nx_rfc_dir_ingest.nx source
↩ module page · 69 lines · 2495 B
1// nx_rfc_dir_ingest.nx -- ingest every RFC section header in the
2// corpus at nxc2/_offc/rfc_corpus.txt.
3
4// nx_safety_envelope:
5// intended_use: AUTO_APPLIED -- primitive-specific tuning queued
6// sil_target: SIL1
7// evidence: [bulk_applied_2026-05-16, see-file-comment-for-detail]
8// verdict: NOT_YET_EVALUATED
9
10import "nx_syscalls.nx"
11import "nx_runtime.nx"
12import "nx_tier.nx"
13import "nx_rfc_ingest.nx"
14import "nx_ingest_runner.nx"
15import "nx_rfc_stream_ingest.nx"
16import "nx_bloom_capacity.nx"
17
18const NX_RFC_DIR_DISK_BUDGET: nx_size = 1073741824
19const NX_RFC_DIR_SHARD_BYTES: nx_size = 104857600
20const NX_RFC_DIR_BLOOM_CAPACITY: nx_int = 1000000
21const NX_RFC_DIR_WATCHDOG_MS: nx_int = 60000
22
23func main() -> nx_exit {
24 let corpus_path: *u8 = "nxc2/_offc/rfc_corpus.txt" as *u8
25 let out_prefix: *u8 = "/tmp/nx_ingest_rfc" as *u8
26 let ckpt_path: *u8 = "/tmp/nx_ingest_rfc.checkpoint" as *u8
27
28 let corpus_len_p: *nx_size = (sys_mmap(NX_SIZEOF_NX_SIZE)) as *nx_size
29 corpus_len_p[0] = 0
30 let corpus: *u8 = sys_read_file(corpus_path, corpus_len_p)
31 if (corpus as nx_size) == 0 {
32 sys_write(NX_FD_STDOUT, "nx_rfc_dir_ingest: cannot read corpus\n" as *u8, 37)
33 return 1
34 }
35 let corpus_n: nx_size = corpus_len_p[0]
36
37 sys_write(NX_FD_STDOUT, "nx_rfc_dir_ingest: corpus bytes=" as *u8, 32)
38 print_i64(corpus_n as i64)
39 sys_write(NX_FD_STDOUT, "\n" as *u8, 1)
40
41 let db: *RfcDb = nx_rfc_ingest_corpus(corpus, corpus_n as nx_int)
42 sys_write(NX_FD_STDOUT, "nx_rfc_dir_ingest: decls parsed=" as *u8, 32)
43 print_i64(db.n_decls as i64)
44 sys_write(NX_FD_STDOUT, "\n" as *u8, 1)
45
46 let run: *NxIngestRun = nx_ingest_run_new(
47 out_prefix,
48 NX_RFC_DIR_DISK_BUDGET,
49 NX_RFC_DIR_SHARD_BYTES,
50 NX_RFC_DIR_BLOOM_CAPACITY,
51 NX_BLOOM_PROFILE_1PCT,
52 ckpt_path,
53 NX_RFC_DIR_WATCHDOG_MS)
54 if run == (0 as *NxIngestRun) { return 2 }
55
56 let row_buf: *u8 = sys_mmap(NX_BUF_HUGE)
57 let n_emitted: nx_int = nx_rfc_stream_offer_all(db, run, row_buf)
58
59 sys_write(NX_FD_STDOUT, "nx_rfc_dir_ingest: emitted=" as *u8, 27)
60 print_i64(n_emitted as i64)
61 sys_write(NX_FD_STDOUT, " duplicate=" as *u8, 11)
62 print_i64(nx_ingest_run_n_duplicate(run) as i64)
63 sys_write(NX_FD_STDOUT, " shards=" as *u8, 8)
64 print_i64(nx_ingest_run_n_shards(run) as i64)
65 sys_write(NX_FD_STDOUT, "\n" as *u8, 1)
66
67 nx_ingest_run_close(run)
68 return 0
69}