@unerr-ai/unerr 0.1.0 → 0.1.2

package/dist/proxy/shell-strategies/error-diagnostic.js

@@ -0,0 +1,796 @@
+/**
+ * Strategy T7 — format-aware error diagnostic compression.
+ * Groups by error code (tsc, eslint, biome) or deduplicates stack traces (node).
+ */
+function detectErrorFormat(command, lines) {
+    if (command) {
+        const cmd = command.toLowerCase();
+        if (cmd.includes("tsc") || cmd.includes("typescript"))
+            return "tsc";
+        if (cmd.includes("eslint"))
+            return "eslint";
+        if (cmd.includes("biome"))
+            return "biome";
+        if (/\b(gcc|g\+\+|clang|clang\+\+)\b/.test(cmd))
+            return "gcc_clang";
+        if (/\b(rustc|cargo clippy|cargo check|cargo build)\b/.test(cmd))
+            return "rustc";
+        if (/\b(go vet|golangci-lint|staticcheck)\b/.test(cmd))
+            return "go";
+        if (/\b(python|python3|mypy|pyright|pylint|flake8|ruff)\b/.test(cmd))
+            return "python_tb";
+        if (/\b(javac|kotlinc)\b/.test(cmd))
+            return "javac";
+    }
+    const sample = lines.slice(0, 30).join("\n");
+    if (/error TS\d{4}:/.test(sample))
+        return "tsc";
+    if (/^\s+\d+:\d+\s+(error|warning)\s+.+\s{2,}\S+/m.test(sample))
+        return "eslint";
+    if (/^\s*×/.test(sample) || /\bdiagnostics?\b/i.test(sample))
+        return "biome";
+    // gcc/clang: file.c:10:5: error: ...
+    if (/^.+:\d+:\d+:\s+(error|warning|note):/m.test(sample))
+        return "gcc_clang";
+    // rustc: error[E0308]: mismatched types
+    if (/^error\[E\d{4}\]:/m.test(sample))
+        return "rustc";
+    // Go: file.go:10:5: ...
+    if (/^.+\.go:\d+:\d+:.*$/m.test(sample) && !/error TS/.test(sample))
+        return "go";
+    // Python traceback
+    if (/^Traceback \(most recent call last\):/m.test(sample) ||
+        /^ {2}File ".+", line \d+/m.test(sample))
+        return "python_tb";
+    // javac: File.java:10: error: ...
+    if (/^.+\.java:\d+: error:/m.test(sample))
+        return "javac";
+    if (/^\s*at\s+/.test(sample) ||
+        /^(Error|TypeError|RangeError|ReferenceError|SyntaxError):/m.test(sample))
+        return "node_stack";
+    // shellcheck: "In file.sh line N:"
+    if (/^In .+ line \d+:/m.test(sample))
+        return "shellcheck";
+    // Generic line diagnostics: file:line:col: message (rubocop, stylelint, vale, etc.)
+    if (/^.+:\d+(?::\d+)?:\s*(?:error|warning|info|note|convention|refactor)\s*:?\s*.+$/m.test(sample))
+        return "line_diagnostic";
+    return "generic";
+}
+function compressTscErrors(lines) {
+    const TSC_ERROR = /^(.+)\((\d+),(\d+)\): error (TS\d{4}): (.+)$/;
+    const SUMMARY = /^Found (\d+) errors?/;
+    const groups = new Map();
+    let summaryLine = "";
+    for (const line of lines) {
+        const m = TSC_ERROR.exec(line);
+        if (m) {
+            const [, filePath, , , code, message] = m;
+            const existing = groups.get(code);
+            if (existing) {
+                existing.count++;
+                existing.files.add(filePath);
+            }
+            else {
+                groups.set(code, {
+                    code: code,
+                    message: message,
+                    firstOccurrence: line,
+                    count: 1,
+                    files: new Set([filePath]),
+                });
+            }
+            continue;
+        }
+        const s = SUMMARY.exec(line);
+        if (s)
+            summaryLine = line;
+    }
+    if (groups.size === 0)
+        return "";
+    const parts = ["_shell_fmt:error_diagnostic"];
+    if (summaryLine)
+        parts.push(summaryLine);
+    const sorted = [...groups.values()].sort((a, b) => b.count - a.count);
+    for (const group of sorted) {
+        parts.push("");
+        parts.push(group.firstOccurrence);
+        if (group.count > 1) {
+            parts.push(`  └─ ${group.code}: ${group.count - 1} more in ${group.files.size} file(s)`);
