Re^2: Replacing commas with pipes

Your second example turns a 3 column record into a 4 column record. Now, it matches the specification as given, but if you go the CSV way, it doesn't seem to be a logical solution.

Considering that Text::CSV_XS allows you to set the separator, it seems more logical to use two Text::CVS_XS objects: one to read, one to write.

Comment on Re^2: Replacing commas with pipes

Replies are listed 'Best First'.
Re^3: Replacing commas with pipes by ww (Archbishop) on May 11, 2009 at 19:48 UTC
re JavaFan's comment about toolic's "...second example": /me thinks that's a typo in data (`d,e,"f\|g"` should be `d,e,"f,g"`), as OP's spec has commas throughout...even inside the quotes. Update: rephrased to clarify first line with attributions. Update2: In fairness, JavaFan's `s/("[^"]")\|,/$1\|\|"\|"/eg;` appears to work with this data: `"x,y,z",red,blue 1,"2,3",4,foo,bar,"blivitz_kung" a,b,"c,d,e,",nodereaper "Super_Search","XP per node",data,"code tags,markup" monk,troll,saint,"holders of unholy powers","Orders of Monks:novice,in +itiate,etc"` [download] (Update3*: bah, humbug, copied wrong data. Fixed above.) producing this output: `"x,y,z"\|red\|blue 1\|"2,3"\|4\|foo\|bar\|"blivitz_kung" a\|b\|"c,d,e,"\|nodereaper "Super_Search"\|"XP per node"\|data\|"code tags,markup" monk\|troll\|saint\|"holders of unholy powers"\|"Orders of Monks:novice,in +itiate,etc"` [download] Nonetheless, given the likelihood of (unknown, unspecified) edge-cases, a well-tested module remains the way to go, IMO.	[reply] [d/l] [select]

Replies are listed 'Best First'.

Re^3: Replacing commas with pipes
by ww (Archbishop) on May 11, 2009 at 19:48 UTC

JavaFan

toolic

/me thinks that's a typo in data (d,e,"f|g" should be d,e,"f,g"), as OP's spec has commas throughout...even inside the quotes.

Update: rephrased to clarify first line with attributions.

Update2: In fairness, JavaFan's s/("[^"]*")|,/$1||"|"/eg; appears to work with this data:

"x,y,z",red,blue
1,"2,3",4,foo,bar,"blivitz_kung"
a,b,"c,d,e,",nodereaper
"Super_Search","XP per node",data,"code tags,markup"
monk,troll,saint,"holders of unholy powers","Orders of Monks:novice,in
+itiate,etc"
[download]

(Update3: bah, humbug, copied wrong data. Fixed above.)

producing this output:

"x,y,z"|red|blue
1|"2,3"|4|foo|bar|"blivitz_kung"
a|b|"c,d,e,"|nodereaper
"Super_Search"|"XP per node"|data|"code tags,markup"
monk|troll|saint|"holders of unholy powers"|"Orders of Monks:novice,in
+itiate,etc"
[download]

Nonetheless, given the likelihood of (unknown, unspecified) edge-cases, a well-tested module remains the way to go, IMO.

[reply]
[d/l]
[select]