csv reworked, and header support added, no memory leaks found, still needs more testing

src/std/parser/csv.c

@@ -67,6 +67,9 @@ void ARC_ParserCSVData_VectorDestroyVectorFn(void *data){
 }
 
 void ARC_ParserCSVData_GetDataTag(ARC_Vector *dataVector, ARC_ParserTagToken *tagToken, ARC_ParserCSVUserData *userData){
+    //cast the csv data back to its original type
+    ARC_ParserCSVUserData *csvUserData = (ARC_ParserCSVUserData *)userData;
+
     //loop through the tags either recursing to next body or adding data to vector
     for(uint32_t index = 0; index < ARC_Vector_GetSize(tagToken->tagTokens); index++){
         ARC_ParserTagToken *childTagToken = (ARC_ParserTagToken *)ARC_Vector_Get(tagToken->tagTokens, index);
@@ -83,14 +86,19 @@ void ARC_ParserCSVData_GetDataTag(ARC_Vector *dataVector, ARC_ParserTagToken *ta
                 ARC_ParserData_HelperRecurseStringAdd(&stringData, childTagToken);
 
                 //move data string and cleanup
-                void *data;
+                void *data = (void *)stringData;
-                userData->castTypeFn(&data, stringData);
-                ARC_String_Destroy(stringData);
 
                 //get the last row vector
                 rowIndex = ARC_Vector_GetSize(dataVector) - 1;
                 row = (ARC_Vector *)ARC_Vector_Get(dataVector, rowIndex);
 
+                //only call the userData cast type callback if it is not a header
+                if(rowIndex != 0 || csvUserData->header == ARC_False){
+                    data = NULL;
+                    userData->castTypeFn(&data, stringData);
+                    ARC_String_Destroy(stringData);
+                }
+
                 //add the data to the row vector
                 ARC_Vector_Add(row, data);
                 continue;
@@ -112,7 +120,8 @@ void ARC_ParserCSVData_RunLineTag(ARC_Vector *dataVector, ARC_ParserTagToken *ta
     for(uint32_t index = 0; index < ARC_Vector_GetSize(tagToken->tagTokens); index++){
         ARC_ParserTagToken *childTagToken = (ARC_ParserTagToken *)ARC_Vector_Get(tagToken->tagTokens, index);
 
-        ARC_Vector *row;
+        ARC_Vector *row        = NULL;
+        uint32_t    rowIndex   = 0;
 
         switch(childTagToken->id){
             //recuse to run the next line
@@ -127,7 +136,14 @@ void ARC_ParserCSVData_RunLineTag(ARC_Vector *dataVector, ARC_ParserTagToken *ta
 
             //add a new row for each new line
             case ARC_PARSER_CSV_CHAR_NEWLINE:
-                //create a new row
+                //check if current is not empty (no need to create another empty row)
+                rowIndex = ARC_Vector_GetSize(dataVector) - 1;
+                row = (ARC_Vector *)ARC_Vector_Get(dataVector, rowIndex);
+                if(ARC_Vector_GetSize(row) == 0){
+                    continue;
+                }
+
+                //create and add a new row
                 ARC_Vector_Create(&row, NULL, NULL);
                 ARC_Vector_Add(dataVector, (void *)row);
                 continue;
@@ -163,90 +179,68 @@ void ARC_ParserCSVData_CreateFn(void **data, ARC_ParserTagToken *parsedData, voi
 
     //recursively add data from the parsedData to the data vector
     ARC_ParserCSVData_RunLineTag(dataVector, parsedData, csvUserData);
-    if(ARC_Vector_GetSize(dataVector) == 0){
+
+    //get the first line to check if it has any values
+    dataRowVector = ARC_Vector_Get(dataVector, 0);
+    if(ARC_Vector_GetSize(dataRowVector) == 0){
         //TODO: iterate and clear the vector
         ARC_Vector_Destroy(dataVector);
         return;
     }
 
-    uint32_t dataStartIndex = 0;
+    //create the data that will be saved
-    if(csvUserData->header == ARC_True){
+    ARC_ParserCSVData *csvData = (ARC_ParserCSVData *)malloc(sizeof(ARC_ParserCSVData));
-        //TODO: headers
+    csvData->hasHeader = csvUserData->header;
-        dataStartIndex++;
+    csvData->headers   = NULL;
+    csvData->height    = 0;
+    csvData->width     = ARC_Vector_GetSize(dataRowVector);
+    csvData->data      = NULL;
+
+    //create the starting index (to offset the header if it exists)
+    uint32_t heightStartIndex = 0;
+
+    //store the headers if they exist
+    if(csvData->hasHeader == ARC_True){
+        heightStartIndex++;
+
+        //copy the headers
+        csvData->headers = (ARC_String **)malloc(sizeof(ARC_String *) * csvData->width);
+        for(uint32_t headerIndex = 0; headerIndex < csvData->width; headerIndex++){
+            csvData->headers[headerIndex] = (ARC_String *)ARC_Vector_Get(dataRowVector, headerIndex);
         }
-
-    //init the height and width of all found rows, height starts at 1 as the first row is already found
-    uint32_t dataHeight = 1;
-    uint32_t dataWidth  = 0;
-
-    //TODO: fix this for headers
-    //get the first non-empty row of dataVector for its width
-    for(; dataStartIndex < ARC_Vector_GetSize(dataVector); dataStartIndex++){
-        dataRowVector = (ARC_Vector *)ARC_Vector_Get(dataVector, dataStartIndex);
-        dataWidth = ARC_Vector_GetSize(dataRowVector);
-
-        //breakout if a valid row is found
-        if(dataWidth != 0){
-            break;
-        }
-    }
-
-    //fix this for headers
-    //check if a valid row if found
-    if(dataWidth == 0){
-        //TODO: iterate and clear the vector
-        //TODO: error here?
-        *data = NULL;
-        ARC_Vector_Destroy(dataVector);
-        return;
     }
 
