rewrite readdb and writedb to read and write in binary format

3 files changed, 129 insertions, 61 deletions

diff --git a/Makefile b/Makefile
index 6e0e769..2bcd015 100644
--- a/Makefile
+++ b/Makefile
@@ -2,7 +2,7 @@ CC = gcc
 BUILD = target
 C_FLAGS = -Wall -lmdb -std=c11
 D_FLAGS = -L$(BUILD) -Wl,-rpath=$(BUILD) -O0
-L_FLAGS = -c -fPIC -Wall -Werror
+L_FLAGS = -c -fPIC -Wall
 
 all: mdb
 mdb: lib
@@ -14,7 +14,7 @@ lib: $(BUILD)
 
 test: dev_lib
 	$(CC) src/test.c $(D_FLAGS) -g $(C_FLAGS) -o $(BUILD)/test.out
-	valgrind --tool=memcheck -s --log-file="mem_dbg" ./$(BUILD)/test.out
+	valgrind --tool=memcheck --leak-check=full --log-file="mem_dbg" ./$(BUILD)/test.out
 	valgrind --tool=drd -s --log-file="thrd_dbg" ./$(BUILD)/test.out
 	rm -rf $(BUILD)
 dev: dev_lib
diff --git a/src/include/engine/file.c b/src/include/engine/file.c
index 4da8b2d..e89dacf 100644
--- a/src/include/engine/file.c
+++ b/src/include/engine/file.c
@@ -4,95 +4,138 @@
 
 #include "engine.h"
 
-static char *getpair(int *c, FILE *fp);
+static int check(char **known, int len, char *str);
+static long getsz(FILE *fp);
+static char *makepair(char *key, char *value);
 
-// TODO: reimplement this to read file in FI format
+// TODO: reimplement this to read file in FI (Fast Infoset) format
+// TODO: make this smaller
 tablist_t *readdb(char *filename)
 {
-  int len = 2;
-  tablist_t *list = calloc(len, sizeof(tablist_t));
-  list[0].len = len;
   FILE *fp;
-  if (filename == NULL ||
-    (fp = fopen(filename, "rb")) == NULL)
+  int llen = 2, tlen = 2;
+  char **ktags = calloc(tlen, sizeof(char *));
+  tablist_t *list = calloc(llen, sizeof(tablist_t));
+  list[0].len = llen;
+
+  if (filename == NULL || (fp = fopen(filename, "rb")) == NULL)
     return list;
 
-  int c, i = 0, open = 0;
-  char *p;
-  while ((c = fgetc(fp)) != EOF) {
-    switch (c) {
-      case 250:
-        p = getpair(&c, fp);
-        if (p == NULL) goto fail;
-        setkeys(&list, i, &p, 1);
-        free(p);
-        break;
-      case 251:
-        if (open == 1) goto fail;
-        open = 1;
-        break;
-      case 254:
-        open = 0;
-        i++;
-        break;
-      default:
-        goto fail;
+  long sz = getsz(fp);
+  char *buf = calloc(sz, sizeof(char));
+  fread(buf, sizeof(char), sz, fp);
+  char *v, *k;
+  for (int i = 0, j = 0, id = 0; i < sz; ++i) {
+    if (buf[i] == '\n')
+      continue;
+    if (j >= tlen)
+      ktags = realloc(ktags, ++tlen * sizeof(char *));
+    if (buf[i] == '<' || buf[i] == ':') {
+      if (buf[i] == '<') {
+        int t;
+        for (t = i + 1; buf[t] != '>'; ++t) ;
+        ktags[j] = calloc(t - i, sizeof(char));
+        strncpy(ktags[j++], buf + i + 1, t - i - 1);
+        k = ktags[j - 1];
+        i += t - i;
+      } else {
+        int t;
+        for (t = i + 1; buf[t] != '\n'; ++t) ;
+        v = calloc(t - i, sizeof(char));
+        strncpy(v, buf + i + 1, t - i - 1);
+        i += t - i;
+        char *pair = makepair(k, v);
+        setkeys(&list, id, &pair, 1);
+        free(pair);
+        free(v);
+      }
+    } else {
+      int idx = 0;
+      memcpy(&idx, buf + i, sizeof(int));
+      i += sizeof(int) - 1;
+      if (idx == 1)
+        id++;
+      k = ktags[idx];
     }
   }
-  fclose(fp);
-  return list;
 
-fail:
+  for (int i = 0; i < tlen - 1; ++i)
+    free(ktags[i]);
+  free(ktags);
+  free(buf);
   fclose(fp);
-  delkeys(list, -1,  NULL, 0);
-  free(list);
-  return NULL;
+  return list;
 }
 
