@contentrain/mcp 1.1.2 → 1.2.1

package/dist/{scanner-DPrbU-oK.mjs → scanner-DUoXnmtI.mjs}

@@ -1,73 +1,36 @@
 import { o as readText } from "./fs-BnIQ4dKj.mjs";
 import { i as autoDetectSourceDirs, o as discoverFiles } from "./scan-config-QgnJ1-52.mjs";
-import { t as parseTsx } from "./tsx-parser-CYVyBWjn.mjs";
+import { t as parseTsx } from "./tsx-parser-BWiMGAIR.mjs";
 import { extname, join } from "node:path";
 //#region src/core/ast-scanner/pre-filter.ts
-const PURE_NUMBER_RE = /^\d+(\.\d+)?$/;
+const PURE_NUMBER_RE = /^-?\d+(\.\d+)?$/;
 const HEX_COLOR_RE = /^#[0-9a-f]{3,8}$/i;
-const FILE_EXT_RE$1 = /\.(png|jpg|jpeg|gif|svg|webp|ico|css|scss|less|js|ts|tsx|jsx|json|md|html|xml|yaml|yml|woff|woff2|ttf|eot|mp4|webm|mp3|wav|pdf)$/i;
-const HTML_PROP_VALUES = new Set([
-	"class",
-	"variant",
-	"secondary",
-	"outline",
-	"ghost",
-	"destructive",
-	"default",
-	"primary",
+const FILE_EXT_RE = /\.(png|jpg|jpeg|gif|svg|webp|ico|css|scss|less|js|ts|tsx|jsx|json|md|html|xml|yaml|yml|woff|woff2|ttf|eot|mp4|webm|mp3|wav|pdf)$/i;
+const SVG_PATH_DATA_RE = /^[Mm][\d\s.,LHVCSQTAZlhvcsqtazmMzZ-]+$/;
+const SVG_VIEWBOX_RE = /^\d+(\.\d+)?\s+\d+(\.\d+)?\s+\d+(\.\d+)?\s+\d+(\.\d+)?$/;
+const I18N_KEY_RE = /^[a-z][a-z0-9_]*(\.[a-z][a-z0-9_]*)+$/;
+const TECHNICAL_IDENTIFIER_RE = /^[_a-z][a-z0-9_-]*$/;
+const ERROR_CODE_RE = /^[A-Z][A-Z0-9_]+$/;
+const PLACEHOLDER_RE = /^\{\d+\}$|^\.{2,}$/;
+const CAMEL_CASE_RE = /^[a-z]+[A-Z]/;
+const LOCALE_CODE_RE = /^[a-z]{2}[-_][A-Z]{2}$/;
+const DIMENSION_RE = /^\d+[x×]\d+$/;
+const REPEAT_CHAR_RE = /^(.)\1{3,}$/;
+const MIME_TYPE_RE = /^(application|text|image|audio|video|multipart|font)\/[\w.+-]+$/;
+const PASCAL_CASE_RE = /^[A-Z][a-z]+[A-Z]/;
+const HTML_TARGETS = new Set([
 	"_blank",
-	"noopener",
-	"noreferrer",
-	"button",
-	"submit",
-	"reset",
-	"text",
-	"numeric",
-	"password",
-	"email",
-	"checkbox",
-	"radio",
-	"hidden",
-	"none",
-	"auto",
-	"inherit",
-	"initial",
-	"sm",
-	"md",
-	"lg",
-	"xl",
-	"xs",
-	"2xl",
-	"3xl",
-	"icon",
-	"icon-sm",
-	"icon-lg"
-]);
-const SLOT_EVENT_NAMES = new Set([
-	"header",
-	"footer",
-	"default",
-	"trigger",
-	"content",
-	"sidebar",
-	"overlay",
-	"body",
-	"actions",
-	"click",
-	"change",
-	"input",
-	"focus",
-	"blur",
-	"submit",
-	"mounted",
-	"unmounted",
-	"updated"
+	"_self",
+	"_parent",
+	"_top"
 ]);
+function isURLLike(str) {
+	if (/^(https?|ftp|file|mailto|data):/.test(str)) return true;
+	if (/^(\.\.?\/|\/|[A-Za-z]:\\)/.test(str)) return true;
