-/* $Id: mandocdb.c,v 1.82 2013/12/27 03:06:17 schwarze Exp $ */
+/* $Id: mandocdb.c,v 1.92 2013/12/31 19:40:20 schwarze Exp $ */
/*
* Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
* Copyright (c) 2011, 2012, 2013 Ingo Schwarze <schwarze@openbsd.org>
};
enum form {
- FORM_SRC, /* format is -man or -mdoc */
- FORM_CAT, /* format is cat */
- FORM_NONE /* format is unknown */
+ FORM_NONE, /* format is unknown */
+ FORM_SRC, /* format is -man or -mdoc */
+ FORM_CAT /* format is cat */
};
struct str {
struct mlink {
char file[PATH_MAX]; /* filename rel. to manpath */
enum form dform; /* format from directory */
- enum form sform; /* format from file name suffix */
+ enum form fform; /* format from file name suffix */
char *dsec; /* section from directory */
char *arch; /* architecture from directory */
char *name; /* name from file name (not empty) */
char *fsec; /* section from file name suffix */
+ struct mlink *next; /* singly linked list */
};
struct title {
static void dbindex(const struct mpage *, struct mchars *);
static int dbopen(int);
static void dbprune(void);
-static void fileadd(struct mlink *);
-static int filecheck(const char *);
static void filescan(const char *);
static void *hash_alloc(size_t, void *);
static void hash_free(void *, size_t, void *);
static void *hash_halloc(size_t, void *);
-static void inoadd(const struct stat *, struct mpage *);
-static int inocheck(const struct stat *);
-static void ofadd(int, const char *, const char *, const char *,
- const char *, const char *, const struct stat *);
+static void mlink_add(struct mlink *, const struct stat *);
static void mlink_free(struct mlink *);
+static void mlinks_undupe(struct mpage *);
static void mpages_free(void);
static void mpages_merge(struct mchars *, struct mparse *, int);
static void parse_cat(struct mpage *);
static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
-static int parse_mdoc_In(struct mpage *, const struct mdoc_node *);
static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
-static int parse_mdoc_St(struct mpage *, const struct mdoc_node *);
static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
static void putkey(const struct mpage *,
const char *, uint64_t);
static enum op op; /* operational mode */
static char basedir[PATH_MAX]; /* current base directory */
static struct ohash mpages; /* table of distinct manual pages */
-static struct ohash filenames; /* table of filenames */
+static struct ohash mlinks; /* table of directory entries */
static struct ohash strings; /* table of all strings */
static sqlite3 *db = NULL; /* current database */
static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
{ parse_mdoc_Fn, 0 }, /* Fn */
{ NULL, TYPE_Ft }, /* Ft */
{ NULL, TYPE_Ic }, /* Ic */
- { parse_mdoc_In, TYPE_In }, /* In */
+ { NULL, TYPE_In }, /* In */
{ NULL, TYPE_Li }, /* Li */
{ parse_mdoc_Nd, TYPE_Nd }, /* Nd */
{ parse_mdoc_Nm, TYPE_Nm }, /* Nm */
{ NULL, 0 }, /* Ot */
{ NULL, TYPE_Pa }, /* Pa */
{ NULL, 0 }, /* Rv */
- { parse_mdoc_St, 0 }, /* St */
+ { NULL, TYPE_St }, /* St */
{ NULL, TYPE_Va }, /* Va */
{ parse_mdoc_body, TYPE_Va }, /* Vt */
{ parse_mdoc_Xr, 0 }, /* Xr */
struct mchars *mc;
struct manpaths dirs;
struct mparse *mp;
- struct ohash_info mpages_info, filename_info;
+ struct ohash_info mpages_info, mlinks_info;
memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
memset(&dirs, 0, sizeof(struct manpaths));
- mpages_info.alloc = filename_info.alloc = hash_alloc;
- mpages_info.halloc = filename_info.halloc = hash_halloc;
- mpages_info.hfree = filename_info.hfree = hash_free;
+ mpages_info.alloc = mlinks_info.alloc = hash_alloc;
+ mpages_info.halloc = mlinks_info.halloc = hash_halloc;
+ mpages_info.hfree = mlinks_info.hfree = hash_free;
mpages_info.key_offset = offsetof(struct mpage, inodev);
- filename_info.key_offset = offsetof(struct mlink, file);
+ mlinks_info.key_offset = offsetof(struct mlink, file);
progname = strrchr(argv[0], '/');
if (progname == NULL)
mc = mchars_alloc();
ohash_init(&mpages, 6, &mpages_info);
- ohash_init(&filenames, 6, &filename_info);
+ ohash_init(&mlinks, 6, &mlinks_info);
if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
/*
if (j) {
ohash_init(&mpages, 6, &mpages_info);
- ohash_init(&filenames, 6, &filename_info);
+ ohash_init(&mlinks, 6, &mlinks_info);
}
if (0 == set_basedir(dirs.paths[j]))
if (j + 1 < dirs.sz) {
mpages_free();
ohash_delete(&mpages);
- ohash_delete(&filenames);
+ ohash_delete(&mlinks);
}
}
}
mparse_free(mp);
mpages_free();
ohash_delete(&mpages);
- ohash_delete(&filenames);
+ ohash_delete(&mlinks);
return(exitcode);
usage:
fprintf(stderr, "usage: %s [-anvW] [-C file]\n"
{
FTS *f;
FTSENT *ff;
+ struct mlink *mlink;
int dform;
- char *sec;
+ char *fsec;
const char *dsec, *arch, *cp, *path;
const char *argv[2];
while (NULL != (ff = fts_read(f))) {
path = ff->fts_path + 2;
/*
- * If we're a regular file, add an mpage by using the
+ * If we're a regular file, add an mlink by using the
* stored directory data and handling the filename.
