From adc6d41101547d2d20b64c38634e5159cd21cd20 Mon Sep 17 00:00:00 2001
From: Clortox <Clortox@protonmail.com>
Date: Sun, 25 Jul 2021 19:07:40 -0400
Subject: [PATCH] Fix CDATA not parsing

---
 src/main.cpp    |  2 ++
 src/options.cpp | 40 +++++++++++++++++++++++------
 src/rss.cpp     | 68 +++++++++++++++++++++++++++++++++----------------
 src/rss.hpp     |  4 +++
 src/rss_out.cpp |  5 ++++
 5 files changed, 89 insertions(+), 30 deletions(-)

diff --git a/src/main.cpp b/src/main.cpp
index 08bd87e..acb67dd 100644
--- a/src/main.cpp
+++ b/src/main.cpp
@@ -43,6 +43,8 @@ int main(int argc, char** argv) {
         output += "\n" + rss_utils::rss_to_items(feed, opts);
     }
 
+    std::cout << "Item count: " << feed.getItemCount() << std::endl;
+
     std::cout << output << std::endl;
 
     delete opts;
diff --git a/src/options.cpp b/src/options.cpp
index 6997946..c2bb7d1 100644
--- a/src/options.cpp
+++ b/src/options.cpp
@@ -93,13 +93,22 @@ option_flags* parse_options(int argc, char** argv) {
                     current_item->description ^= 1;
                 break;
             case 'e':
-                ret->language ^= 1;
+                if(current_item == nullptr)
+                    ret->language ^= 1;
+                else
+                    std::cerr << "-e option not understood in context of --item; ignoring" << std::endl;
                 break;
             case 'm':
-                ret->webmaster ^= 1;
+                if(current_item == nullptr)
+                    ret->webmaster ^= 1;
+                else
+                    std::cerr << "-m option not understood in context of --item; ignoring" << std::endl;
                 break;
             case 'c':
-                ret->copyright ^= 1;
+                if(current_item == nullptr)
+                    ret->copyright ^= 1;
+                else
+                    std::cerr << "-c option not understood in context of --item; ignoring" << std::endl;
                 break;
             case 'p':
                 if(current_item == nullptr)
@@ -108,19 +117,34 @@ option_flags* parse_options(int argc, char** argv) {
                     current_item->pubdate ^= 1;
                 break;
             case 'q':
-                ret->managingeditor ^= 1;
+                if(current_item == nullptr)
+                    ret->managingeditor ^= 1;
+                else
+                    std::cerr << "-q option not understood in context of --item; ignoring" << std::endl;
                 break;
             case 'g':
-                ret->generator ^= 1;
+                if(current_item == nullptr)
+                    ret->generator ^= 1;
+                else
+                    std::cerr << "-g option not understood in context of --item; ignoring" << std::endl;
                 break;
             case 'o':
-                ret->docs ^= 1;
+                if(current_item == nullptr)
+                    ret->docs ^= 1;
+                else
+                    std::cerr << "-o option not understood in context of --item; ignoring" << std::endl;
                 break;
             case 'w':
-                ret->ttl ^= 1;
+                if(current_item == nullptr)
+                    ret->ttl ^= 1;
+                else
+                    std::cerr << "-w option not understood in context of --item; ignoring" << std::endl;
                 break;
             case 'b':
-                ret->builddate ^= 1;
+                if(current_item == nullptr)
+                    ret->builddate ^= 1;
+                else
+                    std::cerr << "-b option not understood in context of --item; ignoring" << std::endl;
                 break;
             case 'i':
                 if(ret->items == nullptr){
diff --git a/src/rss.cpp b/src/rss.cpp
index 184639f..0ae8440 100644
--- a/src/rss.cpp
+++ b/src/rss.cpp
@@ -106,7 +106,7 @@ std::string rss::getTitle() const {
     rapidxml::xml_node<> *tmp = _item_node->first_node("title");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 std::string rss::getLink() const {
@@ -116,7 +116,7 @@ std::string rss::getLink() const {
     rapidxml::xml_node<> *tmp = _item_node->first_node("link");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 std::string rss::getDescription() const {
@@ -126,7 +126,7 @@ std::string rss::getDescription() const {
     rapidxml::xml_node<> *tmp = _item_node->first_node("description");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 std::string rss::getLanguage() const {
@@ -136,7 +136,7 @@ std::string rss::getLanguage() const {
     rapidxml::xml_node<> *tmp = _item_node->first_node("language");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 std::string rss::getWebMaster() const {
@@ -146,7 +146,7 @@ std::string rss::getWebMaster() const {
     rapidxml::xml_node<> *tmp = _item_node->first_node("webMaster");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 std::string rss::getCopyright() const {
@@ -156,7 +156,7 @@ std::string rss::getCopyright() const {
     rapidxml::xml_node<> *tmp = _item_node->first_node("copyright");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 std::string rss::getPubDate() const {
@@ -166,7 +166,7 @@ std::string rss::getPubDate() const {
     rapidxml::xml_node<> *tmp = _item_node->first_node("pubDate");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 
 }
 