+	if (/^['"]?[@a-z][\w-]*/.test(str.toLowerCase()) && !str.includes(" ") && (str.includes("/") || str.includes("."))) return true;
+	return false;
+}
 const TAILWIND_SEGMENT_RE = /^(?:bg-|text-|border-|flex|grid|p-|px-|py-|pt-|pb-|pl-|pr-|m-|mx-|my-|mt-|mb-|ml-|mr-|rounded|shadow|w-|h-|min-|max-|gap-|space-|items-|justify-|self-|overflow-|z-|opacity-|transition|duration-|ease-|animate-|font-|leading-|tracking-|decoration-|underline|line-through|uppercase|lowercase|capitalize|truncate|whitespace-|break-|sr-only|not-sr-only|hover:|focus:|active:|disabled:|dark:|sm:|md:|lg:|xl:|2xl:|group-|peer-|ring-|outline-|divide-|table-|col-|row-|aspect-|object-|inset-|top-|right-|bottom-|left-|translate-|rotate-|scale-|skew-|origin-|cursor-|select-|resize-|fill-|stroke-|block|inline|absolute|relative|fixed|sticky|static|float-|clear-|isolate|visible|invisible|grow|shrink|basis-|order-|place-)/;
-/**
-* Returns true if the string looks like a CSS class list (Tailwind or similar).
-* Requires 2+ space-separated segments with majority matching utility patterns.
-*/
 function isCssClassList(value) {
 	const segments = value.trim().split(/\s+/);
 	if (segments.length < 2) return false;
@@ -75,116 +38,223 @@ function isCssClassList(value) {
 	for (const seg of segments) if (TAILWIND_SEGMENT_RE.test(seg)) matched++;
 	return matched / segments.length >= .5;
 }
-/**
-* Returns true if the string is a single CSS utility token (e.g. "bg-blue-500").
-*/
 function isSingleCssUtility(value) {
 	const trimmed = value.trim();
 	if (trimmed.includes(" ")) return false;
 	return TAILWIND_SEGMENT_RE.test(trimmed);
 }
-const PRE_FILTER_RULES = [
-	{
-		context: "import_path",
-		reason: "import_path"
-	},
-	{
-		context: "type_annotation",
-		reason: "type_annotation"
-	},
-	{
-		context: "css_class",
-		reason: "css_class"
-	},
-	{
-		context: "css_utility_call",
-		reason: "css_utility_call"
-	},
-	{
-		context: "console_call",
-		reason: "console_call"
-	},
-	{
-		context: "test_assertion",
-		reason: "test_assertion"
-	},
-	{
-		context: "switch_case",
-		reason: "switch_case"
-	},
-	{
-		condition: (v) => v.length <= 1,
-		reason: "single_char"
-	},
-	{
-		condition: (v) => PURE_NUMBER_RE.test(v),
-		reason: "pure_number"
-	},
-	{
-		condition: (v) => v.startsWith("--"),
-		reason: "cli_flag"
-	},
-	{
-		condition: (v) => HEX_COLOR_RE.test(v),
-		reason: "hex_color"
-	},
-	{
-		condition: (v) => FILE_EXT_RE$1.test(v),
