add: csv_read() improved

Error messages are more specific and now columns are read as arrays of
strings.

2 files changed, 94 insertions, 40 deletions

diff --git a/src/parse.c b/src/parse.c
index 98ac379..2c21b7e 100644
--- a/src/parse.c
+++ b/src/parse.c
@@ -17,6 +17,8 @@ static void csv_readline_values(
         char *line_buffer, size_t line_number,
         char separator);
 
+static void csv_keys2cols(size_t cols[], char *keys[], size_t keys_size);
+
 void json_read(
         FILE *fp,
         Array *input, Array *out,
@@ -79,25 +81,38 @@ json_read_error:
 void csv_read(
         FILE *fp,
         Array *input, Array *out,
-        size_t in_cols[], size_t in_cols_size,
-        size_t out_cols[], size_t out_cols_size,
+        char *in_keys[], size_t n_in_cols,
+        char *out_keys[], size_t n_out_cols,
         bool read_output,
-        char separator
-        )
+        bool has_header, //TODO
+        char separator)
 {
     char line_buffer[1024];
     char *line_ptr;
     double *num_buffer;
     size_t line = 0, num_buffer_length = 1; 
+    size_t *in_cols, *out_cols;
     int ret;
 
+    if (fp == NULL) die("csv_read() Error:");
+
+    in_cols = ecalloc(n_in_cols, sizeof(size_t));
+    csv_keys2cols(in_cols, in_keys, n_in_cols);
+
+    if (read_output) {
+        out_cols = ecalloc(n_out_cols, sizeof(size_t));
+        csv_keys2cols(out_cols, out_keys, n_out_cols);
+    }
+
     input->shape[0] = 1;
-    input->shape[1] = in_cols_size;
+    input->shape[1] = n_in_cols;
     input->data = ecalloc(input->shape[1], sizeof(double));
 
-    out->shape[0] = 1;
-    out->shape[1] = out_cols_size;
-    out->data = ecalloc(input->shape[1], sizeof(double));
+    if (read_output) {
+        out->shape[0] = 1;
+        out->shape[1] = n_out_cols;
+        out->data = ecalloc(input->shape[1], sizeof(double));
+    }
 
     fgets(line_buffer, 1024, fp);
     for (line_ptr = line_buffer; *line_ptr != '\0'; line_ptr++) {
@@ -105,28 +120,30 @@ void csv_read(
             num_buffer_length++;
         }
     }
+
     num_buffer = ecalloc(num_buffer_length, sizeof(double));
-    
+
     csv_readline_values(num_buffer, num_buffer_length, line_buffer, 1, separator);
-    csv_columns_select(input->data + line * input->shape[1], num_buffer, in_cols, in_cols_size, num_buffer_length);
-    csv_columns_select(out->data + line * out->shape[1], num_buffer, out_cols, out_cols_size, num_buffer_length);
+    csv_columns_select(input->data + line * input->shape[1], num_buffer, in_cols, n_in_cols, num_buffer_length);
+    if (read_output) csv_columns_select(out->data + line * out->shape[1], num_buffer, out_cols, n_out_cols, num_buffer_length);
+
     for (line = 1; fgets(line_buffer, 1024, fp) != NULL; line++) {
+        csv_readline_values(num_buffer, num_buffer_length, line_buffer, line+1, separator);
 
         input->shape[0]++;
-        out->shape[0]++;
-
         input->data = erealloc(input->data, input->shape[0] * input->shape[1] * sizeof(double));
-        out->data = erealloc(out->data, out->shape[0] * out->shape[1] * sizeof(double));
+        csv_columns_select(input->data + line * input->shape[1], num_buffer, in_cols, n_in_cols, num_buffer_length);
 
-        csv_readline_values(num_buffer, num_buffer_length, line_buffer, line+1, separator);
-        csv_columns_select(input->data + line * input->shape[1], num_buffer, in_cols, in_cols_size, num_buffer_length);
-        csv_columns_select(out->data + line * out->shape[1], num_buffer, out_cols, out_cols_size, num_buffer_length);
+        if (read_output) {
+            out->shape[0]++;
+            out->data = erealloc(out->data, out->shape[0] * out->shape[1] * sizeof(double));
+            csv_columns_select(out->data + line * out->shape[1], num_buffer, out_cols, n_out_cols, num_buffer_length);
+        }
     }
     free(num_buffer);
+    free(in_cols);
+    free(out_cols);
     return;
-
-csv_read_error:
-    die("csv_read() error on line %zu: ", line + 1);
 }
 
