basically functional, but still a work in progress

1 files changed, 184 insertions, 0 deletions

diff --git a/parse_csv.c b/parse_csv.c
new file mode 100644
index 0000000..12e98f3
--- /dev/null
+++ b/parse_csv.c
@@ -0,0 +1,184 @@
+#include <istream>
+#include <string>
+#include <vector>
+#include <stdexcept>
+#include "parse_csv.h"
+using namespace std;
+
+/*************
+
+Format defined:
+  https://tools.ietf.org/html/rfc4180
+See also:
+  https://en.wikipedia.org/wiki/Comma-separated_values
+
+We depart from the RFC by ignoring \r, so that lines can
+end with either \r\n (dos style) or simply \n (unix style).
+
+We check for malformed quoted fields and throw exceptions.
+
+The last line of the input .csv file may or may not
+contain a newline:
+:; echo -en "xxx\n" | ./parse_csv_demo - | od -b -Anone
+ 170 170 170 012
+:; echo -en "xxx" | ./parse_csv_demo - | od -b -Anone
+ 170 170 170 012
+
+A line containing nothing but the newline has one field
+with a zero-length datum:
+:; echo -en "\n" | ./parse_csv_demo - | od -b -Anone
+ 012
+
+A line containing no data /and/ no newline isn't a line
+at all.  No line, no fields, no data:
+:; echo -en "" | ./parse_csv_demo - | wc
+      0       0       0
+
+************/
+
+string const Q("\"");
+
+enum class CSVState {
+    Start,
+    UnquotedField,
+    QuotedField,
+    QuoteInQuote
+};
+
+// Read one row of .csv table.
+// Return a zero-length vector if we encounter EoF
+//   before seeing any data.
+// Note that a table row can extend across multiple .csv file lines,
+//   if there are embedded newlines.
+//
+// Within quoted fields, we accept commas, quotes, and newlines,
+//   e.g "Hughie, ""Louie"", Dewey".
+template <class datum = qstring>
+vector<datum> readCSVRow(istream& input) {
+  using namespace std;
+  if (input.eof()) return vector<datum>(0);
+  CSVState state = CSVState::Start;
+  vector<datum> fields(0);
+  char c;
+  for (;;) {
+    input.read(&c, 1);
+    if (input.eof()) {
+      if (state == CSVState::QuotedField)
+        throw runtime_error("Unterminated quote in .csv file in field# "
+                               + to_string(fields.size())
+                    + "\n... namely <"
+                    + Q + fields.back() + ">");
+      break;
+    }
+    if (!input.good()) throw runtime_error("input error in .csv file");
+
+// if we made it this far, the line contains at least one datum:
+    if (fields.size() == 0) fields.push_back(datum());
+    switch (state) {
+      case CSVState::Start:
+        switch (c) {
+          case ',':     // zero-length field
+            fields.push_back(datum());
+            break;
+          case '"':
+            state = CSVState::QuotedField;
+// This is no-op if the datum is not of type qstring:
+            set_q(fields.back(), 1);
+            break;
+          case '\r':
+            break;
+          case '\n':
+            goto EndRecord;
+          default:
+            state = CSVState::UnquotedField;
+            fields.back().push_back(c);
+            break;
+        }
+        break;
+      case CSVState::UnquotedField:
+        switch (c) {
+          case ',': // start a new field:
+            state = CSVState::Start;
+            fields.push_back(datum());
+            break;
+          case '"': throw runtime_error
+            ("Stray quote in .csv file in field# "
+               + to_string(fields.size())
+               + "\n... namely <"
+               + fields.back() + Q + ">"
+            );
+            break;
+          case '\r': break;
+          case '\n': goto EndRecord;
+          default:  fields.back().push_back(c);
+            break;
+        }
+        break;
+      case CSVState::QuotedField:
+        switch (c) {
+          case '"': state = CSVState::QuoteInQuote;
+            break;
+// Accept anything except quote here;
+// that includes comma, \r, and \n.
+          default:  fields.back().push_back(c);
+            break;
+        }
+        break;
+// Previous state saw a quote; we consider what follows:
+      case CSVState::QuoteInQuote:
+        switch (c) {
+          case ',':     // comma after closing quote
+            state = CSVState::Start;
+            fields.push_back(datum());
+            break;
+          case '"':     // double ", map to plain "
+            fields.back().push_back('"');
+            state = CSVState::QuotedField;
+            break;
+          case '\r': break;
+          case '\n': goto EndRecord;
+          default:  throw runtime_error
+              ("Extraneous verbiage in .csv file after quoted field# "
+                    + to_string(fields.size())
+                    + "\n... namely <"
+                    + Q + fields.back() + Q
+                    + string(1,c) +">");
+            break;
+        }
+        break;
+    }
+  }
+  EndRecord:;;;;
+  return fields;
+}
+
+// Read entire CSV file.
+// Within quoted fields, we accept commas, quotes, and newlines,
+// e.g "Hughie, ""Louie"", Dewey".
+template <class datum = qstring>
+vector<vector<datum> > readCSV(istream &in) {
+  using namespace std;
+  vector<vector<datum> > table;
+  datum row;
+  while (!in.eof()) {
+    auto fields = readCSVRow<datum>(in);
+    if (fields.size() == 0) break;
+    table.push_back(fields);
+  }
+  return table;
+}
+
+// Explicit instantiation for the <qstring> version.
+// If you want some other version, you'll have to instantiate it.
+template vector<qstring> readCSVRow<qstring>(istream& input);
+template vector<vector<qstring> >readCSV<qstring>(istream &in);
+// Apparently it's not necessary to mention get_q() here,
+// but probably more portable to do it anyway:
+////template int get_q<qstring>(qstring const&);
+
+#if 1
+// Explicit instantiation for the <string> version.
+template vector<string> readCSVRow<string>(istream& input);
+template vector<vector<string> >readCSV<string>(istream &in);
+/////template int get_q<string>(string const&);
+#endif