-		reason: "file_extension"
-	},
-	{
-		condition: (v) => HTML_PROP_VALUES.has(v.toLowerCase()),
-		reason: "html_prop_value"
-	},
-	{
-		condition: (v) => isCssClassList(v),
-		reason: "css_class_list"
-	},
-	{
-		condition: (v) => isSingleCssUtility(v),
-		reason: "css_utility_token"
-	},
-	{
-		condition: (v) => {
-			const lower = v.toLowerCase();
-			return v === lower && !v.includes(" ") && SLOT_EVENT_NAMES.has(lower);
-		},
-		reason: "slot_event_name"
+const SVG_TECHNICAL_ATTRIBUTES = new Set([
+	"d",
+	"viewBox",
+	"points",
+	"transform",
+	"pathLength",
+	"xmlns",
+	"preserveAspectRatio",
+	"stroke-linecap",
+	"stroke-linejoin",
+	"stroke-width",
+	"stroke-dasharray",
+	"stroke-dashoffset",
+	"stroke-miterlimit",
+	"fill-rule",
+	"clip-rule"
+]);
+const SVG_GRAPHIC_ELEMENTS = new Set([
+	"svg",
+	"path",
+	"circle",
+	"rect",
+	"line",
+	"polyline",
+	"polygon",
+	"ellipse",
+	"g",
+	"defs",
+	"use",
+	"symbol",
+	"clipPath",
+	"mask",
+	"pattern",
+	"linearGradient",
+	"radialGradient",
+	"stop",
+	"marker",
+	"animate",
+	"animateTransform",
+	"image"
+]);
+const I18N_FUNCTIONS = new Set([
+	"t",
+	"$t",
+	"i18n",
+	"translate",
+	"formatMessage",
+	"msg"
+]);
+const EMIT_FUNCTIONS = new Set(["emit", "$emit"]);
+const TRANSLATABLE_ATTRIBUTES = new Set([
+	"title",
+	"alt",
+	"placeholder",
+	"label",
+	"summary",
+	"caption",
+	"abbr",
+	"accesskey",
+	"content",
+	"description",
+	"aria-label",
+	"aria-description",
+	"aria-placeholder",
+	"aria-roledescription",
+	"aria-valuetext",
+	"accessibilityLabel",
+	"accessibilityHint",
+	"accessibilityValue",
+	"heading",
+	"subheading",
+	"message",
+	"hint",
+	"tooltip",
+	"helper-text",
+	"error-message",
+	"success-message",
+	"confirm-text",
+	"cancel-text",
+	"empty-text",
+	"loading-text",
+	"no-data-text",
+	"no-results-text"
+]);
+const TRANSLATABLE_PROPERTIES = new Set([
+	"label",
+	"title",
+	"description",
+	"text",
+	"message",
+	"placeholder",
+	"caption",
+	"summary",
+	"heading",
+	"subheading",
+	"subtitle",
+	"tooltip",
+	"hint",
+	"helpText",
+	"errorMessage",
+	"successMessage",
+	"name"
+]);
+/**
+* Determines if a string is definitely NOT user-visible content.
+* Returns skip reason if it should be filtered, null if it should proceed to scoring.
+*
+* Conservative for template_text/jsx_text (tag-between text is almost always content).
+* Aggressive for everything else (technical tokens, config values, framework artifacts).
