hiperf_txt_parser 1.0.0 → 1.0.2

package/README.md

@@ -25,12 +25,14 @@ import {
   parsePerfData,
   formatPerfDataToText,
   formatPerfDataToJson,
+  filterByTgid,
 } from "hiperf_txt_parser";
 ```
 
 - `parsePerfData(text: string): PerfData`
 - `formatPerfDataToText(data: PerfData): string`
 - `formatPerfDataToJson(data: PerfData): Array<{ issuce: "unknow"; call_chain: string }>`
+- `filterByTgid(data: PerfData, tgid: number): PerfData`（仅保留 `pid === tgid` 的 RecordSample）
 
 ## 快速示例
 
@@ -40,7 +42,8 @@ import { parsePerfData, formatPerfDataToJson, formatPerfDataToText } from "hiper
 const input = `record sample: type 9, misc 2, size 520\n  sample_type: 0x8000107e7\n  ID 13`;
 
 const parsed = parsePerfData(input);
-const jsonArray = formatPerfDataToJson(parsed);
+const filtered = filterByTgid(parsed, 1234);
+const jsonArray = formatPerfDataToJson(filtered);
 const txt = formatPerfDataToText(parsed);
 ```
 

package/dist/index.d.ts

@@ -1,4 +1,4 @@
-export { parsePerfData } from "./parser.js";
+export { parsePerfData, filterByTgid } from "./parser.js";
 export { formatPerfDataToText, formatPerfDataToJson } from "./serializer.js";
 export type { PerfData, RecordSample } from "./types.js";
 export type { RecordSampleJsonExportItem } from "./serializer.js";

package/dist/index.js

@@ -1,2 +1,2 @@
-export { parsePerfData } from "./parser.js";
+export { parsePerfData, filterByTgid } from "./parser.js";
 export { formatPerfDataToText, formatPerfDataToJson } from "./serializer.js";

package/dist/parser.d.ts

@@ -3,3 +3,7 @@ import type { PerfData } from "./types.js";
  * 解析 perf data 文本，仅返回 record sample 结构数组
  */
 export declare function parsePerfData(text: string): PerfData;
+/**
+ * 按 tgid 过滤 RecordSample（当前以 pid 字段作为 tgid）
+ */
+export declare function filterByTgid(data: PerfData, tgid: number): PerfData;