@@ -177,7 +177,7 @@ std::string rss::getManagingEditor() const {
     rapidxml::xml_node<> *tmp = _item_node->first_node("managingEditor");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 std::string rss::getGenerator() const {
@@ -187,7 +187,7 @@ std::string rss::getGenerator() const {
     rapidxml::xml_node<> *tmp = _item_node->first_node("generator");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 std::string rss::getDocs() const {
@@ -197,7 +197,7 @@ std::string rss::getDocs() const {
     rapidxml::xml_node<> *tmp = _item_node->first_node("docs");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 std::string rss::getTTL() const {
@@ -207,7 +207,7 @@ std::string rss::getTTL() const {
     rapidxml::xml_node<> *tmp = _item_node->first_node("ttl");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 
 }
 
@@ -218,7 +218,7 @@ std::string rss::getLastBuildDate() const {
     rapidxml::xml_node<> *tmp = _item_node->first_node("lastBuildDate");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 int rss::getItemCount() const {
@@ -324,11 +324,35 @@ bool rss::parse(const std::string& rss_str){
     return _ok;
 }
 
+std::string rss::cdata_to_string(const rapidxml::xml_node<>* node) const{
+    //this will dig till were past the cdata
+    std::cout << "in cdata" << std::endl;
+    const rapidxml::xml_node<>* tmp = node;
+    while(tmp->value()[0] == '\0'){ //if string is empty
+        tmp = tmp->first_node();
+        if(tmp == NULL)
+            return "";
+    }
+    return tmp->value();
+}
+
 size_t rss_utils::write_to_string(void* ptr, size_t size, size_t nmemb, std::string* s){
     s->append(static_cast<char *>(ptr), size * nmemb);
     return size * nmemb;
 }
 
+std::string item::cdata_to_string(const rapidxml::xml_node<>* node) const{
+    //this will dig till were past the cdata
+    std::cout << "in cdata" << std::endl;
+    const rapidxml::xml_node<>* tmp = node;
+    while(tmp->value()[0] == '\0'){ //if string is empty
+        tmp = tmp->first_node();
+        if(tmp == NULL)
+            return "";
+    }
+    return tmp->value();
+}
+
 item::item() {}
 item::item(rapidxml::xml_node<>* node){
     _item = node;
@@ -356,42 +380,42 @@ std::string item::getTitle() const {
     rapidxml::xml_node<>* tmp = _item->first_node("title");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 std::string item::getLink() const {
     rapidxml::xml_node<>* tmp = _item->first_node("link");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 std::string item::getDescription() const {
     rapidxml::xml_node<>* tmp = _item->first_node("description");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 std::string item::getAuthor() const {
     rapidxml::xml_node<>* tmp = _item->first_node("author");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 std::string item::getCategory() const {
     rapidxml::xml_node<>* tmp = _item->first_node("category");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 std::string item::getComments() const {
     rapidxml::xml_node<>* tmp = _item->first_node("comments");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 /*
@@ -399,7 +423,7 @@ std::string item::getEnclosure() const {
     rapidxml::xml_node<>* tmp = _item->first_node("enclosure");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 */
 
@@ -407,20 +431,20 @@ std::string item::getGuid() const {
     rapidxml::xml_node<>* tmp = _item->first_node("guid");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 std::string item::getPubDate() const {
     rapidxml::xml_node<>* tmp = _item->first_node("pubDate");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
 std::string item::getSource() const {
     rapidxml::xml_node<>* tmp = _item->first_node("source");
     if(tmp == 0)
         return "";
-    return tmp->value();
+    return cdata_to_string(tmp);
 }
 
diff --git a/src/rss.hpp b/src/rss.hpp
index 3daa9f4..ddcb6b2 100644
--- a/src/rss.hpp
+++ b/src/rss.hpp
@@ -11,6 +11,7 @@
 #endif
 
 #include <string>
+#include <sstream>
 #include <vector>
 #include <map>
 #include <iostream>
@@ -69,6 +70,7 @@ namespace rss_utils {
     private:
         bool parse(const std::string&);
         std::vector<rapidxml::xml_node<>*> parseItems();
+        std::string cdata_to_string(const rapidxml::xml_node<>*) const;
 
         bool _ok;
         std::string _uri;
@@ -99,6 +101,8 @@ namespace rss_utils {
 
 
     private:
+        std::string cdata_to_string(const rapidxml::xml_node<>*) const;
+
         rapidxml::xml_node<> *_item;
     };
 
diff --git a/src/rss_out.cpp b/src/rss_out.cpp
index 6278984..e88c4bb 100644
--- a/src/rss_out.cpp
+++ b/src/rss_out.cpp
@@ -49,6 +49,11 @@ std::string rss_utils::rss_to_items(const rss& rss_obj, const option_flags* flag
     int maxItem = rss_obj.getItemCount();
     std::string ret;
 
+    if(maxItem == -1){
+        ret = "<No items in feed!>";
+        return ret;
+    }
+
     for(unsigned int i=0; i < flags->item_count; ++i, ++items){
         //if has a valid index
         if(items->index >= 0 && items->index < maxItem){