+        }
+    }
+    return parts.join("\n");
+}
+function compressEslintErrors(lines) {
+    const FILE_HEADER = /^(\/[^\s]+\.(ts|js|tsx|jsx|mts|mjs|cjs|cts))$/;
+    const ERROR_LINE = /^\s+(\d+):(\d+)\s+(error|warning)\s+(.+?)\s{2,}(\S+)\s*$/;
+    const SUMMARY = /^✖ (\d+) problems?/;
+    const ruleGroups = new Map();
+    let currentFile = "";
+    let summaryLine = "";
+    for (const line of lines) {
+        const fh = FILE_HEADER.exec(line);
+        if (fh) {
+            currentFile = fh[1];
+            continue;
+        }
+        const el = ERROR_LINE.exec(line);
+        if (el) {
+            const [, lineNum, , , message, rule] = el;
+            const existing = ruleGroups.get(rule);
+            if (existing) {
+                existing.count++;
+            }
+            else {
+                ruleGroups.set(rule, {
+                    message: message?.trim() ?? "",
+                    count: 1,
+                    firstFile: currentFile,
+                    firstLine: lineNum,
+                });
+            }
+            continue;
+        }
+        const s = SUMMARY.exec(line);
+        if (s)
+            summaryLine = line;
+    }
+    if (ruleGroups.size === 0)
+        return "";
+    const parts = ["_shell_fmt:error_diagnostic"];
+    if (summaryLine)
+        parts.push(summaryLine);
+    const sorted = [...ruleGroups.entries()].sort((a, b) => b[1].count - a[1].count);
+    for (const [rule, group] of sorted) {
+        if (group.count === 1) {
+            parts.push(`  ${group.firstFile}:${group.firstLine}  ${group.message}  ${rule}`);
+        }
+        else {
+            parts.push(`  ${rule}: ${group.message} (${group.count} occurrences)`);
+            parts.push(`    first: ${group.firstFile}:${group.firstLine}`);
+        }
+    }
+    return parts.join("\n");
+}
+function filterStackFrames(frames) {
+    return frames.filter((f, i) => {
+        if (i === 0)
+            return true;
+        if (/node:internal/.test(f))
+            return false;
+        if (/node_modules/.test(f)) {
+            const isFirstNM = i === 0 || !/node_modules/.test(frames[i - 1] ?? "");
+            const isLastNM = i === frames.length - 1 || !/node_modules/.test(frames[i + 1] ?? "");
+            return isFirstNM || isLastNM;
+        }
+        return true;
+    });
+}
+function compressNodeStack(lines) {
+    const parts = ["_shell_fmt:error_diagnostic"];
+    const seenFingerprints = new Map();
+    let currentStack = [];
+    let currentError = "";
+    const flushStack = () => {
+        if (currentStack.length === 0)
+            return;
+        const filtered = filterStackFrames(currentStack);
+        const fingerprint = filtered.join("\n");
+        const existing = seenFingerprints.get(fingerprint) ?? 0;
+        seenFingerprints.set(fingerprint, existing + 1);
+        if (existing === 0) {
+            if (currentError)
+                parts.push(currentError);
+            parts.push(...filtered);
+        }
+        currentStack = [];
+        currentError = "";
+    };
+    for (const line of lines) {
+        const isFrame = /^\s+at\s/.test(line);
+        if (!isFrame && currentStack.length > 0) {
+            flushStack();
+        }
+        if (isFrame) {
+            currentStack.push(line);
+        }
+        else if (/^(Error|TypeError|RangeError|ReferenceError|SyntaxError|Caused by):/.test(line)) {
+            currentError = line;
+            parts.push(line);
+        }
+        else {
+            if (line.trim())
+                parts.push(line);
+        }
+    }
+    flushStack();
+    for (const [, count] of seenFingerprints) {
+        if (count > 1) {
+            parts.push(`  (identical stack repeated ×${count})`);
+        }
+    }
+    return parts.join("\n");
+}
+/**
+ * gcc/clang: file.c:10:5: error: ... — group by error message, dedup across files.
