initial commit

1 files changed, 69 insertions, 0 deletions

diff --git a/comp2521/tf_idf/io.c b/comp2521/tf_idf/io.c
new file mode 100644
index 0000000..566b1b9
--- /dev/null
+++ b/comp2521/tf_idf/io.c
@@ -0,0 +1,69 @@
+#include "io.h"
+
+// Returns a malloc'ed char* of the next word in the file. NULL otherwise.
+// Undefined behaviour if the word is greater than 100 chars.
+// We will assume that the specification is referring to length as
+// LENGTH + \0, so 101 bytes should be allocated.
+char *getNextWord(FILE *const fptr) {
+    // We can assume a maximum word length of 100.
+    char *ret = malloc(sizeof(char) * 101);
+    // Read until there is a non-space.
+    for (int c = fgetc(fptr); c != EOF; c = fgetc(fptr)) {
+        if (isspace(c)) {
+            continue;
+        }
+        // Put the char back into the stream so that we may read it correctly
+        // in the next loop.
+        ungetc(c, fptr);
+        break;
+    }
+    int str_pos = 0;
+    // Read the word and write it into ret, until there is another space char.
+    for (int c = fgetc(fptr); c != EOF; c = fgetc(fptr)) {
+        if (isspace(c)) {
+            break;
+        }
+        ret[str_pos++] = (char)c;
+    }
+    ret[str_pos] = '\0';
+    // Only return a value if ret has been written to.
+    if (!str_pos) {
+        free(ret);
+        return NULL;
+    }
+    // Requires +2 as str_pos is (size -1) AND null terminator.
+    return realloc(ret, (size_t)str_pos + 1);
+}
+
+// Returns true if the string in the second argument exists past the fptr in a
+// file. Preserves original position of the fptr.
+bool exists_after(FILE *const fptr, char *const string) {
+    bool exists = false;
+    // Save the position and restore it when the function ends.
+    long pos = ftell(fptr);
+    char *word = NULL;
+    while ((word = getNextWord(fptr))) {
+        word = normaliseWord(word);
+        if (!strcmp(word, string)) {
+            exists = true;
+            break;
+        }
+    }
+    rewind(fptr);
+    fseek(fptr, pos, 0);
+    return exists;
+}
+
+// Outputs the contents of an inverted index to the file specified by fptr.
+void writeInvertedIndex(FILE *const fptr, struct InvertedIndexNode *node) {
+    if (node == NULL) {
+        return;
+    }
+    writeInvertedIndex(fptr, node->left);
+    fprintf(fptr, "%s ", node->word);
+    for (struct FileListNode *i = node->fileList; i != NULL; i = i->next) {
+        fprintf(fptr, "%s (%lf) ", i->filename, i->tf);
+    }
+    fprintf(fptr, "\n");
+    writeInvertedIndex(fptr, node->right);
+}