small bug fixed & tab-seperator support

src/lib/csv/csv.c

@@ -69,7 +69,7 @@ csv_parse(struct csv *csv, const char *s, const char *end)
         // in some cases buffer is not used
         switch(state) {
         case 0: //leading spaces
-            if(p == end || !(*p == ' ' || *p == '\t')) {
+            if(p == end || *p != ' ') {
                 state = 1;
                 field_begin = p;
             }
@@ -77,7 +77,7 @@ csv_parse(struct csv *csv, const char *s, const char *end)
         case 1: //common case without buffer
             if(isendl || *p == csv->csv_delim) {
                 state = 0;
-                for(field_end = p - 1; field_end > field_begin && isspace(*field_end); field_end--);
+                for(field_end = p - 1; field_end > field_begin && *field_end == ' '; field_end--);
                 field_end++;
                 csv->emit_field(csv->emit_ctx, field_begin, field_end);
             } else if(*p == csv->csv_quote) {
@@ -106,7 +106,7 @@ csv_parse(struct csv *csv, const char *s, const char *end)
                     state = 4;
                 } else {
                    field_end = p;
-                   for(p++;p < end && isspace(*p) && *p != '\n' && *p !='\r'; p++);
+                   for(p++;p < end && *p == ' '; p++);
                    if(*p == csv->csv_delim || *p == '\n' || *p == '\r') {
                        state = 0;
                        isendl = (*p == '\n' || *p == '\r');
@@ -126,7 +126,7 @@ csv_parse(struct csv *csv, const char *s, const char *end)
             }
             if(isendl || *p == csv->csv_delim) {
                 state = 0;
-                for(bufp--; bufp > csv->buf && isspace(*bufp); bufp--);
+                for(bufp--; bufp > csv->buf && *bufp == ' '; bufp--);
                 bufp++;
                 csv->emit_field(csv->emit_ctx, csv->buf, bufp);
             } else if(*p == csv->csv_quote) {
@@ -164,22 +164,6 @@ csv_parse(struct csv *csv, const char *s, const char *end)
             }
             break;
         }
-        /*if(isendl) {
-            bufp = csv->buf;
-            if(p == end)
-                csv->emit_row(csv->emit_ctx);
-            while(p != end && (*p == '\n' || *p == '\r')) {
-                if(p != end && *p != *(p + 1) && (*(p + 1) == '\n' || *(p + 1) == '\r')) {
-                    p++; //\r\n - is only 1 endl
-                    csv->emit_row(csv->emit_ctx);
-                } else {
-                    csv->emit_row(csv->emit_ctx);
-                }
-                p++;
-            }
-            if(p != end)
-                p--;
-        }*/
         if(isendl) {
             assert(state == 0);
             bufp = csv->buf;
@@ -189,6 +173,11 @@ csv_parse(struct csv *csv, const char *s, const char *end)
         }
         p++;
     } while(p != end + 1);
+
+    if(csv->buf) {
+        free(csv->buf);
+        csv->buf = 0;
+    }
 }
 
 

test/unit/csv.c

@@ -58,6 +58,18 @@ void test4() {
                       "1, 12  34, 56, \"quote , \", 66\nok");
     footer();
 }
+void test5() {
+    header();
+    const char * const s = "abc\tlonglonglonglonglonglonglonglonglonglonglonglonglonglonglonglonglonglong\t0\n"
+        "123\t456\t\n" "0\t\t\n";
+    struct csv csv;
+    csv_create(&csv);
+    csv.emit_field = print_field;
+    csv.emit_row = print_endl;
+    csv.csv_delim = '\t';
+    csv_parse(&csv, s, s + strlen(s));
+    footer();
+}
 
 void test_chunk(const char* const s)
 {
@@ -148,6 +160,7 @@ int main()
     test2();
     test3();
     test4();
+    test5();
     test_chunk("123 , 5  ,       92    , 0, "
                " 0\n1, 12  34, 56, \"quote , \", 66\nok");
     big_chunk_separated_test();

test/unit/csv.result

@@ -18,6 +18,11 @@
 |1|	|12  34|	|56|	|quote , |	|66|	
 |ok|	
 	*** test4: done ***
+ 	*** test5 ***
+|abc|	|longlonglonglonglonglonglonglonglonglonglonglonglonglonglonglonglonglong|	|0|	
+|123|	|456|	||	
+|0|	||	||	
+	*** test5: done ***
  	*** test_chunk ***
 tail: ok
 	*** test_chunk: done ***