+ * Also handles: warning, note (notes are dropped unless attached to first occurrence).
+ */
+function compressGccClangErrors(lines) {
+    // Pattern: file:line:col: severity: message
+    const DIAG_RE = /^(.+):(\d+):(\d+):\s+(error|warning|note):\s+(.+)$/;
+    const groups = new Map();
+    let summaryErrors = 0;
+    let summaryWarnings = 0;
+    for (const line of lines) {
+        const m = DIAG_RE.exec(line);
+        if (m) {
+            const [, filePath, , , severity, message] = m;
+            if (severity === "note")
+                continue; // drop notes
+            if (severity === "error")
+                summaryErrors++;
+            else
+                summaryWarnings++;
+            const key = `${severity}:${message}`;
+            const existing = groups.get(key);
+            if (existing) {
+                existing.count++;
+                existing.files.add(filePath);
+            }
+            else {
+                groups.set(key, {
+                    code: severity,
+                    message: message,
+                    firstOccurrence: line,
+                    count: 1,
+                    files: new Set([filePath]),
+                });
+            }
+        }
+    }
+    if (groups.size === 0)
+        return "";
+    const parts = ["_shell_fmt:error_diagnostic"];
+    parts.push(`${summaryErrors} error(s), ${summaryWarnings} warning(s)`);
+    const sorted = [...groups.values()].sort((a, b) => b.count - a.count);
+    for (const group of sorted) {
+        parts.push("");
+        parts.push(group.firstOccurrence);
+        if (group.count > 1) {
+            parts.push(`  └─ ${group.count - 1} more in ${group.files.size} file(s)`);
+        }
+    }
+    return parts.join("\n");
+}
+/**
+ * rustc / cargo clippy: error[E0308]: message — group by error code.
+ */
+function compressRustcErrors(lines) {
+    // error[E0308]: mismatched types
+    const RUSTC_ERROR = /^(error|warning)\[([A-Z]\d{4})\]:\s+(.+)$/;
+    // --> file.rs:10:5
+    const LOCATION = /^\s*-->\s+(.+):(\d+):(\d+)$/;
+    const SUMMARY = /^(error|warning): .+ generated (\d+) warnings?/;
+    const ABORT = /^error: aborting due to (\d+)/;
+    const groups = new Map();
+    let currentCode = "";
+    let summaryLine = "";
+    for (const line of lines) {
+        const m = RUSTC_ERROR.exec(line);
+        if (m) {
+            const [, , code, message] = m;
+            currentCode = code;
+            const existing = groups.get(code);
+            if (existing) {
+                existing.count++;
+            }
+            else {
+                groups.set(code, {
+                    code: code,
+                    message: message,
+                    firstOccurrence: line,
+                    count: 1,
+                    files: new Set(),
+                });
+            }
+            continue;
+        }
+        const loc = LOCATION.exec(line);
+        if (loc && currentCode) {
+            groups.get(currentCode)?.files.add(loc[1]);
+        }
+        if (SUMMARY.test(line) || ABORT.test(line)) {
+            summaryLine = line;
+        }
+    }
+    if (groups.size === 0)
+        return "";
+    const parts = ["_shell_fmt:error_diagnostic"];
+    if (summaryLine)
+        parts.push(summaryLine);
+    const sorted = [...groups.values()].sort((a, b) => b.count - a.count);
+    for (const group of sorted) {
+        parts.push("");
+        parts.push(group.firstOccurrence);
+        if (group.files.size > 0) {
+            const fileList = [...group.files].slice(0, 5).join(", ");
+            const more = group.files.size > 5 ? ` +${group.files.size - 5} more` : "";
+            parts.push(`  └─ in: ${fileList}${more}`);
+        }
+        if (group.count > 1) {
+            parts.push(`  └─ ${group.count} total occurrences`);
+        }
+    }
+    return parts.join("\n");
+}
+/**
+ * Go: file.go:10:5: message — group by message, dedup across files.
+ * Handles go vet, golangci-lint, staticcheck output.
