rapidjson_regex_example.cpp

#include "rapidjson/document.h"
#include "rapidjson/writer.h"
#include "rapidjson/stringbuffer.h"
#include <cstdio>
#include <string>
#include <iostream>
//#include <regex>
#include <boost/regex.hpp>

using namespace rapidjson;

int main() {
    Document d;
    StringBuffer buffer;
    Writer<StringBuffer> writer(buffer);

    boost::regex r1("\\{\\s*\"\\$oid\"\\s*:\\s*\"([a-f0-9]+)\"\\s*\\}");
    boost::regex r2("\\{\\s*\"\\$numberLong\"\\s*:\\s*\"([\\-0-9]+)\"\\s*\\}");
    boost::regex r3("\\{\\s*\"\\$date\"\\s*:\\s*\"([0-9\\-\\:T\\.Z\\+]+)\"\\s*\\}");
    boost::regex r4("NumberLong\\(([\\-0-9]+)\\)");
    boost::regex r5("(\"\\[object Object\\]\")");
    boost::regex r6("(\"_id\")");
    std::string date_limit("2015-04-01");

    std::string line;
    while (std::cin.good()) {
        std::getline(std::cin, line);
        if (!std::cin.good())
            break;
        if (line.size() == 0)
            continue;

        line = boost::regex_replace(line, r1, "\"$1\"");
        line = boost::regex_replace(line, r2, "$1");
        line = boost::regex_replace(line, r3, "\"$1\"");
        line = boost::regex_replace(line, r4, "$1");
        line = boost::regex_replace(line, r5, "null");
        line = boost::regex_replace(line, r6, "\"bson_id\"");

        d.Parse(line.c_str());

        Value& vt = d["t"];
        std::string t = vt.GetString();
        std::string year = t.substr(0, 4);

        if (t.compare(date_limit) > 0)
            continue;

        buffer.Clear();
        writer.Reset(buffer);
        d.Accept(writer);
        std::cout << year << "\t" << buffer.GetString() << std::endl;

    }

    return 0;
}

// vim: ts=8 sts=4 sw=4 expandtab