Fix #2844 VAPOR fails to reliably identify file formats of files supp…

…lied on the command line (#2970) * Fix #2844 * clang-format pre-push hook * Add BOV logic to reject header files > 1MB in size * clang-format * Add -ftype option * clang-format pre-push hook * fix typo Co-authored-by: Scott Pearse <pearse@ucar.edu>
NCAR · Jan 19, 2022 · dd7ac4e · dd7ac4e
1 parent f9d3615
commit dd7ac4e
Show file tree

Hide file tree

Showing 4 changed files with 47 additions and 23 deletions.
diff --git a/apps/vaporgui/MainForm.cpp b/apps/vaporgui/MainForm.cpp
@@ -321,7 +321,7 @@ class ProgressStatusBar : public QWidget {
 
 // Only the main program should call the constructor:
 //
-MainForm::MainForm(vector<QString> files, QApplication *app, bool interactive, QWidget *parent) : QMainWindow(parent)
+MainForm::MainForm(vector<QString> files, QApplication *app, bool interactive, string filesType, QWidget *parent) : QMainWindow(parent)
 {
     _initMembers();
 
@@ -465,12 +465,18 @@ MainForm::MainForm(vector<QString> files, QApplication *app, bool interactive, Q
         for (auto &f : files) paths.push_back(f.toStdString());
 
         string fmt;
-        if (determineDatasetFormat(paths, &fmt)) {
-            loadDataHelper("", paths, "", "", fmt, true, ReplaceFirst);
+
+        if (filesType == "auto") {
+            if (!determineDatasetFormat(paths, &fmt)) {
+                fmt = "";
+                MSG_ERR("Could not determine dataset format for command line parameters");
+            }
         } else {
-            MSG_ERR("Could not determine dataset format for command line parameters");
+            fmt = filesType;
         }
 
+        if (!fmt.empty()) loadDataHelper("", paths, "", "", fmt, true, ReplaceFirst);
+
         _stateChangeCB();
     }
 
@@ -559,23 +565,21 @@ template<class T> bool MainForm::isDatasetValidFormat(const std::vector<std::str
 
 bool MainForm::determineDatasetFormat(const std::vector<std::string> &paths, std::string *fmt) const
 {
-    if (isDatasetValidFormat<VDCNetCDF>(paths))
-        *fmt = "vdc";
-    else if (isDatasetValidFormat<DCWRF>(paths))
-        *fmt = "wrf";
-    else if (isDatasetValidFormat<DCMPAS>(paths))
-        *fmt = "mpas";
-    else if (isDatasetValidFormat<DCP>(paths))
-        *fmt = "dcp";
-    else if (isDatasetValidFormat<DCUGRID>(paths))
-        *fmt = "ugrid";
-    else if (isDatasetValidFormat<DCCF>(paths))
-        *fmt = "cf";
-    else if (isDatasetValidFormat<DCBOV>(paths))
-        *fmt = "bov";
-    else
-        return false;
-    return true;
+    vector<pair<string, bool>> formats = {
+        {"vdc", isDatasetValidFormat<VDCNetCDF>(paths)}, {"wrf", isDatasetValidFormat<DCWRF>(paths)}, {"mpas", isDatasetValidFormat<DCMPAS>(paths)}, {"dcp", isDatasetValidFormat<DCP>(paths)},
+        {"ugrid", isDatasetValidFormat<DCUGRID>(paths)}, {"cf", isDatasetValidFormat<DCCF>(paths)},   {"bov", isDatasetValidFormat<DCBOV>(paths)},
+    };
+
+    int nOk = 0;
+    for (auto &f : formats) {
+        if (f.second) {
+            nOk++;
+            *fmt = f.first;
+        }
+    }
+    if (nOk == 1) return true;
+    MyBase::SetErrMsg("Unable to confidently determine the dataset format. Please load it manually in the GUI");
+    return false;
 }
 
 void MainForm::CheckForUpdates()

diff --git a/apps/vaporgui/MainForm.h b/apps/vaporgui/MainForm.h
@@ -77,7 +77,7 @@ class MainForm : public QMainWindow {
     Q_OBJECT
 