+*/
+function shouldSkip(str) {
+	if (str.context === "import_path") return "import_path";
+	if (str.context === "type_annotation") return "type_annotation";
+	if (str.context === "css_class") return "css_class";
+	if (str.context === "css_utility_call") return "css_utility_call";
+	if (str.context === "console_call") return "console_call";
+	if (str.context === "test_assertion") return "test_assertion";
+	if (str.context === "switch_case") return "switch_case";
+	const v = str.value;
+	if (v.length <= 1) return "single_char";
+	if (/^\s+$/.test(v)) return "whitespace";
+	if (PURE_NUMBER_RE.test(v)) return "pure_number";
+	if (HEX_COLOR_RE.test(v)) return "hex_color";
+	if (FILE_EXT_RE.test(v)) return "file_extension";
+	if (v.startsWith("--")) return "cli_flag";
+	if (I18N_KEY_RE.test(v)) return "i18n_key";
+	if (MIME_TYPE_RE.test(v)) return "mime_type";
+	if (isURLLike(v)) return "url_path";
+	if (isCssClassList(v)) return "css_class_list";
+	if (isSingleCssUtility(v)) return "css_utility_token";
+	if (v.length > 3 && SVG_PATH_DATA_RE.test(v)) return "svg_path_data";
+	if (SVG_VIEWBOX_RE.test(v)) return "svg_viewbox";
+	if (str.parentProperty !== void 0 && SVG_TECHNICAL_ATTRIBUTES.has(str.parentProperty)) return "svg_technical_attr";
+	if (str.context === "template_attribute" && SVG_GRAPHIC_ELEMENTS.has(str.parent)) return "svg_element_attr";
+	if (v.startsWith("update:")) return "vue_emit_event";
+	if (PLACEHOLDER_RE.test(v)) return "placeholder";
+	if (LOCALE_CODE_RE.test(v)) return "locale_code";
+	if (DIMENSION_RE.test(v)) return "dimension";
+	if (REPEAT_CHAR_RE.test(v)) return "repeat_chars";
+	if (HTML_TARGETS.has(v)) return "html_target";
+	if (str.context === "function_argument" && I18N_FUNCTIONS.has(str.parent)) {
+		if (/^[a-z][a-z0-9_.-]*$/.test(v)) return "i18n_function_arg";
+	}
+	if (str.context === "function_argument" && EMIT_FUNCTIONS.has(str.parent)) return "emit_event_arg";
+	if (str.context !== "template_text" && str.context !== "jsx_text") {
+		if (TECHNICAL_IDENTIFIER_RE.test(v) && v.length < 30) return "technical_identifier";
+	}
+	if (ERROR_CODE_RE.test(v) && v.includes("_") && v.length > 3) return "error_code";
+	return null;
+}
+/**
+* Calculates a content confidence score (0-1) for a string that passed shouldSkip.
+* Uses AST context metadata (our advantage over offset-based tools) combined with
+* value-based signals proven by i18next-cli.
+*
+* Base score: 0.5. Boosted/penalized by context and value characteristics.
+*/
+function calculateContentScore(str) {
+	let score = .5;
+	if (str.context === "template_text" || str.context === "jsx_text") score += .3;
+	if (str.context === "template_attribute" || str.context === "jsx_attribute") if (str.parentProperty && TRANSLATABLE_ATTRIBUTES.has(str.parentProperty)) score += .2;
+	else score -= .2;
+	if (str.context === "object_property") {
+		if (str.parentProperty && TRANSLATABLE_PROPERTIES.has(str.parentProperty)) score += .25;
 	}
-];
+	const wordCount = str.value.split(/\s+/).length;
+	if (wordCount >= 3) score += .2;
+	else if (wordCount === 2) score += .1;
+	if (/[.!?:;]$/.test(str.value)) score += .1;
+	if (/[\u0080-\uFFFF]/.test(str.value)) score += .15;
+	if (/^[A-Z]/.test(str.value) && /[a-z]/.test(str.value)) score += .1;
+	if (CAMEL_CASE_RE.test(str.value)) score -= .3;
+	if (PASCAL_CASE_RE.test(str.value) && !str.value.includes(" ")) score -= .25;
+	if (/^[A-Z]{2,5}$/.test(str.value)) score -= .15;
+	if (str.value.includes("/") && !str.value.includes(" ")) score -= .2;
+	return Math.max(0, Math.min(1, score));
+}
 /**
-* Structural pre-filter: removes strings that are 100% NOT content.
+* Two-phase pre-filter:
+* 1. shouldSkip(): Binary removal of definite non-content
+* 2. calculateContentScore(): 0-1 confidence scoring for ambiguous strings
 *
-* Conservative — when in doubt, INCLUDE the string.
-* Returns candidates (passed), count of filtered, and reason breakdown.
+* Returns candidates that passed both phases, with content scores attached.
