admin.cpp

一个C++写的RSS解析器,需要libxml2.rpm支持

#include "string.h"
#include "stdio.h"
#include <stdlib.h>

#include "admin.h"
#include "utils.h"
#include "database.h"




#define XML_PREFIX		"tempdata"

int processRss(char* rss, int class_id, struct Outline* outline, int is_insert_news)
{
	struct Channel* channel = NULL;
	struct Channel* pCurrent = NULL;
	
	struct ITEM* pItem = NULL;

	int ret = 0;
	int count = 0;

	channel = createRss(rss, outline);

	if(channel == NULL)
	{
		return -1;
	}
	
	pCurrent = channel;
	
	while(pCurrent)
	{
		pItem = pCurrent->pItem;
		while(pItem)
		{
			pItem = pItem->pNext;
		}
	
		pCurrent = pCurrent->pNext;
	}

	ret = db_init();
	if(ret == 0)
	{
		debug("连接数据库成功\n");
		
		int source_id = query_sourceid(class_id, outline->title);
		if(source_id < 0)
		{
			db_fini();
			error("获取souceid失败,%s还没有添加到数据库\n", outline->title);
			return -1;
		}

		debug("准备插入到t_superlink表\n");
		count = set_superlink(outline, class_id, source_id, channel);
		debug("插入到t_superlink表结束：插入数量%d\n", count);
		ret = count;
		if(is_insert_news == 1 && count > 0)
		{//仅仅插入新闻链接到t_news表中
			debug("准备插入到t_news表\n");
			int cnt = set_news(count, outline, class_id, source_id, channel);	
			debug("插入到t_news表结束：插入数量%d\n", cnt);			
		}

		db_fini();
		
		debug("关闭数据库连接\n");
	}else
	{
		ret = -2;
		error("db_init failed.\n");
	}

	destroyRss(channel);
	
	return ret;
}

int processOutline(int class_id, struct Outline* outline, int is_insert_news)
{
	char rss_file[32];

	sprintf(rss_file, "%s.xml", XML_PREFIX);
		
	debug("准备下载:%s\n", outline->xmlUrl);

	char command[256];
	sprintf(command, "rm %s -f && wget -t1 -q -O %s %s", rss_file, rss_file, outline->xmlUrl);	
	system(command);
	
	debug("下载结束\n");

	if(filesize(rss_file) <= 0)
		return -1;

	int ret = processRss(rss_file, class_id, outline, is_insert_news);
	if(ret == -1)
		debug("format of %s is wrong.\n", outline->xmlUrl);
	if(ret == -2)
		debug("connection to database failed.\n");

	return ret;
	
}

int processSource(int key, struct Source* source)
{
	struct Outline* pNext = NULL;
	struct Outline* pCur = source->outline;
	
	int class_id = 0;
	int source_id = 0;
	int is_insert_news = 0;
	
	int ret = db_init();
	if(ret == 0)
	{
		class_id = query_classid(key, source->subkey);
		db_fini();
	}
	
	if(class_id < 0)
		return -1;
		
	if(source->subkey == 0)
		is_insert_news = 1;//仅仅插入新闻条目

	while(pCur)
	{
		pNext = pCur->pNext;
		
		ret = processOutline(class_id, pCur, is_insert_news);
		
		info("  来源:%s; 类型: %d; 链接:%s; 新增条目: %d\n", pCur->title, pCur->kind, pCur->xmlUrl, ret);

		pCur = pNext;
	}

	return 0;
}

int processChannel(struct ChannelCuster* channelCuster)
{
	struct Source* pNext = NULL;
	struct Source* pCur = channelCuster->source;

	while(pCur)
	{
		info("%s(共%d条): \n", pCur->name, pCur->outline_count);
		
		pNext = pCur->pNext;
		processSource(channelCuster->key, pCur);

		pCur = pNext;
	}

	return 0;
}

int work(struct ChannelCuster* channelCuster)
{
	struct ChannelCuster* pNext = NULL;
	struct ChannelCuster* pCur = channelCuster;
	int count = 1;

	while(pCur)
	{
		info("------------%s-----------\n", pCur->name);
		
		pNext = pCur->pNext;

		processChannel(pCur);

		count++;

		pCur = pNext;
	}
	
	
	return 0;
}