@iris-code/core 0.1.0

package/dist/csharp/lexer.js

@@ -0,0 +1,182 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.transformCSharp = transformCSharp;
+exports.stripCSharpComments = stripCSharpComments;
+exports.csharpCodeOnly = csharpCodeOnly;
+/**
+ * The C# lexical layer, in its own module because TWO callers need it: the C#
+ * analyser (every structural rule) and `detectParseErrorReason` in guards.ts
+ * (the delimiter balance check that decides whether a file is analysed at all).
+ *
+ * That second caller is the reason this is not a private helper. The generic
+ * sanitiser treats `\"` as an escaped quote, so the everyday C# verbatim string
+ * `@"C:\logs\"` read as UNTERMINATED, every brace after it landed at the wrong
+ * depth, and the file was reported as a parse error with all findings zeroed.
+ * A Windows path in a verbatim string is not an edge case in C#.
+ *
+ * ONE scanner serves both "strip comments" and "strip comments and strings":
+ * C# has five string forms and keeping two copies of that lexer is how they
+ * drift. Both modes are LENGTH-PRESERVING and keep every CR/LF, so a finding's
+ * offset still maps to its native line with no remapping layer.
+ *
+ * Handled forms: line and block comments, char literals, regular ("..." with
+ * backslash escapes), verbatim (@"..." where "" is the escape and a backslash is
+ * literal), interpolated ($"...", $@"...", @$"..."), and C# 11 raw strings
+ * ("""...""", any delimiter length, $-interpolated included). Raw strings are the
+ * trap: scanned naively, a `"""` opens and closes on its own first two quotes and
+ * every following brace lands in the wrong nesting level.
+ */
+function transformCSharp(source, blankStrings) {
+    let result = '';
+    let i = 0;
+    const blank = (ch) => (ch === '\n' ? '\n' : ' ');
+    const emit = (ch) => { result += blankStrings ? blank(ch) : ch; };
+    while (i < source.length) {
+        const ch = source[i];
+        if (ch === '/' && source[i + 1] === '/') {
+            while (i < source.length && source[i] !== '\n') {
+                result += ' ';
+                i++;
+            }
+            continue;
+        }
+        if (ch === '/' && source[i + 1] === '*') {
+            result += '  ';
+            i += 2;
+            while (i < source.length) {
+                if (source[i] === '*' && source[i + 1] === '/') {
+                    result += '  ';
+                    i += 2;
+                    break;
+                }
+                result += blank(source[i]);
+                i++;
+            }
+            continue;
+        }
+        if (ch === '"' || ch === '$' || ch === '@') {
+            // Read any $ / @ prefix, then the run of quotes that opens the literal.
+            let j = i;
+            let verbatim = false;
+            while (source[j] === '$' || source[j] === '@') {
+                if (source[j] === '@')
+                    verbatim = true;
+                j++;
+            }
+            if (source[j] !== '"') {
+                // A bare $ or @: an escaped identifier (@class, @event) or an
+                // interpolation hole, neither of which opens a literal.
+                result += ch;
+                i++;
+                continue;
+            }
+            let quoteRun = 0;
+            while (source[j + quoteRun] === '"')
+                quoteRun++;
+            // Two quotes open an EMPTY regular/verbatim string, not a raw one: only a
+            // run of three or more is a raw-string delimiter.
+            const openLength = quoteRun >= 3 ? quoteRun : 1;
+            // Prefix and opening delimiter are never content, so they are blanked in
+            // code mode and preserved in comment-only mode (the SQL interpolation
+            // check reads the `$"` marker off that output).
+            for (let k = i; k < j + openLength; k++)
+                emit(source[k]);
+            i = j + openLength;
+            if (quoteRun >= 3) {
+                // Raw string: ends at the next run of at least `quoteRun` quotes.
+                while (i < source.length) {
+                    if (source[i] === '"') {
+                        let run = 0;
+                        while (source[i + run] === '"')
+                            run++;
+                        for (let k = 0; k < run; k++)
+                            emit('"');
+                        i += run;
+                        if (run >= quoteRun)
+                            break;
+                        continue;
+                    }
+                    emit(source[i]);
+                    i++;
+                }
+                continue;
+            }
+            if (verbatim) {
+                // Verbatim: "" is the escape, a backslash is literal, newlines are part
+                // of the string.
+                while (i < source.length) {
+                    if (source[i] === '"' && source[i + 1] === '"') {
+                        emit('"');
+                        emit('"');
+                        i += 2;
+                        continue;
+                    }
+                    if (source[i] === '"') {
+                        emit('"');
+                        i++;
+                        break;
+                    }
+                    emit(source[i]);
+                    i++;
+                }
+                continue;
+            }
+            // Regular or interpolated string. A newline ends it: the file is already
+            // broken at that point, and running to EOF would blank the rest of it.
