From: Tom Peters <thopeter@cisco.com>
Date: Tue, 1 Jul 2014 17:37:57 +0000 (-0400)
Subject: URI table refactoring
X-Git-Tag: 3.0.0-233~1449
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=f8fb7284ae6dc46b684d9f4e25f77e4fe68b9db1;p=thirdparty%2Fsnort3.git

URI table refactoring
---

diff --git a/src/service_inspectors/nhttp_inspect/nhttp_enum.h b/src/service_inspectors/nhttp_inspect/nhttp_enum.h
index cd5be5189..8da00127f 100644
--- a/src/service_inspectors/nhttp_inspect/nhttp_enum.h
+++ b/src/service_inspectors/nhttp_inspect/nhttp_enum.h
@@ -87,7 +87,7 @@ typedef enum {
 typedef enum { NORM_NULL, NORM_FIELD, NORM_INT64, NORM_ENUM64, NORM_ENUM64LIST } NormFormat;
 
 // Types of character for URI scanning
-typedef enum { CHAR_NORMAL=2, CHAR_PERCENT, CHAR_SLASH, CHAR_BACKSLASH, CHAR_PERIOD, CHAR_INVALID, CHAR_EIGHTBIT } CharAction;
+typedef enum { CHAR_NORMAL=2, CHAR_PERCENT, CHAR_PATH, CHAR_INVALID, CHAR_EIGHTBIT } CharAction;
 
 // Transfer codings
 typedef enum { TRANSCODE__OTHER=1, TRANSCODE_CHUNKED, TRANSCODE_IDENTITY, TRANSCODE_GZIP, TRANSCODE_COMPRESS, TRANSCODE_DEFLATE } Transcoding;
diff --git a/src/service_inspectors/nhttp_inspect/nhttp_uri_norm.cc b/src/service_inspectors/nhttp_inspect/nhttp_uri_norm.cc
index d1dd9da81..3e46622bc 100644
--- a/src/service_inspectors/nhttp_inspect/nhttp_uri_norm.cc
+++ b/src/service_inspectors/nhttp_inspect/nhttp_uri_norm.cc
@@ -80,7 +80,7 @@ void UriNormalizer::normalize(const field &input, field &result, ScratchPad &scr
 
 bool UriNormalizer::noPathCheck(const uint8_t* inBuf, int32_t inLength, uint64_t& infractions) const {
     for (int32_t k = 0; k < inLength; k++) {
-        if (nonPathChar[inBuf[k]] == CHAR_NORMAL) continue;
+        if ((uriChar[inBuf[k]] == CHAR_NORMAL) || (uriChar[inBuf[k]] == CHAR_PATH)) continue;
         infractions |= INF_URINEEDNORM;
         return false;
     }
@@ -89,7 +89,7 @@ bool UriNormalizer::noPathCheck(const uint8_t* inBuf, int32_t inLength, uint64_t
 
 bool UriNormalizer::pathCheck(const uint8_t* inBuf, int32_t inLength, uint64_t& infractions) const {
     for (int32_t k = 0; k < inLength; k++) {
-        if (pathChar[inBuf[k]] == CHAR_NORMAL) continue;
+        if (uriChar[inBuf[k]] == CHAR_NORMAL) continue;
         if ((inBuf[k] == '/') && ((k == 0) || (inBuf[k-1] != '/'))) continue;
         infractions |= INF_URINEEDNORM;
         return false;
@@ -100,8 +100,9 @@ bool UriNormalizer::pathCheck(const uint8_t* inBuf, int32_t inLength, uint64_t&
 int32_t UriNormalizer::normCharClean(const uint8_t* inBuf, int32_t inLength, uint8_t *outBuf, uint64_t& infractions, const void *) const {
     int32_t length = 0;
     for (int32_t k = 0; k < inLength; k++) {
-        switch (nonPathChar[inBuf[k]]) {
+        switch (uriChar[inBuf[k]]) {
           case CHAR_NORMAL:
+          case CHAR_PATH:
             outBuf[length++] = inBuf[k];
             break;
           case CHAR_INVALID:
@@ -124,7 +125,7 @@ int32_t UriNormalizer::normCharClean(const uint8_t* inBuf, int32_t inLength, uin
                     else {
                         // Suspicious % escape of an ASCII character that does not need to be escaped
                         infractions |= INF_URIPERCENTASCII;
-                        if (nonPathChar[value] == CHAR_INVALID) infractions |= INF_URIBADCHAR;
+                        if (uriChar[value] == CHAR_INVALID) infractions |= INF_URIBADCHAR;
                         outBuf[length++] = value;
                         k += 2;
                     }
@@ -147,9 +148,6 @@ int32_t UriNormalizer::normCharClean(const uint8_t* inBuf, int32_t inLength, uin
                 outBuf[length++] = '%';
             }
             break;
-          default:
-            assert(0);
-            break;
         }
     }
     return length;
diff --git a/src/service_inspectors/nhttp_inspect/nhttp_uri_norm.h b/src/service_inspectors/nhttp_inspect/nhttp_uri_norm.h
index 9251feb6a..860ce218d 100644
--- a/src/service_inspectors/nhttp_inspect/nhttp_uri_norm.h
+++ b/src/service_inspectors/nhttp_inspect/nhttp_uri_norm.h
@@ -37,8 +37,7 @@ public:
     void normalize(const field &input, field &result, ScratchPad &scratchPad, uint64_t &infractions) const;
 
 private:
-    static const NHttpEnums::CharAction pathChar[256];
-    static const NHttpEnums::CharAction nonPathChar[256];
+    static const NHttpEnums::CharAction uriChar[256];
     static const int8_t asHex[256];
     static const bool goodPercent[256];
 
diff --git a/src/service_inspectors/nhttp_inspect/nhttp_uri_tables.cc b/src/service_inspectors/nhttp_inspect/nhttp_uri_tables.cc
index c147115fd..9b334bb9f 100644
--- a/src/service_inspectors/nhttp_inspect/nhttp_uri_tables.cc
+++ b/src/service_inspectors/nhttp_inspect/nhttp_uri_tables.cc
@@ -35,62 +35,21 @@
 
 using namespace NHttpEnums;
 
-const CharAction UriNormalizer::pathChar[256] = {
+const CharAction UriNormalizer::uriChar[256] = {
    CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,
    CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,
    CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,
    CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,
 
    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_PERCENT,   CHAR_NORMAL,    CHAR_NORMAL,
-   CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_PERIOD,    CHAR_SLASH,
+   CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_PATH,      CHAR_PATH,
    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
 
    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
-   CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_BACKSLASH, CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
-
-   CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
-   CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
-   CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
-   CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_INVALID,
-
-   CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,
-   CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,
-   CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,
-   CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,
-
-   CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,
-   CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,
-   CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,
-   CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,
-
-   CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,
-   CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,
-   CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,
-   CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,
-
-   CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,
-   CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,
-   CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,
-   CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT,  CHAR_EIGHTBIT };
-
-const CharAction UriNormalizer::nonPathChar[256] = {
-   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,
-   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,
-   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,
-   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,   CHAR_INVALID,
-
-   CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_PERCENT,   CHAR_NORMAL,    CHAR_NORMAL,
-   CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
-   CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
-   CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
-
-   CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
-   CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
-   CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
-   CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
+   CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_PATH,      CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
 
    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,
    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,    CHAR_NORMAL,