package/dist/parser.js

@@ -1,12 +1,5 @@
 const RECORD_SAMPLE_PREFIX = "record sample:";
 const RECORD_COMM_PREFIX = "record comm";
-/**
- * 获取行首空格数量（缩进）
- */
-function getIndent(line) {
-    const m = line.match(/^(\s*)/);
-    return m ? m[1].length : 0;
-}
 /**
  * 解析 "record sample: type 9, misc 2, size 520" 行
  */
@@ -23,6 +16,20 @@ function parseHeaderLine(line) {
 /**
  * 解析单个 record sample 块（已按行切分好的行数组）
  */
+function isTopLevelField(trimmed) {
+    return (trimmed.startsWith("sample_type:") ||
+        trimmed.startsWith("ID ") ||
+        trimmed.startsWith("ip ") ||
+        (trimmed.startsWith("pid ") && trimmed.includes(", tid ")) ||
+        trimmed.startsWith("time ") ||
+        trimmed.startsWith("stream_id ") ||
+        (trimmed.startsWith("cpu ") && trimmed.includes(", res ")) ||
+        trimmed.startsWith("period ") ||
+        trimmed.startsWith("callchain nr=") ||
+        trimmed.startsWith("raw size=") ||
+        trimmed.startsWith("server nr=") ||
+        trimmed.startsWith("callchain: "));
+}
 function parseOneBlock(lines) {
     if (lines.length === 0 || !lines[0].trimStart().startsWith(RECORD_SAMPLE_PREFIX)) {
         throw new Error("Invalid record sample block");
@@ -41,108 +48,157 @@ function parseOneBlock(lines) {
         res: 0,
         period: 0,
     };
+    let mode = null;
     let i = 1;
     while (i < lines.length) {
         const line = lines[i];
-        const indent = getIndent(line);
-        const trimmed = line.trimStart();
-        if (indent === 0 && trimmed.length > 0) {
+        const trimmed = line.trim();
+        if (!trimmed) {
+            i++;
+            continue;
+        }
+        if (trimmed.startsWith(RECORD_SAMPLE_PREFIX) ||
+            trimmed.startsWith(RECORD_COMM_PREFIX)) {
             break;
         }
-        if (indent === 2) {
-            if (trimmed.startsWith("sample_type:")) {
-                sample.sample_type = trimmed.replace(/^sample_type:\s*/, "").trim();
-            }
-            else if (trimmed.startsWith("ID ")) {
-                sample.id = parseInt(trimmed.replace(/^ID\s+/, ""), 10) || 0;
-            }
-            else if (trimmed.startsWith("ip ")) {
-                sample.ip = trimmed.replace(/^ip\s+/, "").trim();
-            }
-            else if (trimmed.startsWith("pid ") && trimmed.includes(", tid ")) {
-                const pidMatch = trimmed.match(/pid\s+(\d+).*tid\s+(\d+)/);
-                if (pidMatch) {
-                    sample.pid = parseInt(pidMatch[1], 10);
-                    sample.tid = parseInt(pidMatch[2], 10);
-                }
-            }
-            else if (trimmed.startsWith("time ")) {
-                sample.time = parseInt(trimmed.replace(/^time\s+/, ""), 10) || 0;
-            }
-            else if (trimmed.startsWith("stream_id ")) {
-                sample.stream_id = parseInt(trimmed.replace(/^stream_id\s+/, ""), 10) || 0;
-            }
-            else if (trimmed.startsWith("cpu ") && trimmed.includes(", res ")) {
-                const cpuMatch = trimmed.match(/cpu\s+(\d+).*res\s+(\d+)/);
-                if (cpuMatch) {
-                    sample.cpu = parseInt(cpuMatch[1], 10);
-                    sample.res = parseInt(cpuMatch[2], 10);
-                }
+        if (mode && isTopLevelField(trimmed)) {
+            mode = null;
+            continue;
+        }
+        if (trimmed.startsWith("sample_type:")) {
+            sample.sample_type = trimmed.replace(/^sample_type:\s*/, "").trim();
+            mode = null;
+            i++;
+            continue;
+        }
+        if (trimmed.startsWith("ID ")) {
+            sample.id = parseInt(trimmed.replace(/^ID\s+/, ""), 10) || 0;
+            mode = null;
+            i++;
+            continue;
+        }
+        if (trimmed.startsWith("ip ")) {
+            sample.ip = trimmed.replace(/^ip\s+/, "").trim();
+            mode = null;
+            i++;
+            continue;
+        }
+        if (trimmed.startsWith("pid ") && trimmed.includes(", tid ")) {
+            const pidMatch = trimmed.match(/pid\s+(\d+).*tid\s+(\d+)/);
+            if (pidMatch) {
+                sample.pid = parseInt(pidMatch[1], 10);
+                sample.tid = parseInt(pidMatch[2], 10);
             }
-            else if (trimmed.startsWith("period ")) {
