bk2site.c

一百个病毒的源代码 包括熊猫烧香等 极其具有研究价值

// bk2site by Jose M. Vidal
// http://jmvidal.ece.sc.edu  [e-mail]
//
// Converts the Netscape bookmark.html file into a yahoo-like
// web site.
//
// Copyright 1999 Jose M. Vidal
// Jose M. Vidal, [e-mail], http://jmvidal.ece.sc.edu
// This program is free software.  You can redistribute it and/or modify
// it under the terms of the GNU General Public License
//
// CONTRIBUTORS:
//  See the AUTHORS file
//
//
// $Id: bk2site.C,v 1.49 2000/12/15 23:52:56 jmvidal Exp $
// -*- mode:C++; tab-width:4; c-basic-offset:2; indent-tabs-mode:nil -*- 

//TO DO:
// - Escape accented charactes in name2filename (name2url)
// - Add ALIASID= and ALIASOF= options to directory comments, so as
//   to allow more flexible directory aliasing.
// - Add support for separators <HR>
// - urllog should have "time url IP" and bk2site should count only one hit
///   per IP.
//
// BUGS
// - Make search.pl ignore html marking....how????

#ifdef HAVE_CONFIG_H
#include <config.h>
#endif

#include <iostream>
#include <fstream>
#include <iomanip>
#include <cstdlib>
#include <string>

// #if TIME_WITH_SYS_TIME
// # include <sys/time.h>
// # include <time.h>
// #else
// # if HAVE_SYS_TIME_H
// #  include <sys/time.h>
// # else
// #  include <time.h>
// # endif
// #endif 
#include <ctime>

#include "channel.H" //this include must appear before #include "reference.H"
#include "reference.H"


//defined in reference.C
extern string trim(const string & s);
extern int cmpNoCase(const string& s, const string &s2);
extern int replaceAllNC(string & s, string s1, string s2);
extern int replaceAll(string & s, string s1, string s2);

const string version = VERSION;
using namespace std;

void printUsage(const string & s)
{
  cout << "Usage: " << s << ": [OPTIONS] [main-bookmark-file] [bookmarkfile]*" << endl 
       << "\t If main-bookmark-file is omitted we use ~/.netscape/bookmarks.html" << endl
       << "\t Any extra bookmark files given are merged with main, see the README" << endl << endl;
  cout << "Builds a yahoo-like website based on one or more Navigator bookmark files" << endl << endl
       << "-nd\t\t - don't use a subdirectory hierachy. Use ugly long filenames" << endl
       << "\t\t  it's useful for tree depth limited providers like geocities" << endl
    //	 << "-bf <book file>\t - File with bookmarks (~/.netscape/bookmarks.html)" << endl
       << "-f <config file> - Name of configuration file (~/.bk2siterc)" << endl
       << "-d <dest dir>\t - Local directory for generated files." << endl
       << "\t\t   Default is /home/httpd/html/" << endl
       << "-t <name>\t - Name of top-level bookmark folder to use." << endl
       << "\t\t   If ommited will convert whole bookmark file." << endl
       << "-f1 <file>\t - Index template file for top index. " << endl
       << "-f2 <file>\t - Template for the other pages. " << endl 
       << "-o <file>\t - Output bookmark file." << endl
			 << "-xbel <file>\t - Output an xbel file." << endl 
       << "-ns\t\t - Do NOT generate the website. " << endl 
       << "-nc\t\t - Do NOT put HTML comments in the output." << endl
       << "-old <int>\t - Number of days (30) after which a bookmark is no longer new." << endl << endl
       << "Version " << version << endl << endl;
  exit(1);
}

void readConfigStream(istream & configStream, string varValues[],
											vector<string> & extraFileBase, vector<string> & extraFileName,
											channelContainer & channels)
{
	string curVarName;
	string curVarValue;

	string channelName ="";
	string channelUrl;
	string channelFile;
	while (configStream >> curVarName) {
		getline(configStream, curVarValue);
		curVarValue = trim(curVarValue);
		if (curVarName[0] == '#') continue;
		if (curVarValue[0] == '"') //get rid of surrounding "'s
	    curVarValue.erase(0,1);
		if (curVarValue[curVarValue.size() -1] == '"')
	    curVarValue.erase(curVarValue.size() - 1,1);

		if (getenv("HOME") != NULL)
			replaceAll(curVarValue,"$HOME", getenv("HOME"));
		else
			replaceAll(curVarValue,"$HOME", "");

		if (getenv("USER") != NULL)
			replaceAll(curVarValue,"$USER", getenv("USER"));
		else
			replaceAll(curVarValue,"$USER", "");

		if (getenv("LOGNAME") != NULL)
			replaceAll(curVarValue,"$LOGNAME", getenv("LOGNAME"));
		else
			replaceAll(curVarValue,"$LOGNAME", "");

		int i, found = 0;
		for (i = 0; i < NUMVARS; i++){
	    if (cmpNoCase(varNames[i],curVarName) == 0){
				varValues[i] = curVarValue;
				//	cout << varNames[i] << ":" << varValues[i] << endl;
				found = 1;
				break;
	    }
		};
		//these two are special, since they can appear many times.
		if (cmpNoCase("EXTRAFILEBASE",curVarName) == 0){
			extraFileBase.push_back(curVarValue);
			found = 1;
		}
		else if (cmpNoCase("EXTRAFILENAME",curVarName) == 0){
			extraFileName.push_back(curVarValue);
			found = 1;
		};