+            while (i < source.length) {
+                const current = source[i];
+                if (current === '\n')
+                    break;
+                if (current === '\\') {
+                    emit(current);
+                    i++;
+                    if (i < source.length && source[i] !== '\n') {
+                        emit(source[i]);
+                        i++;
+                    }
+                    continue;
+                }
+                emit(current);
+                i++;
+                if (current === '"')
+                    break;
+            }
+            continue;
+        }
+        if (ch === "'") {
+            emit(ch);
+            i++;
+            while (i < source.length) {
+                const current = source[i];
+                if (current === '\n')
+                    break;
+                if (current === '\\') {
+                    emit(current);
+                    i++;
+                    if (i < source.length) {
+                        emit(source[i]);
+                        i++;
+                    }
+                    continue;
+                }
+                emit(current);
+                i++;
+                if (current === "'")
+                    break;
+            }
+            continue;
+        }
+        result += ch;
+        i++;
+    }
+    return result;
+}
+/** Comments blanked, string contents intact. */
+function stripCSharpComments(source) {
+    return transformCSharp(source, false);
+}
+/** Comments AND string contents blanked: the input for anything structural. */
+function csharpCodeOnly(source) {
+    return transformCSharp(source, true);
+}
+//# sourceMappingURL=lexer.js.map

package/dist/csharp/lexer.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"lexer.js","sourceRoot":"","sources":["../../src/csharp/lexer.ts"],"names":[],"mappings":";;AAuBA,0CAyHC;AAGD,kDAEC;AAGD,wCAEC;AA1JD;;;;;;;;;;;;;;;;;;;;;;GAsBG;AACH,SAAgB,eAAe,CAAC,MAAc,EAAE,YAAqB;IACnE,IAAI,MAAM,GAAG,EAAE,CAAA;IACf,IAAI,CAAC,GAAG,CAAC,CAAA;IACT,MAAM,KAAK,GAAG,CAAC,EAAU,EAAU,EAAE,CAAC,CAAC,EAAE,KAAK,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAA;IAChE,MAAM,IAAI,GAAG,CAAC,EAAU,EAAQ,EAAE,GAAG,MAAM,IAAI,YAAY,CAAC,CAAC,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,CAAA,CAAC,CAAC,CAAA;IAE9E,OAAO,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC;QACzB,MAAM,EAAE,GAAG,MAAM,CAAC,CAAC,CAAC,CAAA;QAEpB,IAAI,EAAE,KAAK,GAAG,IAAI,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;YACxC,OAAO,CAAC,GAAG,MAAM,CAAC,MAAM,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC;gBAAC,MAAM,IAAI,GAAG,CAAC;gBAAC,CAAC,EAAE,CAAA;YAAC,CAAC;YACtE,SAAQ;QACV,CAAC;QACD,IAAI,EAAE,KAAK,GAAG,IAAI,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;YACxC,MAAM,IAAI,IAAI,CAAA;YACd,CAAC,IAAI,CAAC,CAAA;YACN,OAAO,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC;gBACzB,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;oBAAC,MAAM,IAAI,IAAI,CAAC;oBAAC,CAAC,IAAI,CAAC,CAAC;oBAAC,MAAK;gBAAC,CAAC;gBACjF,MAAM,IAAI,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAA;gBAC1B,CAAC,EAAE,CAAA;YACL,CAAC;YACD,SAAQ;QACV,CAAC;QAED,IAAI,EAAE,KAAK,GAAG,IAAI,EAAE,KAAK,GAAG,IAAI,EAAE,KAAK,GAAG,EAAE,CAAC;YAC3C,wEAAwE;YACxE,IAAI,CAAC,GAAG,CAAC,CAAA;YACT,IAAI,QAAQ,GAAG,KAAK,CAAA;YACpB,OAAO,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;gBAC9C,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG;oBAAE,QAAQ,GAAG,IAAI,CAAA;gBACtC,CAAC,EAAE,CAAA;YACL,CAAC;YACD,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;gBACtB,8DAA8D;gBAC9D,wDAAwD;gBACxD,MAAM,IAAI,EAAE,CAAA;gBACZ,CAAC,EAAE,CAAA;gBACH,SAAQ;YACV,CAAC;YACD,IAAI,QAAQ,GAAG,CAAC,CAAA;YAChB,OAAO,MAAM,CAAC,CAAC,GAAG,QAAQ,CAAC,KAAK,GAAG;gBAAE,QAAQ,EAAE,CAAA;YAC/C,0EAA0E;YAC1E,kDAAkD;YAClD,MAAM,UAAU,GAAG,QAAQ,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAA;YAE/C,yEAAyE;YACzE,sEAAsE;YACtE,gDAAgD;YAChD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,GAAG,UAAU,EAAE,CAAC,EAAE;gBAAE,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAA;YACxD,CAAC,GAAG,CAAC,GAAG,UAAU,CAAA;YAElB,IAAI,QAAQ,IAAI,CAAC,EAAE,CAAC;gBAClB,kEAAkE;gBAClE,OAAO,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC;oBACzB,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;wBACtB,IAAI,GAAG,GAAG,CAAC,CAAA;wBACX,OAAO,MAAM,CAAC,CAAC,GAAG,GAAG,CAAC,KAAK,GAAG;4BAAE,GAAG,EAAE,CAAA;wBACrC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,EAAE,CAAC,EAAE;4BAAE,IAAI,CAAC,GAAG,CAAC,CAAA;wBACvC,CAAC,IAAI,GAAG,CAAA;wBACR,IAAI,GAAG,IAAI,QAAQ;4BAAE,MAAK;wBAC1B,SAAQ;oBACV,CAAC;oBACD,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAA;oBACf,CAAC,EAAE,CAAA;gBACL,CAAC;gBACD,SAAQ;YACV,CAAC;YAED,IAAI,QAAQ,EAAE,CAAC;gBACb,wEAAwE;gBACxE,iBAAiB;gBACjB,OAAO,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC;oBACzB,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;wBAAC,IAAI,CAAC,GAAG,CAAC,CAAC;wBAAC,IAAI,CAAC,GAAG,CAAC,CAAC;wBAAC,CAAC,IAAI,CAAC,CAAC;wBAAC,SAAQ;oBAAC,CAAC;oBAC1F,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,GAAG,EAAE,CAAC;wBAAC,IAAI,CAAC,GAAG,CAAC,CAAC;wBAAC,CAAC,EAAE,CAAC;wBAAC,MAAK;oBAAC,CAAC;oBAChD,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAA;oBACf,CAAC,EAAE,CAAA;gBACL,CAAC;gBACD,SAAQ;YACV,CAAC;YAED,yEAAyE;YACzE,uEAAuE;YACvE,OAAO,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC;gBACzB,MAAM,OAAO,GAAG,MAAM,CAAC,CAAC,CAAC,CAAA;gBACzB,IAAI,OAAO,KAAK,IAAI;oBAAE,MAAK;gBAC3B,IAAI,OAAO,KAAK,IAAI,EAAE,CAAC;oBACrB,IAAI,CAAC,OAAO,CAAC,CAAA;oBACb,CAAC,EAAE,CAAA;oBACH,IAAI,CAAC,GAAG,MAAM,CAAC,MAAM,IAAI,MAAM,CAAC,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC;wBAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;wBAAC,CAAC,EAAE,CAAA;oBAAC,CAAC;oBACrE,SAAQ;gBACV,CAAC;gBACD,IAAI,CAAC,OAAO,CAAC,CAAA;gBACb,CAAC,EAAE,CAAA;gBACH,IAAI,OAAO,KAAK,GAAG;oBAAE,MAAK;YAC5B,CAAC;YACD,SAAQ;QACV,CAAC;QAED,IAAI,EAAE,KAAK,GAAG,EAAE,CAAC;YACf,IAAI,CAAC,EAAE,CAAC,CAAA;YACR,CAAC,EAAE,CAAA;YACH,OAAO,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC;gBACzB,MAAM,OAAO,GAAG,MAAM,CAAC,CAAC,CAAC,CAAA;gBACzB,IAAI,OAAO,KAAK,IAAI;oBAAE,MAAK;gBAC3B,IAAI,OAAO,KAAK,IAAI,EAAE,CAAC;oBACrB,IAAI,CAAC,OAAO,CAAC,CAAA;oBACb,CAAC,EAAE,CAAA;oBACH,IAAI,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC;wBAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;wBAAC,CAAC,EAAE,CAAA;oBAAC,CAAC;oBAC/C,SAAQ;gBACV,CAAC;gBACD,IAAI,CAAC,OAAO,CAAC,CAAA;gBACb,CAAC,EAAE,CAAA;gBACH,IAAI,OAAO,KAAK,GAAG;oBAAE,MAAK;YAC5B,CAAC;YACD,SAAQ;QACV,CAAC;QAED,MAAM,IAAI,EAAE,CAAA;QACZ,CAAC,EAAE,CAAA;IACL,CAAC;IACD,OAAO,MAAM,CAAA;AACf,CAAC;AAED,gDAAgD;AAChD,SAAgB,mBAAmB,CAAC,MAAc;IAChD,OAAO,eAAe,CAAC,MAAM,EAAE,KAAK,CAAC,CAAA;AACvC,CAAC;AAED,+EAA+E;AAC/E,SAAgB,cAAc,CAAC,MAAc;IAC3C,OAAO,eAAe,CAAC,MAAM,EAAE,IAAI,CAAC,CAAA;AACtC,CAAC"}

package/dist/duplicates.d.ts

@@ -0,0 +1,53 @@
+import { DuplicateBlockFinding } from './types';
+export interface DuplicateLocation {
+    file: string;
+    startLine: number;
