pgmin

A program that will minimize those pesky SQL queries for you so you can focus on more important things like reading Hacker News.

Usage

Basic usage is:

$ ./pgmin file.sql > file.min.sql
$ psql < file.sql
$ psql < file.min.sql  # same results!

The generated file can only be run in psql at this point, which certainly is more secure than whatever ridiculous method you were going to use already. Plus psql is really what Real Hackers(tm) use, and you want to be a Real Hacker(tm), don't you? Of course you do, that's why you're reading Hacker News.

Example

Here is a sample query before pgmin:

SELECT
  current_database(), schemaname, tablename, /*reltuples::bigint, relpages::bigint, otta,*/
  ROUND((CASE WHEN otta=0 THEN 0.0 ELSE sml.relpages::float/otta END)::numeric,1) AS tbloat,
  CASE WHEN relpages < otta THEN 0 ELSE bs*(sml.relpages-otta)::BIGINT END AS wastedbytes,
  iname, /*ituples::bigint, ipages::bigint, iotta,*/
  ROUND((CASE WHEN iotta=0 OR ipages=0 THEN 0.0 ELSE ipages::float/iotta END)::numeric,1) AS ibloat,
  CASE WHEN ipages < iotta THEN 0 ELSE bs*(ipages-iotta) END AS wastedibytes
FROM (
  SELECT
    schemaname, tablename, cc.reltuples, cc.relpages, bs,
    CEIL((cc.reltuples*((datahdr+ma-
      (CASE WHEN datahdr%ma=0 THEN ma ELSE datahdr%ma END))+nullhdr2+4))/(bs-20::float)) AS otta,
    COALESCE(c2.relname,'?') AS iname, COALESCE(c2.reltuples,0) AS ituples, COALESCE(c2.relpages,0) AS ipages,
    COALESCE(CEIL((c2.reltuples*(datahdr-12))/(bs-20::float)),0) AS iotta -- very rough approximation, assumes all cols
  FROM (
    SELECT
      ma,bs,schemaname,tablename,
      (datawidth+(hdr+ma-(case when hdr%ma=0 THEN ma ELSE hdr%ma END)))::numeric AS datahdr,
      (maxfracsum*(nullhdr+ma-(case when nullhdr%ma=0 THEN ma ELSE nullhdr%ma END))) AS nullhdr2
    FROM (
      SELECT
        schemaname, tablename, hdr, ma, bs,
        SUM((1-null_frac)*avg_width) AS datawidth,
        MAX(null_frac) AS maxfracsum,
        hdr+(
          SELECT 1+count(*)/8
          FROM pg_stats s2
          WHERE null_frac<>0 AND s2.schemaname = s.schemaname AND s2.tablename = s.tablename
        ) AS nullhdr
      FROM pg_stats s, (
        SELECT
          (SELECT current_setting('block_size')::numeric) AS bs,
          CASE WHEN substring(v,12,3) IN ('8.0','8.1','8.2') THEN 27 ELSE 23 END AS hdr,
          CASE WHEN v ~ 'mingw32' THEN 8 ELSE 4 END AS ma
        FROM (SELECT version() AS v) AS foo
      ) AS constants
      GROUP BY 1,2,3,4,5
    ) AS foo
  ) AS rs
  JOIN pg_class cc ON cc.relname = rs.tablename
  JOIN pg_namespace nn ON cc.relnamespace = nn.oid AND nn.nspname = rs.schemaname AND nn.nspname <> 'information_schema'
  LEFT JOIN pg_index i ON indrelid = cc.oid
  LEFT JOIN pg_class c2 ON c2.oid = i.indexrelid
) AS sml
ORDER BY wastedbytes DESC

And this same query after pgmin:

