How to implement multi keyword search in C? - c

Question

I want to implement a case-insensitive text search which supports parallel testing of multiple keywords. I was already able to achieve this in a way which to me does not seem to be efficient in terms of performance.
The function "strcasestr" (Link to Linux man page) seems to be doing a good job when searching for one keyword, but when you want to simultaneously test multiple keywords - in my understanding - you want to iterate the characters of the text (Haystack) only one single time to find an occurrence of the keywords (Needles).
Using "strcasestr" multiple times would cause - how I understand it - multiple iterations over the text (Haystack), which might not be the fastest solution. An example:
#define _GNU_SOURCE
#include <stdio.h>
#include <string.h>
int main (void) {
// Text to search in
char *str = "This is a test!";
char *result = strcasestr(str, "not_found1");
if (result == NULL) {
result = strcasestr(str, "NOT_FOUND2");
}
if (result == NULL) {
result = strcasestr(str, "TEST!");
}
printf("Result pointer: %s\n", result );
return 0;
}
Is there a way to get the position of the first occurrence of one of the (case-insensitive) keywords in the text in a faster way than I did it?
I would appreciate it if the solution would be extensible so that I could continue looping over the text to find all positions of the occurrences of the keywords, because I am working on a full-text search with a result rating system. Frameworks and small hints to put me in the right direction are also very welcome.

Answer 1

After a long time of learning and testing I found a solution which is working well for me. I tested a one-keyword version of it and the performance was comparable to the function "strcasestr" (Tested with ca. 500 MB of text).
To explain what the below code does:
First the text (Haystack) and the keywords (Needles) are defined. Then the keywords are already converted into lowercase for good performance. iter is an Array of numbers which reflect how many characters the current text progress is in match with each keyword. The program linearly iterates over each character of text until it finds a match in one of the keywords - in this case, the program ends and the result is "True". If it does not find a match (=0), the result if "False".
I welcome tips in the comments for better code quality or higher performance.
#include <stdio.h>
#include <string.h>
#include <ctype.h>
int main (void) {
int i, j;
int match = 0;
// Haystack
char *text = "This is a test!";
// Needles
int keywords_len = 3;
char keywords[][12] = {
"not_found1",
"NOT_FOUND2",
"TEST!"
};
// Make needles lowercase
for (i = 0; i < keywords_len; i++)
for (j = 0; keywords[i][j]; j++)
keywords[i][j] = tolower(keywords[i][j]);
// Define counters for keywords matches
int iter[] = { 0, 0, 0 };
// Loop over all characters and test match
char ptext;
while (ptext = *text++)
// Compare matches
// NOTE: (x | 32) means case-insensitive
if (!match)
for (i = 0; i < keywords_len; i++)
if ((ptext | 32) == keywords[i][iter[i]]) {
if (keywords[i][++(iter[i])] == '\0') {
match = 1;
break;
}
} else
iter[i] = 0;
else
break;
printf("Result: %s\n", match ? "True" : "False");
return 0;
}