neil/interpret
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

KJV updated to test.

Browse Source
This commit is contained in:
Neil 2022-12-27 12:28:04 -08:00
parent 3d334cd169
commit afec7f716d
4 changed files with 136 additions and 83 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
Makefile
View File

@ -36,7 +36,7 @@ default: $(projects)
# success # success
bin/text: build/text.o build/test_text.o bin/text: build/text.o build/test_text.o
bin/kjv: build/text.o build/kjv.o bin/kjv: build/text.o build/kjv.o build/test_kjv.o
bin/%: bin/%:
# linking test $@ # linking test $@

30
src/kjv.h
View File

@ -1,3 +1,4 @@
#ifndef KJV_OMIT_BASE /* <!-- base */
#define BOOKS \ #define BOOKS \
X(Genesis),\ X(Genesis),\
X(Exodus),\ X(Exodus),\
@ -74,11 +75,34 @@ enum kjv_book { BOOKS };
static const char *kjv_book_string[] = { BOOKS }; static const char *kjv_book_string[] = { BOOKS };
#undef X #undef X
#undef BOOKS #undef BOOKS
#include <stdint.h> /* C99 */ #include <stdint.h> /* C99 */
/* A set of verses. */
union kjvcite { union kjvcite {
struct { unsigned verse : 12, chapter : 13, book : 7; }; /* C11 */ struct { unsigned verse : 12, chapter : 13, book : 7; }; /* C11 */
uint32_t u32; uint32_t u32;
}; };
#else /* base --><!-- !base */
#undef KJV_OMIT_BASE
#endif /* !base --> */
#ifndef KJV_OMIT_VERSES /* <!-- verses: For external inclusion. */
struct table_verse_bucket;
struct verse_table {
struct table_verse_bucket *buckets;
uint32_t log_capacity, size, top;
};
#else /* verses --><!-- !verses */
#undef KJV_OMIT_VERSES
#endif /* !verses --> */
#ifndef KJV_OMIT_PROTO /* <!-- proto */
#include <stddef.h>
struct kjv { struct verse_table verses; size_t words; };
struct kjv kjv(void);
void kjv_(struct kjv *);
unsigned kjv_words(/*const*/ struct kjv *const kjv, const union kjvcite cite);
const char *kjv_entire_to_string(const struct kjv *const kjv);
#else /* proto --><!-- !proto */
#undef KJV_OMIT_PROTO
#endif /* !proto --> */