		//these can also appear many times
		if (cmpNoCase("CHANNELNAME",curVarName) == 0){
			if (channelName != ""){ //push back old one
				channels.addNewChannel(channelName, channelUrl, channelFile);
			}
			else
				channelName = curVarValue;
			found = 1;
		}
		else if (cmpNoCase("CHANNELURL",curVarName) == 0){
			channelUrl = curVarValue;
			found = 1;
		}
		else if (cmpNoCase("CHANNELFILE",curVarName) == 0){
			channelFile = curVarValue;
			channels.addNewChannel(channelName, channelUrl, channelFile);
			channelName = channelFile = channelUrl = "";
			found = 1;
		};
		if (found == 0)
	    cerr << "ERROR: variable " << curVarName << " is not a recognized name." << endl;
	}
	if (extraFileName.size() != extraFileBase.size()){
		cerr << "ERROR: the number of extrafilename declarations is not the same as the " << endl
				 << "       number of extrafilebase declarations. Exiting." << endl;
		exit(1);
	}
}

int main (int argc, char **argv)
{
	//initiallize default values.
	string varValues[NUMVARS];
	varValues[NEWGIF] = "<img src=\"/icons/new.gif\" alt=\"Added %MONTHCRE1/%DAYCRE/%YEARCRE\">";
	varValues[DESTDIR] = "";
	varValues[TOPFOLDER] = ""; 
	varValues[BOOKMARKFILE] = "bookmarks.html";
	varValues[INDEXFILENAME] = "indexbase.html";
	varValues[OTHERFILENAME] = "otherbase.html";
	varValues[TIMECUTOFF] = "30";
	varValues[TOP] = "Top";
	varValues[INDEX] = "index.html";
	varValues[SEARCH] = "";
	varValues[SEARCHTOROOTPATH] = "../";
	varValues[URLLOGFILE] = "";
	varValues[SUBDIRSEP] = "/";
	varValues[BACKDIRSEP] ="../";
	varValues[SEARCHURLTEMPLATE] = "<A HREF=\"%URL\">%TITLE</A> %NEW %CONDDASH %COMMENT";
	varValues[NEWSTOPFOLDER] = "";
	varValues[TITLE] = "My Bookmarks";
	varValues[OUTPUTBOOKMARKFILE] = "";
	varValues[COMMENTS] = "yes";
	varValues[FOLDERCREATION] = "maxdescendants";
	varValues[TOPHITSINCLUDENEWS] = "no";

	string xbelFileName = "";

	string homeDir = getenv("HOME");

	vector<string> extraFileBase;
	vector<string> extraFileName;

	vector<string> bookmarkfiles;
	int numBookmarkfiles = 0;
	int generateSite = 1;

    //read /etc/bk2site/dot.bk2siterc
    //    string globalConfigFile("CONFIGFILE");
    //    cout << "Checking " << "CONFIGFILE" << endl; //the -D on the compiler does not work. (egcs?)
	string globalConfigFile(CONFIGFILE);
	// cout << "Checking " << CONFIGFILE << endl;
	//the -D on the compiler does not work. (egcs?)
	// string globalConfigFile("/etc/bk2site/dot.bk2siterc");
	//string globalConfigFile("/etc/bk2site/dot.bk2siterc");
	ifstream globalConfigStream(globalConfigFile.c_str(), ios::in);
	channelContainer channels;
	if (globalConfigStream) 
		readConfigStream(globalConfigStream, varValues, extraFileBase, extraFileName, channels);

	int readHomeConfig = 1;
	for (int i = 1; i < argc; i++){
		string s(argv[i]);
		if (s == "-f")
			readHomeConfig = 0;
	}

	//read ~/.bk2siterc
	string configFile = homeDir + "/.bk2siterc";
	if (readHomeConfig == 1) {
		cout << "Reading " << configFile << endl;
		ifstream configStream(configFile.c_str(), ios::in);
		if (configStream) 
			readConfigStream(configStream, varValues, extraFileBase, extraFileName, channels);
	};

	//read the file  in -f
	for (int i = 1; i < argc; i++){
		string s(argv[i]);
		if (s == "-f")
	    if (++i < argc){
				configFile = argv[i];
				cout << "Reading " << configFile << endl;
				ifstream configStream2(configFile.c_str(), ios::in);
				if (configStream2)
					readConfigStream(configStream2, varValues, extraFileBase, extraFileName, channels);
				else {