version 1.65, 2013/06/06 19:00:28 |
version 1.66, 2013/06/07 01:01:34 |
|
|
struct str { |
struct str { |
char *utf8; /* key in UTF-8 form */ |
char *utf8; /* key in UTF-8 form */ |
const struct of *of; /* if set, the owning parse */ |
const struct of *of; /* if set, the owning parse */ |
struct str *next; /* next in owning parse sequence */ |
|
uint64_t mask; /* bitmask in sequence */ |
uint64_t mask; /* bitmask in sequence */ |
char key[]; /* the string itself */ |
char key[]; /* the string itself */ |
}; |
}; |
|
|
enum form dform; /* path-cued form */ |
enum form dform; /* path-cued form */ |
enum form sform; /* suffix-cued form */ |
enum form sform; /* suffix-cued form */ |
char file[PATH_MAX]; /* filename rel. to manpath */ |
char file[PATH_MAX]; /* filename rel. to manpath */ |
const char *desc; /* parsed description */ |
char *desc; /* parsed description */ |
const char *sec; /* suffix-cued section (or empty) */ |
char *name; /* name (from filename) (not empty) */ |
const char *dsec; /* path-cued section (or empty) */ |
char *sec; /* suffix-cued section (or empty) */ |
const char *arch; /* path-cued arch. (or empty) */ |
char *dsec; /* path-cued section (or empty) */ |
const char *name; /* name (from filename) (not empty) */ |
char *arch; /* path-cued arch. (or empty) */ |
}; |
}; |
|
|
enum stmt { |
enum stmt { |
Line 138 static int inocheck(const struct stat *); |
|
Line 137 static int inocheck(const struct stat *); |
|
static void ofadd(int, const char *, const char *, const char *, |
static void ofadd(int, const char *, const char *, const char *, |
const char *, const char *, const struct stat *); |
const char *, const char *, const struct stat *); |
static void offree(void); |
static void offree(void); |
static void ofmerge(struct mchars *, struct mparse *); |
static void ofmerge(struct mchars *, struct mparse *, |
|
struct ohash_info*); |
static void parse_catpage(struct of *); |
static void parse_catpage(struct of *); |
static void parse_man(struct of *, const struct man_node *); |
static void parse_man(struct of *, const struct man_node *); |
static void parse_mdoc(struct of *, const struct mdoc_node *); |
static void parse_mdoc(struct of *, const struct mdoc_node *); |
Line 160 static void putkeys(const struct of *, |
|
Line 160 static void putkeys(const struct of *, |
|
static void putmdockey(const struct of *, |
static void putmdockey(const struct of *, |
const struct mdoc_node *, uint64_t); |
const struct mdoc_node *, uint64_t); |
static void say(const char *, const char *, ...); |
static void say(const char *, const char *, ...); |
static char *stradd(const char *); |
|
static char *stradds(const char *, size_t); |
|
static int treescan(void); |
static int treescan(void); |
static size_t utf8(unsigned int, char [7]); |
static size_t utf8(unsigned int, char [7]); |
static void utf8key(struct mchars *, struct str *); |
static void utf8key(struct mchars *, struct str *); |
Line 178 static struct ohash inos; /* table of inodes/devices |
|
Line 176 static struct ohash inos; /* table of inodes/devices |
|
static struct ohash filenames; /* table of filenames */ |
static struct ohash filenames; /* table of filenames */ |
static struct ohash strings; /* table of all strings */ |