158
src/kjv.re_c.c
View File

@ -8,13 +8,65 @@
@std C11 */ @std C11 */
#include "../src/text.h" #include "../src/text.h"
#include "../src/kjv.h" #define KJV_OMIT_VERSES
#include <stdlib.h> #define KJV_OMIT_PROTO
#include "../src/kjv.h" /* Just the base data. */
#include <stdio.h> #include <stdio.h>
#include <assert.h> #include <stdlib.h>
#include <errno.h> #include <errno.h>
#include <assert.h>
#include <dirent.h> /* opendir readdir closedir */ #include <dirent.h> /* opendir readdir closedir */
#include <unistd.h> /* chdir (POSIX) (because I'm lazy) */ #include <unistd.h> /* chdir (POSIX) (because I'm lazy) */
/* #include <cmph.h> No; overkill. */
/* Reversible hash map. */
/** <https://nullprogram.com/blog/2018/07/31/>
<https://github.com/skeeto/hash-prospector> on `x`. */
static uint32_t lowbias32(uint32_t x) {
x ^= x >> 16;
x *= 0x7feb352dU;
x ^= x >> 15;
x *= 0x846ca68bU;
x ^= x >> 16;
return x;
}
/* Inverts `x`. */
static uint32_t lowbias32_r(uint32_t x) {
x ^= x >> 16;
x *= 0x43021123U;
x ^= x >> 15 ^ x >> 30;
x *= 0x1d69e2a5U;
x ^= x >> 16;
return x;
}
static uint32_t kjv_hash(const union kjvcite x) { return lowbias32(x.u32); }
static union kjvcite kjv_unhash(const uint32_t x)
{ union kjvcite k; k.u32 = lowbias32_r(x); return k; }
static void kjv_to_string(const union kjvcite x, char (*const a)[12])
{ sprintf(*a, "%.4s%u:%u", kjv_book_string[x.book],
(x.chapter + 1) % 1000, (x.verse + 1) % 1000); }
#define TABLE_NAME kjv
#define TABLE_KEY union kjvcite
#define TABLE_UINT uint32_t
#define TABLE_INVERSE
#define TABLE_TO_STRING
#include "../src/table.h"
/* Derived information on verse word count. */
static uint32_t verse_hash(const union kjvcite x) { return kjv_hash(x); }
static union kjvcite verse_unhash(const uint32_t x) { return kjv_unhash(x); }
static void verse_to_string(const union kjvcite x, const unsigned count,
char (*const a)[12]) { (void)count; kjv_to_string(x, a); }
#define TABLE_NAME verse
#define TABLE_KEY union kjvcite
#define TABLE_UINT uint32_t
#define TABLE_VALUE unsigned /* Count words. */
#define TABLE_INVERSE
#define TABLE_DEFAULT 0
#define TABLE_TO_STRING
#include "../src/table.h"
/** Helper to parse unsigned; [`s`,`e`) => `n`. */ /** Helper to parse unsigned; [`s`,`e`) => `n`. */
@ -101,69 +153,24 @@ scan:
} }
/* Reversible hash map. */ #define KJV_OMIT_BASE
#include <stdint.h> /* C99 */ #define KJV_OMIT_VERSES
/** <https://nullprogram.com/blog/2018/07/31/> #include "../src/kjv.h" /* Just the kjv and prototypes. */
<https://github.com/skeeto/hash-prospector> on `x`. */
static uint32_t lowbias32(uint32_t x) { void kjv_(struct kjv *const kjv) {
x ^= x >> 16; if(!kjv) return;
x *= 0x7feb352dU; verse_table_(&kjv->verses);
x ^= x >> 15;
x *= 0x846ca68bU;
x ^= x >> 16;
return x;
}
/* Inverts `x`. */
static uint32_t lowbias32_r(uint32_t x) {
x ^= x >> 16;
x *= 0x43021123U;
x ^= x >> 15 ^ x >> 30;
x *= 0x1d69e2a5U;
x ^= x >> 16;
return x;
} }
static uint32_t kjv_hash(const union kjvcite x) { return lowbias32(x.u32); } struct kjv kjv(void) {
static union kjvcite kjv_unhash(const uint32_t x)
{ union kjvcite k; k.u32 = lowbias32_r(x); return k; }
static void kjv_to_string(const union kjvcite x, char (*const a)[12])
{ sprintf(*a, "%.4s%u:%u", kjv_book_string[x.book],
(x.chapter + 1) % 1000, (x.verse + 1) % 1000); }
#define TABLE_NAME kjv
#define TABLE_KEY union kjvcite
#define TABLE_UINT uint32_t
#define TABLE_INVERSE
#define TABLE_TO_STRING
#include "../src/table.h"
/* Derived information on verse word count. */
static uint32_t verse_hash(const union kjvcite x) { return kjv_hash(x); }
static union kjvcite verse_unhash(const uint32_t x) { return kjv_unhash(x); }
static void verse_to_string(const union kjvcite x, const unsigned count,
char (*const a)[12]) { (void)count; kjv_to_string(x, a); }
#define TABLE_NAME verse
#define TABLE_KEY union kjvcite
#define TABLE_UINT uint32_t
#define TABLE_VALUE unsigned /* Count words. */
#define TABLE_INVERSE
#define TABLE_DEFAULT 0
#define TABLE_TO_STRING
#include "../src/table.h"
int main(void) {
const char *const dir_kjv = "kjv"; const char *const dir_kjv = "kjv";
struct { struct text backing;
struct text backing; struct kjv kjv = { 0 };
struct verse_table verses;
size_t words;
} kjv = { 0 };
DIR *dir = 0; DIR *dir = 0;
struct dirent *de = 0; struct dirent *de = 0;
struct { size_t offset; int is; } build[KJV_BOOK_SIZE] = { 0 }; struct { size_t offset; int is; } build[KJV_BOOK_SIZE] = { 0 };
enum kjv_book b = 0; enum kjv_book b = 0;
int success = EXIT_SUCCESS, attempted_closedir = 0; int attempted_closedir = 0;
errno = 0;
/* For all files in directory KJV with <#>*.txt, read into backing. */ /* For all files in directory KJV with <#>*.txt, read into backing. */
if(chdir(dir_kjv) == -1 || !(dir = opendir("."))) goto catch; if(chdir(dir_kjv) == -1 || !(dir = opendir("."))) goto catch;
@ -176,10 +183,10 @@ int main(void) {
{ errno = ERANGE; goto catch; } /* Not in range. */ { errno = ERANGE; goto catch; } /* Not in range. */
if(build[b = ordinal - 1].is) /* Convert to zero-based. */ if(build[b = ordinal - 1].is) /* Convert to zero-based. */
{ errno = EDOM; goto catch; } /* Is duplicate. */ { errno = EDOM; goto catch; } /* Is duplicate. */
if(!(unstable_book = text_append_file(&kjv.backing, de->d_name))) if(!(unstable_book = text_append_file(&backing, de->d_name)))
goto catch; goto catch;
build[b].is = 1; build[b].is = 1;
build[b].offset = (size_t)(unstable_book - kjv.backing.a.data); build[b].offset = (size_t)(unstable_book - backing.a.data);
} }
if(attempted_closedir = 1, closedir(dir) == -1) goto catch; dir = 0; if(attempted_closedir = 1, closedir(dir) == -1) goto catch; dir = 0;
@ -188,7 +195,7 @@ int main(void) {
struct lex x; struct lex x;
if(!build[b].is) { fprintf(stderr, "Missing book [%u]%s.\n", if(!build[b].is) { fprintf(stderr, "Missing book [%u]%s.\n",
b + 1, kjv_book_string[b]); errno = EDOM; goto catch; } b + 1, kjv_book_string[b]); errno = EDOM; goto catch; }
x = lex(kjv.backing.a.data + build[b].offset); x = lex(backing.a.data + build[b].offset);
while(lex_next_verse(&x)) { while(lex_next_verse(&x)) {
const union kjvcite cite const union kjvcite cite
= { .book = b, .chapter = x.chapter, .verse = x.verse }; = { .book = b, .chapter = x.chapter, .verse = x.verse };
@ -204,27 +211,20 @@ int main(void) {
if(x.error) { fprintf(stderr, "[%u]%s on line %zu\n", if(x.error) { fprintf(stderr, "[%u]%s on line %zu\n",
b + 1, kjv_book_string[b], x.line); goto catch; } b + 1, kjv_book_string[b], x.line); goto catch; }
} }
printf("words: %s\n", verse_table_to_string(&kjv.verses));
printf("kjv: %zu total words\n", kjv.words);
{
union kjvcite c;
struct verse_table_iterator it = verse_table_begin(&kjv.verses);
unsigned *w;
while(verse_table_next(&it, &c, &w))
printf("%s %u:%u -> %u\n",
kjv_book_string[c.book], c.chapter, c.verse, *w);
c = (union kjvcite){ .book = Genesis, .chapter = 1, .verse = 1 };
printf("1:1:1 -> %u\n", verse_table_get(&kjv.verses, c));
}
goto finally; goto finally;
catch: catch:
success = EXIT_FAILURE;
if(de) fprintf(stderr, "While reading %s.\n", de->d_name); if(de) fprintf(stderr, "While reading %s.\n", de->d_name);
perror(de ? de->d_name : dir_kjv); perror(de ? de->d_name : dir_kjv);
if(dir && !attempted_closedir && closedir(dir) == -1) perror(dir_kjv); if(dir && !attempted_closedir && closedir(dir) == -1) perror(dir_kjv);
kjv_(&kjv);
finally: finally:
verse_table_(&kjv.verses); text_(&backing);
text_(&kjv.backing); return kjv;
return success;
} }
unsigned kjv_words(/*const*/ struct kjv *const kjv, const union kjvcite cite) {
return kjv && verse_table_get(&kjv->verses, cite);
}
const char *kjv_entire_to_string(const struct kjv *const kjv)
{ return kjv ? verse_table_to_string(&kjv->verses) : ""; }

