BUG: to_sql with sqlite fallback rewrites table scheme even with if_exists='append'

[jorisvandenbossche]

From SO: http://stackoverflow.com/questions/24064509/to-sql-pandas-method-changes-the-scheme-of-sqlite-tables

The reason is that get_table is not implemented for fallback mode (https://github.com/pydata/pandas/blob/master/pandas/io/sql.py#L1072), so a create_table_statement is always called (https://github.com/pydata/pandas/blob/master/pandas/io/sql.py#L525).

Should look at it further to see if this is solvable, or a restriction of sqlite fallback.

[jorisvandenbossche]

I was a little bit too quick. As far as I can see, if if_exists='append' self.create() should normally never be called.

And I also cannot reproduce this: see notebook example http://nbviewer.ipython.org/gist/jorisvandenbossche/fe408096e798230d1c08

[drastega]

I suppose it was my mistake. Now I have no problems with or without using the sqlalchemy engine.

[jorisvandenbossche]

@drastega OK, no problem! Better to report to much than nothing. If you encounter other issues, certainly let know (it's for a large part new code, so there can certainly be some missings).

[Kare-c]

Sir,Has this problem been solved?

[Kare-c]

先生，这个问题解决了吗？
sorry，I found out that if you use pandas.io.sql.to_sql it will change the table structure, but if you use datas.to_sql it will not, but I encountered a new problem, insert the field error: duplicate entry, sir, how can I solve it?

[Kare-c]

先生，这个问题解决了吗？

Is there a way to insert data in pandas like ingnore, or repalce?