+ */
+function compressGoErrors(lines) {
+    // file.go:10:5: message  OR  file.go:10: message
+    const GO_DIAG = /^(.+\.go):(\d+)(?::\d+)?:\s+(.+)$/;
+    // golangci-lint: file.go:10:5: message (linter)
+    const GOLANGCI = /^(.+\.go):(\d+)(?::\d+)?:\s+(.+?)\s+\((\w+)\)$/;
+    const groups = new Map();
+    for (const line of lines) {
+        const gl = GOLANGCI.exec(line);
+        if (gl) {
+            const [, filePath, , message, linter] = gl;
+            const key = `${linter}:${message}`;
+            const existing = groups.get(key);
+            if (existing) {
+                existing.count++;
+                existing.files.add(filePath);
+            }
+            else {
+                groups.set(key, {
+                    code: linter,
+                    message: message,
+                    firstOccurrence: line,
+                    count: 1,
+                    files: new Set([filePath]),
+                });
+            }
+            continue;
+        }
+        const m = GO_DIAG.exec(line);
+        if (m) {
+            const [, filePath, , message] = m;
+            const existing = groups.get(message);
+            if (existing) {
+                existing.count++;
+                existing.files.add(filePath);
+            }
+            else {
+                groups.set(message, {
+                    code: "",
+                    message: message,
+                    firstOccurrence: line,
+                    count: 1,
+                    files: new Set([filePath]),
+                });
+            }
+        }
+    }
+    if (groups.size === 0)
+        return "";
+    const parts = ["_shell_fmt:error_diagnostic"];
+    const sorted = [...groups.values()].sort((a, b) => b.count - a.count);
+    for (const group of sorted) {
+        parts.push("");
+        parts.push(group.firstOccurrence);
+        if (group.count > 1) {
+            parts.push(`  └─ ${group.count - 1} more in ${group.files.size} file(s)`);
+        }
+    }
+    return parts.join("\n");
+}
+/**
+ * Python tracebacks: dedup identical stack traces, keep assertion details.
+ * Also handles mypy/pyright/pylint/ruff/flake8 line-based diagnostics.
+ */
+function compressPythonErrors(lines) {
+    // mypy/pyright: file.py:10: error: Message  [error-code]
+    const MYPY_RE = /^(.+\.py):(\d+):\s+(error|warning|note):\s+(.+?)(?:\s+\[(.+)\])?$/;
+    // ruff/flake8: file.py:10:5: E501 line too long
+    const RUFF_RE = /^(.+\.py):(\d+):\d+:\s+([A-Z]\d+)\s+(.+)$/;
+    // pylint: file.py:10:0: C0114: Missing module docstring (missing-module-docstring)
+    const PYLINT_RE = /^(.+\.py):(\d+):\d+:\s+([A-Z]\d{4}):\s+(.+?)(?:\s+\((.+)\))?$/;
+    // Try line-based diagnostics first (mypy, ruff, flake8, pylint)
+    const diagGroups = new Map();
+    let diagCount = 0;
+    for (const line of lines) {
+        let code = "";
+        let message = "";
+        let filePath = "";
+        const mp = MYPY_RE.exec(line);
+        if (mp) {
+            filePath = mp[1] ?? "";
+            message = mp[4] ?? "";
+            code = mp[5] ?? "error";
+        }
+        const rp = !mp ? RUFF_RE.exec(line) : null;
+        if (rp) {
+            filePath = rp[1] ?? "";
+            code = rp[3] ?? "";
+            message = rp[4] ?? "";
+        }
+        const pp = !mp && !rp ? PYLINT_RE.exec(line) : null;
+        if (pp) {
+            filePath = pp[1] ?? "";
+            code = pp[3] ?? "";
+            message = pp[4] ?? "";
+        }
+        if (code && message) {
+            diagCount++;
+            const key = `${code}:${message}`;
+            const existing = diagGroups.get(key);
+            if (existing) {
+                existing.count++;
+                existing.files.add(filePath);
+            }
+            else {
+                diagGroups.set(key, {
+                    code,
+                    message,
+                    firstOccurrence: line,
+                    count: 1,
+                    files: new Set([filePath]),
+                });
+            }
+        }
+    }
+    if (diagGroups.size > 0) {
+        const parts = ["_shell_fmt:error_diagnostic"];
