X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Ftrav.c;h=d055ad58d3fa3fe39c7f631be44d53242bf65e7d;hb=162fa86044b5aa303c03fa26f1ecf140b34060d5;hp=27a803c3cbe20b2495848809c52253056e03a618;hpb=51a52e1b014d5237581f74379916b8ce92cafef7;p=idzebra-moved-to-github.git diff --git a/index/trav.c b/index/trav.c index 27a803c..d055ad5 100644 --- a/index/trav.c +++ b/index/trav.c @@ -1,10 +1,54 @@ /* - * Copyright (C) 1995, Index Data I/S + * Copyright (C) 1994-1995, Index Data I/S * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * * $Log: trav.c,v $ - * Revision 1.1 1995-09-01 14:06:36 adam + * Revision 1.13 1995-11-28 09:09:46 adam + * Zebra config renamed. + * Use setting 'recordId' to identify record now. + * Bug fix in recindex.c: rec_release_blocks was invokeded even + * though the blocks were already released. + * File traversal properly deletes records when needed. + * + * Revision 1.12 1995/11/24 11:31:37 adam + * Commands add & del read filenames from stdin if source directory is + * empty. + * Match criteria supports 'constant' strings. + * + * Revision 1.11 1995/11/22 17:19:19 adam + * Record management uses the bfile system. + * + * Revision 1.10 1995/11/21 15:01:16 adam + * New general match criteria implemented. + * New feature: document groups. + * + * Revision 1.9 1995/11/21 09:20:32 adam + * Yet more work on record match. + * + * Revision 1.8 1995/11/20 16:59:46 adam + * New update method: the 'old' keys are saved for each records. + * + * Revision 1.7 1995/11/20 11:56:28 adam + * Work on new traversal. + * + * Revision 1.6 1995/11/17 15:54:42 adam + * Started work on virtual directory structure. + * + * Revision 1.5 1995/10/17 18:02:09 adam + * New feature: databases. Implemented as prefix to words in dictionary. + * + * Revision 1.4 1995/09/28 09:19:46 adam + * xfree/xmalloc used everywhere. + * Extract/retrieve method seems to work for text records. + * + * Revision 1.3 1995/09/06 16:11:18 adam + * Option: only one word key per file. + * + * Revision 1.2 1995/09/04 12:33:43 adam + * Various cleanup. YAZ util used instead. + * + * Revision 1.1 1995/09/01 14:06:36 adam * Split of work into more files. * */ @@ -17,14 +61,21 @@ #include #include -#include +#include #include "index.h" -static void repository_extract_r (int cmd, char *rep) +static int repComp (const char *a, const char *b, size_t len) +{ + if (!len) + return 0; + return memcmp (a, b, len); +} + +static void repositoryExtractR (int deleteFlag, char *rep, + struct recordGroup *rGroup) { struct dir_entry *e; int i; - struct stat fs; size_t rep_len = strlen (rep); e = dir_open (rep); @@ -37,256 +88,246 @@ static void repository_extract_r (int cmd, char *rep) for (i=0; e[i].name; i++) { strcpy (rep +rep_len+1, e[i].name); - stat (rep, &fs); - switch (fs.st_mode & S_IFMT) + switch (e[i].kind) { - case S_IFREG: - file_extract (cmd, rep, rep); + case dirs_file: + fileExtract (NULL, rep, rGroup, deleteFlag); break; - case S_IFDIR: - repository_extract_r (cmd, rep); + case dirs_dir: + repositoryExtractR (deleteFlag, rep, rGroup); break; } } dir_free (&e); -} - -void copy_file (const char *dst, const char *src) -{ - int d_fd = open (dst, O_WRONLY|O_CREAT, 0666); - int s_fd = open (src, O_RDONLY); - char *buf; - size_t i, r, w; - if (d_fd == -1) - { - log (LOG_FATAL|LOG_ERRNO, "Cannot create %s", dst); - exit (1); - } - if (s_fd == -1) - { - log (LOG_FATAL|LOG_ERRNO, "Cannot open %s", src); - exit (1); - } - if (!