diff options
| author | Onur Gündüz <ogunduz@gmail.com> | 2026-06-15 08:40:11 +0200 |
|---|---|---|
| committer | Onur Gündüz <ogunduz@gmail.com> | 2026-06-15 08:40:11 +0200 |
| commit | c4a1c97a09da6d8d88c2ac834bf6234506fc41bc (patch) | |
| tree | 81ce5675ce886aad0309a41f4d6a696826e0a061 /tests/tok_fixture.h | |
| parent | b377ba8656bfb807ed8dda6c9f26942b50ce626d (diff) | |
| download | ffwd-c4a1c97a09da6d8d88c2ac834bf6234506fc41bc.tar.gz ffwd-c4a1c97a09da6d8d88c2ac834bf6234506fc41bc.zip | |
chore: run clang-format
Diffstat (limited to 'tests/tok_fixture.h')
| -rw-r--r-- | tests/tok_fixture.h | 36 |
1 files changed, 18 insertions, 18 deletions
diff --git a/tests/tok_fixture.h b/tests/tok_fixture.h index 7890927..24e1563 100644 --- a/tests/tok_fixture.h +++ b/tests/tok_fixture.h @@ -10,15 +10,13 @@ #include <string.h> /* GPT-2 byte -> unicode codepoint mapping (mirrors the tokenizer's rules). */ -static int tf_byte_to_cp(int b) -{ +static int tf_byte_to_cp(int b) { static int table[256]; static int init = 0; if (!init) { int n = 0; for (int i = 0; i < 256; i++) { - int normal = (i >= 33 && i <= 126) || (i >= 161 && i <= 172) || - (i >= 174 && i <= 255); + int normal = (i >= 33 && i <= 126) || (i >= 161 && i <= 172) || (i >= 174 && i <= 255); table[i] = normal ? i : 256 + n++; } init = 1; @@ -26,9 +24,12 @@ static int tf_byte_to_cp(int b) return table[b]; } -static int tf_cp_to_utf8(int cp, char *out) -{ - if (cp < 0x80) { out[0] = (char)cp; out[1] = '\0'; return 1; } +static int tf_cp_to_utf8(int cp, char *out) { + if (cp < 0x80) { + out[0] = (char)cp; + out[1] = '\0'; + return 1; + } out[0] = (char)(0xC0 | (cp >> 6)); out[1] = (char)(0x80 | (cp & 0x3F)); out[2] = '\0'; @@ -36,11 +37,11 @@ static int tf_cp_to_utf8(int cp, char *out) } /* JSON-escape a byte-unicode token (the byte alphabet needs only \" and \\). */ -static const char *tf_json_escape(const char *s, char *buf, size_t cap) -{ +static const char *tf_json_escape(const char *s, char *buf, size_t cap) { size_t o = 0; for (; *s && o + 2 < cap; s++) { - if (*s == '"' || *s == '\\') buf[o++] = '\\'; + if (*s == '"' || *s == '\\') + buf[o++] = '\\'; buf[o++] = *s; } buf[o] = '\0'; @@ -52,8 +53,7 @@ static const char *TF_MERGED[] = { "he", "ll", "hell", "hello", "\xC4\xA0w", "or", "ld", "\xC4\xA0wor", }; static const char *TF_MERGES[] = { - "h e", "l l", "he ll", "hell o", "\xC4\xA0 w", "o r", "l d", - "\xC4\xA0w or", + "h e", "l l", "he ll", "hell o", "\xC4\xA0 w", "o r", "l d", "\xC4\xA0w or", }; enum { TF_N_MERGED = sizeof(TF_MERGED) / sizeof(TF_MERGED[0]) }; /* Special tokens go last so contextual and late-interaction ids resolve @@ -65,17 +65,16 @@ enum { TF_DOCUMENT_ID = TF_QUERY_ID + 1 }; enum { TF_VOCAB_SIZE = TF_DOCUMENT_ID + 1 }; /* Write vocab.json + merges.txt into dir. Returns 0 on success. */ -static int tf_write_vocab(const char *dir) -{ +static int tf_write_vocab(const char *dir) { char path[2048], tok[4], esc[16]; snprintf(path, sizeof(path), "%s/vocab.json", dir); FILE *f = fopen(path, "w"); - if (!f) return -1; + if (!f) + return -1; fputs("{", f); for (int b = 0; b < 256; b++) { tf_cp_to_utf8(tf_byte_to_cp(b), tok); - fprintf(f, "%s\"%s\": %d", b ? ", " : "", - tf_json_escape(tok, esc, sizeof(esc)), b); + fprintf(f, "%s\"%s\": %d", b ? ", " : "", tf_json_escape(tok, esc, sizeof(esc)), b); } for (int i = 0; i < TF_N_MERGED; i++) fprintf(f, ", \"%s\": %d", TF_MERGED[i], 256 + i); @@ -88,7 +87,8 @@ static int tf_write_vocab(const char *dir) snprintf(path, sizeof(path), "%s/merges.txt", dir); f = fopen(path, "w"); - if (!f) return -1; + if (!f) + return -1; for (size_t i = 0; i < sizeof(TF_MERGES) / sizeof(TF_MERGES[0]); i++) fprintf(f, "%s\n", TF_MERGES[i]); fclose(f); |
