Add spellcheck example

One of the benefits of implementing system call support from scratch is
that we're able to have embedded zip filesystem support which trickles
into libraries such as stdio, without unportable symbolic interposition.
It's also be great if we could say open("gs://bucket/object", O_RDONLY)
for seamless GCS, similar to Java NIO, but abstracted by the C library.
main
Justine Tunney 2020-06-27 12:00:43 -07:00
parent 0ad0408ac6
commit b5b60015f5
6 changed files with 180 additions and 1 deletions

View File

@ -114,6 +114,18 @@ o/$(MODE)/examples/hellojs.com.dbg: \
$(APE) $(APE)
@$(APELINK) @$(APELINK)
o/$(MODE)/examples/ispell.com.dbg: \
$(EXAMPLES_DEPS) \
o/$(MODE)/examples/ispell.o \
o/$(MODE)/usr/share/dict/words.zip.o \
o/$(MODE)/examples/examples.pkg \
$(CRT) \
$(APE)
@$(APELINK)
o/$(MODE)/usr/share/dict/words: usr/share/dict/words.gz
$(GZ) $(ZFLAGS) <$< >$@
.PHONY: o/$(MODE)/examples .PHONY: o/$(MODE)/examples
o/$(MODE)/examples: \ o/$(MODE)/examples: \
o/$(MODE)/examples/package \ o/$(MODE)/examples/package \

157
examples/ispell.c 100644
View File

@ -0,0 +1,157 @@
#if 0
/*─────────────────────────────────────────────────────────────────╗
To the extent possible under law, Justine Tunney has waived
all copyright and related or neighboring rights to this file,
as it is written in the following disclaimers:
http://unlicense.org/ │
http://creativecommons.org/publicdomain/zero/1.0/ │
*/
#endif
#include "libc/alg/alg.h"
#include "libc/alg/arraylist.h"
#include "libc/bits/bits.h"
#include "libc/calls/calls.h"
#include "libc/conv/conv.h"
#include "libc/log/check.h"
#include "libc/log/log.h"
#include "libc/macros.h"
#include "libc/mem/mem.h"
#include "libc/stdio/stdio.h"
#include "libc/str/str.h"
#include "libc/str/tpdecode.h"
#include "libc/sysv/consts/fileno.h"
#include "libc/x/x.h"
/**
* @fileoverview Simple Interactive Spell Checker.
*
* This is an attempt to get spell checking to work in Emacs across
* platforms. While the computer science behind spell checking is very
* simple, unfortunately Emacs probes all these System Five spell progs
* similar to how websites have been known to probe User-Agent strings.
*
* Here's how we believe the repl works:
*
* $ make -j8 o//examples/ispell.com
* $ o//examples/ispell.com
* @(#) Cosmopolitan Ispell Version 3.4.00 8 Feb 2015
* word: hello
* ok
* word: héllo
* how about: hills, hello, hilly, jello
* word: lova
* how about: diva, dona, dora, dove, elva, fora, hove, iota
* word: hecruhecrue
* not found
*
* The dictionary for this program is embedded as a text file within the
* zip structure of the binary. It can be edited after distribution.
*
* It's possible to go even fancier than what this code is doing, by
* using cmudict phonemes, bloom filters, unicode tables e.g. e vs. é,
* and even doing ML similar to Google's online spell checker.
*
* TODO: Figure out why Emacs rejects this interface.
*/
#define MISSING_LETTER_DISTANCE 5
#define MAX_NEARBY_WORD_DISTANCE 6
#define MAX_NEARBY_RESULTS 8
FILE *f;
char *line;
size_t linesize;
const char *query;
struct critbit0 words; /* does O(log 𝑛) fast path lookup */
struct NearbyWords {
size_t i, n;
struct WordMatch {
long dist;
char *word;
} * p;
} nearby;
long WordDistance(const char *a, const char *b) {
long dist;
int gota, gotb;
unsigned long h, p;
wint_t chara, charb;
dist = p = 0;
for (;;) {
gota = abs(tpdecode(a, &chara)); /* parses utf-8 multibyte characters */
gotb = abs(tpdecode(b, &charb)); /* abs() handles -1, always yields <20> */
if (!chara && !charb) break;
if (!chara || !charb) {
dist += MISSING_LETTER_DISTANCE;
} else if ((h = hamming(chara, charb))) {
dist += h + p++; /* penalize multiple mismatched letters */
}
if (chara) a += gota;
if (charb) b += gotb;
}
return dist;
}
intptr_t ConsiderWord(const char *word, void *arg) {
long dist;
if ((dist = WordDistance(word, query)) < MAX_NEARBY_WORD_DISTANCE) {
append(&nearby, &((struct WordMatch){dist, word}));
}
return 0;
}
int CompareWords(const struct WordMatch *a, const struct WordMatch *b) {
return a->dist < b->dist;
}
void FindNearbyWords(void) {
nearby.i = 0;
critbit0_allprefixed(&words, "", ConsiderWord, NULL);
qsort(nearby.p, nearby.i, sizeof(struct WordMatch), (void *)CompareWords);
}
void SpellChecker(void) {
int i;
printf("@(#) Cosmopolitan Ispell Version 3.4.00 8 Feb 2015\r\n");
while (!feof(stdin)) {
printf("word: ");
fflush(stdout);
if (getline(&line, &linesize, stdin) > 0) {
query = strtolower(chomp(line));
if (critbit0_contains(&words, query)) {
printf("ok\r\n");
} else {
FindNearbyWords();
if (nearby.i) {
printf("how about: ");
for (i = 0; i < MIN(MAX_NEARBY_RESULTS, nearby.i); ++i) {
if (i) printf(", ");
fputs(nearby.p[i].word, stdout);
}
printf("\r\n");
} else {
printf("not found\r\n");
}
}
}
printf("\r\n");
}
CHECK_NE(-1, fclose(stdin));
CHECK_NE(-1, fclose(stdout));
}
void LoadWords(void) {
CHECK_NOTNULL((f = fopen("zip:usr/share/dict/words", "r")));
while (getline(&line, &linesize, f) > 0) {
critbit0_insert(&words, strtolower(chomp(line)));
}
CHECK_NE(-1, fclose(f));
}
int main(int argc, char *argv[]) {
showcrashreports();
LoadWords();
SpellChecker();
return 0;
}