+        parts.push(`${diagCount} diagnostic(s), ${diagGroups.size} unique`);
+        const sorted = [...diagGroups.values()].sort((a, b) => b.count - a.count);
+        for (const group of sorted.slice(0, 30)) {
+            parts.push("");
+            parts.push(group.firstOccurrence);
+            if (group.count > 1) {
+                parts.push(`  └─ ${group.code}: ${group.count - 1} more in ${group.files.size} file(s)`);
+            }
+        }
+        if (sorted.length > 30) {
+            parts.push(`\n… and ${sorted.length - 30} more diagnostic types`);
+        }
+        return parts.join("\n");
+    }
+    // Fall back to traceback compression (similar to node_stack)
+    const parts = ["_shell_fmt:error_diagnostic"];
+    const tracebacks = [];
+    let currentFrames = [];
+    let currentError = "";
+    let inTraceback = false;
+    for (const line of lines) {
+        if (/^Traceback \(most recent call last\):/.test(line)) {
+            if (currentFrames.length > 0) {
+                tracebacks.push({ error: currentError, frames: currentFrames });
+            }
+            currentFrames = [];
+            currentError = "";
+            inTraceback = true;
+            continue;
+        }
+        if (inTraceback && /^ {2}File ".+", line \d+/.test(line)) {
+            currentFrames.push(line);
+            continue;
+        }
+        if (inTraceback && /^\w+Error:|^\w+Exception:/.test(line)) {
+            currentError = line;
+            inTraceback = false;
+            continue;
+        }
+        if (!inTraceback && line.trim()) {
+            parts.push(line);
+        }
+    }
+    if (currentFrames.length > 0 || currentError) {
+        tracebacks.push({ error: currentError, frames: currentFrames });
+    }
+    // Dedup identical tracebacks
+    const seen = new Map();
+    for (const tb of tracebacks) {
+        const fingerprint = `${tb.frames.join("\n")}\n${tb.error}`;
+        const prev = seen.get(fingerprint) ?? 0;
+        seen.set(fingerprint, prev + 1);
+        if (prev === 0) {
+            parts.push("");
+            parts.push("Traceback (most recent call last):");
+            // Keep first + last 3 frames max
+            const frames = tb.frames;
+            if (frames.length > 6) {
+                parts.push(...frames.slice(0, 3));
+                parts.push(`  ... ${frames.length - 6} frames omitted ...`);
+                parts.push(...frames.slice(-3));
+            }
+            else {
+                parts.push(...frames);
+            }
+            if (tb.error)
+                parts.push(tb.error);
+        }
+    }
+    for (const [, count] of seen) {
+        if (count > 1)
+            parts.push(`  (identical traceback repeated ×${count})`);
+    }
+    return parts.length > 1 ? parts.join("\n") : "";
+}
+/**
+ * javac: File.java:10: error: message — group by message.
+ */
+function compressJavacErrors(lines) {
+    // File.java:10: error: message
+    const JAVAC_RE = /^(.+\.(?:java|kt|scala)):(\d+):\s+(error|warning):\s+(.+)$/;
+    const SUMMARY = /^(\d+) errors?$/;
+    const groups = new Map();
+    let summaryLine = "";
+    for (const line of lines) {
+        const m = JAVAC_RE.exec(line);
+        if (m) {
+            const [, filePath, , severity, message] = m;
+            const key = `${severity}:${message}`;
+            const existing = groups.get(key);
+            if (existing) {
+                existing.count++;
+                existing.files.add(filePath);
+            }
+            else {
+                groups.set(key, {
+                    code: severity,
+                    message: message,
+                    firstOccurrence: line,
+                    count: 1,
+                    files: new Set([filePath]),
+                });
+            }
+            continue;
+        }
+        if (SUMMARY.test(line))
+            summaryLine = line;
+    }
+    if (groups.size === 0)
+        return "";
+    const parts = ["_shell_fmt:error_diagnostic"];
+    if (summaryLine)
+        parts.push(summaryLine);
+    const sorted = [...groups.values()].sort((a, b) => b.count - a.count);
+    for (const group of sorted) {
+        parts.push("");