-                sample.period = parseInt(trimmed.replace(/^period\s+/, ""), 10) || 0;
+            mode = null;
+            i++;
+            continue;
+        }
+        if (trimmed.startsWith("time ")) {
+            sample.time = parseInt(trimmed.replace(/^time\s+/, ""), 10) || 0;
+            mode = null;
+            i++;
+            continue;
+        }
+        if (trimmed.startsWith("stream_id ")) {
+            sample.stream_id = parseInt(trimmed.replace(/^stream_id\s+/, ""), 10) || 0;
+            mode = null;
+            i++;
+            continue;
+        }
+        if (trimmed.startsWith("cpu ") && trimmed.includes(", res ")) {
+            const cpuMatch = trimmed.match(/cpu\s+(\d+).*res\s+(\d+)/);
+            if (cpuMatch) {
+                sample.cpu = parseInt(cpuMatch[1], 10);
+                sample.res = parseInt(cpuMatch[2], 10);
             }
-            else if (trimmed.startsWith("callchain nr=")) {
-                const nrMatch = trimmed.match(/callchain\s+nr=(\d+)/);
-                const nr = nrMatch ? parseInt(nrMatch[1], 10) : 0;
-                const addresses = [];
+            mode = null;
+            i++;
+            continue;
+        }
+        if (trimmed.startsWith("period ")) {
+            sample.period = parseInt(trimmed.replace(/^period\s+/, ""), 10) || 0;
+            mode = null;
+            i++;
+            continue;
+        }
+        if (trimmed.startsWith("callchain nr=")) {
+            const nrMatch = trimmed.match(/callchain\s+nr=(\d+)/);
+            const nr = nrMatch ? parseInt(nrMatch[1], 10) : 0;
+            sample.callchain = { nr, addresses: [] };
+            mode = "callchainAddr";
+            i++;
+            continue;
+        }
+        if (trimmed.startsWith("raw size=")) {
+            const sizeMatch = trimmed.match(/raw\s+size=(\d+)/);
+            const size = sizeMatch ? parseInt(sizeMatch[1], 10) : 0;
+            sample.raw = { size, lines: [] };
+            mode = "raw";
+            i++;
+            continue;
+        }
+        if (trimmed.startsWith("server nr=")) {
+            const nrMatch = trimmed.match(/server\s+nr=(\d+)/);
+            const nr = nrMatch ? parseInt(nrMatch[1], 10) : 0;
+            sample.server = { nr, pids: [] };
+            mode = "server";
+            i++;
+            continue;
+        }
+        if (trimmed.startsWith("callchain: ")) {
+            const countMatch = trimmed.match(/callchain:\s*(\d+)/);
+            const count = countMatch ? parseInt(countMatch[1], 10) : 0;
+            sample.callchainFrames = { count, frames: [] };
+            mode = "frames";
+            i++;
+            continue;
+        }
+        if (mode === "callchainAddr") {
+            if (/^0x[0-9a-fA-F]+$/.test(trimmed) && sample.callchain) {
+                sample.callchain.addresses.push(trimmed);
                 i++;
-                while (i < lines.length && getIndent(lines[i]) >= 4) {
-                    const addr = lines[i].trim();
-                    if (addr && /^0x[0-9a-fA-F]+$/.test(addr)) {
-                        addresses.push(addr);
-                    }
-                    i++;
-                }
-                sample.callchain = { nr, addresses };
                 continue;
             }
-            else if (trimmed.startsWith("raw size=")) {
-                const sizeMatch = trimmed.match(/raw\s+size=(\d+)/);
-                const size = sizeMatch ? parseInt(sizeMatch[1], 10) : 0;
-                const entries = [];
-                i++;
-                while (i < lines.length && getIndent(lines[i]) >= 4) {
-                    const rawLine = lines[i].trim();
-                    const hexShort = rawLine.match(/^(0x[0-9a-fA-F]+)\s*\(([0-9a-fA-F]+)\)$/);
-                    if (hexShort) {
-                        entries.push({ hex: hexShort[1], short: hexShort[2] });
-                    }
-                    else if (rawLine.startsWith("0x")) {
-                        entries.push({ hex: rawLine });
-                    }
+            mode = null;
+            continue;
+        }
+        if (mode === "raw") {
+            if (sample.raw) {
+                const hexShort = trimmed.match(/^(0x[0-9a-fA-F]+)\s*\(([0-9a-fA-F]+)\)$/);
+                if (hexShort) {
+                    sample.raw.lines.push({ hex: hexShort[1], short: hexShort[2] });