static struct ohash strings; /* table of all strings */ |
static struct of *ofs = NULL; /* vector of files to parse */ |
static struct of *ofs = NULL; /* vector of files to parse */ |
static struct str *words = NULL; /* word list in current parse */ |
|
static sqlite3 *db = NULL; /* current database */ |
static sqlite3 *db = NULL; /* current database */ |
static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */ |
static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */ |
|
|
|
|
main(int argc, char *argv[]) |
main(int argc, char *argv[]) |
{ |
{ |
int ch, i; |
int ch, i; |
unsigned int slot; |
|
size_t j, sz; |
size_t j, sz; |
const char *path_arg; |
const char *path_arg; |
struct str *s; |
|
struct mchars *mc; |
struct mchars *mc; |
struct manpaths dirs; |
struct manpaths dirs; |
struct mparse *mp; |
struct mparse *mp; |
Line 406 main(int argc, char *argv[]) |
|
Line 401 main(int argc, char *argv[]) |
|
MANDOCLEVEL_FATAL, NULL, NULL, NULL); |
MANDOCLEVEL_FATAL, NULL, NULL, NULL); |
mc = mchars_alloc(); |
mc = mchars_alloc(); |
|
|
ohash_init(&strings, 6, &str_info); |
|
ohash_init(&inos, 6, &ino_info); |
ohash_init(&inos, 6, &ino_info); |
ohash_init(&filenames, 6, &filename_info); |
ohash_init(&filenames, 6, &filename_info); |
|
|
Line 430 main(int argc, char *argv[]) |
|
Line 424 main(int argc, char *argv[]) |
|
if (OP_TEST != op) |
if (OP_TEST != op) |
dbprune(); |
dbprune(); |
if (OP_DELETE != op) |
if (OP_DELETE != op) |
ofmerge(mc, mp); |
ofmerge(mc, mp, &str_info); |
dbclose(1); |
dbclose(1); |
} else { |
} else { |
/* |
/* |
Line 460 main(int argc, char *argv[]) |
|
Line 454 main(int argc, char *argv[]) |
|
dirs.paths[j][--sz] = '\0'; |
dirs.paths[j][--sz] = '\0'; |
if (0 == sz) |
if (0 == sz) |
continue; |
continue; |
|
|
|
if (j) { |
|
ohash_init(&inos, 6, &ino_info); |
|
ohash_init(&filenames, 6, &filename_info); |
|
} |
|
|
if (0 == set_basedir(dirs.paths[j])) |
if (0 == set_basedir(dirs.paths[j])) |
goto out; |
goto out; |
if (0 == treescan()) |
if (0 == treescan()) |
Line 478 main(int argc, char *argv[]) |
|
Line 478 main(int argc, char *argv[]) |
|
SQL_EXEC("PRAGMA synchronous = OFF"); |
SQL_EXEC("PRAGMA synchronous = OFF"); |
#endif |
#endif |
|
|
ofmerge(mc, mp); |
ofmerge(mc, mp, &str_info); |
dbclose(0); |
dbclose(0); |
offree(); |
|
ohash_delete(&inos); |
if (j + 1 < dirs.sz) { |
ohash_init(&inos, 6, &ino_info); |
ohash_delete(&inos); |
ohash_delete(&filenames); |
ohash_delete(&filenames); |
ohash_init(&filenames, 6, &filename_info); |
offree(); |
|
} |
} |
} |
} |
} |
out: |
out: |
|
|
manpath_free(&dirs); |
manpath_free(&dirs); |
mchars_free(mc); |
mchars_free(mc); |
mparse_free(mp); |
mparse_free(mp); |
for (s = ohash_first(&strings, &slot); NULL != s; |
|
s = ohash_next(&strings, &slot)) { |
|
if (s->utf8 != s->key) |
|
free(s->utf8); |
|
free(s); |
|
} |
|
ohash_delete(&strings); |
|
ohash_delete(&inos); |
ohash_delete(&inos); |
ohash_delete(&filenames); |
ohash_delete(&filenames); |
offree(); |
offree(); |
|
|
FTSENT *ff; |
FTSENT *ff; |
int dform; |
int dform; |
char *sec; |
char *sec; |
const char *dsec, *arch, *cp, *name, *path; |
const char *dsec, *arch, *cp, *path; |
const char *argv[2]; |
const char *argv[2]; |
|
|
argv[0] = "."; |
argv[0] = "."; |
|
|
if (warnings) |
if (warnings) |
