getNextCSV() and setQuotedCSV(): library to process CSV files.
More...
#include <string>
#include <iostream>
#include <vector>
#include <stdlib.h>
Go to the source code of this file.
Namespaces | |
| namespace | std |
| Defined by the C++ standard library. | |
Defines | |
| #define | English_dox "Doxygen English documentation" |
| Doxygen English documentation. | |
Functions | |
| void | setQuotedCSV (std::string &res, const std::string &value) |
Prepares value for output into a CSV file. | |
| bool | getNextCSV (std::string &csv, std::istream &CIN) |
Scans input stream CIN and returns the next CSV value. | |
| void | trim (std::string &str) |
Deletes leading and trailing whitespace from "str". | |
| void | trimCSV (std::string &str) |
| Converts an incorrect CSV field value into its probably correct value. | |
| void | chop (std::string &str, char ch=0) |
Deletes ch when it is the trailing character in str. | |
getNextCSV() and setQuotedCSV(): library to process CSV files.
CSV: Comma Separated Value. The CSV file format have been better defined by IETF with RFC-4180.
getNextCSV().void setQuotedCSV() and store the resulting string.
Class CSV_line is a wrapper around these routines, but requires that no quoted Line Feed characters "\n" appear within each line in a CSV file.
Definition in file CSV.h.
| #define English_dox "Doxygen English documentation" |
| void setQuotedCSV | ( | std::string & | res, | |
| const std::string & | value | |||
| ) |
Prepares value for output into a CSV file.
res.value has whitespace.value has double-quotes.value has commas ",".'"' within value with 2 double-quotes char, not tested for wchar_t.
{{ // test::setQuotedCSV()
std::string res;
setQuotedCSV( res, "," ); assertTrue( res == "\",\"" ); // [","]
setQuotedCSV( res, "2" ); assertTrue( res == "2" ); // [2]
setQuotedCSV( res, "" ); assertTrue( res == "" ); // []
setQuotedCSV( res, "4,5" ); assertTrue( res == "\"4,5\"" ); // ["4,5"]
setQuotedCSV( res, "K\"" ); assertTrue( res == "\"K\"\"\"" ); // ["K"""]
setQuotedCSV( res, "\r\n" ); assertTrue( res == "\"\r\n\"" ); // ["\r\n"]
}}
| bool getNextCSV | ( | std::string & | csv, | |
| std::istream & | CIN | |||
| ) |
Scans input stream CIN and returns the next CSV value.
CIN should be open in std::ios::binary mode as chars are extracted one by one, using CIN.get(ch).CIN gets stored into csv.char, not tested for wchar_t.csv the trailing (CR+LF or LF) ==> "\r\n" o "\n".
"\n" (LF -> LineFeed).{{ // test::getNextCSV()
VEC.clear(); // std::vector<std::string> VEC;
std::string csv;
bool eol_CIN = false; // stop when the end of line is reached
std::istringstream ist( str , std::ios::binary );
while ( ! eol_CIN && ! ist.fail() ) { // ! ist.eof() pitfall!
eol_CIN = getNextCSV( csv, ist );
VEC.push_back( csv );
}
return;
// Using std::ios::binary ensures that no CR+LF chars are discarded
}}
| void trim | ( | std::string & | str | ) |
Deletes leading and trailing whitespace from "str".
" \f\n\r\t\v".isspace(ch) to find out if a letter is whitespace.
{{ // test::trim()
std::string str;
str = " a b "; trim(str); assertTrue( str == "a b" );
str = " a\nb "; trim(str); assertTrue( str == "a\nb" );
str = ""; trim(str); assertTrue( str == "" );
str = "\r\t\n "; trim(str); assertTrue( str == "" );
str = " a b "; trim(str); assertTrue( str == "a b" );
str = " ab " ; trim(str); assertTrue( str == "ab" );
}}
| void trimCSV | ( | std::string & | str | ) |
Converts an incorrect CSV field value into its probably correct value.
trim().
Sometimes a FILE.csv has quoted fields surrounded by whitespace. As these field values do not comply with RFC-4180, they are extracted by getNextCSV() as they come, with no whitespace removed and with their double quotes pairs intact. In the following example the string is enclosed in square parenthesis [..] instad of double quotes ["] for legibility:
["zero", "if "" 1" , , " 3xt" \r\n] [....0.,........ 1..,2,.........3...] csv field getNextCSV() trimCSV() +------------------+----------------+----------+ | ["zero"] | [zero] | [zero] | | [, "if "" 1" ] | [ "if "" 1" ] | [if " 1] | | [, ] | [ ] | [] | | [, " 3xt" \r\n] | [ " 3xt" ] | [ 3xt] | +------------------+----------------+----------+
By common sense, the programmer would expect that these strings be returned as they appear in the trimCSV() column, but the fact of the matter is that the only one that complies with RFC-4180 is the first one. After using trimCSV() on the value returned by getNextCSV() the result is what is reasonbly expected.
"\r" or carriage returns "\n" are probably processed in a way different form what it is expected, even before they are passed as arguments to trimCSV(). It is wiser no to trust this routine as a complete solution to process CSV files that do not fully comply with RFC-4180.
{{ // test::trimCSV()
CSV_line csv("\"zero\", \"if \"\" 1\" , , \" 3xt\" \r\f"); std::string s;
s=csv[0]; assertTrue( s == "zero" ); trimCSV(s); assertTrue( s == "zero" );
s=csv[1]; assertTrue( s == " \"if \"\" 1\" " ); trimCSV(s); assertTrue( s == "if \" 1" );
s=csv[2]; assertTrue( s == " " ); trimCSV(s); assertTrue( s == "" );
s=csv[3]; assertTrue( s == " \" 3xt\" \r\f" ); trimCSV(s); assertTrue( s == " 3xt" );
}}
| void chop | ( | std::string & | str, | |
| char | ch = 0 | |||
| ) |
Deletes ch when it is the trailing character in str.
ch.
{{ // test::chop()
std::string str; char ch;
str = "12345"; assertTrue( str == "12345" );
chop(str,'0'); assertTrue( str == "12345" );
for ( ch='5'; ch != '0'; --ch ) {
assertTrue( str[str.size()-1] == ch );
chop(str,ch);
}
assertTrue( str == "" );
chop(str,'3'); assertTrue( str == "" );
}}
1.5.8