string s = "";
int c;
while ((c = getchar()) != EOF)
    s.push_back((char)c);

Correct read input C++

Try your generated //CPP code with MSVC to check if there's any compile error.
If you don't have MSVC installed, just submit the *generated* code.

Even after reading all the posts here. Please help.
1. assert program is shorter than input
2. assert no long lines in source code
3. split string literal into vectors
4. assert no '\r' but '\n' after reading input
5. output with std::cout and '\n'
6. many tests and asserts

===this is an example of archive===
might be overkill to use Huffman and base64
shorten variable names in actual archive
===================================
//CPP
#include <iostream>
#include <string>
#include <unordered_map>
#include <vector>

int length = 37;
std::vector<std::string> encoded{
"7VKh",
"7lg="
};
std::unordered_map<std::string, char> table{
{"00", 'o'},
{"01", 'l'},
{"100", 'w'},
{"1010", ' '},
{"1011", '!'},
{"1100", 'd'},
{"1101", 'e'},
{"1110", 'h'},
{"1111", 'r'}
};

struct HuffmanTree {
    char ch;
    struct HuffmanTree* left;
    struct HuffmanTree* right;
    ~HuffmanTree() {
        delete left;
        delete right;
    }
};

void build_tree(HuffmanTree* root, std::string code, char ch)
{
    auto node{root};
    for (auto x : code)
        if (x == '1') {
            if (node->right == nullptr)
                node->right = new HuffmanTree();
            node = node->right;
        } else {
            if (node->left == nullptr)
                node->left = new HuffmanTree();
            node = node->left;
        }
    node->ch = ch;
}

char lookup(std::vector<bool>::iterator& it, const HuffmanTree* tree)
{
    auto node{tree};
    while (true) {
        if (node->left == nullptr)
            break;
        if (*it)
            node = node->right;
        else
            node = node->left;
        it++;
    }
    return node->ch;
}

void huffman_decode(std::vector<bool> bits, const HuffmanTree* tree)
{
    auto it{bits.begin()};
    while (it != bits.end())
        std::cout << lookup(it, tree);
}

int six(char c)
{
    if (c >= 'A' and c <= 'Z')
        return c-'A';
    if (c >= 'a' and c <= 'z')
        return c-'a' + 26;
    if (c >= '0' and c <= '9')
        return c-'0' + 52;
    if (c == '+')
        return 62;
    if (c == '/')
        return 63;
    return 0;
}

std::vector<std::string> base64_decode(std::vector<std::string> encoded)
{
    std::vector<std::string> decoded;
    for (auto s : encoded) {
        std::string t;
        auto i{0u};
        while (i < s.size()) {
            auto p0{six(s[i])};
            auto p1{six(s[i+1])};
            auto p2{six(s[i+2])};
            auto p3{six(s[i+3])};
            t.push_back((p0<<2) + ((p1&0x30)>>4));
            if (s[i+2] != '=')
                t.push_back(((p1&0x0f)<<4) + ((p2&0x3c)>>2));
            if (s[i+3] != '=')
                t.push_back(((p2&0x03)<<6) + p3);
            i += 4;
        }
        decoded.push_back(t);
    }
    return decoded;
}

std::vector<bool> string_to_bits(int n, std::vector<std::string> v)
{
    std::vector<bool> bits;
    for (auto s : v)
        for (auto byte : s)
            for (auto i{7}; i >= 0; i--)
                bits.push_back((byte >> i) & 1);
    while (int(bits.size()) > n)
        bits.pop_back();
    return bits;
}

int main()
{
    auto decoded{base64_decode(encoded)};
    auto bits{string_to_bits(length, decoded)};
    auto root{new HuffmanTree()};
    for (auto [code, ch] : table)
        build_tree(root, code, ch);
    huffman_decode(bits, root);
}

0. I wanted to implement universal approach with Huffman coding + compressing binary code to 64 base code. Not just eng text property (I was upset, because for so interesting task so stupid solution works).

