gudusoft.gsqlparser.pp2.token

Class TokenEquivalence

Object

gudusoft.gsqlparser.pp2.token.TokenEquivalence

public final class TokenEquivalence
extends Object

Compares two SQL inputs for token-level equivalence modulo formatting.

The contract — verified by 12 hand-built cases in TokenEquivalenceTest — is that two SQL strings are equivalent when:

• Their solid + comment token sequences (in order) are equal character-by-character.
• Whitespace differences (spaces, tabs, linebreaks, indent) are ignored.
• Comment position may shift between the two streams — the comments must appear in the same relative order, but their surrounding whitespace does not matter.
• Case differences on keywords and unquoted identifiers are tolerated when caseInsensitive=true. Quoted identifiers, string literals, and operator/punctuation tokens are always compared byte-exact because case changes there alter semantics.

This helper is the safety net for several downstream slices:

• S13 (GuardedAstDelegate) — verifies that FormatterFactory.pp()'s output preserves every solid input token.
• S32 golden tests — sanity-checks goldens against their inputs.
• S33 overlay annotator — exercises the equivalence helper on real corpus inputs.
• S34 content-preservation — the strongest property test on the 50-SQL corpus.
• S35 per-vendor smoke — confirms every vendor preserves content through the pp2 pipeline.

The helper deliberately operates on raw SQL strings (not Pp2TokenStream instances) because the comparison is between an arbitrary input and an arbitrary output — there is no single pp2 stream shared between the two. The internal tokenization uses the GSP lexer directly with the same overlap-skip discipline as Pp2TokenStreamBuilder.

Changes the user explicitly asked for (variable or delimited-name case, compact mode's comment rewrite, comment removal) are accepted in their exact expected form when a TransformPolicy allows them; see that class. Pp2FormatOptions.transformPolicy carries it to pp2's guards.

Plan reference: §7.3/S10, §7.4/S10, Q8.

Method Summary

Modifier and Type	Method and Description
static boolean	commentsPreserved(String left, String right, EDbVendor vendor) The comment half of equalsModuloFormatting(String, String, Pp2FormatOptions, EDbVendor, boolean): the comments of left appear in right in the same relative order with byte-exact text; their position relative to the solid tokens may differ.
static boolean	commentsPreserved(String left, String right, EDbVendor vendor, TransformPolicy policy) commentsPreserved(String, String, EDbVendor) that also accepts the comment transformations policy allows.
static boolean	equalsModuloFormatting(String left, String right, Pp2FormatOptions opts) Case-insensitive comparison using the default Oracle dialect.
static boolean	equalsModuloFormatting(String left, String right, Pp2FormatOptions opts, EDbVendor vendor, boolean caseInsensitive) Full-control comparison.
static boolean	equalsModuloFormatting(String left, String right, Pp2FormatOptions opts, EDbVendor vendor, boolean caseInsensitive, boolean stringsAtStart) equalsModuloFormatting(String, String, Pp2FormatOptions, EDbVendor, boolean) for two texts that follow other statements: stringsAtStart is true when one of those turned the quoted-identifier option off, so that every double-quoted text of both may be a string (see DoubleQuotedStrings).
static boolean	solidTokensEquivalent(String left, String right, EDbVendor vendor, boolean caseInsensitive) The solid-token half of equalsModuloFormatting(String, String, Pp2FormatOptions, EDbVendor, boolean): every non-comment token of left appears in right once, in order, with the same type and (case-insensitively for keywords and unquoted identifiers when caseInsensitive) the same text.
static boolean	solidTokensEquivalent(String left, String right, EDbVendor vendor, boolean caseInsensitive, TransformPolicy policy) solidTokensEquivalent(String, String, EDbVendor, boolean) that also accepts the solid-token transformations policy allows.

Methods inherited from class Object

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

Method Detail

equalsModuloFormatting

public static boolean equalsModuloFormatting(String left,
                                             String right,
                                             Pp2FormatOptions opts)

Case-insensitive comparison using the default Oracle dialect. The "case insensitive" mode tolerates case differences on keywords and unquoted identifiers; literals and quoted identifiers are always byte-exact.

equalsModuloFormatting

public static boolean equalsModuloFormatting(String left,
                                             String right,
                                             Pp2FormatOptions opts,
                                             EDbVendor vendor,
                                             boolean caseInsensitive)

Full-control comparison.

Parameters:

left - first SQL string; must not be null

right - second SQL string; must not be null

opts - pp2 options; its transformPolicy allows the requested transformations; may be null (none allowed)

vendor - dialect to tokenize with; must not be null

caseInsensitive - when true, keyword and unquoted identifier comparisons are case-insensitive; when false, all comparisons are byte-exact on text

Returns:

true iff the two strings have the same token sequence under the rules described in the class Javadoc

Throws:

NullPointerException - if left, right, or vendor is null

equalsModuloFormatting

public static boolean equalsModuloFormatting(String left,
                                             String right,
                                             Pp2FormatOptions opts,
                                             EDbVendor vendor,
                                             boolean caseInsensitive,
                                             boolean stringsAtStart)

equalsModuloFormatting(String, String, Pp2FormatOptions, EDbVendor, boolean) for two texts that follow other statements: stringsAtStart is true when one of those turned the quoted-identifier option off, so that every double-quoted text of both may be a string (see DoubleQuotedStrings).

solidTokensEquivalent

public static boolean solidTokensEquivalent(String left,
                                            String right,
                                            EDbVendor vendor,
                                            boolean caseInsensitive)

The solid-token half of equalsModuloFormatting(String, String, Pp2FormatOptions, EDbVendor, boolean): every non-comment token of left appears in right once, in order, with the same type and (case-insensitively for keywords and unquoted identifiers when caseInsensitive) the same text. Exposed so that callers reporting the two invariants separately (the format worker) apply the same policy as the combined check.

solidTokensEquivalent

public static boolean solidTokensEquivalent(String left,
                                            String right,
                                            EDbVendor vendor,
                                            boolean caseInsensitive,
                                            TransformPolicy policy)

solidTokensEquivalent(String, String, EDbVendor, boolean) that also accepts the solid-token transformations policy allows.

commentsPreserved

public static boolean commentsPreserved(String left,
                                        String right,
                                        EDbVendor vendor)

The comment half of equalsModuloFormatting(String, String, Pp2FormatOptions, EDbVendor, boolean): the comments of left appear in right in the same relative order with byte-exact text; their position relative to the solid tokens may differ.

commentsPreserved

public static boolean commentsPreserved(String left,
                                        String right,
                                        EDbVendor vendor,
                                        TransformPolicy policy)

commentsPreserved(String, String, EDbVendor) that also accepts the comment transformations policy allows.