 */
-function applyPreFilter(strings) {
+function applyPreFilter(strings, minScore = .4) {
 	const candidates = [];
-	const filterReasons = {};
-	let filtered = 0;
+	const skipReasons = {};
+	let skipped = 0;
+	let lowConfidence = 0;
 	for (const str of strings) {
-		const matchedRule = findMatchingRule(str);
-		if (matchedRule) {
-			filtered++;
-			filterReasons[matchedRule.reason] = (filterReasons[matchedRule.reason] ?? 0) + 1;
-		} else candidates.push(str);
+		const skipReason = shouldSkip(str);
+		if (skipReason) {
+			skipped++;
+			skipReasons[skipReason] = (skipReasons[skipReason] ?? 0) + 1;
+			continue;
+		}
+		const contentScore = calculateContentScore(str);
+		if (contentScore < minScore) {
+			lowConfidence++;
+			skipReasons["low_confidence"] = (skipReasons["low_confidence"] ?? 0) + 1;
+			continue;
+		}
+		str.contentScore = contentScore;
+		candidates.push(str);
 	}
 	return {
 		candidates,
-		filtered,
-		filterReasons
+		skipped,
+		lowConfidence,
+		skipReasons
 	};
 }
-/**
-* Find the first pre-filter rule that matches this string.
-* Returns the rule if matched, undefined if the string should pass through.
-*/
-function findMatchingRule(str) {
-	for (const rule of PRE_FILTER_RULES) {
-		if (rule.context !== void 0 && str.context === rule.context) return rule;
-		if (rule.condition !== void 0 && rule.condition(str.value)) return rule;
-	}
-}
 //#endregion
 //#region src/core/ast-scanner/index.ts
 const TSX_EXTENSIONS = new Set([
@@ -203,7 +273,7 @@ const ASTRO_EXTENSIONS = new Set([".astro"]);
 */
 async function loadVueParser() {
 	try {
-		return (await import("./vue-parser-CjS4YAoQ.mjs")).parseVue;
+		return (await import("./vue-parser-DTPTN3Yb.mjs")).parseVue;
 	} catch {
 		return;
 	}
@@ -214,7 +284,7 @@ async function loadVueParser() {
 */
 async function loadSvelteParser() {
 	try {
-		return (await import("./svelte-parser-COTNG92w.mjs")).parseSvelte;
+		return (await import("./svelte-parser-1bBcMx3T.mjs")).parseSvelte;
 	} catch {
 		return;
 	}
@@ -225,7 +295,7 @@ async function loadSvelteParser() {
 */
 async function loadAstroParser() {
 	try {
-		return (await import("./astro-parser-RofKmKQi.mjs")).parseAstro;
+		return (await import("./astro-parser-BJ-54JGe.mjs")).parseAstro;
 	} catch {
 		return;
 	}
@@ -343,6 +413,7 @@ const DEFAULT_LIMIT = 50;
 const DEFAULT_OFFSET = 0;
 const DEFAULT_MIN_LENGTH = 2;
 const DEFAULT_MAX_LENGTH = 500;
+const DEFAULT_MIN_SCORE = .4;
 const SUMMARY_SAMPLE_SIZE = 10;
 const CONTEXT_MAP = {
 	"template_text": "template_text",
@@ -364,142 +435,17 @@ const CONTEXT_MAP = {
 	"console_call": "other",
 	"test_assertion": "other"
 };
-const IMPORT_PATH_RE = /^(\.\/|\.\.\/|@\/|@[a-z]|[a-z][a-z0-9-]*\/)/;
-const CSS_CLASS_PREFIX_RE = /^(flex|grid|block|inline|hidden|absolute|relative|fixed|sticky|p-|m-|w-|h-|bg-|text-|border-|rounded|shadow|gap-|space-|font-|leading-|tracking-|opacity-|z-|overflow-|cursor-|transition|transform|animate-|hover:|focus:|active:|dark:|sm:|md:|lg:|xl:)/;
-const URL_ROUTE_RE = /^(https?:\/\/|\/api\/|\/v[0-9]|#\/|mailto:|tel:)/;
-const ROUTE_LIKE_RE = /^\/[a-z0-9/:_-]*$/i;
-const COLOR_RE = /^(#[0-9a-f]{3,8}|rgba?\(|hsla?\(|transparent|inherit|currentColor)$/i;
-const CAMEL_CASE_RE = /^[a-z][a-zA-Z0-9]*$/;
-const SCREAMING_SNAKE_RE = /^[A-Z][A-Z0-9_]+$/;
-const KEBAB_TECH_RE = /^[a-z][a-z0-9]*(-[a-z0-9]+)+$/;