                     i++;
+                    continue;
                 }
-                sample.raw = { size, lines: entries };
-                continue;
-            }
-            else if (trimmed.startsWith("server nr=")) {
-                const nrMatch = trimmed.match(/server\s+nr=(\d+)/);
-                const nr = nrMatch ? parseInt(nrMatch[1], 10) : 0;
-                const pids = [];
-                i++;
-                while (i < lines.length && getIndent(lines[i]) >= 4) {
-                    const pidMatch = lines[i].trim().match(/pid:\s*(\d+)/);
-                    if (pidMatch) {
-                        pids.push(parseInt(pidMatch[1], 10));
-                    }
+                if (/^0x[0-9a-fA-F]+$/.test(trimmed)) {
+                    sample.raw.lines.push({ hex: trimmed });
                     i++;
+                    continue;
                 }
-                sample.server = { nr, pids };
-                continue;
             }
-            else if (trimmed.startsWith("callchain: ")) {
-                const countMatch = trimmed.match(/callchain:\s*(\d+)/);
-                const count = countMatch ? parseInt(countMatch[1], 10) : 0;
-                const frames = [];
-                i++;
-                while (i < lines.length && getIndent(lines[i]) >= 4) {
-                    frames.push(lines[i].trim());
+            mode = null;
+            continue;
+        }
+        if (mode === "server") {
+            if (sample.server) {
+                const pidMatch = trimmed.match(/^pid:\s*(\d+)$/);
+                if (pidMatch) {
+                    sample.server.pids.push(parseInt(pidMatch[1], 10));
                     i++;
+                    continue;
                 }
-                sample.callchainFrames = { count, frames };
+            }
+            mode = null;
+            continue;
+        }
+        if (mode === "frames") {
+            if (sample.callchainFrames) {
+                sample.callchainFrames.frames.push(trimmed);
+                i++;
                 continue;
             }
+            mode = null;
+            continue;
         }
         i++;
     }
@@ -158,15 +214,14 @@ function extractRecordSampleBlocks(text) {
     for (let i = 0; i < lines.length; i++) {
         const line = lines[i];
         const trimmed = line.trimStart();
-        const indent = getIndent(line);
-        if (trimmed.startsWith(RECORD_SAMPLE_PREFIX) && indent === 0) {
+        if (trimmed.startsWith(RECORD_SAMPLE_PREFIX)) {
             if (current.length > 0) {
                 blocks.push(current);
             }
             current = [line];
             continue;
         }
-        if (trimmed.startsWith(RECORD_COMM_PREFIX) && indent === 0) {
+        if (trimmed.startsWith(RECORD_COMM_PREFIX)) {
             if (current.length > 0) {
                 blocks.push(current);
                 current = [];
@@ -202,3 +257,11 @@ export function parsePerfData(text) {
     }
     return { recordSamples };
 }
+/**
+ * 按 tgid 过滤 RecordSample（当前以 pid 字段作为 tgid）
+ */
+export function filterByTgid(data, tgid) {
+    return {
+        recordSamples: data.recordSamples.filter((sample) => sample.pid === tgid),
+    };
+}

package/dist/serializer.js

@@ -33,9 +33,9 @@ function serializeOneSample(sample) {
     }
     if (sample.callchainFrames) {
         lines.push(`  `);
-        lines.push(`  callchain: ${sample.callchainFrames.count}`);
+        lines.push(` callchain: ${sample.callchainFrames.count}`);
         for (const frame of sample.callchainFrames.frames) {
-            lines.push(`    ${frame}`);
+            lines.push(`  ${frame}`);
         }
     }
     return lines.join("\n");

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "hiperf_txt_parser",
-  "version": "1.0.0",
+  "version": "1.0.2",
   "description": "Parse perf data.txt and output structured TypeScript data",
   "type": "module",
   "main": "dist/index.js",
@@ -16,9 +16,14 @@
   ],
   "scripts": {
     "build": "tsc",
-    "test": "npm run build && node --test dist/__tests__/parser.test.js"
+    "prepublishOnly": "npm run build",
+    "test": "npm run build && node --test tests/*.test.mjs"
   },
-  "keywords": ["perf", "parser", "library"],
+  "keywords": [
+    "perf",
+    "parser",
+    "library"
+  ],
   "author": "",
   "license": "MIT",
   "devDependencies": {