1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
use clap::Parser;
use std::path::PathBuf;
#[global_allocator]
static GLOBAL: mimalloc::MiMalloc = mimalloc::MiMalloc;
mod commands;
mod config;
mod convert;
mod dump;
mod output;
mod script;
mod table;
#[derive(Parser)]
#[command(name = "qql", about = "Qdrant Query Language CLI")]
struct Cli {
/// Qdrant REST URL. Overrides QDRANT_URL when supplied.
#[arg(long, global = true)]
url: Option<String>,
/// Execute supported commands against the configured in-process edge backend.
#[arg(long, global = true)]
edge: bool,
#[command(subcommand)]
command: Command,
}
#[derive(clap::Subcommand)]
enum Command {
/// Execute a QQL query
Exec {
/// QQL query string (e.g., "QUERY 'hello' FROM docs LIMIT 5")
query: String,
/// Output as JSON
#[arg(long)]
json: bool,
/// Quiet mode
#[arg(long, short)]
quiet: bool,
},
/// Execute multiple QQL queries from a file
Execute {
/// Path to .qql script file
file: String,
/// Stop on first error
#[arg(long)]
stop_on_error: bool,
},
/// Explain a QQL query (show execution plan)
Explain {
query: String,
/// Output as JSON
#[arg(long)]
json: bool,
/// Quiet mode
#[arg(long, short)]
quiet: bool,
},
/// Start interactive REPL connected to Qdrant
Connect,
/// Convert REST JSON payload to QQL
Convert {
/// Path to JSON file (or stdin if omitted)
file: Option<String>,
},
/// Dump collection to .qql file
Dump {
collection: String,
output: String,
#[arg(long, default_value = "100")]
batch_size: u32,
/// Output as JSON
#[arg(long)]
json: bool,
/// Quiet mode
#[arg(long, short)]
quiet: bool,
},
/// Check Qdrant connection health
Doctor {
/// Output as JSON
#[arg(long)]
json: bool,
/// Quiet mode
#[arg(long, short)]
quiet: bool,
},
/// Configure persistent CLI settings.
Config {
#[command(subcommand)]
command: Box<ConfigCommand>,
},
/// Show version
Version,
}
#[derive(clap::Subcommand)]
enum ConfigCommand {
/// Configure the local qdrant-edge backend used by --edge.
Edge {
/// Directory for persistent qdrant-edge data.
#[arg(long)]
data_dir: Option<PathBuf>,
/// Keep payloads in memory instead of persisting them to disk.
#[arg(long)]
in_memory: bool,
/// Embedding backend: fastembed or an OpenAI-compatible HTTP endpoint.
#[arg(long, default_value = "fastembed")]
embedder: String,
/// Local FastEmbed dense model name or alias.
#[arg(long)]
model: Option<String>,
/// Offline sparse model for fastembed (e.g. splade, bge-m3).
#[arg(long)]
sparse_model: Option<String>,
/// Offline multivector model for fastembed (e.g. bge-m3).
#[arg(long)]
multi_model: Option<String>,
/// Offline CLIP vision model for fastembed (e.g. clip-vision).
#[arg(long)]
image_model: Option<String>,
/// Offline cross-encoder model (e.g. bge-reranker-base).
#[arg(long)]
reranker_model: Option<String>,
/// Directory used for downloaded FastEmbed models.
#[arg(long)]
cache_dir: Option<PathBuf>,
/// Show model download progress.
#[arg(long)]
show_download_progress: bool,
/// OpenAI-compatible embedding endpoint used by the HTTP backend.
#[arg(long)]
embed_url: Option<String>,
/// API key used by the HTTP embedding backend.
#[arg(long, default_value = "")]
embed_key: String,
/// Model name sent to the HTTP embedding backend.
#[arg(long, default_value = "nomic-embed-text")]
embed_model: String,
/// Expected HTTP embedding dimension.
#[arg(long, default_value_t = 768)]
embed_dim: usize,
/// Optional multi/ColBERT HTTP embedding endpoint.