View File

@ -36,6 +36,7 @@ uint64_t bitreverse64(uint64_t) libcesque pureconst;
unsigned long roundup2pow(unsigned long) libcesque pureconst; unsigned long roundup2pow(unsigned long) libcesque pureconst;
unsigned long roundup2log(unsigned long) libcesque pureconst; unsigned long roundup2log(unsigned long) libcesque pureconst;
unsigned long rounddown2pow(unsigned long) libcesque pureconst; unsigned long rounddown2pow(unsigned long) libcesque pureconst;
unsigned long hamming(unsigned long, unsigned long) pureconst;
/*───────────────────────────────────────────────────────────────────────────│─╗ /*───────────────────────────────────────────────────────────────────────────│─╗
cosmopolitan § bits » no assembly required cosmopolitan § bits » no assembly required

View File

@ -19,6 +19,9 @@
*/ */
#include "libc/bits/bits.h" #include "libc/bits/bits.h"
unsigned long(hamming)(unsigned long x, unsigned long y) { /**
* Counts number of different bits.
*/
unsigned long hamming(unsigned long x, unsigned long y) {
return popcount(x ^ y); return popcount(x ^ y);
} }

View File

@ -30,6 +30,12 @@
wint_t wc; wint_t wc;
TEST(tpdecode, testEmptyString_consumesNulTerminator) {
wc = 123;
EXPECT_EQ(1, tpdecode("", &wc));
EXPECT_EQ(0, wc);
}
TEST(tpdecode, testGlyph) { TEST(tpdecode, testGlyph) {
EXPECT_EQ(u'', PROGN(ASSERT_EQ(3, tpdecode("", &wc)), wc)); EXPECT_EQ(u'', PROGN(ASSERT_EQ(3, tpdecode("", &wc)), wc));
EXPECT_EQ(L'𐌰', PROGN(ASSERT_EQ(4, tpdecode("𐌰𐌱𐌲𐌳", &wc)), wc)); EXPECT_EQ(L'𐌰', PROGN(ASSERT_EQ(4, tpdecode("𐌰𐌱𐌲𐌳", &wc)), wc));

Binary file not shown.