misc/mysql_splitter/splitter.cpp

#include <iostream>
#include <sstream>
#include <string>
#include <vector>

#include <cstdio>

#include <boost/algorithm/string.hpp>
#include <boost/filesystem.hpp>
#include <boost/program_options.hpp>

namespace po = boost::program_options;

using namespace std;

const int BUFSIZE = 1024*1024;

string remove_comments(string in) {
        bool isComment = false;
        ostringstream out;
        for (unsigned int i=0; i<in.length() -1; i++) {
                if (in.at(i) == '/' && in.at(i+1) == '*')
                        isComment = true;

                if (in.at(i) == '*' && in.at(i+1) == '/') {
                        i++;
                        isComment = false;
                        continue;
                }

                if (isComment == false)
                        out << in.at(i) ;
        }

        return out.str();
}

string get_db_name(string input) {
        input = remove_comments(input);

        boost::erase_all(input,"`");
        boost::erase_all(input, ";");
        boost::trim(input);

        vector<string> words;
        words = boost::split(words, input, boost::is_any_of(" ") );

        string last = words.back();
        boost::trim(last);

        return last;
}


int main(int argc, char** argv) {

        po::options_description desc("Mysql dump file splitter:\nAllowed options");
        desc.add_options()
        ("help", "produce help message")
        ("data", "ignore ceate tablestatements")
                ("input-file", po::value< string >(), "input file")
        ;

        po::positional_options_description p;
        p.add("input-file", -1);


        po::variables_map vm;
        po::store( po::command_line_parser(argc, argv).options(desc).positional(p).run(), vm);
        po::notify(vm);

        if (vm.count("help")) {
        cout << desc << "\n";
        return 1;
        }


        if ( vm.count("input-file") == 0) {
                cout << desc << endl;
                return 1;
        }


        char linebuf[BUFSIZE];

        string inputfile = vm["input-file"].as< string >();
        if (! boost::filesystem::exists(inputfile) ) {
                cout << "No file named " << inputfile << endl;
                return 1;
        }


        FILE* in = fopen( inputfile.c_str(), "r" );
        FILE* out = NULL;
        //ifstream in( inputfile.c_str() );
        //ofstream out;

        ostringstream header;

        if ( in == NULL ) {
                cout << "Could not open " << argv[1] << endl;
                return 2;
        }

        time_t start = time(NULL);

        const char* SEARCH = "CREATE DATABASE";
        const int SEARCHLEN = strlen(SEARCH);

        while ( feof(in) == 0 && ferror(in) == 0 ) {
                fgets(linebuf, BUFSIZE, in);


                //if (line.substr(0, 15) == "CREATE DATABASE" ) {
                //if ( boost::starts_with(line, "CREATE DATABASE") ) {
                if (strncmp(linebuf,SEARCH, SEARCHLEN) == 0) {

            string line(linebuf);
                        
                        if (out != NULL ) {
                                fclose(out);
                        }

                        boost::trim(line);
                        string dbname = get_db_name(line);
                        cout << ">" << dbname << endl;


                        ostringstream oss;
                        oss << dbname << "_dump.sql";

            string filename = oss.str();

                        bool did_exist = boost::filesystem::exists(filename);

                        //out.open( filename.c_str(), ios::app );
                        out  = fopen(filename.c_str(), "a");


                        if ( out == NULL ) {
                                cout << "could not create outfile " << filename << endl;
                                return 3;
                        }

                        if (!did_exist) {
                                fputs(header.str().c_str(), out);
                                //out << header.str() << endl; //write preamble in new file
                        }

                }


                if (out != NULL ) {
                        fputs(linebuf, out);
                        //fputs("\n", out);
                        //out << linebuf << endl;
                } else {
                        header << linebuf; //collect preamble for later use
                }

        }


        if (out != NULL)
                fclose(out); 