- * Disallow duplicate (hard-linked) files.
*/
if (FTS_F == ff->fts_info) {
if (0 == strcmp(path, MANDOC_DB))
if (warnings)
say(path, "Extraneous file");
continue;
- } else if (inocheck(ff->fts_statp)) {
- if (warnings)
- say(path, "Duplicate file");
- continue;
- } else if (NULL == (sec =
+ } else if (NULL == (fsec =
strrchr(ff->fts_name, '.'))) {
if ( ! use_all) {
if (warnings)
"No filename suffix");
continue;
}
- } else if (0 == strcmp(++sec, "html")) {
+ } else if (0 == strcmp(++fsec, "html")) {
if (warnings)
say(path, "Skip html");
continue;
- } else if (0 == strcmp(sec, "gz")) {
+ } else if (0 == strcmp(fsec, "gz")) {
if (warnings)
say(path, "Skip gz");
continue;
- } else if (0 == strcmp(sec, "ps")) {
+ } else if (0 == strcmp(fsec, "ps")) {
if (warnings)
say(path, "Skip ps");
continue;
- } else if (0 == strcmp(sec, "pdf")) {
+ } else if (0 == strcmp(fsec, "pdf")) {
if (warnings)
say(path, "Skip pdf");
continue;
} else if ( ! use_all &&
- ((FORM_SRC == dform && strcmp(sec, dsec)) ||
- (FORM_CAT == dform && strcmp(sec, "0")))) {
+ ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
+ (FORM_CAT == dform && strcmp(fsec, "0")))) {
if (warnings)
say(path, "Wrong filename suffix");
continue;
} else
- sec[-1] = '\0';
- ofadd(dform, path, ff->fts_name, dsec, sec,
- arch, ff->fts_statp);
+ fsec[-1] = '\0';
+ mlink = mandoc_calloc(1, sizeof(struct mlink));
+ strlcpy(mlink->file, path, sizeof(mlink->file));
+ mlink->dform = dform;
+ if (NULL != dsec)
+ mlink->dsec = mandoc_strdup(dsec);
+ if (NULL != arch)
+ mlink->arch = mandoc_strdup(arch);
+ mlink->name = mandoc_strdup(ff->fts_name);
+ if (NULL != fsec)
+ mlink->fsec = mandoc_strdup(fsec);
+ mlink_add(mlink, ff->fts_statp);
continue;
- } else if (FTS_D != ff->fts_info &&
+ } else if (FTS_D != ff->fts_info &&
FTS_DP != ff->fts_info) {
if (warnings)
say(path, "Not a regular file");
}
/*
- * Add a file to the file vector.
+ * Add a file to the mlinks table.
* Do not verify that it's a "valid" looking manpage (we'll do that
* later).
*
* or
* [./]cat<section>[/<arch>]/<name>.0
*
- * Stuff this information directly into the mpage vector.
* See treescan() for the fts(3) version of this.
