf_state *st = self->state;
     *out = NULL;
 
-    int ci = tf_batch_col_index(in, st->column);
-    if (ci < 0) return TF_OK;
+    if (acf_check_schema(st, in) != TF_OK) return TF_ERROR;
+    if (st->ignore || st->force_null || st->col_idx < 0) return TF_OK;
 
     for (size_t r = 0; r < in->n_rows; r++) {
-        if (tf_batch_is_null(in, r, ci)) continue;
-        double val = get_numeric(in, r, ci);
+        if (tf_batch_is_null(in, r, (size_t)st->col_idx)) continue;
+        double val = get_numeric(in, r, st->col_idx);
 
         if (st->n_values >= st->cap_values) {
-            size_t newcap = st->cap_values ? st->cap_values * 2 : 256;
-            double *tmp = realloc(st->values, newcap * sizeof(double));
+            size_t min_cap = 0, newcap = 0;
+            if (tf_size_add(st->n_values, 1, &min_cap) != TF_OK ||
+                tf_size_grow_pow2(st->cap_values, min_cap, 256, &newcap) != TF_OK) {
+                return TF_ERROR;
+            }
+            double *tmp = tf_reallocarray_checked(st->values, newcap, sizeof(double));
             if (!tmp) return TF_ERROR;
             st->values = tmp;
             st->cap_values = newcap;
@@ -53,23 +147,48 @@ static int acf_process(tf_step *self, tf_batch *in, tf_batch **out,
     return TF_OK;
 }
 
+static int acf_emit_nulls(acf_state *st, tf_batch **out) {
+    size_t out_rows = 0;
+    if (tf_size_add((size_t)st->lags, 1, &out_rows) != TF_OK) return TF_ERROR;
+    tf_batch *ob = tf_batch_create(2, out_rows);
+    if (!ob) return TF_ERROR;
+    if (tf_batch_set_schema(ob, 0, "lag", TF_TYPE_INT64) != TF_OK ||
+        tf_batch_set_schema(ob, 1, "acf", TF_TYPE_FLOAT64) != TF_OK) {
+        tf_batch_free(ob);
+        return TF_ERROR;
+    }
+    for (int k = 0; k <= st->lags; k++) {
+        if (tf_batch_set_int64(ob, (size_t)k, 0, k) != TF_OK) {
+            tf_batch_free(ob);
+            return TF_ERROR;
+        }
+        if (tf_batch_expose_row(ob, (size_t)k) != TF_OK) {
+            tf_batch_free(ob);
+            return TF_ERROR;
+        }
+    }
+    *out = ob;
+    return TF_OK;
+}
+
 static int acf_flush(tf_step *self, tf_batch **out, tf_side_channels *side) {
     (void)side;
     acf_state *st = self->state;
     *out = NULL;
 
+    if (st->ignore) return TF_OK;
+    if (st->force_null) return acf_emit_nulls(st, out);
+
     size_t n = st->n_values;
     if (n < 2) return TF_OK;
 
     int max_lag = st->lags;
     if ((size_t)max_lag >= n) max_lag = (int)(n - 1);
 
-    /* Compute mean */
     double mean = 0;
     for (size_t i = 0; i < n; i++) mean += st->values[i];
     mean /= (double)n;
 
-    /* Compute variance (denominator for ACF) */
     double var = 0;
     for (size_t i = 0; i < n; i++) {
         double d = st->values[i] - mean;
@@ -77,12 +196,15 @@ static int acf_flush(tf_step *self, tf_batch **out, tf_side_channels *side) {
     }
     if (var == 0) return TF_OK;
 
-    /* Output: lag+1 rows (lag 0 to max_lag) */
-    size_t out_rows = (size_t)(max_lag + 1);
+    size_t out_rows = 0;
+    if (tf_size_add((size_t)max_lag, 1, &out_rows) != TF_OK) return TF_ERROR;
     tf_batch *ob = tf_batch_create(2, out_rows);
     if (!ob) return TF_ERROR;
-    tf_batch_set_schema(ob, 0, "lag", TF_TYPE_INT64);
-    tf_batch_set_schema(ob, 1, "acf", TF_TYPE_FLOAT64);
+    if (tf_batch_set_schema(ob, 0, "lag", TF_TYPE_INT64) != TF_OK ||
+        tf_batch_set_schema(ob, 1, "acf", TF_TYPE_FLOAT64) != TF_OK) {
+        tf_batch_free(ob);
+        return TF_ERROR;
+    }
 
     for (int k = 0; k <= max_lag; k++) {
         double cov = 0;
@@ -91,9 +213,15 @@ static int acf_flush(tf_step *self, tf_batch **out, tf_side_channels *side) {
         }
         double acf_val = cov / var;
 
-        tf_batch_set_int64(ob, k, 0, k);
-        tf_batch_set_float64(ob, k, 1, acf_val);
-        ob->n_rows = k + 1;
+        if (tf_batch_set_int64(ob, (size_t)k, 0, k) != TF_OK ||
+            tf_batch_set_float64(ob, (size_t)k, 1, acf_val) != TF_OK) {
+            tf_batch_free(ob);
+            return TF_ERROR;
+        }
+        if (tf_batch_expose_row(ob, (size_t)k) != TF_OK) {
+            tf_batch_free(ob);
+            return TF_ERROR;
+        }
     }
 
     *out = ob;
@@ -101,7 +229,7 @@ static int acf_flush(tf_step *self, tf_batch **out, tf_side_channels *side) {
 }
 
 static void acf_destroy(tf_step *self) {
-    acf_state *st = self->state;
+    acf_state *st = self ? self->state : NULL;
     if (st) {
         free(st->column);
         free(st->values);
@@ -110,20 +238,42 @@ static void acf_destroy(tf_step *self) {
     free(self);
 }
 
+static int parse_lags(const cJSON *lags_j, int *out) {
+    *out = 20;
+    if (!lags_j) return TF_OK;
+    size_t parsed_lags = 0;
+    if (tf_json_size_value(lags_j, "lags", 1, (size_t)INT_MAX,
+                           &parsed_lags, "acf") < 0) {
+        return TF_ERROR;
+    }
+    *out = (int)parsed_lags;
+    return TF_OK;
+}
+
 tf_step *tf_acf_create(const cJSON *args) {
     if (!args) return NULL;
     cJSON *col_j = cJSON_GetObjectItemCaseSensitive(args, "column");
-    if (!cJSON_IsString(col_j)) return NULL;
+    if (!cJSON_IsString(col_j) || !col_j->valuestring[0]) {
+        tf_set_last_error("acf: column is required");
+        return NULL;
+    }
 
     acf_state *st = calloc(1, sizeof(acf_state));
     if (!st) return NULL;
+    st->col_idx = -1;
     st->column = strdup(col_j->valuestring);
+    if (!st->column) { free(st); return NULL; }
 
     cJSON *lags_j = cJSON_GetObjectItemCaseSensitive(args, "lags");
-    st->lags = cJSON_IsNumber(lags_j) ? lags_j->valueint : 20;
-    if (st->lags < 1) st->lags = 1;
+    if (parse_lags(lags_j, &st->lags) != TF_OK ||
+        acf_parse_missing_policy(args, &st->missing) != TF_OK ||
+        acf_parse_type_policy(args, &st->on_type_error) != TF_OK) {
+        free(st->column);
+        free(st);
+        return NULL;
+    }
 
-    tf_step *step = malloc(sizeof(tf_step));
+    tf_step *step = calloc(1, sizeof(tf_step));
     if (!step) { free(st->column); free(st); return NULL; }
     step->process = acf_process;
     step->flush = acf_flush;