29
test/test_kjv.c Normal file
View File

@ -0,0 +1,29 @@
#include "../src/kjv.h"
#include <stdio.h>
#include <inttypes.h> /* C99 */
#include <stdlib.h>
int main(void) {
struct kjv k = kjv();
union kjvcite c;
/*struct verse_table_iterator it = verse_table_begin(&kjv.verses);*/
unsigned *verses;
/* printf("%%{\n"
"#include \"../src/kjv.h\"\n"
"%%}\n"
"struct kjvverse { uint32_t name; unsigned words; };\n"
"%%%%\n"); I don't even think this is possible 1.44 bits/key. */
/*while(verse_table_next(&it, &c, &verses))
printf("%" PRIu32 ", %u\n", c.u32, *verses);*/
/*printf("%s %u:%u -> %u\n",
kjv_book_string[c.book], c.chapter, c.verse, *verses);*/
/*c = (union kjvcite){ .book = Genesis, .chapter = 1, .verse = 1 };
printf("1:1:1 -> %u\n", verse_table_get(&kjv.verses, c));*/
/*c = (union kjvcite){ .book = Genesis, .chapter = 1, .verse = 1 };
printf("1:1:1 -> %u\n", verse_table_get(&kjv.verses, c));*/
fprintf(stderr, "%zu total words, %s\n", k.words, kjv_entire_to_string(&k));
kjv_(&k);
return EXIT_SUCCESS;
}