Create QgsGdalUtils::pathIsCheapToOpen

Returns true if a dataset is considered "cheap" to open. The logic is very rudimentary for now, and considers only the file size of a few hardcoded formats.
qgis · Jul 23, 2021 · f818a62 · f818a62
1 parent 5455e91
commit f818a62
Show file tree

Hide file tree

Showing 3 changed files with 60 additions and 0 deletions.
diff --git a/src/core/qgsgdalutils.cpp b/src/core/qgsgdalutils.cpp
@@ -26,6 +26,7 @@
 #include <QNetworkProxy>
 #include <QString>
 #include <QImage>
+#include <QFileInfo>
 
 bool QgsGdalUtils::supportsRasterCreate( GDALDriverH driver )
 {
@@ -514,4 +515,31 @@ void QgsGdalUtils::setupProxy()
     }
   }
 }
+
+bool QgsGdalUtils::pathIsCheapToOpen( const QString &path, int smallFileSizeLimit )
+{
+  const QFileInfo info( path );
+  const long long size = info.size();
+
+  // if size could not be determined, safest to flag path as expensive
+  if ( size == 0 )
+    return false;
+
+  const QString suffix = info.suffix().toLower();
+  static QStringList sFileSizeDependentExtensions
+  {
+    QStringLiteral( "xlsx" ),
+    QStringLiteral( "ods" ),
+    QStringLiteral( "csv" )
+  };
+  if ( sFileSizeDependentExtensions.contains( suffix ) )
+  {
+    // path corresponds to a file type which is only cheap to open for small files
+    return size < smallFileSizeLimit;
+  }
+
+  // treat all other formats as expensive.
+  // TODO -- flag formats which only require a quick header parse as cheap
+  return false;
+}
 #endif
diff --git a/src/core/qgsgdalutils.h b/src/core/qgsgdalutils.h
@@ -132,6 +132,22 @@ class CORE_EXPORT QgsGdalUtils
     static void setupProxy();
 #endif
 
+    /**
+     * Returns TRUE if the dataset at the specified \a path is considered "cheap" to open.
+     *
+     * Datasets which are considered cheap to open may correspond to very small file sizes, or data types
+     * which only require some inexpensive header parsing to open.
+     *
+     * One use case for this method is to test whether a remote dataset can be safely opened
+     * to resolve the geometry types and other metadata without causing undue network traffic.
+     *
+     * The \a smallFileSizeLimit argument specifies the maximum file size (in bytes) which will
+     * be considered as small.
+     *
+     * \since QGIS 3.22
+     */
+    static bool pathIsCheapToOpen( const QString &path, int smallFileSizeLimit = 50000 );
+
     friend class TestQgsGdalUtils;
 };
 

diff --git a/tests/src/core/testqgsgdalutils.cpp b/tests/src/core/testqgsgdalutils.cpp
@@ -41,6 +41,7 @@ class TestQgsGdalUtils: public QObject
     void testResampleSingleBandRaster();
     void testImageToDataset();
     void testResampleImageToImage();
+    void testPathIsCheapToOpen();
 
   private:
 
@@ -268,6 +269,21 @@ void TestQgsGdalUtils::testResampleImageToImage()
   QCOMPARE( qAlpha( res.pixel( 40, 40 ) ), 255 );
 }
 
+void TestQgsGdalUtils::testPathIsCheapToOpen()
+{
+  // should be safe and report false paths which don't exist
+  QVERIFY( !QgsGdalUtils::pathIsCheapToOpen( "/not/a/file" ) );
+
+  // for now, tiff aren't marked as cheap to open
+  QVERIFY( !QgsGdalUtils::pathIsCheapToOpen( QStringLiteral( TEST_DATA_DIR ) + "/big_raster.tif" ) );
+
+  // a csv is considered cheap to open
+  QVERIFY( QgsGdalUtils::pathIsCheapToOpen( QStringLiteral( TEST_DATA_DIR ) + "/delimitedtext/test.csv" ) );
+
+  // ... unless it's larger than the specified file size limit (500 bytes)
+  QVERIFY( !QgsGdalUtils::pathIsCheapToOpen( QStringLiteral( TEST_DATA_DIR ) + "/delimitedtext/testdms.csv", 500 ) );
+}
+
 double TestQgsGdalUtils::identify( GDALDatasetH dataset, int band, int px, int py )
 {
   GDALRasterBandH hBand = GDALGetRasterBand( dataset, band );