github/openclaw
1
0
Fork 0
mirror of https://github.com/openclaw/openclaw.git synced 2026-05-09 15:34:31 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix text attachment MIME misclassification (#3628)

Browse Source 
* Fix text file attachment detection

* Add file attachment extraction tests
This commit is contained in:
Frank Yang
2026-01-28 18:33:03 -08:00
committed by GitHub
parent a109b7f1a9
commit cb18ce7a85
4 changed files with 364 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

77
src/media-understanding/apply.test.ts
View File

@@ -41,7 +41,7 @@ describe("applyMediaUnderstanding", () => {
mockedResolveApiKey.mockClear();
mockedFetchRemoteMedia.mockReset();
mockedFetchRemoteMedia.mockResolvedValue({
buffer: Buffer.from("audio-bytes"),
buffer: Buffer.from([0, 255, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]),
contentType: "audio/ogg",
fileName: "note.ogg",
});
@@ -51,7 +51,7 @@ describe("applyMediaUnderstanding", () => {
const { applyMediaUnderstanding } = await loadApply();
const dir = await fs.mkdtemp(path.join(os.tmpdir(), "moltbot-media-"));
const audioPath = path.join(dir, "note.ogg");
await fs.writeFile(audioPath, "hello");
await fs.writeFile(audioPath, Buffer.from([0, 255, 0, 1, 2, 3, 4, 5, 6, 7, 8]));
const ctx: MsgContext = {
Body: "<media:audio>",
@@ -94,7 +94,7 @@ describe("applyMediaUnderstanding", () => {
const { applyMediaUnderstanding } = await loadApply();
const dir = await fs.mkdtemp(path.join(os.tmpdir(), "moltbot-media-"));
const audioPath = path.join(dir, "note.ogg");
await fs.writeFile(audioPath, "hello");
await fs.writeFile(audioPath, Buffer.from([0, 255, 0, 1, 2, 3, 4, 5, 6, 7, 8]));
const ctx: MsgContext = {
Body: "<media:audio> /capture status",
@@ -176,7 +176,7 @@ describe("applyMediaUnderstanding", () => {
const { applyMediaUnderstanding } = await loadApply();
const dir = await fs.mkdtemp(path.join(os.tmpdir(), "moltbot-media-"));
const audioPath = path.join(dir, "large.wav");
await fs.writeFile(audioPath, "0123456789");
await fs.writeFile(audioPath, Buffer.from([0, 255, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10]));
const ctx: MsgContext = {
Body: "<media:audio>",
@@ -211,7 +211,7 @@ describe("applyMediaUnderstanding", () => {
const { applyMediaUnderstanding } = await loadApply();
const dir = await fs.mkdtemp(path.join(os.tmpdir(), "moltbot-media-"));
const audioPath = path.join(dir, "note.ogg");
await fs.writeFile(audioPath, "hello");
await fs.writeFile(audioPath, Buffer.from([0, 255, 0, 1, 2, 3, 4, 5, 6, 7, 8]));
const ctx: MsgContext = {
Body: "<media:audio>",
@@ -352,7 +352,7 @@ describe("applyMediaUnderstanding", () => {
const { applyMediaUnderstanding } = await loadApply();
const dir = await fs.mkdtemp(path.join(os.tmpdir(), "moltbot-media-"));
const audioPath = path.join(dir, "fallback.ogg");
await fs.writeFile(audioPath, "hello");
await fs.writeFile(audioPath, Buffer.from([0, 255, 0, 1, 2, 3, 4, 5, 6]));
const ctx: MsgContext = {
Body: "<media:audio>",
@@ -390,8 +390,8 @@ describe("applyMediaUnderstanding", () => {
const dir = await fs.mkdtemp(path.join(os.tmpdir(), "moltbot-media-"));
const audioPathA = path.join(dir, "note-a.ogg");
const audioPathB = path.join(dir, "note-b.ogg");
await fs.writeFile(audioPathA, "hello");
await fs.writeFile(audioPathB, "world");
await fs.writeFile(audioPathA, Buffer.from([200, 201, 202, 203, 204, 205, 206, 207, 208]));
await fs.writeFile(audioPathB, Buffer.from([200, 201, 202, 203, 204, 205, 206, 207, 208]));
const ctx: MsgContext = {
Body: "<media:audio>",
@@ -435,7 +435,7 @@ describe("applyMediaUnderstanding", () => {
const audioPath = path.join(dir, "note.ogg");
const videoPath = path.join(dir, "clip.mp4");
await fs.writeFile(imagePath, "image-bytes");
await fs.writeFile(audioPath, "audio-bytes");
await fs.writeFile(audioPath, Buffer.from([200, 201, 202, 203, 204, 205, 206, 207, 208]));
await fs.writeFile(videoPath, "video-bytes");
const ctx: MsgContext = {
@@ -487,4 +487,63 @@ describe("applyMediaUnderstanding", () => {
expect(ctx.CommandBody).toBe("audio ok");
expect(ctx.BodyForCommands).toBe("audio ok");
});
it("treats text-like audio attachments as CSV (comma wins over tabs)", async () => {
const { applyMediaUnderstanding } = await loadApply();
const dir = await fs.mkdtemp(path.join(os.tmpdir(), "moltbot-media-"));
const csvPath = path.join(dir, "data.mp3");
const csvText = '"a","b"\t"c"\n"1","2"\t"3"';
const csvBuffer = Buffer.concat([Buffer.from([0xff, 0xfe]), Buffer.from(csvText, "utf16le")]);
await fs.writeFile(csvPath, csvBuffer);
const ctx: MsgContext = {
Body: "<media:audio>",
MediaPath: csvPath,
MediaType: "audio/mpeg",
};
const cfg: MoltbotConfig = {
tools: {
media: {
audio: { enabled: false },
image: { enabled: false },
video: { enabled: false },
},
},
};
const result = await applyMediaUnderstanding({ ctx, cfg });
expect(result.appliedFile).toBe(true);
expect(ctx.Body).toContain('<file name="data.mp3" mime="text/csv">');
expect(ctx.Body).toContain('"a","b"\t"c"');
});
it("infers TSV when tabs are present without commas", async () => {
const { applyMediaUnderstanding } = await loadApply();
const dir = await fs.mkdtemp(path.join(os.tmpdir(), "moltbot-media-"));
const tsvPath = path.join(dir, "report.mp3");
const tsvText = "a\tb\tc\n1\t2\t3";
await fs.writeFile(tsvPath, tsvText);
const ctx: MsgContext = {
Body: "<media:audio>",
MediaPath: tsvPath,
MediaType: "audio/mpeg",
};
const cfg: MoltbotConfig = {
tools: {
media: {
audio: { enabled: false },
image: { enabled: false },
video: { enabled: false },
},
},
};
const result = await applyMediaUnderstanding({ ctx, cfg });
expect(result.appliedFile).toBe(true);
expect(ctx.Body).toContain('<file name="report.mp3" mime="text/tab-separated-values">');
expect(ctx.Body).toContain("a\tb\tc");
});
});