say(path, "Wrong filename suffix"); |
say(path, "Wrong filename suffix"); |
continue; |
continue; |
} else { |
} else |
sec[-1] = '\0'; |
sec[-1] = '\0'; |
sec = stradd(sec); |
ofadd(dform, path, ff->fts_name, dsec, sec, |
} |
arch, ff->fts_statp); |
name = stradd(ff->fts_name); |
|
ofadd(dform, path, |
|
name, dsec, sec, arch, ff->fts_statp); |
|
continue; |
continue; |
} else if (FTS_D != ff->fts_info && |
} else if (FTS_D != ff->fts_info && |
FTS_DP != ff->fts_info) { |
FTS_DP != ff->fts_info) { |
|
|
|
|
if (0 == strncmp(cp, "man", 3)) { |
if (0 == strncmp(cp, "man", 3)) { |
dform = FORM_SRC; |
dform = FORM_SRC; |
dsec = stradd(cp + 3); |
dsec = cp + 3; |
} else if (0 == strncmp(cp, "cat", 3)) { |
} else if (0 == strncmp(cp, "cat", 3)) { |
dform = FORM_CAT; |
dform = FORM_CAT; |
dsec = stradd(cp + 3); |
dsec = cp + 3; |
} |
} |
|
|
if (NULL != dsec || use_all) |
if (NULL != dsec || use_all) |
|
|
*/ |
*/ |
arch = NULL; |
arch = NULL; |
if (FTS_DP != ff->fts_info && NULL != dsec) |
if (FTS_DP != ff->fts_info && NULL != dsec) |
arch = stradd(ff->fts_name); |
arch = ff->fts_name; |
break; |
break; |
default: |
default: |
if (FTS_DP == ff->fts_info || use_all) |
if (FTS_DP == ff->fts_info || use_all) |
Line 867 ofadd(int dform, const char *file, const char *name, c |
|
Line 858 ofadd(int dform, const char *file, const char *name, c |
|
|
|
of = mandoc_calloc(1, sizeof(struct of)); |
of = mandoc_calloc(1, sizeof(struct of)); |
strlcpy(of->file, file, PATH_MAX); |
strlcpy(of->file, file, PATH_MAX); |
of->name = name; |
of->name = mandoc_strdup(name); |
of->sec = sec; |
of->sec = mandoc_strdup(sec); |
of->dsec = dsec; |
of->dsec = mandoc_strdup(dsec); |
of->arch = arch; |
of->arch = mandoc_strdup(arch); |
of->sform = sform; |
of->sform = sform; |
of->dform = dform; |
of->dform = dform; |
of->next = ofs; |
of->next = ofs; |
|
|
|
|
while (NULL != (of = ofs)) { |
while (NULL != (of = ofs)) { |
ofs = of->next; |
ofs = of->next; |
|
free(of->name); |
|
free(of->sec); |
|
free(of->dsec); |
|
free(of->arch); |
free(of); |
free(of); |
} |
} |
} |
} |
|
|
* and filename to determine whether the file is parsable or not. |
* and filename to determine whether the file is parsable or not. |
*/ |
*/ |
static void |
static void |
ofmerge(struct mchars *mc, struct mparse *mp) |
ofmerge(struct mchars *mc, struct mparse *mp, |
|
struct ohash_info *infop) |
{ |
{ |
int form; |
int form; |
size_t sz; |
size_t sz; |
Line 945 ofmerge(struct mchars *mc, struct mparse *mp) |
|
Line 941 ofmerge(struct mchars *mc, struct mparse *mp) |
|
} |
} |
} |
} |
|
|
words = NULL; |
ohash_init(&strings, 6, infop); |
mparse_reset(mp); |
mparse_reset(mp); |
mdoc = NULL; |
mdoc = NULL; |
man = NULL; |
man = NULL; |
Line 1023 ofmerge(struct mchars *mc, struct mparse *mp) |
|
Line 1019 ofmerge(struct mchars *mc, struct mparse *mp) |
|
if (NULL != mdoc) { |
if (NULL != mdoc) { |
if (NULL != (cp = mdoc_meta(mdoc)->name)) |
if (NULL != (cp = mdoc_meta(mdoc)->name)) |
putkey(of, cp, TYPE_Nm); |
putkey(of, cp, TYPE_Nm); |
|
assert(NULL == of->desc); |
parse_mdoc(of, mdoc_node(mdoc)); |
parse_mdoc(of, mdoc_node(mdoc)); |
|
putkey(of, NULL != of->desc ? |
|
of->desc : of->name, TYPE_Nd); |
} else if (NULL != man) |
} else if (NULL != man) |
parse_man(of, man_node(man)); |
