Re: Using COPY to import large xml file

Anto Aravinth <anto.aravinth.cse@xxxxxxxxx> · Mon, 25 Jun 2018 21:00:47 +0530

On Mon, Jun 25, 2018 at 8:54 PM, Anto Aravinth <anto.aravinth.cse@xxxxxxxxx> wrote:

On Mon, Jun 25, 2018 at 8:20 PM, Nicolas Paris <niparisco@xxxxxxxxx> wrote:

2018-06-25 16:25 GMT+02:00 Anto Aravinth <anto.aravinth.cse@xxxxxxxxx>:
Thanks a lot. But I do got lot of challenges! Looks like SO data contains lot of tabs within itself.. So tabs delimiter didn't work for me. I thought I can give a special demiliter but looks like Postrgesql copy allow only one character as delimiter :(
Sad, I guess only way is to insert or do a through serialization of my data into something that COPY can understand. 

easiest way would be:
xml -> csv -> \copy

by csv, I mean regular quoted csv (Simply wrap csv field with double quote, and escape
 enventually contained quotes with an other double quote.).

I tried but no luck. Here is the sample csv, I wrote from my xml convertor:

1       "Are questions about animations or comics inspired by Japanese culture or styles considered on-topic?"  "pExamples include a href="" href="http://www.imdb.com/title/tt0417299/" target="_blank">http://www.imdb.com/title/tt0417299/"" rel=""nofollow""Avatar/a, a href="" href="http://www.imdb.com/title/tt1695360/" target="_blank">http://www.imdb.com/title/tt1695360/"" rel=""nofollow""Korra/a and, to some extent, a href="" href="http://www.imdb.com/title/tt0278238/" target="_blank">http://www.imdb.com/title/tt0278238/"" rel=""nofollow""Samurai Jack/a. They're all widely popular American cartoons, sometimes even referred to as ema href="" href="https://en.wikipedia.org/wiki/Anime-influenced_animation" target="_blank">https://en.wikipedia.org/wiki/Anime-influenced_animation"" rel=""nofollow""Amerime/a/em./p

pAre questions about these series on-topic?/p
"       "pExamples include a href="" href="http://www.imdb.com/title/tt0417299/" target="_blank">http://www.imdb.com/title/tt0417299/"" rel=""nofollow""Avatar/a, a href="" href="http://www.imdb.com/title/tt1695360/" target="_blank">http://www.imdb.com/title/tt1695360/"" rel=""nofollow""Korra/a and, to some extent, a href="" href="http://www.imdb.com/title/tt0278238/" target="_blank">http://www.imdb.com/title/tt0278238/"" rel=""nofollow""Samurai Jack/a. They're all widely popular American cartoons, sometimes even referred to as ema href="" href="https://en.wikipedia.org/wiki/Anime-influenced_animation" target="_blank">https://en.wikipedia.org/wiki/Anime-influenced_animation"" rel=""nofollow""Amerime/a/em./p

pAre questions about these series on-topic?/p
"       "null"

the schema of my table is:

  CREATE TABLE so2 (
    id  INTEGER NOT NULL PRIMARY KEY,
    title varchar(1000) NULL,
    posts text,
    body TSVECTOR,
    parent_id INTEGER NULL,
    FOREIGN KEY (parent_id) REFERENCES so1(id)
);

and when I run:

COPY so2 from '/Users/user/programs/js/node-mbox/file.csv';

I get:

ERROR:  missing data for column "body"
CONTEXT:  COPY so2, line 1: "1	"Are questions about animations or comics inspired by Japanese culture or styles considered on-top..."

CONTEXT:  COPY so2, line 1: "1	"Are questions about animations or comics inspired by Japanese culture or styles considered on-top..."

Not sure what I'm missing. Not sure the above csv is breaking because I have newlines within my content. But the error message is very hard to debug. 

Postgresql copy csv parser is one of the most robust I ever tested
before. 