fix path

kuzudb · Mar 21, 2024 · 02f8798 · 02f8798
1 parent 4a7b109
commit 02f8798
Showing 1 changed file with 58 additions and 4 deletions.
diff --git a/src/binder/bind/bind_import_database.cpp b/src/binder/bind/bind_import_database.cpp
@@ -4,6 +4,9 @@
 #include "binder/copy/bound_import_database.h"
 #include "common/exception/binder.h"
 #include "common/file_system/virtual_file_system.h"
+#include "common/string_utils.h"
+#include "parser/copy.h"
+#include "parser/parser.h"
 #include "parser/port_db.h"
 
 using namespace kuzu::common;
@@ -12,7 +15,38 @@ using namespace kuzu::parser;
 namespace kuzu {
 namespace binder {
 
-std::string getFilePath(
+std::string getPathSeparator() {
+    auto separatorStr = "/";
+#ifdef _WIN32
+    separatorStr = "\\";
+#endif
+    return separatorStr;
+}
+
+std::string convertSeparators(const std::string& path) {
+    auto separatorStr = getPathSeparator();
+    char separator = separatorStr[0];
+    if (separator == '/') {
+        // on unix-based systems we only accept / as a separator
+        return path;
+    }
+    // on windows-based systems we accept both
+    auto returnPath = path;
+    StringUtils::replaceAll(returnPath, separatorStr, "/");
+    return returnPath;
+}
+
+std::string extractTableName(const std::string& path) {
+    if (path.empty()) {
+        return std::string();
+    }
+    auto normalizedPath = convertSeparators(path);
+    auto splits = StringUtils::split(normalizedPath, "/");
+    KU_ASSERT(!splits.empty());
+    return splits.back();
+}
+
+std::string getQueryFromFile(
     common::VirtualFileSystem* vfs, const std::string boundFilePath, const std::string fileName) {
     auto filePath = vfs->joinPath(boundFilePath, fileName);
     if (!vfs->fileOrPathExists(filePath)) {
@@ -37,9 +71,29 @@ std::unique_ptr<BoundStatement> Binder::bindImportDatabaseClause(const Statement
         throw BinderException(stringFormat("Directory {} does not exist.", boundFilePath));
     }
     std::string finalQueryStatements;
-    finalQueryStatements += getFilePath(fs, boundFilePath, ImportDBConstants::SCHEMA_NAME);
-    finalQueryStatements += getFilePath(fs, boundFilePath, ImportDBConstants::COPY_NAME);
-    finalQueryStatements += getFilePath(fs, boundFilePath, ImportDBConstants::MACRO_NAME);
+    finalQueryStatements += getQueryFromFile(fs, boundFilePath, ImportDBConstants::SCHEMA_NAME);
+    auto copyQuery = getQueryFromFile(fs, boundFilePath, ImportDBConstants::COPY_NAME);
+    auto parsedStatements = Parser::parseQuery(copyQuery);
+    // replace the path in copy from statement with the bound path
+    for (auto& parsedStatement : parsedStatements) {
+        KU_ASSERT(parsedStatement->getStatementType() == StatementType::COPY_FROM);
+        auto copyFromStatement =
+            ku_dynamic_cast<const Statement*, const CopyFrom*>(parsedStatement.get());
+        KU_ASSERT(copyFromStatement->getSource()->type == common::ScanSourceType::FILE);
+        auto filePaths = ku_dynamic_cast<parser::BaseScanSource*, parser::FileScanSource*>(
+            copyFromStatement->getSource())
+                             ->filePaths;
+        KU_ASSERT(filePaths.size() == 1);
+        auto csvFilePath = filePaths[0];
+        auto extractedFileName = extractTableName(csvFilePath);
+        csvFilePath = boundFilePath + "/" + extractedFileName;
+        auto csvConfig = CSVReaderConfig::construct(
+            bindParsingOptions(copyFromStatement->getParsingOptionsRef()));
+        auto csvQuery = "COPY " + copyFromStatement->getTableName() + " FROM '" + csvFilePath +
+                        "' " + csvConfig.option.toCypher() + "\n";
+        finalQueryStatements += csvQuery;
+    }
+    finalQueryStatements += getQueryFromFile(fs, boundFilePath, ImportDBConstants::MACRO_NAME);
     return std::make_unique<BoundImportDatabase>(boundFilePath, finalQueryStatements);
 }
 } // namespace binder