parse_man(of, man_node(man)); |
else |
else |
parse_catpage(of); |
parse_catpage(of); |
|
|
dbindex(mc, form, of); |
dbindex(mc, form, of); |
|
ohash_delete(&strings); |
} |
} |
} |
} |
|
|
Line 1096 parse_catpage(struct of *of) |
|
Line 1096 parse_catpage(struct of *of) |
|
if (NULL == title || '\0' == *title) { |
if (NULL == title || '\0' == *title) { |
if (warnings) |
if (warnings) |
say(of->file, "Cannot find NAME section"); |
say(of->file, "Cannot find NAME section"); |
|
assert(NULL == of->desc); |
|
of->desc = mandoc_strdup(of->name); |
putkey(of, of->name, TYPE_Nd); |
putkey(of, of->name, TYPE_Nd); |
fclose(stream); |
fclose(stream); |
free(title); |
free(title); |
Line 1134 parse_catpage(struct of *of) |
|
Line 1136 parse_catpage(struct of *of) |
|
plen -= 2; |
plen -= 2; |
} |
} |
|
|
of->desc = stradd(p); |
assert(NULL == of->desc); |
putkey(of, p, TYPE_Nd); |
of->desc = mandoc_strdup(p); |
|
putkey(of, of->desc, TYPE_Nd); |
fclose(stream); |
fclose(stream); |
free(title); |
free(title); |
} |
} |
Line 1299 parse_man(struct of *of, const struct man_node *n) |
|
Line 1302 parse_man(struct of *of, const struct man_node *n) |
|
start++; |
start++; |
|
|
assert(NULL == of->desc); |
assert(NULL == of->desc); |
of->desc = stradd(start); |
of->desc = mandoc_strdup(start); |
putkey(of, start, TYPE_Nd); |
putkey(of, of->desc, TYPE_Nd); |
free(title); |
free(title); |
return; |
return; |
} |
} |
|
|
parse_mdoc_Nd(struct of *of, const struct mdoc_node *n) |
parse_mdoc_Nd(struct of *of, const struct mdoc_node *n) |
{ |
{ |
size_t sz; |
size_t sz; |
char *sv, *desc; |
|
|
|
if (MDOC_BODY != n->type) |
if (MDOC_BODY != n->type) |
return(0); |
return(0); |
Line 1465 parse_mdoc_Nd(struct of *of, const struct mdoc_node *n |
|
Line 1467 parse_mdoc_Nd(struct of *of, const struct mdoc_node *n |
|
* into the document table. |
* into the document table. |
*/ |
*/ |
|
|
desc = NULL; |
|
for (n = n->child; NULL != n; n = n->next) { |
for (n = n->child; NULL != n; n = n->next) { |
if (MDOC_TEXT == n->type) { |
if (MDOC_TEXT == n->type) { |
sz = strlen(n->string) + 1; |
if (NULL != of->desc) { |
if (NULL != (sv = desc)) |
sz = strlen(of->desc) + |
sz += strlen(desc) + 1; |
strlen(n->string) + 2; |
desc = mandoc_realloc(desc, sz); |
of->desc = mandoc_realloc(of->desc, sz); |
if (NULL != sv) |
strlcat(of->desc, " ", sz); |
strlcat(desc, " ", sz); |
strlcat(of->desc, n->string, sz); |
else |
} else |
*desc = '\0'; |
of->desc = mandoc_strdup(n->string); |
strlcat(desc, n->string, sz); |
|
} |
} |
if (NULL != n->child) |
if (NULL != n->child) |
parse_mdoc_Nd(of, n); |
parse_mdoc_Nd(of, n); |
} |
} |
|
|
of->desc = NULL != desc ? stradd(desc) : NULL; |
|
free(desc); |
|
return(1); |
return(1); |
} |
} |
|
|
Line 1521 parse_mdoc_body(struct of *of, const struct mdoc_node |
|
Line 1518 parse_mdoc_body(struct of *of, const struct mdoc_node |
|
} |
} |
|
|
/* |
/* |
* See stradds(). |
* Add a string to the hash table for the current manual. |
|
* Each string has a bitmask telling which macros it belongs to. |
|
* When we finish the manual, we'll dump the table. |
*/ |
*/ |
static char * |
|
stradd(const char *cp) |
|
{ |
|
|
|
return(stradds(cp, strlen(cp))); |
|
} |
|
|
|
/* |
|
* This looks up or adds a string to the string table. |