+        parts.push(group.firstOccurrence);
+        if (group.count > 1) {
+            parts.push(`  └─ ${group.count - 1} more in ${group.files.size} file(s)`);
+        }
+    }
+    return parts.join("\n");
+}
+/** Compress shellcheck output: group by SC code, show first example of each. */
+function compressShellcheck(lines) {
+    // Shellcheck format: 3-line blocks:
+    //   In file.sh line N:
+    //     <code snippet>
+    //     ^-- SCXXXX (severity): message
+    const SC_HEADER = /^In (.+) line (\d+):$/;
+    const SC_CODE = /\^--\s*(SC\d{4})\s*\((\w+)\):\s*(.+)$/;
+    const groups = new Map();
+    let i = 0;
+    while (i < lines.length) {
+        const headerMatch = SC_HEADER.exec(lines[i]);
+        if (!headerMatch) {
+            i++;
+            continue;
+        }
+        const file = headerMatch[1];
+        const blockLines = [lines[i]];
+        i++;
+        // Collect lines until next header or end (max 5 lines per block)
+        let codeFound = "";
+        let severity = "";
+        let message = "";
+        while (i < lines.length && blockLines.length < 6) {
+            const line = lines[i];
+            if (SC_HEADER.test(line))
+                break;
+            blockLines.push(line);
+            const codeMatch = SC_CODE.exec(line);
+            if (codeMatch) {
+                codeFound = codeMatch[1];
+                severity = codeMatch[2];
+                message = codeMatch[3];
+            }
+            i++;
+        }
+        if (!codeFound)
+            continue;
+        const existing = groups.get(codeFound);
+        if (existing) {
+            existing.count++;
+            existing.files.add(file);
+        }
+        else {
+            groups.set(codeFound, {
+                code: codeFound,
+                severity,
+                message,
+                count: 1,
+                files: new Set([file]),
+                firstBlock: blockLines,
+            });
+        }
+    }
+    if (groups.size === 0)
+        return "";
+    const parts = ["_shell_fmt:error_diagnostic"];
+    const total = [...groups.values()].reduce((s, g) => s + g.count, 0);
+    parts.push(`(${total} issues, ${groups.size} unique codes)`);
+    const sorted = [...groups.values()].sort((a, b) => b.count - a.count);
+    for (const g of sorted) {
+        parts.push("");
+        parts.push(`${g.code} (${g.severity}): ${g.message}  [×${g.count} in ${g.files.size} file${g.files.size > 1 ? "s" : ""}]`);
+        // Show first example block indented
+        for (const bl of g.firstBlock) {
+            parts.push(`  ${bl}`);
+        }
+    }
+    return parts.join("\n");
+}
+/** Compress generic line diagnostics (rubocop, stylelint, vale, markdownlint, etc.) */
+function compressLineDiagnostics(lines) {
+    // Pattern: file:line:col: severity: message  OR  file:line: [code] message
+    const DIAG_RE = /^(.+?):(\d+)(?::(\d+))?:\s*(?:(error|warning|info|note|convention|refactor)\s*:?\s*)?(.+)$/;
+    const groups = new Map();
+    const nonDiagLines = [];
+    for (const line of lines) {
+        const m = DIAG_RE.exec(line);
+        if (!m) {
+            if (line.trim())
+                nonDiagLines.push(line);
+            continue;
+        }
+        const [, filePath, , , severity, message] = m;
+        // Normalize message: strip trailing whitespace and file-specific references
+        const normMsg = message
+            ?.trim()
+            .replace(/`[^`]+`/g, "`<id>`")
+            .replace(/'\S+'/, "'<id>'");
+        const key = `${severity ?? "warning"}:${normMsg}`;
+        const existing = groups.get(key);
+        if (existing) {
+            existing.count++;
+            existing.files.add(filePath);
+        }
+        else {
+            groups.set(key, {
+                message: message?.trim() ?? "",
+                severity: severity ?? "warning",
+                count: 1,
+                files: new Set([filePath]),
+                firstLine: line,
+            });
+        }
+    }
+    if (groups.size === 0)
+        return "";
+    const parts = ["_shell_fmt:error_diagnostic"];
+    const total = [...groups.values()].reduce((s, g) => s + g.count, 0);