-// TODO: rewrite this function to write data in FI format
+// TODO: Make this write to a file in the FI (Fast Infoset) format
+// TODO: Make this smaller
 void writedb(char *filename, tablist_t *list)
 {
   FILE *fp = fopen(filename, "wb");
   if (fp == NULL)
     return;
-  for (int i = 0; i < list[0].len; ++i) {
-    fputc(0xFB, fp);
-    tablist_t *indexes = getkeys(list, i, NULL, 0);
-    for (int j = 0; indexes[0].tab[j].flag; ++j) {
-      fprintf(fp, "\xfa%s:", indexes[0].tab[j].key);
-      switch (indexes[0].tab[j].flag) {
+  int len = 2;
+  char **ktags = calloc(len, sizeof(char *));
+
+  fprintf(fp, "<documents>\n");
+  fprintf(fp, "<document>\n");
+  ktags[0] = "documents";
+  ktags[1] = "document";
+  tablist_t *indexes = getkeys(list, -1, NULL, 0);
+  for (int i = 0, j = 2; i < indexes[0].len; ++i) {
+    int n = 1;
+    if (i > 0)
+      fwrite(&n, sizeof(int), 1, fp);
+    for (int k = 0; indexes[i].tab[k].flag; ++k) {
+      if ((n = check(ktags, len, indexes[i].tab[k].key)) >= 0)
+        fwrite(&n, sizeof(int), 1, fp);
+      else {
+        fprintf(fp, "<%s>", indexes[i].tab[k].key);
+        if (j >= len)
+          ktags = realloc(ktags, ++len * sizeof(char *));
+        ktags[j] = calloc(strlen(indexes[i].tab[k].key) + 1, sizeof(char));
+        strcpy(ktags[j++], indexes[i].tab[k].key);
+      }
+      switch (indexes[i].tab[k].flag) {
         case 1:
-          fprintf(fp, "%.2lf\xfc", indexes[0].tab[j].value.num);
+          fprintf(fp, ":%g\n", indexes[i].tab[k].value.num);
           break;
         case 2:
-          fprintf(fp, "%s\xfc", indexes[0].tab[j].value.boolean ?
-                  "true" : "false");
+          fprintf(fp, ":%s\n", indexes[i].tab[k].value.boolean ? "true" : "false");
           break;
         case 3:
-          fprintf(fp, "%s\xfc", indexes[0].tab[j].value.str);
+          fprintf(fp, ":%s\n", indexes[i].tab[k].value.str);
           break;
       }
     }
-    free(indexes);
-    fputc(0xFE, fp);
   }
+  for (int i = 2; i < len; ++i)
+    free(ktags[i]);
+  free(ktags);
+  free(indexes);
   fclose(fp);
 }
 
-static char *getpair(int *c, FILE *fp)
+static int check(char **known, int len, char *str)
 {
-  char *pair = calloc(3, sizeof(char));
-  int i = 0, len = 3;
-  while ((*c = fgetc(fp)) != 252 && *c != EOF) {
-    if (i >= len)
-      pair = realloc(pair, ++len * sizeof(char));
-    pair[i++] = *c;
-  }
-  pair = realloc(pair, ++len * sizeof(char));
-  pair[i] = '\0';
-  if (*c != 252) {
-    free(pair);
-    return NULL;
+  for (int i = 0; i < len; ++i) {
+    if (!strcmp(str, known[i]))
+      return i;
   }
+  return -1;
+}
+
+static long getsz(FILE *fp)
+{
+  fseek(fp, 0, SEEK_END);
+  long sz = ftell(fp);
+  fseek(fp, 0, SEEK_SET);
+  return sz;
+}
+
+static char *makepair(char *key, char *value)
+{
+  char *pair = calloc(strlen(key) + strlen(value) + 2, sizeof(char));
+  sprintf(pair, "%s:%s", key, value);
   return pair;
 }
diff --git a/src/test.c b/src/test.c
index ee4b118..6a00fa6 100644
--- a/src/test.c
+++ b/src/test.c
@@ -3,6 +3,29 @@
 
 #include "include/mdb.h"
 
+void test_writedb(void)
+{
+  tablist_t *list = readdb("dbs/new.db");
+  writedb("dbs/test.db", list);
+  delkeys(list, -1, NULL, 0);
+  free(list);
+}
+
+void test_readdb(void)
+{
+  tablist_t *list = readdb("dbs/new.db");
+  tablist_t *indices = getkeys(list, -1, NULL, 0);
+  for (int i = 0; i < indices[0].len; ++i) {
+    printf("id: %d\n", i);
+    for (int j = 0; indices[i].tab[j].flag; ++j)
+      printf("%s\n", indices[i].tab[j].key);
+  }
+  free(indices);
+  delkeys(list, -1, NULL, 0);
+  free(list);
+}
+
+
 void test_setkeys(void)
 {
   tablist_t *list = readdb("dbs/test.db");
@@ -178,6 +201,8 @@ void test_getkeys(void)
 
 int main(void)
 {
+  test_writedb();
+  test_readdb();
   test_getkeys();
   test_getkeys_multi();
   test_getkeys_multi_fail();