Stronger dimension name checking.

io/TextReader.cpp

@@ -88,10 +88,12 @@ QuickInfo TextReader::inspect()
 }
 
 
+// Make sure we have a header line.
 void TextReader::checkHeader(const std::string& header)
 {
     auto it = std::find_if(header.begin(), header.end(),
         [](char c){ return std::isalpha(c); });
+
     if (it == header.end())
         log()->get(LogLevel::Warning) << getName() <<
             ": file '" << m_filename <<
@@ -100,6 +102,43 @@ void TextReader::checkHeader(const std::string& header)
 
 
 void TextReader::parseHeader(const std::string& header)
+{
+    // If the first character is a double quote, assume that we have quoted
+    // field names.
+    if (header[0] == '"')
+        parseQuotedHeader(header);
+    else
+        parseUnquotedHeader(header);
+}
+
+
+void TextReader::parseQuotedHeader(const std::string& header)
+{
+    if (m_separatorArg->set())
+        throwError("Separator option not supported with a header line "
+            "containing quoted dimension names.");
+
+    // We know there's a double quote at position 0.
+    std::string::size_type pos = 1;
+    while (true)
+    {
+        size_t count = Dimension::extractName(header, pos);
+        m_dimNames.push_back(header.substr(pos, count));
+        pos += count;
+        if (header[pos] != '"')
+            throwError("Invalid character '" + std::string(1, header[pos]) +
+                "' found while parsing quoted header line.");
+        pos++; // Skip ending quote.
+
+        // Skip everything other than a double quote.
+        count = Utils::extract(header, pos, [](char c){ return c != '"'; });
+        pos += count;
+        if (header[pos++] != '"')
+            break;
+    }
+}
+
+void TextReader::parseUnquotedHeader(const std::string& header)
 {
     auto isspecial = [](char c)
         { return (!std::isalnum(c)); };
@@ -108,7 +147,7 @@ void TextReader::parseHeader(const std::string& header)
     // from the header line.
     if (!m_separatorArg->set())
     {
-        // Scan string for some character not a number, space or letter.
+        // Scan string for some character not a number or letter.
         for (size_t i = 0; i < header.size(); ++i)
             if (isspecial(header[i]))
             {
@@ -121,6 +160,13 @@ void TextReader::parseHeader(const std::string& header)
         m_dimNames = Utils::split(header, m_separator);
     else
         m_dimNames = Utils::split2(header, m_separator);
+    for (auto& s : m_dimNames)
+    {
+        size_t cnt = Dimension::extractName(s, 0);
+        if (cnt != s.size())
+            throwError("Invalid character '" + std::string(1, s[cnt]) +
+                "' in dimension name.");
+    }
 }
 
 

io/TextReader.hpp

@@ -117,10 +117,23 @@ class PDAL_DLL TextReader : public Reader, public Streamable
       Parse a header line into a list of dimension names.
 
       \param header  Header line to parse.
-      \return  List of dimension names.
     */
     void parseHeader(const std::string& header);
 
+    /**
+      Parse a header line that starts with a quote.
+
+      \param header Header line to parse.
+    */
+    void parseQuotedHeader(const std::string& header);
+
+    /**
+      Parse a header line that doesn't start with a quote.
+
+      \param header Header line to parse.
+    */
+    void parseUnquotedHeader(const std::string& header);
+
     /**
       Check a header line to see if it appears header-like.  Display a
       warning if it doesn't look like a header.

test/unit/io/TextReaderTest.cpp

@@ -278,13 +278,59 @@ TEST(TextReaderTest, insertHeader)
 
     PointTable table;
     reader.prepare(table);
-    PointViewSet pointViewSet = reader.execute(table);
-    PointViewPtr pointViewPtr = *pointViewSet.begin();
+    PointViewSet s = reader.execute(table);
+    PointViewPtr v = *s.begin();
 
-    EXPECT_EQ(pointViewPtr->size(), 11U);
+    EXPECT_EQ(v->size(), 11U);
     PointLayoutPtr layout = table.layout();
     EXPECT_TRUE(layout->findDim("A") != Dimension::Id::Unknown);
     EXPECT_TRUE(layout->findDim("B") != Dimension::Id::Unknown);
     EXPECT_TRUE(layout->findDim("C") != Dimension::Id::Unknown);
     EXPECT_TRUE(layout->findDim("G") != Dimension::Id::Unknown);
 }
+
+TEST(TextReaderTest, quotedHeader)
+{
+    auto testme = [](Options& options)
+    {
+        TextReader reader;
+        options.add("filename", Support::datapath("text/quoted.txt"));
+        reader.setOptions(options);
+
+        PointTable table;
+        reader.prepare(table);
+        PointViewSet s = reader.execute(table);
+        PointViewPtr v = *s.begin();
+        EXPECT_EQ(v->size(), 9U);
+        PointLayoutPtr layout = table.layout();
+        EXPECT_TRUE(layout->findDim("X") != Dimension::Id::Unknown);
+        EXPECT_TRUE(layout->findDim("Y") != Dimension::Id::Unknown);
+        EXPECT_TRUE(layout->findDim("Z") != Dimension::Id::Unknown);
+    };
+
+    {
+        Options opts;
+        testme(opts);
+    }
+
+    {
+        Options opts;
+        opts.add("header", "\"X\"\"Y\"   \"Z\"  ");
+        opts.add("skip", 1);
+        testme(opts);
+    }
+
+    {
+        Options opts;
+        opts.add("header", "\"X\"\"Y\"   \"  ");
+        opts.add("skip", 1);
+        EXPECT_THROW(testme(opts), pdal_error);
+    }
+
+    {
+        Options opts;
+        opts.add("header", "  \"X\"\"Y\"   \"  ");
+        opts.add("skip", 1);
+        EXPECT_THROW(testme(opts), pdal_error);
+    }
+}