-const FILE_EXT_RE = /\.(png|jpg|jpeg|gif|svg|webp|ico|css|scss|less|js|ts|tsx|jsx|json|md|html|xml|yaml|yml|woff|woff2|ttf|eot|mp4|webm|mp3|wav|pdf)$/i;
-const REGEX_LIKE_RE = /^\^|^\(.*\)$|\$$|\\d|\\w|\\s|\.\*|\.\+/;
-const CONSOLE_RE = /console\.(log|warn|error|debug|info|trace)\s*\(/;
-const CONTENT_WORDS = new Set([
-	"sign",
-	"up",
-	"log",
-	"in",
-	"out",
-	"get",
-	"started",
-	"learn",
-	"more",
-	"read",
-	"click",
-	"here",
-	"go",
-	"back",
-	"next",
-	"prev",
-	"previous",
-	"see",
-	"all",
-	"view",
-	"show",
-	"hide",
-	"open",
-	"close",
-	"add",
-	"new",
-	"save",
-	"edit",
-	"delete",
-	"remove",
-	"submit",
-	"cancel",
-	"confirm",
-	"welcome",
-	"hello",
-	"thank",
-	"you",
-	"good",
-	"bye",
-	"hi",
-	"hey",
-	"no",
-	"yes",
-	"ok",
-	"sorry",
-	"please",
-	"help",
-	"about",
-	"contact",
-	"home",
-	"page",
-	"not",
-	"found",
-	"error",
-	"success",
-	"loading"
-]);
-/**
-* Determines if a string is non-content (technical) and should be filtered out.
-* Pure function — no side effects, no semantic decisions.
-* Conservative: if in doubt, returns false (keeps the string).
-*
-* @deprecated Kept for backward compatibility. Scanner v2 uses structural pre-filter instead.
-*/
-function isNonContent(value, lineContext, minLength = DEFAULT_MIN_LENGTH, maxLength = DEFAULT_MAX_LENGTH) {
-	if (value.length < minLength) return true;
-	if (value.length > maxLength) return true;
-	if (IMPORT_PATH_RE.test(value) && !value.includes(" ")) return true;
-	if (CSS_CLASS_PREFIX_RE.test(value) && !value.includes(" ")) return true;
-	if (value.includes(" ") && looksLikeCssClasses(value)) return true;
-	if (URL_ROUTE_RE.test(value)) return true;
-	if (value.startsWith("/") && ROUTE_LIKE_RE.test(value)) return true;
-	if (COLOR_RE.test(value)) return true;
-	if (isTechnicalIdentifier(value)) return true;
-	if (FILE_EXT_RE.test(value)) return true;
-	if (REGEX_LIKE_RE.test(value)) return true;
-	if (CONSOLE_RE.test(lineContext)) return true;
-	return false;
-}
-function isTechnicalIdentifier(value) {
-	if (value.includes(" ")) return false;
-	if (CAMEL_CASE_RE.test(value) && /[A-Z]/.test(value)) return true;
-	if (SCREAMING_SNAKE_RE.test(value) && value.includes("_")) return true;
-	if (KEBAB_TECH_RE.test(value)) {
-		if (!value.split("-").some((p) => CONTENT_WORDS.has(p))) return true;
-	}
-	if (/^[a-z][a-z0-9-]*$/.test(value) && value.length < 40 && !value.includes(" ")) {
-		if (value.includes("/")) return true;
-	}
-	return false;
-}
-function looksLikeCssClasses(value) {
-	const tokens = value.split(/\s+/);
-	if (tokens.length < 2) return false;
-	return tokens.filter((t) => isCssClassToken(t)).length === tokens.length;
-}
-function isCssClassToken(token) {
-	if (CSS_CLASS_PREFIX_RE.test(token)) return true;
-	if (/^(flex|grid|block|inline|hidden|container|relative|absolute|fixed|sticky|static|isolate|overflow-hidden|overflow-auto|overflow-scroll|underline|italic|uppercase|lowercase|capitalize|truncate|break-words|break-all|whitespace-nowrap|sr-only|not-sr-only)$/.test(token)) return true;
-	if (/^[a-z]+-\[.+\]$/.test(token)) return true;
-	if (/^(sm|md|lg|xl|2xl|hover|focus|active|disabled|first|last|odd|even|dark|group-hover|peer-checked):/.test(token)) return true;
-	return false;
-}
-/**
-* Apply legacy isNonContent filter as a secondary pass on v2 candidates.
