I have a scenario where I'm exporting tables to S3 using SELECT INTO S3... query.
This is the sample command(python) that I am using to export:
export_to_s3_sql = f"SELECT * FROM {database}.{table} " \
f"INTO S3 '{s3_full_path}/{database}/{table}/{table}' " \
f"CONFIG '{json.dumps(export_config)}' " \
f"CREDENTIALS '{json.dumps(export_creds)}' " \
f"FIELDS TERMINATED BY '\\t' ENCLOSED BY '\"' ESCAPED BY '\\\\' " \
f"LINES TERMINATED BY '\\r'"
Once export is complete, I read the tables using Spark. My tables are big in size(~2TB) and sometimes they contain newline characters in the column values.
Since they contain new line characters, I am forced to use multiline: true when reading the CSV via Spark. This causes Spark to read one file(~80GB) via one core with num partition equal to number of csv files.
My aim is to identify whether my CSV in S3 contains newline characters as column value or not when reading via Spark. So that I can remove multiline option from some of my spark jobs
One way to verify is to save the row counts during export. but a table might get updated during, before, or after the query execution.
My first question is, what will happen if the table gets updated during the execution of the above query?
During the export is there any simple way to know if any of the columns contain newline characters?
Can I somehow save row count in a separate file? Given that a table may get update instructions anytime.
During the export, Is there any way to replace all the newline characters in column values with empty strings?
Is there any way I can add a new column with some default value? I will use this flag to detect if my CSV has new lines characters as
Try adding a condition \\n with lines terminated by clause and also add FIELDS TERMINATED BY '\\t\\r\\n' so it can help skip empty lines and reduce extra space from the table.
export_to_s3_sql = f"SELECT * FROM {database}.{table} " \
f"INTO S3 '{s3_full_path}/{database}/{table}/{table}' " \
f"CONFIG '{json.dumps(export_config)}' " \
f"CREDENTIALS '{json.dumps(export_creds)}' " \
f"FIELDS TERMINATED BY '\\t\\r\\n' ENCLOSED BY '\"' ESCAPED BY '\\\\' " \
f"LINES TERMINATED BY '\\r\\n'"
Related
I'm querying some tables on Athena (Presto SAS) and then downloading the generated CSV file to use locally. Opening the file, I realised the data contains new line characters that doesn't appear on AWS interface, only in the CSV and need to get rid of them. Tried using the function replace(string, search, replace) → varchar to skip the newline char replacing \n for \\n without success:
SELECT
p.recvepoch, replace(p.description, '\n', '\\n') AS description
FROM
product p
LIMIT 1000
How can I achieve that?
The problem was that the underlying table data doesn't actually contains \n anywhere, instead, the actual newline character, which is represented by char(10). I was able to achieve the expected behaviour using the replace function passing it as parameter:
SELECT
p.recvepoch, replace(p.description, chr(10), '\n') AS description
FROM
product p
LIMIT 1000
I've been successfully exporting GCloud SQL to CSV with its default delimiter ",". I want to import this CSV to Google Big Query and I've succeed to do this.
However, I'm experiencing a little problem. There's "," in some of my cell/field. It causes Big Query import process not working properly. For Example:
"Budi", "19", "Want to be hero, and knight"
My questions are:
Is it possible to export Google Cloud SQL with custom delimiter e.g. "|"?
If not, how to make above sample data to be imported in Google Big Query and become 3 field/cell?
Cheers.
Is it possible to export Google Cloud SQL with custom delimiter e.g. "|"?
Yes it's, See the documentation page of BigQuery how to set load options provided in this link
You will need to add --field_delimiter = '|' to your command
From the documentation:
(Optional) The separator for fields in a CSV file. The separator can be any ISO-8859-1 single-byte character. To use a character in the range 128-255, you must encode the character as UTF8. BigQuery converts the string to ISO-8859-1 encoding, and uses the first byte of the encoded string to split the data in its raw, binary state. BigQuery also supports the escape sequence "\t" to specify a tab separator. The default value is a comma (,).
