From 93696c29854cc04bbd4bb8877a70aed2f5f883d4 Mon Sep 17 00:00:00 2001
From: SJ <sj@acts.hu>
Date: Sun, 24 Mar 2013 01:20:12 +0100
Subject: [PATCH] parser fixes

---
 src/config.h  |  2 +-
 src/extract.c | 11 ++++++-----
 src/parser.c  | 24 +++++++++++++-----------
 src/test.c    |  3 +++
 4 files changed, 23 insertions(+), 17 deletions(-)

diff --git a/src/config.h b/src/config.h
index 717709a8..9f0c3f78 100644
--- a/src/config.h
+++ b/src/config.h
@@ -13,7 +13,7 @@
 
 #define VERSION "0.1.23-master-branch"
 
-#define BUILD 777
+#define BUILD 782
 
 #define HOSTID "mailarchiver"
 
diff --git a/src/extract.c b/src/extract.c
index 12f37f0c..0b630c73 100644
--- a/src/extract.c
+++ b/src/extract.c
@@ -45,6 +45,7 @@ void remove_xml(char *buf, int *html){
 #ifdef HAVE_ZIP
 int extract_opendocument(struct session_data *sdata, struct _state *state, char *filename, char *prefix){
    int errorp, i=0, len=0, html=0;
+   int len2;
    char buf[MAXBUFSIZE];
    struct zip *z;
    struct zip_stat sb;
@@ -60,14 +61,14 @@ int extract_opendocument(struct session_data *sdata, struct _state *state, char
 
          zf = zip_fopen_index(z, i, 0);
          if(zf){
-            while((len = zip_fread(zf, buf, sizeof(buf))) > 0){
+            while((len = zip_fread(zf, buf, sizeof(buf)-2)) > 0){
 
                remove_xml(buf, &html);
-               len = strlen(buf);
+               len2 = strlen(buf);
 
-               if(state->bodylen < BIGBUFSIZE-len-1){
-                  memcpy(&(state->b_body[state->bodylen]), buf, len);
-                  state->bodylen += len;
+               if(len2 > 0 && state->bodylen < BIGBUFSIZE-len2-1){
+                  memcpy(&(state->b_body[state->bodylen]), buf, len2);
+                  state->bodylen += len2;
                }
 
                memset(buf, 0, sizeof(buf));
diff --git a/src/parser.c b/src/parser.c
index 27056580..431c65d9 100644
--- a/src/parser.c
+++ b/src/parser.c
@@ -320,19 +320,17 @@ int parse_line(char *buf, struct _state *state, struct session_data *sdata, int
 
    if(state->is_header == 0 && buf[0] != ' ' && buf[0] != '\t') state->message_state = MSG_BODY;
 
-   if((state->content_type_is_set == 0 || state->is_header == 1) && strncasecmp(buf, "Content-Type:", strlen("Content-Type:")) == 0) state->message_state = MSG_CONTENT_TYPE;
-   else if(strncasecmp(buf, "Content-Transfer-Encoding:", strlen("Content-Transfer-Encoding:")) == 0) state->message_state = MSG_CONTENT_TRANSFER_ENCODING;
-   else if(strncasecmp(buf, "Content-Disposition:", strlen("Content-Disposition:")) == 0) state->message_state = MSG_CONTENT_DISPOSITION;
-
-
-   if(state->message_state == MSG_CONTENT_TYPE || state->message_state == MSG_CONTENT_TRANSFER_ENCODING) state->is_header = 1;
-
 
    /* header checks */
 
    if(state->is_header == 1){
 
       if(strncasecmp(buf, "From:", strlen("From:")) == 0) state->message_state = MSG_FROM;
+
+      else if(strncasecmp(buf, "Content-Type:", strlen("Content-Type:")) == 0) state->message_state = MSG_CONTENT_TYPE;
+      else if(strncasecmp(buf, "Content-Transfer-Encoding:", strlen("Content-Transfer-Encoding:")) == 0) state->message_state = MSG_CONTENT_TRANSFER_ENCODING;
+      else if(strncasecmp(buf, "Content-Disposition:", strlen("Content-Disposition:")) == 0) state->message_state = MSG_CONTENT_DISPOSITION;
+
       else if(strncasecmp(buf, "To:", 3) == 0) state->message_state = MSG_TO;
       else if(strncasecmp(buf, "Cc:", 3) == 0) state->message_state = MSG_CC;
       else if(strncasecmp(buf, "Bcc:", 4) == 0) state->message_state = MSG_CC;
@@ -358,8 +356,10 @@ int parse_line(char *buf, struct _state *state, struct session_data *sdata, int
    }
 
 
-   if((p = strcasestr(buf, "boundary"))){
-      extract_boundary(p, state);
+   if(state->message_state == MSG_CONTENT_TYPE){
+      if((p = strcasestr(buf, "boundary"))){
+         extract_boundary(p, state);
+      }
    }
 
 
@@ -426,7 +426,7 @@ int parse_line(char *buf, struct _state *state, struct session_data *sdata, int
          p++;
          if(*p == ' ' || *p == '\t') p++;
          snprintf(state->type, TINYBUFSIZE-1, "%s", p);
-         state->content_type_is_set = 1;
+         //state->content_type_is_set = 1;
          p = strchr(state->type, ';');
          if(p) *p = '\0';
       }
@@ -476,7 +476,9 @@ int parse_line(char *buf, struct _state *state, struct session_data *sdata, int
    boundary_line = is_item_on_string(state->boundaries, buf);
 
    if(!strstr(buf, "boundary=") && !strstr(buf, "boundary =") && boundary_line == 1){
-      state->content_type_is_set = 0;
+      state->is_header = 1;
+
+      //state->content_type_is_set = 0;
 
       if(state->has_to_dump == 1){
          if(take_into_pieces == 1 && state->fd != -1){
diff --git a/src/test.c b/src/test.c
index 08a60226..5cb99968 100644
--- a/src/test.c
+++ b/src/test.c
@@ -72,7 +72,10 @@ int main(int argc, char **argv){
    snprintf(sdata.filename, SMALLBUFSIZE-1, "%s", argv[1]);
    snprintf(sdata.tmpframe, SMALLBUFSIZE-1, "%s.m", argv[1]);
 
+   printf("parsing...\n");
    state = parse_message(&sdata, 1, &data, &cfg);
+
+   printf("post parsing...\n");
    post_parse(&sdata, &state, &cfg);
 
    printf("message-id: %s\n", state.message_id);