From 87071200872c2450c947047350132aee493033c1 Mon Sep 17 00:00:00 2001
From: Dominick Allen <djallen@librehumanitas.org>
Date: Thu, 2 Jan 2025 15:11:51 -0600
Subject: Get basic CSV parser operating.

---
 include/fud_csv.hpp | 34 +++++++++++++++++++++-------------
 1 file changed, 21 insertions(+), 13 deletions(-)

(limited to 'include/fud_csv.hpp')

diff --git a/include/fud_csv.hpp b/include/fud_csv.hpp
index efd37e6..38b1b81 100644
--- a/include/fud_csv.hpp
+++ b/include/fud_csv.hpp
@@ -22,6 +22,7 @@
 #include "fud_status.hpp"
 #include "fud_string_view.hpp"
 #include "fud_text.hpp"
+#include "fud_utf8.hpp"
 #include "fud_vector.hpp"
 
 #include <functional> // reference_wrapper
@@ -29,52 +30,59 @@
 namespace fud {
 
 using TextBuffer = Vector<std::byte>;
-using CsvBuffer = Vector<std::byte>;
-using CsvLine = Vector<StringView>;
 
 struct Csv {
     /** \brief The number of lines of data in the CSV. */
-    size_t numLines;
+    size_t numLines{0};
 
     /** \brief The number of columns in the CSV. */
-    size_t numColumns;
+    size_t numColumns{0};
 
     /** \brief Buffer for each line with numColumns of StringView. */
-    Vector<CsvLine> lines;
+    Vector<StringView> entries;
 
     /** \brief Backing buffer for data. */
-    CsvBuffer buffer;
+    Vector<utf8> buffer;
 
     /** \separator for each column */
     Utf8 columnDelimiter{Ascii{','}};
 
+    Utf8 quoteCharacter{Ascii{'"'}};
+
     /** \separator for each line */
     NewlineRepr newlineDelimiter{NewlineRepr::Posix};
 
-    bool strict;
+    bool strictUtf8{true};
+
+    bool strictColumns{true};
+
+    bool strictQuote{false};
+
+    bool skipInitialSpace{false};
 
     /** \brief Uses global Fud allocator for lines and backing buffer. */
     static Csv makeDefault();
 
     /** \brief Specify allocator to use for both lines and backing buffer. */
-    static Csv makeSingleAllocator(Allocator& allocator);
+    static Csv makeWithSingleAllocator(Allocator& allocator);
 
     /** \brief Specify allocator. */
-    static Csv make(Allocator& lineAllocator, Allocator& bufferAllocator);
+    static Csv make(Allocator& entryAllocator, Allocator& bufferAllocator);
 
     /** Consume and return the CSV. */
-    static FudStatus parseCsvFromFilename(
+    static FudStatus parseFromFilename(
         Csv& csv,
         Option<TextBuffer&&> bufferOption,
         StringView filename,
         OpenFlags flags = OpenFlags{},
-        Option<int> dirFdOption = NullOpt);
+        Option<int> dirFdOption = NullOpt,
+        Option<size_t> maxExtraAttempts = NullOpt);
 
     // assumes file is at start
-    static FudStatus parseCsvFromUnbufferedFile(Csv& csv, RegularFile&& file);
+    static FudStatus parseFromUnbufferedFile(Csv& csv, RegularFile&& file, Option<size_t> maxExtraAttempts);
 
     // assumes file is at start
-    static FudStatus parseCsvFromBufferedFile(Csv& csv, BufferedRegularFile& file);
+    static FudStatus parseFromBufferedFile(Csv& csv, BufferedRegularFile& file, Option<size_t> maxExtraAttempts);
 };
 
 } // namespace fud
-- 
cgit v1.2.3