Initial commit

.gitignore

@@ -0,0 +1 @@
+/.ccls-cache/

Makefile

@@ -0,0 +1,18 @@
+##
+# massurl
+#
+# @file Makefile
+# @version 0.1
+
+VERSION = 0.1
+CC = gcc
+CFLAGS = -g -w
+SRC = linkedlist.c urlparse.c tree.c massurl.c
+
+all: massurl
+
+massurl: $(SRC)
+	${CC} $(SRC) -o massurl $(CFLAGS)
+
+
+# end

linkedlist.c

@@ -0,0 +1,45 @@
+/*
+ * linkedlist.c
+ *
+ * Created by Yigit Colakoglu on 07/06/2021.
+ * Copyright yigit@yigitcolakoglu.com. 2021. All rights reserved.
+ */
+
+#include "linkedlist.h"
+#include <stdlib.h>
+#include <string.h>
+#include <stdio.h>
+
+
+LinkedList *linkedlistalloc(void){
+    return (LinkedList *) malloc(sizeof(LinkedList));
+}
+
+int linkedlistfind(LinkedList *p, char *str) {
+    int count = 0;
+    while(p != NULL){
+        if(!strcmp(p->data, str))
+            return count;
+        count++;
+        p = p->next;
+    }
+    return -1;
+}
+
+LinkedList *linkedlistadd(LinkedList *p, char *data){
+    if(p == NULL){
+        p = linkedlistalloc();
+        p->next = NULL;
+        p->data = data;
+    }else
+        p->next = linkedlistadd(p->next, data);
+    return p;
+}
+
+void linkedlistprint(LinkedList *p, FILE *out){
+    if(p != NULL){
+        (p->data == NULL) ? fprintf(out, "NULL") : fprintf(out, "%s", p->data);
+        (p->next == NULL) ? : fprintf(out, "%c",'&');
+        linkedlistprint(p->next, out);
+    }
+}

linkedlist.h

@@ -0,0 +1,22 @@
+/*
+ * linkedlist.h
+ *
+ * Created by Yigit Colakoglu on 07/06/2021.
+ * Copyright yigit@yigitcolakoglu.com. 2021. All rights reserved.
+ */
+
+#ifndef linkedlist_h
+#define linkedlist_h
+
+#include <stdio.h>
+
+typedef struct {
+    struct linkedlist *next;
+    char *data;
+} LinkedList;
+
+LinkedList *linkedlistalloc(void);
+int linkedlistfind(LinkedList *p, char *str);
+LinkedList*linkedlistadd(LinkedList *p, char *data);
+void linkedlistprint(LinkedList *p, FILE *out);
+#endif /* Symbol’s value as variable is void: \. */

massurl.c

@@ -0,0 +1,68 @@
+/*
+ * massurl.c
+ *
+ * Created by Yigit Colakoglu on 07/06/2021.
+ * Copyright yigit@yigitcolakoglu.com. 2021. All rights reserved.
+ */
+
+#include <stdio.h>
+#include "urlparse.h"
+#include "tree.h"
+#include "strings.h"
+#define MAXURL 100000
+
+static void usage(void){
+    fputs("\n", stderr);
+}
+
+enum outformat{ PLAIN = 01 };
+
+int main(int argc, char *argv[]) {
+
+    FILE *fin = stdin, *fout = stdout;
+    char *param, urlstr[MAXURL];
+
+    while(--argc > 0){
+        param= *++argv;
+        if(param[0] == '-'){
+            param++;
+            argc--;
+            switch(*param){
+                case 'o':
+                  if ((fout = fopen(*++argv, "w")) == NULL) {
+                    fprintf(stderr, "Can't open output file for writing.\n");
+                    return 1;
+                  }
+                  if(ferror(fout)){
+                    fprintf(stderr, "Can't open output file for writing.\n");
+                    return 1;
+                  }
+                break;
+                default:
+                  fprintf(stderr, "Parameter -%c does not exist!", *param);
+            }
+        }else{
+            if((fin = fopen(param, "r")) == NULL){
+                fprintf(stderr, "Can't open file %s\n", param);
+                return 1;
+            }
+        }
+    }
+
+    TreeNode *urltree = treealloc();
+    URL *url;
+    while(fgets(urlstr, MAXURL, fin) != NULL){
+        if((url = parseurl(urlstr)) == NULL){
+            fprintf(stderr, "Malformed URL %s", urlstr);
+            continue;
+        }
+        if(urltree->path == NULL){
+            urltree->path = url->base;
+            urltree->params = url->params;
+        }else{
+            urltree = addtree(urltree, url);
+        }
+    }
+    printtree(urltree, fout);
+    return 0;
+}

test.data

@@ -0,0 +1,8 @@
+http://test.com?param4=var4
+http://test.com?param3=var3
+http://abc.com?abcpar=123&asdasd=asdas
+http://abc.com?abcpar123=123
+http://test.com/path1?param3=var3
+http://test.com/path1?param1=var2&param2=var2
+http://bc.com
+http://test.com?param1=var2&param2=var2

