gudusoft.gsqlparser.demos.prettyscore

Class CorpusRunner

Object

gudusoft.gsqlparser.demos.prettyscore.CorpusRunner

public final class CorpusRunner
extends Object

Runs pp and/or pp2 over a benchmark manifest, scores every output with the fidelity gate on, optionally adds seeded syntax-degraded samples (pp2 only), and produces the run record the CLI writes as JSON.

Run record layout (a published contract for --diff-run / --diff-baseline / --backlog):

 { "version": "prettyscore-run/0.1", "vendor", "profile", "engines": [...],
   "entries": [ {file, engine, source: clean|degraded, degrade, parse_ok, gate_passed,
                 score, attainable, ratio, dimensions: {id: score}, findings: [...], formatted} ],
   "summary": { "": {files, scored, mean_score, mean_ratio, min_ratio, gate_failures, parse_failures},
                "degraded": {samples, pp2_recovery_ratio, pp2_gate_failures} } }
 

Field Summary

Fields

Modifier and Type	Field and Description
static String	RUN_VERSION

Constructor Summary

Constructors

Constructor and Description
CorpusRunner(EDbVendor vendor, ScoreProfile profile, boolean keepText)
CorpusRunner(EDbVendor vendor, ScoreProfile profile, boolean keepText, boolean astEnabled)

Method Summary

Modifier and Type	Method and Description
static List<File>	readManifest(File manifest, File sqldir) Read a manifest: non-empty, non-comment lines are paths relative to sqldir.
static String	readText(File f)
Map<String,Object>	run(List<File> files, List<String> engines, int degradeCount, long seed, File relativeTo)
Map<String,Object>	scoreOne(String name, String engine, String source, String degrade, String sql) Format sql with engine and score it against the original.
static Map<String,Object>	summarize(List<Object> entries, List<String> engines)

Methods inherited from class Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

RUN_VERSION

public static final String RUN_VERSION

See Also:

Constant Field Values

Constructor Detail

CorpusRunner

public CorpusRunner(EDbVendor vendor,
                    ScoreProfile profile,
                    boolean keepText)

CorpusRunner

public CorpusRunner(EDbVendor vendor,
                    ScoreProfile profile,
                    boolean keepText,
                    boolean astEnabled)

Method Detail

readManifest

public static List<File> readManifest(File manifest,
                                      File sqldir)
                               throws IOException

Read a manifest: non-empty, non-comment lines are paths relative to sqldir.

Throws:

IOException

readText

public static String readText(File f)
                       throws IOException

Throws:

IOException

run

public Map<String,Object> run(List<File> files,
                              List<String> engines,
                              int degradeCount,
                              long seed,
                              File relativeTo)

Parameters:

files - the corpus files

engines - "pp" and/or "pp2"

degradeCount - number of syntax-degraded samples to derive (0 = none); scored with pp2 only

seed - degrader seed

scoreOne

public Map<String,Object> scoreOne(String name,
                                   String engine,
                                   String source,
                                   String degrade,
                                   String sql)

Format sql with engine and score it against the original.

summarize

public static Map<String,Object> summarize(List<Object> entries,
                                           List<String> engines)