     //check that all the rows are the same size
-    for(uint32_t rowIndex = dataStartIndex + 1; rowIndex < ARC_Vector_GetSize(dataVector); rowIndex++){
+    for(uint32_t rowIndex = heightStartIndex; rowIndex < ARC_Vector_GetSize(dataVector); rowIndex++){
         ARC_Vector *currentRowVector = (ARC_Vector *)ARC_Vector_Get(dataVector, rowIndex);
 
-        //skip an empty line
+        //skip an empty line (this should only be the case for the last line)
         uint32_t currentRowVectorSize = ARC_Vector_GetSize(currentRowVector);
         if(currentRowVectorSize == 0){
             continue;
         }
 
         //a row was found so update the height
-        dataHeight++;
+        csvData->height++;
 
         //TODO: probs want to error
         //cleanup and exit if they don't match
-        if(dataWidth != currentRowVectorSize){
+        if(csvData->width != currentRowVectorSize){
             //TODO: iterate and clear the vector
             ARC_Vector_Destroy(dataVector);
             return;
         }
     }
 
-    //create the data that will be saved
-    ARC_ParserCSVData *csvData = (ARC_ParserCSVData *)malloc(sizeof(ARC_ParserCSVData));
-    csvData->hasHeader = csvUserData->header;
-    //TODO: fix this
-    csvData->headers = NULL;
-    csvData->height  = dataHeight;
-    csvData->width   = dataWidth;
-
-    if(csvData->height == 0 || csvData->width == 0){
-        //TODO: error here?
-        free(csvData);
-        *data = NULL;
-        return;
-    }
-
     //init location to copy data to
     csvData->data = (void ***)malloc(sizeof(void **) * csvData->height);
 
     //copy the data
     uint32_t rowIndex = 0;
     for(uint32_t y = 0; y < csvData->height; y++, rowIndex++){
-        ARC_Vector *currentRowVector = (ARC_Vector *)ARC_Vector_Get(dataVector, rowIndex + dataStartIndex);
+        ARC_Vector *currentRowVector = (ARC_Vector *)ARC_Vector_Get(dataVector, rowIndex + heightStartIndex);
 
-        //skip an empty line
+        //skip an empty line (should only be the case for the last line)
         uint32_t currentRowVectorSize = ARC_Vector_GetSize(currentRowVector);
         if(currentRowVectorSize == 0){
             y--;

tests/std/parser/csv.c

@@ -44,3 +44,38 @@ ARC_TEST(Parser_ParserCSV_BasicTest){
 
     ARC_Parser_Destroy(parser);
 }
+
+ARC_TEST(Parser_ParserCSV_BasicHeaderTest){
+    ARC_Parser *parser;
+    ARC_ParserCSV_CreateAsParser(&parser, ARC_True, TEST_ParserCSV_CastTypeFn, TEST_ParserCSV_DestroyTypeFn);
+
+    const char *tempCString =
+        "a,b,c,d\n"
+        "4,3,2,1\n"
+        "7,3,2,1\n"
+        "4,2,4,1\n"
+        "7,7,7,7\n";
+
+    ARC_String *tempString;
+    ARC_String_CreateWithStrlen(&tempString, (char *)tempCString);
+
+    //this destroys string, so no need for cleanup
+    ARC_Parser_Parse(parser, &tempString);
+    ARC_CHECK(arc_errno == 0);
+
+    ARC_ParserCSVData *data = (ARC_ParserCSVData *)ARC_Parser_GetData(parser);
+
+    for(uint32_t x = 0; x < data->width; x++){
+        printf("%s ", data->headers[x]->data);
+    }
+    printf("\n");
+
+    for(uint32_t y = 0; y < data->height; y++){
+        for(uint32_t x = 0; x < data->width; x++){
+            printf("%d ", *(int32_t *)(data->data[y][x]));
+        }
+        printf("\n");
+    }
+
+    ARC_Parser_Destroy(parser);
+}