sqlplot

This is a python3 reimplementation of https://github.com/bingmann/sqlplot-tools.

The main considerations for writing this reimplementation are:

• more features such as macros for sql expressions
• python needs less code C++ as the main task is to parse text and issue SQL commands

file types

We support tex and python as input, where the keyword are prefixed with %% or ##, respectively. As output, we support

• tex
• python - can be directly used with matplotlib
• csv - can be directly used with plotly via pandas (python3 libraries)
• json - useful for loading in a HTML5 webpage

Permanent Legend

When parsing a tex file, sqlplot maintains a dictionary mapping legend entries (represented internal as tuples) to cycle shift numbers used in pgfplots to determine the shape and the color of a plot entry such that the same entry across several plot has the same color and shape. This dictionary is saved in the file pgf_color_entries.txt, and can be edited to change the cycle shift number of an entry.

new keywords

CONFIG line

A MULTIPLOT command can have a final line

%% CONFIG file={filename} type={filetype} mode=[wa]

This is basically a key-value list, where each key-value is optional. The valid keys are:

• file: write the output in a separate file instead of directly to stdout
• type: use a different file type than the type of the file we are currently parsing
• mode: either 'w' for overwriting or 'a' for appending. Appending is useful if you have several SQL commands that generate data for a single plot
• colorcache=none: do not use the cached scheme written in pgf_color_entries.txt

%% SINGLEPLOT(name) SQL-STATEMENT 

Instead of a multiplot, we plot only a single dataset whose name string we set as a parameter of SINGLELPLOT. The rest is exactly as MULTIPLOT but a regular SQL command (i.e., without MULTIPLOT in the SQL body). The idea of this command is to supplement a MULTIPLOT command from another table that may only contain the benchmarks of a single instance.

%% MATRIX SELECT-STATEMENT containing columns `x`, `y` and `val`

create a tabular with the x attribute as columns, y attribute as rows and val as matrix entries. Example:

%% MATRIX
%% SELECT 
%% "\num{" || prefix || "}" AS x,
%% "\texttt{" || file  || "}" AS y,
%% printf("%.2f", time) AS val
%% FROM stats

%% DEFINE macro(arg1,arg2,...) {body $arg1 ... $arg2 ...} 

Creates the macro macro with arguments arg1, arg2, etc. The arguments must occur in the body preceded by a '$'. The macro can then be used in a SQL expression with $macro(parameter1, paratemer2, ...) You can overwrite the definition of a macro.

%% UNDEF macro

Undefines the macro macro.

Program parameters

• -i <filename> the input file name to parse (required argument)
• -l DEBUG runs the program in debug level logging, issuing all SQL commands that are executed, also unpacking a MULTIPLOT command into multiple SQL commands
• -D databasefile stores in-memory created database in a file in append mode, meaning that it adds tables in case that the file is an existing sqlite database, and assuming that this database does not already contain these tables.