From 7f2f7f7c3e5d4c626d7e8d4d45fee2b11b6d5221 Mon Sep 17 00:00:00 2001
From: Matt Strapp <strap012@umn.edu>
Date: Thu, 24 Sep 2020 07:24:19 -0500
Subject: rearrange

---
 P1/src/mapper.c | 134 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 134 insertions(+)
 create mode 100644 P1/src/mapper.c

(limited to 'P1/src/mapper.c')

diff --git a/P1/src/mapper.c b/P1/src/mapper.c
new file mode 100644
index 0000000..66ac2ef
--- /dev/null
+++ b/P1/src/mapper.c
@@ -0,0 +1,134 @@
+#include "mapper.h"
+
+// combined value list corresponding to a word <1,1,1,1....>
+valueList *createNewValueListNode(char *value){
+	valueList *newNode = (valueList *)malloc (sizeof(valueList));
+	strcpy(newNode -> value, value);
+	newNode -> next = NULL;
+	return newNode;
+}
+
+// insert new count to value list
+valueList *insertNewValueToList(valueList *root, char *count){
+	valueList *tempNode = root;
+	if(root == NULL)
+		return createNewValueListNode(count);
+	while(tempNode -> next != NULL)
+		tempNode = tempNode -> next;
+	tempNode -> next = createNewValueListNode(count);
+	return root;
+}
+
+// free value list
+void freeValueList(valueList *root) {
+	if(root == NULL) return;
+
+	valueList *tempNode = root -> next;;
+	while (tempNode != NULL){
+		free(root);
+		root = tempNode;
+		tempNode = tempNode -> next;
+	}
+}
+
+// create <word, value list>
+intermediateDS *createNewInterDSNode(char *word, char *count){
+	intermediateDS *newNode = (intermediateDS *)malloc (sizeof(intermediateDS));
+	strcpy(newNode -> key, word);
+	newNode -> value = NULL;
+	newNode -> value = insertNewValueToList(newNode -> value, count);
+	newNode -> next = NULL;
+	return newNode;
+}
+
+// insert or update a <word, value> to intermediate DS
+intermediateDS *insertPairToInterDS(intermediateDS *root, char *word, char *count){
+	intermediateDS *tempNode = root;
+	if(root == NULL)
+		return createNewInterDSNode(word, count);
+	while(tempNode -> next != NULL) {
+		if(strcmp(tempNode -> key, word) == 0){
+			tempNode -> value = insertNewValueToList(tempNode -> value, count);
+			return root;
+		}
+		tempNode = tempNode -> next;
+		
+	}
+	if(strcmp(tempNode -> key, word) == 0){
+		tempNode -> value = insertNewValueToList(tempNode -> value, count);
+	} else {
+		tempNode -> next = createNewInterDSNode(word, count);
+	}
+	return root;
+}
+
+// free the DS after usage. Call this once you are done with the writing of DS into file
+void freeInterDS(intermediateDS *root) {
+	if(root == NULL) return;
+
+	intermediateDS *tempNode = root -> next;;
+	while (tempNode != NULL){
+		freeValueList(root -> value);
+		free(root);
+		root = tempNode;
+		tempNode = tempNode -> next;
+	}
+}
+
+// emit the <key, value> into intermediate DS 
+void emit(char *key, char *value) {
+
+}
+
+// map function
+void map(char *chunkData){
+	
+	// you can use getWord to retrieve words from the 
+	// chunkData one by one. Example usage in utils.h
+}
+
+// write intermediate data to separate word.txt files
+// Each file will have only one line : word 1 1 1 1 1 ...
+void writeIntermediateDS() {
+	
+}
+
+int main(int argc, char *argv[]) {
+	
+	if (argc < 2) {
+		printf("Less number of arguments.\n");
+		printf("./mapper mapperID\n");
+		exit(0);
+	}
+	// ###### DO NOT REMOVE ######
+	mapperID = strtol(argv[1], NULL, 10);
+
+	// ###### DO NOT REMOVE ######
+	// create folder specifically for this mapper in output/MapOut
+	// mapOutDir has the path to the folder where the outputs of 
+	// this mapper should be stored
+	mapOutDir = createMapDir(mapperID);
+
+	// ###### DO NOT REMOVE ######
+	while(1) {
+		// create an array of chunkSize=1024B and intialize all 
+		// elements with '\0'
+		char chunkData[chunkSize + 1]; // +1 for '\0'
+		memset(chunkData, '\0', chunkSize + 1);
+
+		char *retChunk = getChunkData(mapperID);
+		if(retChunk == NULL) {
+			break;
+		}
+
+		strcpy(chunkData, retChunk);
+		free(retChunk);
+
+		map(chunkData);
+	}
+
+	// ###### DO NOT REMOVE ######
+	writeIntermediateDS();
+
+	return 0;
+}
\ No newline at end of file
-- 
cgit v1.2.3