From 02c7492f744167fddf7f036404d3a0b27455c8bc Mon Sep 17 00:00:00 2001
From: natjms <natjms@njms.ca>
Date: Fri, 20 Jan 2023 16:11:19 -0800
Subject: [PATCH] Expand the hash table implementation

Our implementation now resizes automatically and supports hashing by
open addressing.
---
 calathea.c | 89 +++++++++++++++++++++++++++++++++++++-----------------
 1 file changed, 62 insertions(+), 27 deletions(-)

diff --git a/calathea.c b/calathea.c
index 8cb05d7..ccd1a7d 100644
--- a/calathea.c
+++ b/calathea.c
@@ -50,6 +50,12 @@ char * read_file(char *filename) {
 }
 
 /*** Hash map implementation ***/
+struct PageMap {
+  struct Page **pages;
+  int capacity;
+  int size;
+};
+
 int helper_hash_polynomial(
     char string[],
     int i,
@@ -65,33 +71,69 @@ int helper_hash_polynomial(
   }
 }
 
-int hash_polynomial(int mapSize, char key[]) {
-  return helper_hash_polynomial(key, 0, strlen(key), mapSize, 0);
+int hash_polynomial(int capacity, char key[]) {
+  return helper_hash_polynomial(key, 0, strlen(key), capacity, 0);
 }
 
-void to_lower_case(char dest[], char str[]) {
+char * to_lower_case(char str[]) {
+  char * lower = malloc((strlen(str) + 1) * sizeof(char));
   int i = 0;
   for (; str[i] != '\0'; i++) {
-    dest[i] = tolower(str[i]);
+    lower[i] = tolower(str[i]);
   }
-  dest[i] = 0;
+  lower[i] = 0;
+
+  return lower;
 }
 
-void map_put(struct Page **map, int mapSize, char title[], struct Page *page) {
-  char lowercased[80];
-  to_lower_case(lowercased, title);
-
-  int index = hash_polynomial(mapSize, lowercased);
-  map[index] = page;
+struct PageMap * map(int capacity) {
+  struct PageMap *map = malloc(sizeof(struct PageMap));
+  map->pages = calloc(capacity, sizeof(struct Page *));
+  map->capacity = capacity;
+  map->size = 0;
+  return map;
 }
 
-struct Page * map_get(struct Page **map, int mapSize, char title[]) {
-  char lowercased[80];
-  to_lower_case(lowercased, title);
+void map_free(struct PageMap *map) {
+  free(map->pages);
+  free(map);
+}
 
-  int index = hash_polynomial(mapSize, lowercased);
-  struct Page *page = map[index];
+void map_put(struct PageMap *map, char title[], struct Page *page) {
+  char *lowercased = to_lower_case(title);
 
+  int index = hash_polynomial(map->capacity, lowercased);
+  while (map->pages[index] != NULL) index++;
+  map->pages[index] = page;
+  map->size++;
+
+  // Resize the map if we're running low on space
+  if (map->size/map->capacity > 0.75) {
+    map->pages = realloc(map->pages, map->capacity * 2);
+    for (int i = map->capacity; i < map->capacity * 2; i++) {
+      map->pages[i] = 0;
+    }
+    map->capacity *= 2;
+  }
+
+  free(lowercased);
+}
+
+struct Page * map_get(struct PageMap *map, char title[]) {
+  char *lowercased = to_lower_case(title);
+
+  int index = hash_polynomial(map->capacity, lowercased);
+  struct Page *page = map->pages[index];
+
+  char *lowercasedFoundTitle = to_lower_case(page->title);
+  while (strcmp(lowercasedFoundTitle, lowercased) != 0) {
+    page = map->pages[++index];
+    free(lowercasedFoundTitle);
+    lowercasedFoundTitle = to_lower_case(page->title);
+  }
+
+  free(lowercased);
+  free(lowercasedFoundTitle);
   return page;
 }
 
@@ -176,7 +218,6 @@ char * substitute_string(char dest[], char sub[], char *start, char *end) {
 
 int main(int argc, char *argv[]) {
   char pagesLocation[256] = "./pages";
-  int mapSize = 1000;
   int initialInboundCapacity = 2;
   char templateFileName[256] = "./template.html";
   char outputDirectoryName[256] = "./build";
@@ -189,11 +230,6 @@ int main(int argc, char *argv[]) {
         i++;
         strcpy(pagesLocation, argv[i]);
       }
-    } else if (strcmp(argv[i], "--table-size") == 0) {
-      if (i + 1 < argc) {
-        i++;
-        mapSize = atoi(argv[i]);
-      }
     } else if (strcmp(argv[i], "--template") == 0) {
       if (i + 1 < argc) {
         i++;
@@ -230,8 +266,7 @@ int main(int argc, char *argv[]) {
     return 1;
   }
 
-  struct Page ** pageMap = malloc(mapSize * sizeof(struct Page *));
-  memset(pageMap, 0, mapSize * sizeof(struct Page *));
+  struct PageMap *pageMap = map(100);
 
   // Contains some information about the current file picked from pagesDir
   struct dirent *fileEntry = readdir(pagesDir);
@@ -310,7 +345,7 @@ int main(int argc, char *argv[]) {
     currentPage->title[min(titleLength, 80)] = 0;
 
     // Insert it into the hash map for lookup later
-    map_put(pageMap, mapSize, currentPage->title, currentPage);
+    map_put(pageMap, currentPage->title, currentPage);
 
     // Get ready to process the next page
     fileEntry = readdir(pagesDir);
@@ -360,14 +395,13 @@ int main(int argc, char *argv[]) {
         strncpy(linkTitle, nextLinkStart + 2, (nextVerticalBar - 1) - (nextLinkStart + 1));
         linkPageTitle = calloc((nextLinkEnd - 2) - nextVerticalBar, sizeof(char));
         strncpy(linkPageTitle, nextVerticalBar + 1, (nextLinkEnd - 1) - nextVerticalBar);
-        puts(linkPageTitle);
       } else {
         // The link is of the form [[page title]]
         linkPageTitle = linkTitle;
         strncpy(linkTitle, nextLinkStart + 2, (linkLength - 2)*sizeof(char));
       }
 
-      struct Page *linkedPage = map_get(pageMap, mapSize, linkPageTitle);
+      struct Page *linkedPage = map_get(pageMap, linkPageTitle);
 
       char *compiledLink = NULL;
 
@@ -499,6 +533,7 @@ int main(int argc, char *argv[]) {
   }
   closedir(pagesDir);
   free(templateContent);
+  map_free(pageMap);
 
   // Deallocate all the pages