webarchive: updated validation

htdocs/archive/addfile.c

@@ -35,12 +35,14 @@ int main()
     const char *query_string = getenv("QUERY_STRING");
     if (query_string == NULL) {
         generatepage("Internal error: empty/invalid data");
-    } else if (strlen(query_string) > 1024) {
+    } else if (strlen(query_string) > MAX_LINE_LEN) {
-        generatepage("Internal error: data size limit exceeded (1024)");
+        char errmsg[100] = {0};
+        sprintf(errmsg, "Internal error: data size limit exceeded (%i)", MAX_LINE_LEN);
+        generatepage(errmsg);
     } else if (NULL != validate(query_string)) {
         generatepage(validate(query_string));
     } else {
-        char data[4096] = {0};
+        char data[MAX_LINE_LEN] = {0};
         unencode(query_string, data);
 
         if (NULL != validate(data)) {

htdocs/archive/deletefile.c

@@ -25,7 +25,7 @@ int main()
     }
     sortdata(data, MAX_RECORDS);
 
-    char name[32] = {0};
+    char name[MAX_NAME_LEN] = {0};
     strcpy(name, getname(query_string));
     int index = -1;
     for (int i = 0; i < MAX_RECORDS && data[i]; i++) {

htdocs/archive/edit.c

@@ -25,7 +25,7 @@ int main()
         return EXIT_SUCCESS;
     }
 
-    char name[32] = {0};
+    char name[MAX_NAME_LEN] = {0};
     strcpy(name, getname(query_string));
     int index = -1;
     for (int i = 0; i < MAX_RECORDS && data[i]; i++) {

htdocs/archive/report.c

@@ -83,7 +83,7 @@ int main()
     if (query_string == NULL || *query_string == '\0') {
         listAll(data);
     } else if (strncmp(query_string, "name=", 5) == 0 && getname(query_string) != NULL) {
-        char name[32] = {0};
+        char name[MAX_NAME_LEN] = {0};
         strcpy(name, getname(query_string));
         listOne(data,name);
     } else {

htdocs/archive/setfiledata.c

@@ -24,7 +24,7 @@ int main()
         return EXIT_SUCCESS;
     }
 
-    char name[32] = {0};
+    char name[MAX_NAME_LEN] = {0};
     strcpy(name, getname(query_string));
     int index = -1;
     for (int i = 0; i < MAX_RECORDS && data[i]; i++) {
@@ -40,7 +40,7 @@ int main()
     }
 
     // cleanup data...
-    char str[1000] = {0};
+    char str[MAX_LINE_LEN] = {0};
     char *dst = str;
     for (const char *src = query_string; *src; src++) {
         *dst = *src;

htdocs/archive/webarchive.h

@@ -1,7 +1,11 @@
 #include <string.h>
 #include <ctype.h>
+#include <stdio.h>
+#include <stdlib.h>
 
-#define MAX_RECORDS 1000
+#define MAX_RECORDS   1000
+#define MAX_LINE_LEN  0xffff
+#define MAX_NAME_LEN  32
 
 static void unencode(const char *src, char *dest)
 {
@@ -12,6 +16,10 @@ static void unencode(const char *src, char *dest)
             int code;
             if (sscanf(src+1, "%2x", &code) != 1)
                 code = '?';
+            if (code == '%' && isxdigit(src[3]) && isxdigit(src[4])) {
+                src += 2;
+                sscanf(src+1, "%2x", &code);
+            }
             *dest = code;
             src += 2;
         } else
@@ -26,7 +34,7 @@ int readdata(char * * const data, int sz)
     if (!f)
         return 0;  // failed
 
-    char line[10000] = {0};
+    char line[MAX_LINE_LEN] = {0};
     int i = 0;
     while (i < sz && fgets(line,sizeof(line)-2,f)) {
         if (strncmp(line, "name=", 5) == 0) {
@@ -163,6 +171,44 @@ const char *validate_name_version_data(const char *data)
     // filedata
     if (strncmp(data+i, "&data=", 6) != 0)
         return "invalid query string: 'data=' not seen at the expected location";
+    i += 6;
+
+    // validate xml
+    char xml[strlen(data+i)];
+    memset(xml, 0, strlen(data+i));
+    unencode(data+i, xml);
+
+    if (strncmp(xml,"<?xml version=\"1.0\"?>",21)!=0)
+        return "invalid query string: XML must start with '&lt;?xml version=\"1.0\"?&gt;'";
+    int linenr = 1;
+    enum {TEXT,ELEMENT} state = TEXT;
+    for (int pos = 21; xml[pos]; pos++) {
+        if (strncmp(&xml[pos], "\r\n", 2)==0) {
+            ++linenr;
+            ++pos;
+        } else if (xml[pos]=='\r' || xml[pos]=='\n') {
+            ++linenr;
+        } else if (xml[pos] == '<') {
+            if (state != TEXT) {
+                static char errmsg[256];
+                sprintf(errmsg, "invalid query string: Invalid XML at line %i", linenr);
+                return errmsg;
+            }
+            state = ELEMENT;
+        } else if (xml[pos] == '>') {
+            if (state != ELEMENT) {
+                static char errmsg[256];
+                sprintf(errmsg, "invalid query string: Invalid XML at line %i", linenr);
+                return errmsg;
+            }
+            state = TEXT;
+        }
+    }
+    if (state != TEXT) {
+        static char errmsg[256];
+        sprintf(errmsg, "invalid query string: Invalid XML at line %i", linenr);
+        return errmsg;
+    }
 
     return NULL;
 }