Use log format parsing to check if standard input matches the log format

Author: dsamarin

Makefile

@@ -2,7 +2,7 @@ CC = gcc
 CFLAGS = -Wall -Werror -std=c99 -pedantic -Os
 LDFLAGS =
 
-SOURCES = log_main.c log_format.c
+SOURCES = log_main.c log_format.c log_read.c
 OBJECTS = $(SOURCES:.c=.o)
 EXE = logtastic
 

log_format.c

@@ -90,7 +90,7 @@ static struct format_token * parse_format_token_variable (char **input) {
 	return token;
 }
 
-struct format_token * parse_format_token (char **input) {
+static struct format_token * parse_format_token (char **input) {
 	struct format_token *token = NULL;
 
 	if (input && *input) {
@@ -107,3 +107,22 @@ struct format_token * parse_format_token (char **input) {
 
 	return token;
 }
+
+
+struct format_token * parse_log_format (char *format) {
+	struct format_token *head = NULL;
+	struct format_token *tail = NULL;
+	struct format_token *item = NULL;
+
+	while ((item = parse_format_token (&format))) {
+		if (!head) {
+			head = item;
+		} else {
+			tail->next = item;
+		}
+		tail = item;
+	}
+
+	return head;
+}
+

log_format.h

@@ -27,6 +27,6 @@ struct format_token {
 	struct format_token *next;
 };
 
-struct format_token * parse_format_token (char **input);
+struct format_token * parse_log_format (char *format);
 
 #endif

log_main.c

@@ -1,94 +1,68 @@
+#include <glib.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 
+#include "log_read.h"
 #include "log_format.h"
 
-/* 
-68.104.1.129 - - [16/Aug/2012:12:30:02 -0700] "GET /testing/finder/FileModel.php?action=read&path=%2Ffiles%2Fsource%2Fcoreutils-7.4%2Fsrc&start=128 HTTP/1.1" 200 99 "http://eboyjr.oftn.org:8080/files/source/coreutils-7.4/src/?pp=tail.c" "Mozilla/5.0 (X11; Linux i686) AppleWebKit/537.1 (KHTML, like Gecko) Chrome/21.0.1180.77 Safari/537.1"
-*/
+static void check_log_message (char *buffer, void *data) {
+	struct format_token *format = data;
+	_Bool matches = 1;
 
-#define BUFFER_SIZE (4096)
+	char *line = buffer;
 
-struct format_token * parse_log_format (char *format) {
-	struct format_token *head = NULL;
-	struct format_token *tail = NULL;
-	struct format_token *item = NULL;
+	if (buffer && format) {
+		while (format && *line) {
+			if (format->type == T_VARIABLE) {
+				char ending = '\0';
+				if (format->next->type != T_CHARACTER) {
+					matches = 0;
+					fprintf (stderr, "Can't have consecutive variables!\n");
+					break;
+				}
+				ending = format->next->data.ch;
 
-	while ((item = parse_format_token (&format))) {
-		if (!head) {
-			head = item;
-		} else {
-			tail->next = item;
+				char *begin = line;
+				size_t length = 0;
+				while (*line != ending && *line != '\0') {
+					length++;
+					line++;
+				}
+				
+				char *copy = g_strndup (begin, length);
+				printf ("\033[33m%d -> %s\033[0m\n", format->data.var, copy);
+				g_free (copy);
+			} else {
+				if (*line != format->data.ch) {
+					matches = 0;
+					break;
+				}
+				line++;
+			}
+			format = format->next;
 		}
-		tail = item;
-	}
 
-	item = head;
-	while (item) {
-		switch (item->type) {
-		case T_CHARACTER:
-			putchar (item->data.ch);
-			break;
-		case T_VARIABLE:
-			printf ("\x1b\x5b\x33\x35\x6d${%d}\x1b\x5b\x33\x39\x6d", item->data.var);
-			break;
+		if (format) {
+			matches = 0;
 		}
-		item = item->next;
+
+		printf ("Line[%s]: %s\n\n",
+			matches ? "\033[32m OK \033[39m" : "\033[31mFAIL\033[39m",
+			buffer);
 	}
-	putchar ('\n');
 
-	return head;
+	free (buffer);
 }
 
-void readlines (FILE *input, void (*emit)(char *)) {
-	int ch;
-
-	char *buffer = NULL;
-	size_t cursor = 0;
-	size_t length = 0;
-
-	setvbuf (input, NULL, _IOLBF, BUFFER_SIZE);
-
-	while ((ch = fgetc (input)) != EOF) {
-
-		if (!buffer) {
-			buffer = malloc (BUFFER_SIZE);
-			if (!buffer) {
-				return;
-			}
-
-			cursor = 0;
-			length = BUFFER_SIZE;
-			memset (buffer, '\0', length);
-		}
-
-		if (ch == '\n') {
-			(*emit) (buffer);
-			buffer = NULL;
-		} else {
-			buffer[cursor++] = ch;
-
-			if (cursor >= length) {
-				char *tmp = NULL;
-				size_t increase = length;
-
-				length = length + increase;
+int main(void) {
+	struct format_token *item = NULL;
 
-				tmp = realloc (buffer, length);
-				if (!tmp) {
-					free (buffer);
-					return;
-				}
+	char *format = "$remote_addr - $remote_user [$time_local] \"$request\" "
+		"$status $body_bytes_sent \"$http_referer\" \"$http_user_agent\"";
 
-				buffer = tmp;
-				memset (buffer + increase, '\0', increase);
-			}
-		}
-	}
-}
+	item = parse_log_format (format);
+	log_read_lines (stdin, check_log_message, (void *) item);
 
-int main(void) {
-	parse_log_format ("$remote_addr - $remote_user [$time_local] \"$request\" $status $body_bytes_sent \"$http_referer\" \"$http_user_agent\"");
 	return 0;
 }

log_read.c

@@ -0,0 +1,59 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "log_read.h"
+
+
+void log_read_lines (FILE *input, void (*emit)(char *, void *), void *data) {
+	int ch;
+
+	char *buffer = NULL;
+	size_t cursor = 0;
+	size_t length = 0;
+
+#define BUFFER_SIZE (4096)
+
+	setvbuf (input, NULL, _IOLBF, BUFFER_SIZE);
+
+	while ((ch = fgetc (input)) != EOF) {
+
+		if (!buffer) {
+			buffer = malloc (BUFFER_SIZE);
+			if (!buffer) {
+				return;
+			}
+
+			cursor = 0;
+			length = BUFFER_SIZE;
+			memset (buffer, '\0', length);
+		}
+
+		if (ch == '\n') {
+			(*emit) (buffer, data);
+			buffer = NULL;
+		} else {
+			buffer[cursor++] = ch;
+
+			if (cursor >= length) {
+				char *tmp = NULL;
+				size_t increase = length;
+
+				length = length + increase;
+
+				tmp = realloc (buffer, length);
+				if (!tmp) {
+					free (buffer);
+					return;
+				}
+
+				buffer = tmp;
+				memset (buffer + increase, '\0', increase);
+			}
+		}
+	}
+
+#undef BUFFER_SIZE
+
+}
+

log_read.h

@@ -0,0 +1,6 @@
+#ifndef LOG_READ_H
+#define LOG_READ_H
+
+void log_read_lines (FILE *input, void (*emit)(char *, void *), void *data);
+
+#endif