\set mJ and \set mw avg_width \set v bigint \set h case \set Cb ceil \set Cc coalesce \set Ti constants \set mR count \set m current_database \set mY current_setting \set CW datahdr \set CO datawidth \set Ts desc \set X else \set u end \set e float \set Ty foo \set Cx from \set TN group \set Cf hdr \set E ibloat \set x iname \set TD indexrelid \set TH indrelid \set V iotta \set g ipages \set CQ ituples \set Tb join \set Tc left \set mk max \set Cu maxfracsum \set TL nspname \set mM null_frac \set CU nullhdr \set Ck nullhdr2 \set o numeric \set TM oid \set Tv order \set K otta \set Tq pg_class \set TR pg_index \set TZ pg_namespace \set mP pg_stats \set CX relname \set Td relnamespace \set a relpages \set CN reltuples \set M round \set T schemaname \set C select \set H sml \set mI substring \set md sum \set P tablename \set Z tbloat \set b then \set TB version \set Y wastedbytes \set Cn wastedibytes \set f when \set mQ where 
 :C :m(),:T,:P,:M((:h :f :K=0 :b 0.0 :X :H.:a:::e/:K :u):::o,1)AS :Z,:h :f :a<:K :b 0 :X bs*(:H.:a-:K):::v :u AS :Y,:x,:M((:h :f :V=0 OR :g=0 :b 0.0 :X :g:::e/:V :u):::o,1)AS :E,:h :f :g<:V :b 0 :X bs*(:g-:V):u AS :Cn :Cx(:C :T,:P,cc.:CN,cc.:a,bs,:Cb((cc.:CN*((:CW+ma-(:h :f :CW%ma=0 :b ma :X :CW%ma :u))+:Ck+4))/(bs-20:::e))AS :K,:Cc(c2.:CX,'?')AS :x,:Cc(c2.:CN,0)AS :CQ,:Cc(c2.:a,0)AS :g,:Cc(:Cb((c2.:CN*(:CW-12))/(bs-20:::e)),0)AS :V :Cx(:C ma,bs,:T,:P,(:CO+(:Cf+ma-(:h :f :Cf%ma=0 :b ma :X :Cf%ma :u))):::o AS :CW,(:Cu*(:CU+ma-(:h :f :CU%ma=0 :b ma :X :CU%ma :u)))AS :Ck :Cx(:C :T,:P,:Cf,ma,bs,:md((1-:mM)*:mw)AS :CO,:mk(:mM)AS :Cu,:Cf+(:C 1+:mR(*)/8 :Cx :mP s2 :mQ :mM<>0 :mJ s2.:T=s.:T :mJ s2.:P=s.:P)AS :CU :Cx :mP s,(:C(:C :mY('block_size'):::o)AS bs,:h :f :mI(v,12,3)IN('8.0','8.1','8.2'):b 27 :X 23 :u AS :Cf,:h :f v~'mingw32' :b 8 :X 4 :u AS ma :Cx(:C :TB()AS v)AS :Ty)AS :Ti :TN BY 1,2,3,4,5)AS :Ty)AS rs :Tb :Tq cc ON cc.:CX=rs.:P :Tb :TZ nn ON cc.:Td=nn.:TM :mJ nn.:TL=rs.:T :mJ nn.:TL<>'information_schema' :Tc :Tb :TR i ON :TH=cc.:TM :Tc :Tb :Tq c2 ON c2.:TM=i.:TD)AS :H :Tv BY :Y :Ts

As you can see, it looks much smaller, which obviously makes this better. Why wouldn't you want to do this? I'm sure the size difference here is significant!

$ wc -c bloat.sql
    2202 bloat.sql
$ wc -c bloat.min.sql
    2039 bloat.min.sql
$ # breathtaking!
$ wc -l bloat.sql
      45 bloat.sql
$ wc -l bloat.min.sql
       2 bloat.min.sql
$ # aweinspiring!
$ rm -r $HOME^C
$ # not recommended!

Readability is overrated!

Limitations/Bugs

There are no bugs, just code inputs I didn't anticipate or undesirable behaviors that I don't know about. That said, there are some possible limitations (I don't care to check them at this time, but assume they are issues):

1. This probably does not handle COPY statements. Due to how our ultra-basic parser works, we do not actually do any deep analysis of contents, so are basically operating on a token level to do our replacements. As such, since psql won't do any such interpolation of its variables in the STDIN portion of the block, the generated output would look like raw :foo tokens, plus we basically discard any whitespace and use single spaces. So while writing this out I've definitely convinced myself that that won't work. So don't do that. Or do write a patch and give it to me and maybe it'll do that then.

2. This will probably not work for any SQL files with psql commands in it, so a minimized file re-minimized probably won't work. I dunno, try it and get back to me; I assume psql will not do the double interpolation properly in the first pass, but it might surprise both of us. Addendum: I have written psql command and non-string psql var handling. As a consequence, you can pipe valid output from pgmin into pgmin and it'll be idempotent. Yay!

3. There are probably parsing quirks that are not accounted for that might need to be, such as scientific notation, E'' quoted strings, etc.

Enhancements

While this is already peak software, I'd welcome alternahancements/feedback. I will probably meh at it, but why not give it a shot? It might be a half-hearted meh followed by a slow nod and an eventual reluctant merge.

Author

David Christensen [email protected], [email protected]