+    endLine: number;
+}
+export interface DuplicateBlock {
+    /** length of the matched run in normalised tokens */
+    tokens: number;
+    a: DuplicateLocation;
+    b: DuplicateLocation;
+}
+export interface DuplicateFileInput {
+    path: string;
+    content: string;
+}
+export interface DuplicateDetectionResult {
+    blocks: DuplicateBlock[];
+    /** files excluded from comparison (too large / minified) — never silently dropped */
+    skippedFiles: string[];
+}
+/** Files above this line count are excluded from duplicate comparison */
+export declare const DUPLICATE_MAX_FILE_LINES = 5000;
+/** A single line longer than this marks the file as generated/minified and excludes it */
+export declare const DUPLICATE_MAX_LINE_LENGTH = 500;
+/** Hard cap on reported blocks per scan (longest first) so boilerplate-heavy repos stay usable */
+export declare const DUPLICATE_MAX_BLOCKS = 500;
+interface Token {
+    norm: string;
+    line: number;
+}
+/** Lex source into normalised tokens with original line numbers. Never throws. */
+export declare function tokeniseNormalised(source: string, filePath: string): Token[];
+/**
+ * Blanks import/re-export statements so they never participate in matching
+ * (newlines are preserved, so reported line numbers stay exact). Import
+ * headers are structural boilerplate - two Next.js pages importing the same
+ * twenty UI components are not copy-paste debt, and established duplication
+ * engines (e.g. SonarQube's CPD) exclude imports for the same reason.
+ */
+export declare function stripImports(source: string, filePath: string): string;
+export declare function detectDuplicates(files: DuplicateFileInput[], minTokens?: number): DuplicateDetectionResult;
+/**
+ * Converts a detected block into per-file smell findings. A within-file block
+ * yields one finding; a cross-file block yields a primary finding on the first
+ * file and a non-primary mirror on the second, so totals never double-count.
+ */
+export declare function duplicateBlockToFindings(block: DuplicateBlock): Array<{
+    file: string;
+    finding: DuplicateBlockFinding;
+}>;
+export {};
+//# sourceMappingURL=duplicates.d.ts.map

package/dist/duplicates.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"duplicates.d.ts","sourceRoot":"","sources":["../src/duplicates.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,qBAAqB,EAAE,MAAM,SAAS,CAAA;AAc/C,MAAM,WAAW,iBAAiB;IAChC,IAAI,EAAE,MAAM,CAAA;IACZ,SAAS,EAAE,MAAM,CAAA;IACjB,OAAO,EAAE,MAAM,CAAA;CAChB;AAED,MAAM,WAAW,cAAc;IAC7B,qDAAqD;IACrD,MAAM,EAAE,MAAM,CAAA;IACd,CAAC,EAAE,iBAAiB,CAAA;IACpB,CAAC,EAAE,iBAAiB,CAAA;CACrB;AAED,MAAM,WAAW,kBAAkB;IACjC,IAAI,EAAE,MAAM,CAAA;IACZ,OAAO,EAAE,MAAM,CAAA;CAChB;AAED,MAAM,WAAW,wBAAwB;IACvC,MAAM,EAAE,cAAc,EAAE,CAAA;IACxB,qFAAqF;IACrF,YAAY,EAAE,MAAM,EAAE,CAAA;CACvB;AAED,yEAAyE;AACzE,eAAO,MAAM,wBAAwB,OAAO,CAAA;AAC5C,0FAA0F;AAC1F,eAAO,MAAM,yBAAyB,MAAM,CAAA;AAC5C,kGAAkG;AAClG,eAAO,MAAM,oBAAoB,MAAM,CAAA;AAyBvC,UAAU,KAAK;IACb,IAAI,EAAE,MAAM,CAAA;IACZ,IAAI,EAAE,MAAM,CAAA;CACb;AAUD,kFAAkF;AAClF,wBAAgB,kBAAkB,CAAC,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,GAAG,KAAK,EAAE,CAiG5E;AAED;;;;;;GAMG;AACH,wBAAgB,YAAY,CAAC,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,GAAG,MAAM,CAuFrE;AAmCD,wBAAgB,gBAAgB,CAAC,KAAK,EAAE,kBAAkB,EAAE,EAAE,SAAS,SAAK,GAAG,wBAAwB,CA6KtG;AAoCD;;;;GAIG;AACH,wBAAgB,wBAAwB,CAAC,KAAK,EAAE,cAAc,GAAG,KAAK,CAAC;IAAE,IAAI,EAAE,MAAM,CAAC;IAAC,OAAO,EAAE,qBAAqB,CAAA;CAAE,CAAC,CA6BvH"}