remove watermark for "pdf.th7.cn/down/files/1407/Real%20World%20OCaml.pdf"

main.cc

/*
 * clang++ -I/usr/local/include -pipe -std=c++11 -stdlib=libc++ -o main.cc.o -c main.cc
 * clang++ -L/usr/local/lib -lpodofo -stdlib=libc++ -o b.out main.cc.o
 *
 */

#include <iostream>
#include <string>
#include <list>
using namespace std;

#include <podofo/podofo.h>

void ShowObjType(const PoDoFo::PdfObject* obj) {
    if (!obj->IsNull()) {
        cout << obj->IsBool() << endl;
        cout << obj->IsNumber() << endl;
        cout << obj->IsReal() << endl;
        cout << obj->IsString() << endl;
        cout << obj->IsHexString() << endl;
        cout << obj->IsName() << endl;
        cout << obj->IsArray() << endl;
        cout << obj->IsDictionary() << endl;
        cout << obj->IsRawData() << endl;
        cout << obj->IsNull() << endl;
        cout << obj->IsReference() << endl;
    }
}

int main() {
    PoDoFo::PdfVecObjects objs;
    PoDoFo::PdfParser parser(&objs);
    parser.ParseFile("/tmp/rwo.pdf", false);

    int idx = 0;
    list<int> badidx;
    for (const auto obj: objs) {
        bool bdel = false;

        // annotate
        if (obj->IsDictionary()) {
            const auto& d0 = obj->GetDictionary();
            const auto& k0 = d0.GetKeys();
            if (k0.size() == 2 && 
                d0.HasKey(PoDoFo::PdfName("Length")) && 
                d0.HasKey(PoDoFo::PdfName("LC"))) {
                const auto stream = (PoDoFo::PdfMemStream*)obj->GetStream();
                if (stream->GetLength() > 0 && 
                    string(stream->Get()).find("www.it-ebooks.info") != string::npos) {
                    bdel = true;
                }
            }
        }
        // URI link
        if (obj->IsDictionary()) {
            const auto& d0 = obj->GetDictionary();
            if (d0.HasKey(PoDoFo::PdfName("A"))) {
                const auto& k0 = d0.GetKey(PoDoFo::PdfName("A"));
                if (k0->IsDictionary()) {
                    const auto& d1 = k0->GetDictionary();
                    if (d1.HasKey(PoDoFo::PdfName("URI"))) {
                        auto str = d1.GetKey(PoDoFo::PdfName("URI"))->GetString().GetString();
                        if (str == string("http://www.it-ebooks.info/")) {
                            bdel = true;
                        }
                    }
                }
            }
        }

        if (bdel) {
            badidx.insert(badidx.begin(), idx);
        }

        idx++;

    }
    cout << "obj count:" << badidx.size() << endl;

    for (auto offset: badidx) {
        auto obj = objs.begin() + offset;
        delete objs.RemoveObject(obj);
    }

    // dangle refernce
    for (const auto obj: objs) {
        auto prune_array = [&badidx](PoDoFo::PdfArray& arr) {
            for (int i = arr.size()-1; i >= 0; --i) {
                const auto& item = arr[i];
                if (item.IsReference()) {
                    bool orphan = false;
                    const auto& num = item.GetReference().ObjectNumber();
                    for (auto offset: badidx) {
                        if (offset+1 == num) {
                            orphan = true;
                            break;
                        }
                    }
                    if (orphan) {
                        arr.erase(arr.begin()+i);
                        cout << "erased" << endl;
                    }
                }
            }
        };
        if (obj->IsArray()) {
            auto& arr = obj->GetArray();
            prune_array(arr);
        }
        if (obj->IsDictionary()) {
            auto& d0 = obj->GetDictionary();
            const auto& k0 = d0.GetKeys();
            auto prune_key = [&](const string& key) {
                if (d0.HasKey(PoDoFo::PdfName(key))) {
                    auto k1 = d0.GetKey(PoDoFo::PdfName(key));
                    if (k1->IsArray()) {
                        auto& arr = k1->GetArray();
                        prune_array(arr);
                        if (arr.empty()) {
                            d0.RemoveKey(PoDoFo::PdfName(key));
                            cout << "removed" << endl;
                        }
                    }
                }
            };
            prune_key("Annots");
            prune_key("Contents");
        }
    }

    objs.Finish();

    // save
    {
        auto ptrailer = parser.GetTrailer();
        PoDoFo::PdfWriter writer(&objs, ptrailer);
        writer.SetPdfVersion(PoDoFo::ePdfVersion_1_6);
        writer.Write("out.pdf");
    }

    return 0;
}