 void csv_columns_select(
@@ -138,7 +155,7 @@ void csv_columns_select(
     for (i = 0; i < cols_size; i++) {
         if (selected_cols[i] >= src_cols_number) {
             die("csv_columns_select() Error: "
-                "selected col %zu is greater than src cols %zu",
+                "selected col '%zu' is greater than src cols '%zu'",
                 selected_cols[i], src_cols_number);
         }
         dst_row[i] = src_row[selected_cols[i]];
@@ -163,39 +180,75 @@ void csv_readline_values(
         if (*(line_ptr + offset) == separator || *(line_ptr + offset) == '\n') {
             offset++;
         } else {
-            die("csv_readline_values() Error on line %zu: format separator must be '%c' not '%c'",
-                line_number, separator, *(line_ptr + 1));
+            die("csv_readline_values() Error: on line %zu format separator must be '%c' not '%c'",
+                line_number, separator, *(line_ptr + offset));
         }
     }
 
-    if (*line_ptr != '\0') {
-        die("csv_readline_values() Error on line %zu: it seems to have more than %zu columns",
+    if (col < num_buffer_length && *line_ptr == '\0') {
+        die("csv_readline_values() Error: line %zu seems to have less than %zu columns",
             line_number, num_buffer_length);
+    } else if (col == num_buffer_length && *line_ptr != '\0') {
+        die("csv_readline_values() Error: line %zu seems to have more than %zu columns",
+            line_number, num_buffer_length);
+    } else if (*line_ptr != '\0') {
+        die("csv_readline_values() Error: "
+            "line %zu format is invalid start checking from column %zu",
+            line_number, col+1);
     }
+}
 
-    if (col < num_buffer_length) {
-        die("csv_readline_values() Error on line %zu: it seems to have less than %zu columns",
-            line_number, num_buffer_length);
+void csv_keys2cols(size_t cols[], char *keys[], size_t keys_size)
+{
+    size_t i;
+    int ret;
+    for (i = 0; i < keys_size; i++) {
+        ret = sscanf(keys[i], "%zu", cols + i);
+        if (ret != 1) die("csv_keys2col() Error: '%s' can not be converted to index", keys[i]);
     }
 }
 
 #ifdef PARSE_TEST
+#include <assert.h>
+#include <string.h>
 // clang -g -DPARSE_TEST -o objs/parse_test src/{utils,parse}.c $(pkg-config --libs-only json-c)
 int main(int argc, char *argv[]) {
-    if (argc != 2) {
-        fprintf(stderr, "usage: parse_test FILENAME\n");
+    FILE *fp;
+    char *filename, separator;
+
+    if (argc < 2 || argc > 3) {
+        fprintf(stderr, "usage: parse_test FILENAME [SEPARATOR]\n");
         return 1;
     }
-    char *filename = argv[1];
-    FILE *fp = fopen(filename, "r");
-    if (fp == NULL) {
-        perror("fopen Error");
-        return 1;
+
+    filename = argv[1];
+    separator = ',';
+    if (argc == 3) {
+        assert(strlen(argv[2]) == 1 && "SEPARATOR must be a character");
+        separator = argv[2][0];
     }
+
+    fp = fopen(filename, "r");
+    if (fp == NULL) die("fopen() Error:");
     Array X, y;
-    size_t in_cols[] = {2, 1};
-    size_t out_cols[] = {0};
-    csv_read(fp, &X, &y, in_cols, 2, out_cols, 1, true, ',');
+    char *in_cols[] = {"1", "2"};
+    char *out_cols[] = {"0"};
+    csv_read(fp, &X, &y, in_cols, 2, out_cols, 1, true, true, separator);
+
+    size_t i, j;
+    for (i = 0; i < X.shape[0]; i++) {
+        for (j = 0; j < X.shape[1]; j++) {
+            printf("%*.2e\t", 4, X.data[i * X.shape[1] + j]);
+        }
+
+        for (j = 0; j < y.shape[1]; j++) {
+            if (j == 0) printf("|\t");
+            printf("%5.2e", y.data[i * y.shape[1] + j]);;
+            if (j < y.shape[1] - 1) printf("\t");
+        }
+        printf("\n");
+
+    }
     return 0;
 }
 #endif
diff --git a/src/parse.h b/src/parse.h
index e50f6bd..05cafa2 100644
--- a/src/parse.h
+++ b/src/parse.h
@@ -20,9 +20,10 @@ void json_read(
 void csv_read(
         FILE *fp,
         Array *input, Array *out,
-        size_t in_cols[], size_t in_cols_size,
-        size_t out_cols[], size_t out_cols_size,
+        char *in_cols[], size_t in_cols_size,
+        char *out_cols[], size_t out_cols_size,
         bool read_output,
+        bool has_header,
         char separator
         );
 #endif