`nx_csv_writer.nx` source

↩ module page · 150 lines · 4849 B
1// csv_writer.nx -- RFC 4180 CSV serialiser.
2//
3// Companion to csv.nx (reader).  Writes cells with quoting that
4// csv.nx can roundtrip:
5//   - Cells without any of \",\\n\\r,\" are emitted raw
6//   - Cells containing any of those are wrapped in \"...\" with
7//     embedded \" doubled to \"\"
8//   - Lines end with CRLF per RFC (not bare LF)
9//
10// Stream API:
11//   csvw_field(out, cap, off, data, n) -> off' (writes cell)
12//   csvw_end_row(out, cap, off)        -> off' (writes CRLF)
13//
14// Caller tracks \"is this the first cell on the line\" by comparing
15// the returned offset to the offset before writing; csvw_field
16// inserts a \",\" separator when writing into the middle of a row.
17// To simplify, we expose csvw_begin_row which resets the comma-
18// separator flag stored on a CsvWriter state struct.
19//
20// Invariants:
21//   CW1  Roundtrip safe: csv.nx parsing our output yields the
22//        original cell bytes (after csv_decode_quoted for quoted
23//        cells).
24//   CW2  Empty string fields are OK -- they write as a bare
25//        \",\" boundary (or \"\"\"\" at end of row for unambiguous
26//        parsing per RFC 4180 §2.4).
27//   CW3  Output uses CRLF (0x0D 0x0A) line endings.
28
29// nx_safety_envelope:
30//   intended_use:       AUTO_APPLIED -- primitive-specific tuning queued
31//   sil_target:         SIL1
32//   evidence:           [bulk_applied_2026-05-16, see-file-comment-for-detail]
33//   verdict:            NOT_YET_EVALUATED
34
35import "nx_syscalls.nx"
36
37const CW_ERR_SHORT: i64 = -1
38
39struct CsvWriter {
40    // 1 if the next cell written should be preceded by a comma,
41    // 0 for first cell in a row.
42    in_row: i64,
43}
44
45func csvw_init(w: *CsvWriter) -> i64 {
46    w.in_row = 0
47    return 0
48}
49
50// Does this cell need quoting?  Quote if any of ',', '\"', LF, CR.
51func cw_needs_quoting(data: *u8, n: i64) -> i64 {
52    var i: i64 = 0
53    while i < n {
54        let b: i64 = data[i]
55        if b == 0x2C { return 1 }           // ','
56        if b == 0x22 { return 1 }           // '\"'
57        if b == 0x0A { return 1 }           // LF
58        if b == 0x0D { return 1 }           // CR
59        i = i + 1
60    }
61    return 0
62}
63
64// Write one cell.  Inserts ',' separator if mid-row.  Returns
65// new offset.
66func csvw_field(out: *u8, cap: i64, off: i64, w: *CsvWriter,
67                data: *u8, n: i64) -> i64 {
68    var cur: i64 = off
69    if w.in_row == 1 {
70        if cur >= cap { return CW_ERR_SHORT }
71        out[cur] = 0x2C
72        cur = cur + 1
73    }
74    if cw_needs_quoting(data, n) == 0 {
75        if cur + n > cap { return CW_ERR_SHORT }
76        var i: i64 = 0
77        while i < n {
78            out[cur + i] = data[i]
79            i = i + 1
80        }
81        cur = cur + n
82    } else {
83        // Quoted, with \" doubled.
84        if cur >= cap { return CW_ERR_SHORT }
85        out[cur] = 0x22
86        cur = cur + 1
87        var i: i64 = 0
88        while i < n {
89            if cur >= cap { return CW_ERR_SHORT }
90            out[cur] = data[i]
91            cur = cur + 1
92            if data[i] == 0x22 {
93                if cur >= cap { return CW_ERR_SHORT }
94                out[cur] = 0x22         // second quote
95                cur = cur + 1
96            }
97            i = i + 1
98        }
99        if cur >= cap { return CW_ERR_SHORT }
100        out[cur] = 0x22
101        cur = cur + 1
102    }
103    w.in_row = 1
104    return cur
105}
106
107// End-of-row: writes CRLF and resets in_row.
108func csvw_end_row(out: *u8, cap: i64, off: i64, w: *CsvWriter) -> i64 {
109    if off + 2 > cap { return CW_ERR_SHORT }
110    out[off]     = 0x0D
111    out[off + 1] = 0x0A
112    w.in_row = 0
113    return off + 2
114}
115
116// Compile-only smoke.
117func main() -> i64 {
118    let out: *u8 = sys_mmap(256)
119    let w_raw: *u8 = sys_mmap(16)
120    let w: *CsvWriter = w_raw as *CsvWriter
121    csvw_init(w)
122    var off: i64 = 0
123
124    off = csvw_field(out, 256, off, w, "a",      1)
125    off = csvw_field(out, 256, off, w, "b",      1)
126    off = csvw_field(out, 256, off, w, "hi,lo",  5)    // needs quote
127    off = csvw_end_row(out, 256, off, w)
128
129    off = csvw_field(out, 256, off, w, "c",        1)
130    off = csvw_field(out, 256, off, w, "say \"hi\"", 8) // embedded \"
131    off = csvw_end_row(out, 256, off, w)
132
133    // Expected:
134    //   a,b,\"hi,lo\"\\r\\n
135    //   c,\"say \"\"hi\"\"\"\\r\\n
136    // Length: 3 + 7 + 2 = 12, then 2 + 12 + 2 = 16.  Total 28.
137    if off != 28 { return 1 }
138    if out[0] != 0x61 { return 2 }        // 'a'
139    if out[1] != 0x2C { return 3 }        // ','
140    if out[2] != 0x62 { return 4 }        // 'b'
141    if out[3] != 0x2C { return 5 }
142    // \"hi,lo\" quoted.
143    if out[4] != 0x22 { return 6 }        // opening '\"'
144    if out[5] != 0x68 { return 7 }        // 'h'
145    // CRLF after first row.
146    if out[10] != 0x22 { return 8 }       // closing '\"'
147    if out[11] != 0x0D { return 9 }       // CR
148    if out[12] != 0x0A { return 10 }      // LF
149    return 0
150}
nx_csv_writer.nx source

`nx_csv_writer.nx` source