-* This catches heuristic patterns (URLs, regex, camelCase identifiers, etc.)
-* that the structural pre-filter does not cover.
-*/
-function applySecondaryFilter(value, _surrounding, minLength, maxLength) {
-	return isNonContent(value, "", minLength, maxLength);
-}
 async function scanCandidates(projectRoot, options) {
 	const limit = options?.limit ?? DEFAULT_LIMIT;
 	const offset = options?.offset ?? DEFAULT_OFFSET;
 	const minLength = options?.min_length ?? DEFAULT_MIN_LENGTH;
 	const maxLength = options?.max_length ?? DEFAULT_MAX_LENGTH;
+	const minScore = options?.min_score ?? DEFAULT_MIN_SCORE;
 	const files = await discoverFiles(projectRoot, {
 		paths: options?.paths ?? await autoDetectSourceDirs(projectRoot),
 		include: options?.include,
 		exclude: options?.exclude
 	});
-	const allCandidates = [];
-	const dupeMap = /* @__PURE__ */ new Map();
-	let rawStringsFound = 0;
 	const filePromises = files.map(async (relPath) => {
 		const filePath = join(projectRoot, relPath);
 		const content = await readText(filePath);
@@ -513,51 +459,89 @@ async function scanCandidates(projectRoot, options) {
 		};
 	});
 	const fileResults = await Promise.all(filePromises);
+	let rawStringsFound = 0;
+	let skippedCount = 0;
+	let lowConfidenceCount = 0;
+	const skipReasons = {};
+	const uniqueMap = /* @__PURE__ */ new Map();
+	const dupeMap = /* @__PURE__ */ new Map();
 	for (const { relPath, extractions } of fileResults) {
 		rawStringsFound += extractions.length;
 		for (const extraction of extractions) {
-			if (applySecondaryFilter(extraction.value, extraction.surrounding, minLength, maxLength)) continue;
+			if (extraction.value.length < minLength || extraction.value.length > maxLength) {
+				skippedCount++;
+				skipReasons["length_filter"] = (skipReasons["length_filter"] ?? 0) + 1;
+				continue;
+			}
+			const contentScore = extraction.contentScore ?? calculateContentScore(extraction);
+			if (contentScore < minScore) {
+				lowConfidenceCount++;
+				skipReasons["low_confidence"] = (skipReasons["low_confidence"] ?? 0) + 1;
+				continue;
+			}
 			const mappedContext = CONTEXT_MAP[extraction.context];
-			const candidate = {
-				file: relPath,
-				line: extraction.line,
-				column: extraction.column,
-				value: extraction.value,
-				context: mappedContext,
-				surrounding: extraction.surrounding
-			};
-			allCandidates.push(candidate);
-			const key = extraction.value;
-			if (!dupeMap.has(key)) dupeMap.set(key, []);
-			dupeMap.get(key).push({
+			const loc = {
 				file: relPath,
 				line: extraction.line
+			};
+			if (!dupeMap.has(extraction.value)) dupeMap.set(extraction.value, []);
+			dupeMap.get(extraction.value).push(loc);
+			if (!uniqueMap.has(extraction.value)) uniqueMap.set(extraction.value, {
+				candidate: {
+					file: relPath,
+					line: extraction.line,
+					column: extraction.column,
+					value: extraction.value,
+					context: mappedContext,
+					surrounding: extraction.surrounding,
+					contentScore,