tree.c

@@ -0,0 +1,51 @@
+/*
+ * tree.c
+ *
+ * Created by Yigit Colakoglu on 07/06/2021.
+ * Copyright yigit@yigitcolakoglu.com. 2021. All rights reserved.
+ */
+
+#include "tree.h"
+#include "linkedlist.h"
+#include "urlparse.h"
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+
+TreeNode *addtree(TreeNode *p, URL *url) {
+  if (p == NULL) {
+    TreeNode *newnode = treealloc();
+    newnode->path = url->base;
+    newnode->params = url->params;
+    newnode->left = newnode->right = NULL;
+    return newnode;
+  }
+  int strdiff = strcmp(url->base, p->path);
+  if (!strdiff) {
+    while(url->params != NULL){
+      if(p->params == NULL || linkedlistfind(p->params, url->params->data) == -1){
+        p->params = linkedlistadd(p->params, url->params->data);
+      }
+      url->params = url->params->next;
+    }
+  } else if (strdiff < 0) {
+    p->left = addtree(p->left, url);
+  } else {
+    p->right = addtree(p->right, url);
+  }
+  return p;
+}
+
+TreeNode *treealloc(void){
+    return (TreeNode *) malloc(sizeof(TreeNode));
+}
+
+void printtree(TreeNode *root, FILE *out){
+    if(root != NULL){
+        printtree(root->left, out);
+        fprintf(out, "%s ", root->path);
+        linkedlistprint(root->params, out);
+        fprintf(out, "%c", '\n');
+        printtree(root->right, out);
+    }
+}

tree.h

@@ -0,0 +1,25 @@
+/*
+ * tree.h
+ *
+ * Created by Yigit Colakoglu on 07/06/2021.
+ * Copyright yigit@yigitcolakoglu.com. 2021. All rights reserved.
+ */
+
+#include "urlparse.h"
+#include <stdio.h>
+
+#ifndef tree_h
+#define tree_h
+
+typedef struct {
+    char *path;
+    LinkedList *params;
+    struct tnode *left;
+    struct tnode *right;
+} TreeNode;
+
+TreeNode *addtree(TreeNode *root, URL *url);
+TreeNode *treealloc(void);
+void printtree(TreeNode *root, FILE *out);
+
+#endif /* Symbol’s value as variable is void: \. */

urlparse.c

@@ -0,0 +1,86 @@
+/*
+ * urlparse.c
+ *
+ * Created by Yigit Colakoglu on 07/06/2021.
+ * Copyright yigit@yigitcolakoglu.com. 2021. All rights reserved.
+ */
+
+#include "urlparse.h"
+#include "linkedlist.h"
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+
+URL *parseurl(char *url) {
+  URL *urlp = urlalloc();
+  urlp->params = NULL;
+  short stage = 0; /* var to keep track of where we are in url */
+  int counter = 0;
+
+  while (*url != '\0' && *url != '\n') {
+    switch (*url++) {
+    case ':':
+      if (stage == 0) {
+        urlp->https = *(url - 2) == 's';
+        if (*(url + 1) == '\0' || *url == '\0' || *url == '\n') /* weird stuff would happen with strings like "http:" */
+          return NULL;
+        url += 2; /* Skip the // after the :*/
+        stage = 1;
+        counter+=4;
+      }
+      break;
+
+    case '?':
+      if (stage == 1) {
+        urlp->base =
+            (char *)malloc(counter); /* +1 for the '\0' in the end */
+        strncpy(urlp->base, url - counter, counter - 1);
+        stage = 2;
+        counter = 1;
+      } else {
+        return NULL;
+      }
+      break;
+
+    case '=':
+      if (stage == 2) {
+        char *foo;
+        foo = (char *)malloc(counter);
+        strncpy(foo, url - counter, counter-1);
+        counter = 1;
+        if (urlp->params == NULL){
+          urlp->params = linkedlistalloc();
+          urlp->params->data = foo;
+        }else
+          urlp->params = linkedlistadd(urlp->params, foo);
+        while(*url != '&' && *url != '\0' && *url != '\n')
+          url++;
+        url++;
+      } else {
+        return NULL;
+      }
+      break;
+
+    default:
+      counter++;
+      break;
+    }
+  }
+
+  switch(stage){
+    case 0:
+      return NULL;
+      break;
+    case 1:
+      urlp->base = (char *)malloc(counter); /* +1 for the '\0' in the end */
+      strncpy(urlp->base, url - (counter-1), counter - 1);
+      break;
+    case 2:
+      break;
+    default:
+      return NULL;
+    }
+  return urlp;
+}
+
+URL *urlalloc(void) { return (URL *)malloc(sizeof(URL)); }

urlparse.h

@@ -0,0 +1,22 @@
+/*
+ * urlparse.h
+ *
+ * Created by Yigit Colakoglu on 07/06/2021.
+ * Copyright yigit@yigitcolakoglu.com. 2021. All rights reserved.
+ */
+
+#include "linkedlist.h"
+
+#ifndef urlparse_h
+#define urlparse_h
+
+typedef struct{
+    unsigned int https : 1;
+    char *base;
+    LinkedList *params;
+} URL;
+
+URL *parseurl(char *urlstr);
+URL *urlalloc(void);
+
+#endif /* Symbol’s value as variable is void: \. */