1. I'm Java developer, and I afraid that Java consume too much memory even for input.
So I have to rewrite all to cpp.

2. Thinks that costs me a lot of time.
Input in cpp. (If you can explain what is wrong. please do )
WRONG
inputData = "";
std::string line = "";
while (std::getline(std::cin, line)) {
    inputData += line;
    //inputData.push_back('\r');//todo maybe we need \r\n
    inputData.push_back('\n');
}


CORRECT
inputData = "";
int c;
while ((c = getchar()) != EOF)
    inputData.push_back((char)c);

3. Escaping characters to produce cpp code:
std::string escape(char ch) {
    switch (ch) {
        case '\'':
            return "\\'";
        case '\"':
            return "\\\"";
        case '\\':
            return "\\\\";
        case '\b':
            return "\\b";
        case '\f':
            return "\\f";
        case '\n':
            return "\\n";
        case '\r':
            return "\\r";
        case '\t':
            return "\\t";
        default:
            return {ch};
    }
}

I finally got AC. So I wish you good luck with this problem and thank to author so interesting problem.

My solutions in the two languages produce the same output on my tests. However, the Go version gets WA1. Please fix this.

I suppose my error is either in the procedure of scanning/printing.
My program receives input and prints output using file pointers.

FILE *fi,*fo;

fi = stdin;
fo = stdout;

And the program reads/writes by fscanf/fprintf. If you have any knowledge concerning the input/output for this problem, please comment.

Edited by author 28.03.2013 07:06

If you're outputting C or C++ codes, then the followings might help:

1. Each line in your program can't be too long. (I chose to add a line break every 950 characters.) Having lines with length exceeding the compiler's limit will make you WA on test 1.

2. Lengths of any string constant (or array?) should be shorter (no longer?) than 65536 bytes. This is yet another unreasonable compiler limitation. Going over the length limit will make you WA on test 3.

I think the statements of this problem must have a full description of compilers used to compile outputs of solutions, otherwise it's nearly impossible to find out what does "WA1" mean. Not to mention that it's just plain wrong to use some magical non-standard-compatible compiler, that accepts "<iostream.h>" and "cout<<i" without using "std" namespace and so on.

Please add to statements hints about code formatting your compiler accepts, don't turn this wonderful problem to boring technical problem "Archiver or Satisfy Our Compiler"

On my dev-c++ (gcc) and on the old Timus compiler, they accept such code:
char a[]="xxx...xxx";  (there are up to 200000 'x')
But after lots of WA1's I tried to run it on VC2008, it won't compile!
It said that sth like "string is to long" (Sorry my VC isn't in English so I can't tell exactly)
So to solve this problem now, you have to think of a more complicated way...

And it took me much effort to escape that terrible WA1..

Check that all of arrays in your source less that 64 KB! (MS VC compiler limitation)

There is my code. My stress works fine. The size of output program is something near 4/5 * length bytes.
Using gets, all endl are written as '@'. There is an assert, that i haven't forgotten some symbols. Please, help!
http://pastebin.com/WnQvat6g

I got WA at #1, but I've tested for thousands of kbs of files and I don't know why I'm wrong at all.
Would you please show me why I got WA,please!!

After trial and trial, I passed the 1st test.
But now, on 3rd test, why do I get WA?
I think my algo will indeed work on the LITERARY text.

ADMIN, would you please tell me what sort of error I got?
CE, OLE(too long), or "Indeed" WA?

Thanks.

ADMINS,what punctuation marks are used?

How should the prog behave?
as i understood we have NORMAL_TEXT as std input.
Then we archivate the text. and what to do next?
output archivated text or what? tell me plz

As all symbols in the input will have codes less that 128
I can convert it to 7-bit char. Theoretically) But what
shall I do?...

Will it me WA or TL?

Если на паскале в конце выводимой программы поставить "readln;", получите WA1

Hello everyone!

First of all, let me say thanks to the author for this wonderful problem. It really hooked my mind for 2 days.

To solve it I used simple bitshift algorithm. My solution is written in C++. I have written 10 tests of different length (starting from 20 000 to more than 200 000 words) and my test system shows that solution is corrent (input=output and source code length is less than original text).

But I always get WA4 when submitting it. WHY? Can you please explain the details? I really have no idea. I will appreciate every comment.

Thank you.

Edited by author 08.08.2010 15:58