From d3caad4f56e481a5f58b5bb6e6e802a8aa3e4f25 Mon Sep 17 00:00:00 2001
From: iaom <18504285112@163.com>
Date: Thu, 15 Apr 2021 09:19:36 +0800
Subject: [PATCH] Add support for pdf in file parser.

---
 debian/control                  |  3 ++-
 libsearch/file-utils.cpp        | 22 ++++++++++++++++++++--
 libsearch/file-utils.h          |  2 ++
 libsearch/index/file-reader.cpp |  5 +++++
 libsearch/index/first-index.h   |  3 ++-
 libsearch/index/inotify-index.h |  3 ++-
 libsearch/libsearch.pro         |  2 +-
 7 files changed, 34 insertions(+), 6 deletions(-)

diff --git a/debian/control b/debian/control
index d58c01b..411e22f 100644
--- a/debian/control
+++ b/debian/control
@@ -16,7 +16,8 @@ Build-Depends: debhelper (>=9.0.0),
                libkf5windowsystem-dev,
                libgsettings-qt-dev,
                libqt5x11extras5-dev,
-               libuchardet-dev
+               libuchardet-dev,
+               libpoppler-qt5-dev
 Standards-Version: 4.5.0
 Homepage: https://www.ukui.org/
 Vcs-Git:  https://github.com/ukui/ukui-search.git
diff --git a/libsearch/file-utils.cpp b/libsearch/file-utils.cpp
index eb827ae..41c5110 100644
--- a/libsearch/file-utils.cpp
+++ b/libsearch/file-utils.cpp
@@ -30,6 +30,7 @@
 #include <QDomDocument>
 #include <QQueue>
 #include "uchardet/uchardet.h"
+#include "poppler-qt5.h"
 
 
 size_t FileUtils::_max_index_count = 0;
@@ -619,7 +620,7 @@ void FileUtils::getPptxTextContent(QString &path, QString &textcontent)
                         while(!ar.isNull())
                         {
                             at = ar.firstChildElement("a:t");
-                            textcontent.append(at.text().replace("\r","")).replace("\t"," ");
+                            textcontent.append(at.text().replace("\r","")).replace("\t","");
                             if(textcontent.length() >= MAX_CONTENT_LENGTH/3)
                             {
                                 file.close();
@@ -678,7 +679,7 @@ void FileUtils::getXlsxTextContent(QString &path, QString &textcontent)
             }
             if(t.isNull())
                 continue;
-            textcontent.append(t.text().replace("\r","").replace("\n"," "));
+            textcontent.append(t.text().replace("\r","").replace("\n",""));
             if(textcontent.length() >= MAX_CONTENT_LENGTH/3)
             {
                 file.close();
@@ -692,6 +693,23 @@ void FileUtils::getXlsxTextContent(QString &path, QString &textcontent)
     return;
 }
 
+void FileUtils::getPdfTextContent(QString &path, QString &textcontent)
+{
+    Poppler::Document *doc = Poppler::Document::load(path);
+    if(doc->isLocked())
+        return;
+    const QRectF qf;
+    int pageNum = doc->numPages();
+    for(int i = 0; i<pageNum; ++i)
+    {
+        textcontent.append(doc->page(i)->text(qf).replace("\n",""));
+        if(textcontent.length() >= MAX_CONTENT_LENGTH/3)
+            break;
+    }
+    delete doc;
+    return;
+}
+
 void FileUtils::getTxtContent(QString &path, QString &textcontent)
 {
     QFile file(path);
diff --git a/libsearch/file-utils.h b/libsearch/file-utils.h
index 172bc9a..76b093e 100644
--- a/libsearch/file-utils.h
+++ b/libsearch/file-utils.h
@@ -35,6 +35,7 @@
 #include <QMimeDatabase>
 #include <QMimeType>
 #include <QDir>
+
 #include "libsearch_global.h"
 //#define INITIAL_STATE 0
 //#define CREATING_INDEX 1
@@ -67,6 +68,7 @@ public:
     static void getDocxTextContent(QString &path, QString &textcontent);
     static void getPptxTextContent(QString &path, QString &textcontent);
     static void getXlsxTextContent(QString &path, QString &textcontent);
+    static void getPdfTextContent(QString &path, QString &textcontent);
     static void getTxtContent(QString &path, QString &textcontent);
     static size_t _max_index_count;
     static size_t _current_index_count; //this one has been Abandoned,do not use it.
diff --git a/libsearch/index/file-reader.cpp b/libsearch/index/file-reader.cpp
index b3dc833..35afefa 100644
--- a/libsearch/index/file-reader.cpp
+++ b/libsearch/index/file-reader.cpp
@@ -55,6 +55,11 @@ void FileReader::getTextContent(QString path, QString &textContent)
             searchdata.RunParser(path,textContent);
         }
     }
+    else if(name == "application/pdf")
+    {
+        if(strsfx.endsWith( "pdf"))
+            FileUtils::getPdfTextContent(path,textContent);
+    }
     else
     {
         qWarning()<<"Unsupport format:["<<path<<"]["<<type.name()<<"]";
diff --git a/libsearch/index/first-index.h b/libsearch/index/first-index.h
index 9bedf1c..2d90ba9 100644
--- a/libsearch/index/first-index.h
+++ b/libsearch/index/first-index.h
@@ -75,7 +75,8 @@ private:
         std::map<QString, bool>::value_type("wps", true),
         std::map<QString, bool>::value_type("pps", true),
         std::map<QString, bool>::value_type("dps", true),
-        std::map<QString, bool>::value_type("et", true)
+        std::map<QString, bool>::value_type("et", true),
+        std::map<QString, bool>::value_type("pdf", true)
     };
 
     //xapian will auto commit per 10,000 changes, donnot change it!!!
diff --git a/libsearch/index/inotify-index.h b/libsearch/index/inotify-index.h
index d50dc99..0bbb7c9 100644
--- a/libsearch/index/inotify-index.h
+++ b/libsearch/index/inotify-index.h
@@ -73,7 +73,8 @@ private:
         std::map<QString, bool>::value_type("wps", true),
         std::map<QString, bool>::value_type("pps", true),
         std::map<QString, bool>::value_type("dps", true),
-        std::map<QString, bool>::value_type("et", true)
+        std::map<QString, bool>::value_type("et", true),
+        std::map<QString, bool>::value_type("pdf", true)
     };
 
 };
diff --git a/libsearch/libsearch.pro b/libsearch/libsearch.pro
index 2c584c5..7a11e61 100644
--- a/libsearch/libsearch.pro
+++ b/libsearch/libsearch.pro
@@ -5,7 +5,7 @@ TARGET = ukui-search
 TEMPLATE = lib
 DEFINES += LIBSEARCH_LIBRARY
 
-PKGCONFIG += gio-2.0 glib-2.0 gio-unix-2.0 gsettings-qt
+PKGCONFIG += gio-2.0 glib-2.0 gio-unix-2.0 gsettings-qt poppler-qt5
 
 CONFIG += c++11 link_pkgconfig no_keywords lrelease