+    parts.push(`(${total} diagnostics, ${groups.size} unique messages)`);
+    const sorted = [...groups.values()].sort((a, b) => b.count - a.count);
+    for (const g of sorted.slice(0, 30)) {
+        const fileInfo = g.files.size > 3 ? `${g.files.size} files` : [...g.files].join(", ");
+        parts.push(g.count > 1
+            ? `${g.severity}: ${g.message}  [×${g.count} in ${fileInfo}]`
+            : g.firstLine);
+    }
+    if (sorted.length > 30) {
+        parts.push(`… ${sorted.length - 30} more unique messages`);
+    }
+    // Keep summary lines (non-diagnostic) at the end
+    const summaryLines = nonDiagLines.filter((l) => /\d+\s*(error|warning|problem|issue|offense)/i.test(l) ||
+        /^(✖|✗|Found|Total:)/i.test(l));
+    if (summaryLines.length > 0) {
+        parts.push("");
+        for (const s of summaryLines.slice(0, 5))
+            parts.push(s);
+    }
+    return parts.join("\n");
+}
+/** Legacy fallback: dedup consecutive identical frames + truncation. */
+function compressFallback(lines) {
+    const out = [];
+    let prevFrame = "";
+    let repeatCount = 0;
+    const flushDup = () => {
+        if (repeatCount > 0) {
+            out.push(`…repeated_frame ×${repeatCount}`);
+            repeatCount = 0;
+        }
+    };
+    for (const line of lines) {
+        const trimmed = line.trimEnd();
+        const isFrame = /^\s*at\s/.test(trimmed);
+        if (isFrame && trimmed === prevFrame) {
+            repeatCount++;
+            continue;
+        }
+        flushDup();
+        prevFrame = isFrame ? trimmed : "";
+        out.push(trimmed);
+    }
+    flushDup();
+    const joined = out.join("\n");
+    const clipped = joined.length > 10_000
+        ? `${joined.slice(0, 6000)}\n…error_mid_omitted…\n${joined.slice(-3000)}`
+        : joined;
+    return `_shell_fmt:error_diagnostic\n${clipped}`;
+}
+/**
+ * R1 — success short-circuit for linters/compilers. If the output has no
+ * "error" / "warning" / "FAIL" markers AND matches a clean-run pattern,
+ * collapse to a one-line summary.
+ */
+const DIAG_ERROR_RE = /\b(error|warning|FAIL(?:ED)?|panic)\b/i;
+const DIAG_SUCCESS_PATTERNS = [
+    /\b0 (?:errors?|problems?)\b/i,
+    /\bno (?:errors|problems|issues|violations|warnings) (?:found|reported)\b/i,
+    /\ball checks pass(?:ed)?\b/i,
+    /\bnothing to report\b/i,
+    /\bclean\b/i,
+];
+function tryDiagShortCircuit(raw, command) {
+    if (DIAG_ERROR_RE.test(raw))
+        return null;
+    for (const pat of DIAG_SUCCESS_PATTERNS) {
+        const m = raw.match(pat);
+        if (m) {
+            const label = command?.split(/\s+/)[0] ?? "lint";
+            return `_shell_fmt:error_diagnostic\n${label} ok — ${m[0]}`;
+        }
+    }
+    return null;
+}
+export function compressErrorDiagnostic(raw, command) {
+    const short = tryDiagShortCircuit(raw, command);
+    if (short)
+        return short;
+    const lines = raw.split("\n");
+    const format = detectErrorFormat(command, lines);
+    let result = "";
+    switch (format) {
+        case "tsc":
+            result = compressTscErrors(lines);
+            break;
+        case "eslint":
+            result = compressEslintErrors(lines);
+            break;
+        case "node_stack":
+            result = compressNodeStack(lines);
+            break;
+        case "gcc_clang":
+            result = compressGccClangErrors(lines);
+            break;
+        case "rustc":
+            result = compressRustcErrors(lines);
+            break;
+        case "go":
+            result = compressGoErrors(lines);
+            break;
+        case "python_tb":
+            result = compressPythonErrors(lines);
+            break;
+        case "javac":
+            result = compressJavacErrors(lines);
+            break;
+        case "shellcheck":
+            result = compressShellcheck(lines);
+            break;
+        case "line_diagnostic":
+            result = compressLineDiagnostics(lines);
+            break;
+        case "biome":
+        case "generic":
+            break;
+    }
+    // Fall back if format-specific parser returned nothing
+    return result || compressFallback(lines);
+}