|
* The string table is a table of all strings encountered during parse |
|
* or file scan. |
|
* In using it, we avoid having thousands of (e.g.) "cat1" string |
|
* allocations for the "of" table. |
|
* We also have a layer atop the string table for keeping track of words |
|
* in a parse sequence (see putkeys()). |
|
*/ |
|
static char * |
|
stradds(const char *cp, size_t sz) |
|
{ |
|
struct str *s; |
|
unsigned int slot; |
|
const char *end; |
|
|
|
end = cp + sz; |
|
slot = ohash_qlookupi(&strings, cp, &end); |
|
if (NULL != (s = ohash_find(&strings, slot))) |
|
return(s->key); |
|
|
|
s = mandoc_calloc(sizeof(struct str) + sz + 1, 1); |
|
memcpy(s->key, cp, sz); |
|
ohash_insert(&strings, slot, s); |
|
return(s->key); |
|
} |
|
|
|
/* |
|
* Add a word to the current parse sequence. |
|
* Within the hashtable of strings, we maintain a list of strings that |
|
* are currently indexed. |
|
* Each of these ("words") has a bitmask modified within the parse. |
|
* When we finish a parse, we'll dump the list, then remove the head |
|
* entry -- since the next parse will have a new "of", it can keep track |
|
* of its entries without conflict. |
|
*/ |
|
static void |
static void |
putkeys(const struct of *of, const char *cp, size_t sz, uint64_t v) |
putkeys(const struct of *of, const char *cp, size_t sz, uint64_t v) |
{ |
{ |
Line 1588 putkeys(const struct of *of, const char *cp, size_t sz |
|
Line 1544 putkeys(const struct of *of, const char *cp, size_t sz |
|
memcpy(s->key, cp, sz); |
memcpy(s->key, cp, sz); |
ohash_insert(&strings, slot, s); |
ohash_insert(&strings, slot, s); |
} |
} |
|
|
s->next = words; |
|
s->of = of; |
s->of = of; |
s->mask = v; |
s->mask = v; |
words = s; |
|
} |
} |
|
|
/* |
/* |
Line 1764 dbindex(struct mchars *mc, int form, const struct of * |
|
Line 1717 dbindex(struct mchars *mc, int form, const struct of * |
|
const char *desc; |
const char *desc; |
int64_t recno; |
int64_t recno; |
size_t i; |
size_t i; |
|
unsigned int slot; |
|
|
if (verb) |
if (verb) |
say(of->file, "Adding to index"); |
say(of->file, "Adding to index"); |
Line 1793 dbindex(struct mchars *mc, int form, const struct of * |
|
Line 1747 dbindex(struct mchars *mc, int form, const struct of * |
|
recno = sqlite3_last_insert_rowid(db); |
recno = sqlite3_last_insert_rowid(db); |
sqlite3_reset(stmts[STMT_INSERT_DOC]); |
sqlite3_reset(stmts[STMT_INSERT_DOC]); |
|
|
for (key = words; NULL != key; key = key->next) { |
for (key = ohash_first(&strings, &slot); NULL != key; |
|
key = ohash_next(&strings, &slot)) { |
assert(key->of == of); |
assert(key->of == of); |
if (NULL == key->utf8) |
if (NULL == key->utf8) |
utf8key(mc, key); |
utf8key(mc, key); |
Line 1803 dbindex(struct mchars *mc, int form, const struct of * |
|
Line 1758 dbindex(struct mchars *mc, int form, const struct of * |
|
SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, recno); |
SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, recno); |
SQL_STEP(stmts[STMT_INSERT_KEY]); |
SQL_STEP(stmts[STMT_INSERT_KEY]); |
sqlite3_reset(stmts[STMT_INSERT_KEY]); |
sqlite3_reset(stmts[STMT_INSERT_KEY]); |
|
if (key->utf8 != key->key) |
|
free(key->utf8); |
|
free(key); |
} |
} |
|
|
SQL_EXEC("END TRANSACTION"); |
SQL_EXEC("END TRANSACTION"); |