 public:
-    MainForm(vector<QString> files, QApplication *app, bool interactive = true, QWidget *parent = 0);
+    MainForm(vector<QString> files, QApplication *app, bool interactive = true, string filesType = "auto", QWidget *parent = 0);
     ~MainForm();
 
     int RenderAndExit(int start, int end, const std::string &baseFile, int width, int height);

diff --git a/apps/vaporgui/main.cpp b/apps/vaporgui/main.cpp
@@ -74,18 +74,21 @@ struct opt_t {
     OptionParser::IntRange_    renderRange;
     OptionParser::Dimension2D_ resolution;
     char *                     outputPath;
+    char *                     datasetType;
 } opt;
 OptionParser::OptDescRec_T set_opts[] = {{"help", 0, "", "Print this message and exit"},
                                          {"render", 0, "", "Render a given session file and exit"},
                                          {"timesteps", 1, "0:0", "Timesteps to render when using -render. Defaults to all timesteps."},
                                          {"resolution", 1, "1920x1080", "Output resolution when using -render"},
                                          {"output", 1, "vapor.tiff", "Output image file when using -render. This will be suffixed with the timestep number. Supports tiff, jpg"},
+                                         {"ftype", 1, "auto", "Specify file format of datasets passed in command line. Valid options are: auto vdc wrf mpas dcp ugrid cf bov"},
                                          {NULL}};
 OptionParser::Option_T     get_options[] = {{"help", Wasp::CvtToBoolean, &opt.help, sizeof(opt.help)},
                                         {"render", Wasp::CvtToBoolean, &opt.render, sizeof(opt.render)},
                                         {"timesteps", Wasp::CvtToIntRange, &opt.renderRange, sizeof(opt.renderRange)},
                                         {"resolution", Wasp::CvtToDimension2D, &opt.resolution, sizeof(opt.resolution)},
                                         {"output", Wasp::CvtToString, &opt.outputPath, sizeof(opt.outputPath)},
+                                        {"ftype", Wasp::CvtToString, &opt.datasetType, sizeof(opt.datasetType)},
                                         {NULL}};
 const char *               ProgName;
 
@@ -175,7 +178,7 @@ int           main(int argc, char **argv)
 
     vector<QString> files;
     for (int i = 1; i < argc; i++) { files.push_back(argv[i]); }
-    MainForm *mw = new MainForm(files, app, !opt.render);
+    MainForm *mw = new MainForm(files, app, !opt.render, opt.datasetType);
 
     // StartupParams* sParams = new StartupParams(0);
 

diff --git a/lib/vdc/BOVCollection.cpp b/lib/vdc/BOVCollection.cpp
@@ -62,6 +62,18 @@ const std::string BOVCollection::_intFormatString = "INT";
 const std::string BOVCollection::_floatFormatString = "FLOAT";
 const std::string BOVCollection::_doubleFormatString = "DOUBLE";
 
+namespace {
+size_t getFileSize(std::string filename)    // path to file
+{
+    FILE *p_file = NULL;
+    p_file = fopen(filename.c_str(), "rb");
+    fseek(p_file, 0, SEEK_END);
+    size_t size = ftell(p_file);
+    fclose(p_file);
+    return size;
+}
+}    // namespace
+
 BOVCollection::BOVCollection()
 : _time(_defaultTime), _dataFile(_defaultFile), _dataFormat(_defaultFormat), _variable(_defaultVar), _byteOffset(_defaultByteOffset), _divideBrick(_defaultDivBrick), _dataEndian(_defaultEndian),
   _centering(_defaultCentering), _dataComponents(_defaultComponents), _tmpDataFormat(_defaultFormat), _tmpByteOffset(_defaultByteOffset), _gridSizeAssigned(false), _formatAssigned(false),
@@ -103,6 +115,11 @@ int BOVCollection::Initialize(const std::vector<std::string> &paths)
 
         header.open(paths[i]);
         if (header.is_open()) {
+            if (getFileSize(paths[i]) > 1000000) {
+                SetErrMsg(("BOV header file larger than 1MB.  This text file shouldn't need to be larger than a few KB." + paths[0]).c_str());
+                return -1;
+            }
+
             rc = _parseHeader(header);
             if (rc < 0) {
                 SetErrMsg(("Error parsing BOV file " + paths[0]).c_str());