+					occurrences: [loc]
+				},
+				maxScore: contentScore
 			});
+			else {
+				const entry = uniqueMap.get(extraction.value);
+				entry.candidate.occurrences.push(loc);
+				if (contentScore > entry.maxScore) {
+					entry.maxScore = contentScore;
+					entry.candidate.contentScore = contentScore;
+					entry.candidate.file = relPath;
+					entry.candidate.line = extraction.line;
+					entry.candidate.column = extraction.column;
+					entry.candidate.context = mappedContext;
+					entry.candidate.surrounding = extraction.surrounding;
+				}
+			}
 		}
 	}
+	const allUniqueCandidates = [...uniqueMap.values()].map((e) => e.candidate).toSorted((a, b) => b.contentScore - a.contentScore);
 	const duplicates = [...dupeMap.entries()].filter(([, occurrences]) => occurrences.length >= 2).map(([value, occurrences]) => ({
 		value,
 		count: occurrences.length,
 		occurrences
 	})).toSorted((a, b) => b.count - a.count);
-	const afterFiltering = allCandidates.length;
-	const paginated = allCandidates.slice(offset, offset + limit);
-	const hasMore = afterFiltering > offset + limit;
+	const uniqueCount = allUniqueCandidates.length;
+	const paginated = allUniqueCandidates.slice(offset, offset + limit);
+	const hasMore = uniqueCount > offset + limit;
 	return {
 		candidates: paginated,
 		duplicates,
 		stats: {
 			files_scanned: files.length,
 			raw_strings_found: rawStringsFound,
-			after_filtering: afterFiltering,
+			skipped: skippedCount,
+			low_confidence: lowConfidenceCount,
+			unique_candidates: uniqueCount,
 			candidates_returned: paginated.length,
-			has_more: hasMore
+			has_more: hasMore,
+			skip_reasons: skipReasons
 		}
 	};
 }
 async function scanSummary(projectRoot, options) {
 	const minLength = options?.min_length ?? DEFAULT_MIN_LENGTH;
 	const maxLength = options?.max_length ?? DEFAULT_MAX_LENGTH;
+	const minScore = options?.min_score ?? DEFAULT_MIN_SCORE;
 	const files = await discoverFiles(projectRoot, {
 		paths: options?.paths ?? await autoDetectSourceDirs(projectRoot),
 		include: options?.include,
@@ -587,7 +571,8 @@ async function scanSummary(projectRoot, options) {
 		});
 		const sampleResults = await Promise.all(samplePromises);
 		for (const extractions of sampleResults) for (const extraction of extractions) {
-			if (applySecondaryFilter(extraction.value, extraction.surrounding, minLength, maxLength)) continue;
+			if (extraction.value.length < minLength || extraction.value.length > maxLength) continue;
+			if ((extraction.contentScore ?? calculateContentScore(extraction)) < minScore) continue;
 			sampleCandidates++;
 			const prev = freqMap.get(extraction.value) ?? 0;
 			freqMap.set(extraction.value, prev + 1);
@@ -614,6 +599,6 @@ async function scanSummary(projectRoot, options) {
 	};
 }
 //#endregion
-export { scanCandidates as n, scanSummary as r, isNonContent as t };
+export { scanSummary as n, scanCandidates as t };
 
-//# sourceMappingURL=scanner-DPrbU-oK.mjs.map
+//# sourceMappingURL=scanner-DUoXnmtI.mjs.map