plugin/media-ebook-plugin-pdf.cpp

   1 /*
   2  * libmedia-service
   3  *
   4  * Copyright (c) 2022 Samsung Electronics Co., Ltd. All rights reserved.
   5  *
   6  * Licensed under the Apache License, Version 2.0 (the "License");
   7  * you may not use this file except in compliance with the License.
   8  * You may obtain a copy of the License at
   9  *
  10  * http://www.apache.org/licenses/LICENSE-2.0
  11  *
  12  * Unless required by applicable law or agreed to in writing, software
  13  * distributed under the License is distributed on an "AS IS" BASIS,
  14  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  15  * See the License for the specific language governing permissions and
  16  * limitations under the License.
  17  *
  18  */
  19
  20 #include "media-ebook-plugin-pdf.h"
  21
  22 #include <dlog.h>
  23 #include <stack>
  24
  25 #ifdef LOG_TAG
  26 #undef LOG_TAG
  27 #endif
  28
  29 #define LOG_TAG "MEDIA_SERVICE"
  30
  31 using namespace EBook;
  32
  33 Pdf::Pdf(std::string path, std::unique_ptr<IRunnable> runner)
  34         : runner(std::move(runner))
  35 {
  36         if (path.empty()) {
  37                 LOGE("invalid path");
  38                 return;
  39         }
  40
  41         LOGD("%s", path.c_str());
  42
  43         try {
  44                 pdf.Load(path.c_str());
  45                 loaded = true;
  46         } catch (const PoDoFo::PdfError& e) {
  47                 LOGE("Initialization failed : %s", e.what());
  48         }
  49 }
  50
  51 bool Pdf::find()
  52 {
  53         if (!loaded)
  54                 return false;
  55
  56         for (int n = 0; n < pdf.GetPageCount(); ++n)
  57                 if (runner->run(parseTextFromPage(n)))
  58                         return true;
  59
  60         return false;
  61 }
  62
  63 void Pdf::insert()
  64 {
  65         if (!loaded)
  66                 return;
  67
  68         for (int n = 0; n < pdf.GetPageCount(); ++n)
  69                 runner->run(parseTextFromPage(n));
  70 }
  71
  72 std::string Pdf::parseTextFromPage(unsigned int index)
  73 {
  74         std::string fullText;
  75
  76         PoDoFo::EPdfContentsType type;
  77         PoDoFo::PdfVariant var;
  78         PoDoFo::PdfFont *cur_font = NULL;
  79         bool text_block = false;
  80         const char* tok;
  81         std::stack<PoDoFo::PdfVariant> stack;
  82         PoDoFo::PdfString unicode;
  83         PoDoFo::PdfArray array;
  84
  85         PoDoFo::PdfPage* page = pdf.GetPage(index);
  86         if (!page)
  87                 return fullText;
  88
  89         PoDoFo::PdfContentsTokenizer tokenizer(page);
  90
  91         while (tokenizer.ReadNext(type, tok, var)) {
  92                 if (type != PoDoFo::ePdfContentsType_Keyword) {
  93                         if (text_block)
  94                                 stack.push(var);
  95
  96                         continue;
  97                 }
  98
  99                 if (!text_block && strcmp(tok, "BT") == 0) {
 100                         text_block = true;
 101                         continue;
 102                 } else if (text_block && strcmp(tok, "ET") == 0) {
 103                         text_block = false;
 104                 }
 105
 106                 if (!text_block)
 107                         continue;
 108
 109                 if (strcmp(tok, "Tf") == 0) {
 110                         if (stack.size() < 2) {
 111                                 cur_font = NULL;
 112                                 continue;
 113                         }
 114
 115                         stack.pop();
 116                         cur_font = pdf.GetFont(page->GetFromResources(PoDoFo::PdfName("Font"), stack.top().GetName()));
 117                 } else if (strcmp(tok, "Tj") == 0 || strcmp(tok, "'") == 0 || strcmp(tok, "\"") == 0) {
 118                         if (stack.empty())
 119                                 continue;
 120
 121                         if (!cur_font || !cur_font->GetEncoding())
 122                                 continue;
 123
 124                         unicode = cur_font->GetEncoding()->ConvertToUnicode(stack.top().GetString(), cur_font);
 125                         fullText += unicode.GetStringUtf8();
 126
 127                         stack.pop();
 128                 } else if (strcmp(tok, "TJ") == 0) {
 129                         if (stack.empty())
 130                                 continue;
 131
 132                         array = stack.top().GetArray();
 133                         stack.pop();
 134
 135                         for (int i = 0; i < static_cast<int>(array.GetSize()); i++) {
 136                                 if (array[i].IsString() || array[i].IsHexString()) {
 137                                         if (!cur_font || !cur_font->GetEncoding())
 138                                                 continue;
 139
 140                                         unicode = cur_font->GetEncoding()->ConvertToUnicode(array[i].GetString(), cur_font);
 141                                         fullText += unicode.GetStringUtf8();
 142                                 }
 143                         }
 144                 }
 145         }
 146
 147         return fullText;
 148 }