As far as I know there's no way of setting a custom delimiter when exporting from CloudSQL to CSV. I attempted to introduce my own delimiter by formulating my select query like so:
select column_1||'|'||column_2 from foo
But this only results in CloudSQL escaping the whole result in the resulting CSV with double quotes. This also aligns with the documentation which states:
Exporting in CSV format is equivalent to running the following SQL statement:
SELECT <query> INTO OUTFILE ... CHARACTER SET 'utf8mb4'
FIELDS TERMINATED BY ',' OPTIONALLY ENCLOSED BY '\"'
ESCAPED BY '\\' LINES TERMINATED BY '\n'
https://cloud.google.com/sql/docs/mysql/import-export/exporting
In the below example, how can I set the skip leading row option?
bq --location=US query --external_table_definition=sales::Region:STRING,Quarter:STRING,Total_sales:INTEGER#CSV=gs://mybucket/sales.csv 'SELECT Region,Total_sales FROM sales;'
Regards,
Sreekanth
Flags options can be found under the installation home folder (I marked in bold below the flag you are looking for)
/google-cloud-sdk/platform/bq/bq.py:
--[no]allow_jagged_rows: Whether to allow missing trailing optional columns in
CSV import data.
--[no]allow_quoted_newlines: Whether to allow quoted newlines in CSV import
data.
-E,--encoding: : The character encoding used by the input
file. Options include:
ISO-8859-1 (also known as Latin-1)
UTF-8
-F,--field_delimiter: The character that indicates the boundary between
columns in the input file. "\t" and "tab" are accepted names for tab.
--[no]ignore_unknown_values: Whether to allow and ignore extra, unrecognized
values in CSV or JSON import data.
--max_bad_records: Maximum number of bad records allowed before the entire job
fails.
(default: '0')
(an integer)
--quote: Quote character to use to enclose records. Default is ". To indicate
no quote character at all, use an empty string.
--[no]replace: If true erase existing contents before loading new data.
(default: 'false')
--schema: Either a filename or a comma-separated list of fields in the form
name[:type].
--skip_leading_rows: The number of rows at the beginning of the source file to
skip.
(an integer)
--source_format: : Format of
source data. Options include:
CSV
NEWLINE_DELIMITED_JSON
DATASTORE_BACKUP
We have fifteen embedded newline characters in the field of a source S3 file. The field size in target table in Redshift is VARCHAR(5096). The field length in the source file is 5089 bytes. We are escaping each of the fifteen newline characters with a backslash \ as required by the ESCAPE option of the copy command. Our expectation with the ESCAPE option is that the backslash \ that has been inserted by us before each newline character will be ignored before loading the target in Redshift. However, when we use copy command with the ESCAPE option we are getting
err_code:1204 - String length exceeds DDL length."
Is there a way in which the added backslash \ characters are not counted for target column loads in Redshift?
Note: When we truncated the above source field in the file to 4000 bytes and inserted the backslash \ before the newline characters, the copy command with ESCAPE option successfully loaded the field in Redshift. Also, the backslash \ characters were not loaded in Redshift as expected.
You cold extend your VARCHAR length to allow for more characters.
Or, you could use the TRUNCATECOLUMNS options to load as much as possible without generating an error.
Our understanding w.r.t the above issue was incorrect. The backslashes "\" that we had inserted were not causing the error "err_code:1204 - String length exceeds DDL length.". The "escape" option with the copy command was actually not counting the inserted backslash characters towards the target limit and was also removing them from the loaded value properly.
The actual issue that were facing was that some of the characters that we were trying to load were multibyte UTF8 characters. Since, we were incorrectly assuming them to be of length 1 byte, the size of the target field was proving to be insufficient. We increased the length of the target field from varchar(5096) to varchar(7096), after which all data was loaded successfully.
I have some columns want to insert a existing table, some columns have contents like this"how, using,list, file",each column is separated by",",so how do I use load data infile to import them?
You didn't indicate which db you're using, So I'll answer you for mysql.
Please see this.
If the lines in such a file are terminated by carriage return/newline pairs, the statement shown here illustrates the field- and line-handling options you would use to load the file:
LOAD DATA INFILE 'data.txt' INTO TABLE tbl_name
FIELDS TERMINATED BY ',' ENCLOSED BY '"'
LINES TERMINATED BY '\r\n'
IGNORE 1 LINES;
And further down the page:
Occurrences of the ENCLOSED BY character within a field value are escaped by prefixing them with the ESCAPED BY character. Also note that if you specify an empty ESCAPED BY value, it is possible to inadvertently generate output that cannot be read properly by LOAD DATA INFILE. For example, the preceding output just shown would appear as follows if the escape character is empty. Observe that the second field in the fourth line contains a comma following the quote, which (erroneously) appears to terminate the field:
1,"a string",100.20
2,"a string containing a , comma",102.20
3,"a string containing a " quote",102.20
4,"a string containing a ", quote and comma",102.20
So, if you have unescaped "" inside columns these data could not be imported in the general case, and you'll have either to export data using correct ENCLOSED BY character, or preprocess the file to escape " first.
If you just have , inside your columns, then it's easy, you'll have to use ENCLOSED BY and TERMINATED BY.