*/
static void
filescan(const char *file)
{
char buf[PATH_MAX];
- const char *sec, *arch, *name, *dsec;
- char *p, *start;
- int dform;
struct stat st;
+ struct mlink *mlink;
+ char *p, *start;
assert(use_all);
exitcode = (int)MANDOCLEVEL_BADARG;
say(file, "Not a regular file");
return;
- } else if (inocheck(&st)) {
- if (warnings)
- say(file, "Duplicate file");
- return;
}
start = buf + strlen(basedir);
- sec = arch = name = dsec = NULL;
- dform = FORM_NONE;
+ mlink = mandoc_calloc(1, sizeof(struct mlink));
+ strlcpy(mlink->file, start, sizeof(mlink->file));
/*
* First try to guess our directory structure.
if (NULL != (p = strchr(start, '/'))) {
*p++ = '\0';
if (0 == strncmp(start, "man", 3)) {
- dform = FORM_SRC;
- dsec = start + 3;
+ mlink->dform = FORM_SRC;
+ mlink->dsec = mandoc_strdup(start + 3);
} else if (0 == strncmp(start, "cat", 3)) {
- dform = FORM_CAT;
- dsec = start + 3;
+ mlink->dform = FORM_CAT;
+ mlink->dsec = mandoc_strdup(start + 3);
}
start = p;
- if (NULL != dsec && NULL != (p = strchr(start, '/'))) {
+ if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
*p++ = '\0';
- arch = start;
+ mlink->arch = mandoc_strdup(start);
start = p;
- }
+ }
}
/*
if ('.' == *p) {
*p++ = '\0';
- sec = p;
+ mlink->fsec = mandoc_strdup(p);
}
/*
* Now try to parse the name.
* Use the filename portion of the path.
*/
- name = start;
+ mlink->name = start;
if (NULL != (p = strrchr(start, '/'))) {
- name = p + 1;
+ mlink->name = p + 1;
*p = '\0';
- }
-
- ofadd(dform, file, name, dsec, sec, arch, &st);
-}
-
-/*
- * See fileadd().
- */
-static int
-filecheck(const char *name)
-{
+ }
+ mlink->name = mandoc_strdup(mlink->name);
- return(NULL != ohash_find(&filenames,
- ohash_qlookup(&filenames, name)));
+ mlink_add(mlink, &st);
}
-/*
- * Use the standard hashing mechanism (K&R) to see if the given filename
- * already exists.
- */
static void
-fileadd(struct mlink *mlink)
-{
- unsigned int slot;
-
- slot = ohash_qlookup(&filenames, mlink->file);
- assert(NULL == ohash_find(&filenames, slot));
- ohash_insert(&filenames, slot, mlink);
-}
-
-/*
- * See inoadd().
- */
-static int
-inocheck(const struct stat *st)
+mlink_add(struct mlink *mlink, const struct stat *st)
{
struct inodev inodev;
- uint32_t hash;
-
- memset(&inodev, 0, sizeof(inodev));
- inodev.st_ino = hash = st->st_ino;
- inodev.st_dev = st->st_dev;
-
- return(NULL != ohash_find(&mpages, ohash_lookup_memory(
- &mpages, (char *)&inodev, sizeof(inodev), hash)));
-}
-
-/*
- * The hashing function used here is quite simple: simply take the inode
- * and use uint32_t of its bits.
- * Then when we do the lookup, use both the inode and device identifier.
- */
-static void
-inoadd(const struct stat *st, struct mpage *mpage)
-{
+ struct mpage *mpage;
unsigned int slot;
- mpage->inodev.st_ino = st->st_ino;
- mpage->inodev.st_dev = st->st_dev;
- slot = ohash_lookup_memory(&mpages, (char *)&mpage->inodev,
- sizeof(struct inodev), st->st_ino);
-
- assert(NULL == ohash_find(&mpages, slot));
- ohash_insert(&mpages, slot, mpage);
-}
-
-static void
-ofadd(int dform, const char *file, const char *name, const char *dsec,
- const char *sec, const char *arch, const struct stat *st)
-{
- struct mpage *mpage;
- struct mlink *mlink;
- int sform;
-
- assert(NULL != file);
-
- if (NULL == name)
- name = "";
- if (NULL == sec)
- sec = "";
- if (NULL == dsec)
- dsec = "";
- if (NULL == arch)
- arch = "";
-
- if ('0' == *sec) {
- sec = dsec;
- sform = FORM_CAT;
- } else if ('1' <= *sec && '9' >= *sec)
- sform = FORM_SRC;
+ assert(NULL != mlink->file);
+
+ if (NULL == mlink->dsec)
+ mlink->dsec = mandoc_strdup("");
+ if (NULL == mlink->arch)
+ mlink->arch = mandoc_strdup("");
+ if (NULL == mlink->name)
+ mlink->name = mandoc_strdup("");
+ if (NULL == mlink->fsec)
+ mlink->fsec = mandoc_strdup("");
+
+ if ('0' == *mlink->fsec) {
+ free(mlink->fsec);
+ mlink->fsec = mandoc_strdup(mlink->dsec);
+ mlink->fform = FORM_CAT;
+ } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
+ mlink->fform = FORM_SRC;
else
- sform = FORM_NONE;
+ mlink->fform = FORM_NONE;
- mlink = mandoc_calloc(1, sizeof(struct mlink));
- strlcpy(mlink->file, file, PATH_MAX);
- mlink->dform = dform;
- mlink->sform = sform;
- mlink->dsec = mandoc_strdup(dsec);
- mlink->arch = mandoc_strdup(arch);
- mlink->name = mandoc_strdup(name);
- mlink->fsec = mandoc_strdup(sec);
-
- mpage = mandoc_calloc(1, sizeof(struct mpage));
- mpage->mlinks = mlink;
+ slot = ohash_qlookup(&mlinks, mlink->file);
+ assert(NULL == ohash_find(&mlinks, slot));
+ ohash_insert(&mlinks, slot, mlink);
- /*
- * Add to unique identifier hash.