#[arg(long)]
multi_embed_url: Option<String>,
/// API key for the multi embedding endpoint.
#[arg(long)]
multi_embed_key: Option<String>,
/// Multi/ColBERT model name for HTTP multi embeds.
#[arg(long)]
multi_embed_model: Option<String>,
/// Per-token dimension for multi embeds (0 = skip check).
#[arg(long, default_value_t = 0)]
multi_embed_dim: usize,
/// Optional image/CLIP vision HTTP embedding endpoint.
#[arg(long)]
image_embed_url: Option<String>,
/// API key for the image embedding endpoint.
#[arg(long)]
image_embed_key: Option<String>,
/// Image/CLIP vision model name for HTTP image embeds.
#[arg(long)]
image_embed_model: Option<String>,
/// Dense dimension for image embeds (CLIP = 512; 0 = use dense dim).
#[arg(long, default_value_t = 0)]
image_embed_dim: usize,
},
}
#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
let cli = Cli::parse();
let use_edge = cli.edge;
let url = cli
.url
.or_else(|| std::env::var("QDRANT_URL").ok())
.unwrap_or_else(|| "http://localhost:6333".to_string());
match cli.command {
Command::Exec { query, json, quiet } => {
commands::handle_exec(&url, use_edge, &query, json, quiet).await
}
Command::Execute {
file,
stop_on_error,
} => commands::handle_execute_file(&url, use_edge, &file, stop_on_error).await,
Command::Explain {
query,
json: _,
quiet: _,
} => commands::handle_explain(&query),
Command::Connect => commands::handle_connect(&url, use_edge).await,
Command::Convert { file } => commands::handle_convert(file.as_deref()),
Command::Dump {
collection,
output,
batch_size,
json,
quiet,
} => {
use std::io::Write;
let progress_fn = |p: dump::DumpProgress| {
eprint!("\rDumped {} points ({} batches)...", p.written, p.batches);
let _ = std::io::stderr().flush();
};
let progress_cb: Option<&(dyn Fn(dump::DumpProgress) + Sync)> = if !json && !quiet {
Some(&progress_fn)
} else {
None
};
let stats = commands::handle_dump(
&url,
use_edge,
&collection,
&output,
batch_size,
progress_cb,
)
.await?;
if !json && !quiet && stats.batches > 0 {
eprintln!();
}
let msg = format!(
"Dumped collection '{}' to {} ({} written, {} skipped, {} batches)",
collection, output, stats.written, stats.skipped, stats.batches
);
if json {
println!(
"{}",
serde_json::json!({
"ok": true,
"operation": "dump",
"collection": collection,
"output": output,
"written": stats.written,
"skipped": stats.skipped,
"batches": stats.batches,
"message": msg,
})
);
} else {
println!("{}", msg);
}
Ok(())
}
Command::Doctor { json, quiet: _ } => commands::handle_doctor(&url, use_edge, json).await,
Command::Config { command } => match *command {
ConfigCommand::Edge {
data_dir,
in_memory,
embedder,
model,
sparse_model,
multi_model,
image_model,
reranker_model,
cache_dir,
show_download_progress,
embed_url,
embed_key,
embed_model,
embed_dim,
multi_embed_url,
multi_embed_key,
multi_embed_model,
multi_embed_dim,
image_embed_url,
image_embed_key,
image_embed_model,
image_embed_dim,
} => commands::handle_configure_edge(config::EdgeConfig {
data_dir: data_dir.unwrap_or_else(|| config::EdgeConfig::default().data_dir),
on_disk_payload: !in_memory,
embedder,
model,
sparse_model,
multi_model,
image_model,
reranker_model,
cache_dir,
show_download_progress,
embed_url,
embed_key,
embed_model,
embed_dimension: embed_dim,
multi_embed_url,
multi_embed_key,
multi_embed_model,
multi_embed_dimension: multi_embed_dim,
image_embed_url,
image_embed_key,
image_embed_model,
image_embed_dimension: image_embed_dim,
}),
},
Command::Version => commands::handle_version(),
}
}