//      if (out.is_open())
//              out.close();

    time_t end = time(NULL);

    cout << "Elapsed " << (end-start) << " seconds" << endl;

        fclose(in);
}
1	#include <iostream>
2	#include <sstream>
3	#include <string>
4	#include <vector>
5
6	#include <cstdio>
7
8	#include <boost/algorithm/string.hpp>
9	#include <boost/filesystem.hpp>
10	#include <boost/program_options.hpp>
11
12	namespace po = boost::program_options;
13
14	using namespace std;
15
16	const int BUFSIZE = 1024*1024;
17
18	string remove_comments(string in) {
19	bool isComment = false;
20	ostringstream out;
21	for (unsigned int i=0; i<in.length() -1; i++) {
22	if (in.at(i) == '/' && in.at(i+1) == '*')
23	isComment = true;
24
25	if (in.at(i) == '*' && in.at(i+1) == '/') {
26	i++;
27	isComment = false;
28	continue;
29	}
30
31	if (isComment == false)
32	out << in.at(i) ;
33	}
34
35	return out.str();
36	}
37
38	string get_db_name(string input) {
39	input = remove_comments(input);
40
41	boost::erase_all(input,"`");
42	boost::erase_all(input, ";");
43	boost::trim(input);
44
45	vector<string> words;
46	words = boost::split(words, input, boost::is_any_of(" ") );
47
48	string last = words.back();
49	boost::trim(last);
50
51	return last;
52	}
53
54
55	int main(int argc, char** argv) {
56
57	po::options_description desc("Mysql dump file splitter:\nAllowed options");
58	desc.add_options()
59	("help", "produce help message")
60	("data", "ignore ceate tablestatements")
61	("input-file", po::value< string >(), "input file")
62	;
63
64	po::positional_options_description p;
65	p.add("input-file", -1);
66
67
68	po::variables_map vm;
69	po::store( po::command_line_parser(argc, argv).options(desc).positional(p).run(), vm);
70	po::notify(vm);
71
72	if (vm.count("help")) {
73	cout << desc << "\n";
74	return 1;
75	}
76
77
78
79	if ( vm.count("input-file") == 0) {
80	cout << desc << endl;
81	return 1;
82	}
83
84
85	char linebuf[BUFSIZE];
86
87	string inputfile = vm["input-file"].as< string >();
88	if (! boost::filesystem::exists(inputfile) ) {
89	cout << "No file named " << inputfile << endl;
90	return 1;
91	}
92
93
94	FILE* in = fopen( inputfile.c_str(), "r" );
95	FILE* out = NULL;
96	//ifstream in( inputfile.c_str() );
97	//ofstream out;
98
99	ostringstream header;
100
101	if ( in == NULL ) {
102	cout << "Could not open " << argv[1] << endl;
103	return 2;
104	}
105
106	time_t start = time(NULL);
107
108	const char* SEARCH = "CREATE DATABASE";
109	const int SEARCHLEN = strlen(SEARCH);
110
111	while ( feof(in) == 0 && ferror(in) == 0 ) {
112	fgets(linebuf, BUFSIZE, in);
113
114
115	//if (line.substr(0, 15) == "CREATE DATABASE" ) {
116	//if ( boost::starts_with(line, "CREATE DATABASE") ) {
117	if (strncmp(linebuf,SEARCH, SEARCHLEN) == 0) {
118
119	string line(linebuf);
120
121	if (out != NULL ) {
122	fclose(out);
123	}
124
125	boost::trim(line);
126	string dbname = get_db_name(line);
127	cout << ">" << dbname << endl;
128
129
130	ostringstream oss;
131	oss << dbname << "_dump.sql";
132
133	string filename = oss.str();
134
135	bool did_exist = boost::filesystem::exists(filename);
136
137	//out.open( filename.c_str(), ios::app );
138	out = fopen(filename.c_str(), "a");
139
140
141
142	if ( out == NULL ) {
143	cout << "could not create outfile " << filename << endl;
144	return 3;
145	}
146
147	if (!did_exist) {
148	fputs(header.str().c_str(), out);
149	//out << header.str() << endl; //write preamble in new file
150	}
151
152	}
153
154
155	if (out != NULL ) {
156	fputs(linebuf, out);
157	//fputs("\n", out);
158	//out << linebuf << endl;
159	} else {
160	header << linebuf; //collect preamble for later use
161	}
162
163	}
164
165
166	if (out != NULL)
167	fclose(out);
168
169	// if (out.is_open())
170	// out.close();
171
172	time_t end = time(NULL);
173
174	cout << "Elapsed " << (end-start) << " seconds" << endl;
175
176	fclose(in);
177	}