- * Then if it's a source manual and we're going to use source in
- * favour of catpages, add it to that hash.
- */
- inoadd(st, mpage);
- fileadd(mpage->mlinks);
+ inodev.st_ino = st->st_ino;
+ inodev.st_dev = st->st_dev;
+ slot = ohash_lookup_memory(&mpages, (char *)&inodev,
+ sizeof(struct inodev), inodev.st_ino);
+ mpage = ohash_find(&mpages, slot);
+ if (NULL == mpage) {
+ mpage = mandoc_calloc(1, sizeof(struct mpage));
+ mpage->inodev.st_ino = inodev.st_ino;
+ mpage->inodev.st_dev = inodev.st_dev;
+ ohash_insert(&mpages, slot, mpage);
+ } else
+ mlink->next = mpage->mlinks;
+ mpage->mlinks = mlink;
}
static void
mpage = ohash_first(&mpages, &slot);
while (NULL != mpage) {
while (NULL != (mlink = mpage->mlinks)) {
- mpage->mlinks = NULL;
+ mpage->mlinks = mlink->next;
mlink_free(mlink);
}
free(mpage->sec);
}
}
+/*
+ * For each mlink to the mpage, check whether the path looks like
+ * it is formatted, and if it does, check whether a source manual
+ * exists by the same name, ignoring the suffix.
+ * If both conditions hold, drop the mlink.
+ */
+static void
+mlinks_undupe(struct mpage *mpage)
+{
+ char buf[PATH_MAX];
+ struct mlink **prev;
+ struct mlink *mlink;
+ char *bufp;
+
+ mpage->form = FORM_CAT;
+ prev = &mpage->mlinks;
+ while (NULL != (mlink = *prev)) {
+ if (FORM_CAT != mlink->dform) {
+ mpage->form = FORM_NONE;
+ goto nextlink;
+ }
+ if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) {
+ if (warnings)
+ say(mlink->file, "Filename too long");
+ goto nextlink;
+ }
+ bufp = strstr(buf, "cat");
+ assert(NULL != bufp);
+ memcpy(bufp, "man", 3);
+ if (NULL != (bufp = strrchr(buf, '.')))
+ *++bufp = '\0';
+ strlcat(buf, mlink->dsec, PATH_MAX);
+ if (NULL == ohash_find(&mlinks,
+ ohash_qlookup(&mlinks, buf)))
+ goto nextlink;
+ if (warnings)
+ say(mlink->file, "Man source exists: %s", buf);
+ if (use_all)
+ goto nextlink;
+ *prev = mlink->next;
+ mlink_free(mlink);
+ continue;
+nextlink:
+ prev = &(*prev)->next;
+ }
+}
+
/*
* Run through the files in the global vector "mpages"
* and add them to the database specified in "basedir".
{
struct ohash title_table;
struct ohash_info title_info, str_info;
- char buf[PATH_MAX];
struct mpage *mpage;
struct mdoc *mdoc;
struct man *man;
struct title *title_entry;
- char *bufp, *title_str;
+ char *title_str;
const char *cp;
- size_t sz;
int match;
unsigned int pslot, tslot;
enum mandoclevel lvl;
mpage = ohash_first(&mpages, &pslot);
while (NULL != mpage) {
- /*
- * If we're a catpage (as defined by our path), then see
- * if a manpage exists by the same name (ignoring the
- * suffix).
- * If it does, then we want to use it instead of our
- * own.
- */
- if ( ! use_all && FORM_CAT == mpage->mlinks->dform) {
- sz = strlcpy(buf, mpage->mlinks->file, PATH_MAX);
- if (sz >= PATH_MAX) {
- if (warnings)
- say(mpage->mlinks->file,
- "Filename too long");
- mpage = ohash_next(&mpages, &pslot);
- continue;
- }
- bufp = strstr(buf, "cat");
- assert(NULL != bufp);
- memcpy(bufp, "man", 3);
- if (NULL != (bufp = strrchr(buf, '.')))
- *++bufp = '\0';
- strlcat(buf, mpage->mlinks->dsec, PATH_MAX);
- if (filecheck(buf)) {
- if (warnings)
- say(mpage->mlinks->file, "Man "
- "source exists: %s", buf);
- mpage = ohash_next(&mpages, &pslot);
- continue;
- }
+ mlinks_undupe(mpage);
+ if (NULL == mpage->mlinks) {
+ mpage = ohash_next(&mpages, &pslot);
+ continue;
}
ohash_init(&strings, 6, &str_info);
* formatted. Fall back to formatted mode.
*/
if (FORM_CAT != mpage->mlinks->dform ||
- FORM_CAT != mpage->mlinks->sform) {
+ FORM_CAT != mpage->mlinks->fform) {
lvl = mparse_readfd(mp, -1, mpage->mlinks->file);
if (lvl < MANDOCLEVEL_FATAL)
mparse_result(mp, &mdoc, &man);
mpage->sec =
mandoc_strdup(mdoc_meta(mdoc)->msec);
mpage->arch = mdoc_meta(mdoc)->arch;
- if (NULL == mpage->arch)
- mpage->arch = "";
- mpage->arch = mandoc_strdup(mpage->arch);
+ mpage->arch = mandoc_strdup(
+ NULL == mpage->arch ? "" : mpage->arch);
mpage->title =
mandoc_strdup(mdoc_meta(mdoc)->title);
} else if (NULL != man) {
if (end > start)
putkeys(mpage, start, end - start + 1, TYPE_In);
- return(1);
-}
-
-static int
-parse_mdoc_In(struct mpage *mpage, const struct mdoc_node *n)
-{
-
- if (NULL != n->child && MDOC_TEXT == n->child->type)
- return(0);
-
- putkey(mpage, n->child->string, TYPE_In);
- return(1);
+ return(0);
}
static int
return(0);
}
-static int
-parse_mdoc_St(struct mpage *mpage, const struct mdoc_node *n)
-{
-
- if (NULL == n->child || MDOC_TEXT != n->child->type)
- return(0);
-
- putkey(mpage, n->child->string, TYPE_St);
- return(1);
-}
-
static int
parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
{
parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
{
- if (SEC_NAME == n->sec)
- return(1);
- else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
- return(0);
-
- return(1);
+ return(SEC_NAME == n->sec ||
+ (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type));
}
static int
/* Read past the slash. */
val++;
- u = 0;
/*
* Parse the escape sequence and see if it's a
static void
dbindex(const struct mpage *mpage, struct mchars *mc)
{
+ struct mlink *mlink;
struct str *key;
const char *desc;
int64_t recno;
recno = sqlite3_last_insert_rowid(db);
sqlite3_reset(stmts[STMT_INSERT_PAGE]);
- i = 1;
- SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mpage->mlinks->dsec);
- SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mpage->mlinks->arch);
- SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mpage->mlinks->file);
- SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, recno);
- SQL_STEP(stmts[STMT_INSERT_LINK]);
- sqlite3_reset(stmts[STMT_INSERT_LINK]);
+ for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
+ i = 1;
+ SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->file);
+ SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
+ SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
+ SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
+ SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, recno);
+ SQL_STEP(stmts[STMT_INSERT_LINK]);
+ sqlite3_reset(stmts[STMT_INSERT_LINK]);
+ }
for (key = ohash_first(&strings, &slot); NULL != key;
key = ohash_next(&strings, &slot)) {
");\n"
"\n"
"CREATE TABLE \"mlinks\" (\n"
+ " \"file\" TEXT NOT NULL,\n"
" \"sec\" TEXT NOT NULL,\n"
" \"arch\" TEXT NOT NULL,\n"
" \"name\" TEXT NOT NULL,\n"
"(file,sec,arch,desc,form) VALUES (?,?,?,?,?)";
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
sql = "INSERT INTO mlinks "
- "(sec,arch,name,pageid) VALUES (?,?,?,?)";
+ "(file,sec,arch,name,pageid) VALUES (?,?,?,?,?)";
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
sql = "INSERT INTO keys "
"(bits,key,pageid) VALUES (?,?,?)";
git.ckatri.com / mandoc.git / blobdiff