(buf = malloc (4096))) - { - log (LOG_FATAL|LOG_ERRNO, "malloc"); - exit (1); - } - while ((r=read (s_fd, buf, 4096))>0) - for (w = 0; w < r; w += i) - { - i = write (d_fd, buf + w, r - w); - if (i == -1) - { - log (LOG_FATAL|LOG_ERRNO, "write"); - exit (1); - } - } - if (r) - { - log (LOG_FATAL|LOG_ERRNO, "read"); - exit (1); - } - free (buf); - close (d_fd); - close (s_fd); } -void del_file (const char *dst) +static void stdinExtractR (int deleteFlag, struct recordGroup *rGroup) { - unlink (dst); -} + char tmppath[256]; -void del_dir (const char *dst) -{ - log (LOG_DEBUG, "rmdir of %s", dst); - if (rmdir (dst) == -1) - log (LOG_ERRNO|LOG_WARN, "rmdir"); + logf (LOG_LOG, "stdinExtractR"); + while (scanf ("%s", tmppath) == 1) + fileExtract (NULL, tmppath, rGroup, deleteFlag); } -void repository_update_r (int cmd, char *dst, char *src); - -void repository_add_tree (int cmd, char *dst, char *src) +static void repositoryDeleteR (struct dirs_info *di, struct dirs_entry *dst, + const char *base, char *src, + struct recordGroup *rGroup) { - mkdir (dst, 0755); - repository_update_r (cmd, dst, src); -} - -void repository_del_tree (int cmd, char *dst, char *src) -{ - size_t dst_len = strlen (dst); + char tmppath[256]; size_t src_len = strlen (src); - struct dir_entry *e_dst; - int i_dst = 0; - struct stat fs_dst; - - e_dst = dir_open (dst); - dir_sort (e_dst); - - if (src[src_len-1] != '/') - src[src_len] = '/'; - else - --src_len; - if (dst[dst_len-1] != '/') - dst[dst_len] = '/'; - else - --dst_len; - while (e_dst[i_dst].name) + while (dst && !repComp (dst->path, src, src_len+1)) { - strcpy (dst +dst_len+1, e_dst[i_dst].name); - strcpy (src +src_len+1, e_dst[i_dst].name); - - stat (dst, &fs_dst); - switch (fs_dst.st_mode & S_IFMT) + switch (dst->kind) { - case S_IFREG: - file_extract ('d', dst, dst); - del_file (dst); + case dirs_file: + sprintf (tmppath, "%s%s", base, dst->path); + fileExtract (&dst->sysno, tmppath, rGroup, 1); + + strcpy (tmppath, dst->path); + dst = dirs_read (di); + dirs_del (di, tmppath); break; - case S_IFDIR: - repository_del_tree (cmd, dst, src); + case dirs_dir: + strcpy (tmppath, dst->path); + dst = dirs_read (di); + dirs_rmdir (di, tmppath); break; + default: + dst = dirs_read (di); } - i_dst++; - } - dir_free (&e_dst); - if (dst_len > 0) - { - dst[dst_len] = '\0'; - del_dir (dst); } } -void repository_update_r (int cmd, char *dst, char *src) +static void repositoryUpdateR (struct dirs_info *di, struct dirs_entry *dst, + const char *base, char *src, + struct recordGroup *rGroup) { - struct dir_entry *e_dst, *e_src; - int i_dst = 0, i_src = 0; - struct stat fs_dst, fs_src; - size_t dst_len = strlen (dst); + struct dir_entry *e_src; + int i_src = 0; + static char tmppath[256]; size_t src_len = strlen (src); - e_dst = dir_open (dst); - e_src = dir_open (src); + sprintf (tmppath, "%s%s", base, src); + e_src = dir_open (tmppath); - if (!e_dst && !e_src) - return; - if (!e_dst) +#if 1 + if (!dst || repComp (dst->path, src, src_len)) +#else + if (!dst || strcmp (dst->path, src)) +#endif { - dir_free (&e_src); - repository_add_tree (cmd, dst, src); - return; + if (!e_src) + return; + if (src_len && src[src_len-1] == '/') + --src_len; + else + src[src_len] = '/'; + src[src_len+1] = '\0'; + dirs_mkdir (di, src, 0); + dst = NULL; } else if (!e_src) { - dir_free (&e_dst); - repository_del_tree (cmd, dst, src); + strcpy (src, dst->path); + repositoryDeleteR (di, dst, base, src, rGroup); return; } - + else + { + if (src_len && src[src_len-1] == '/') + --src_len; + else + src[src_len] = '/'; + src[src_len+1] = '\0'; + dst = dirs_read (di); + } dir_sort (e_src); - dir_sort (e_dst); - if (src[src_len-1] != '/') - src[src_len] = '/'; - else - --src_len; - if (dst[dst_len-1] != '/') - dst[dst_len] = '/'; - else - --dst_len; - while (e_dst[i_dst].name || e_src[i_src].name) + while (1) { int sd; - if (e_dst[i_dst].name && e_src[i_src].name) - sd = strcmp (e_dst[i_dst].name, e_src[i_src].name); + if (dst && !repComp (dst->path, src, src_len+1)) + { + if (e_src[i_src].name) + { + logf (LOG_DEBUG, "dst=%s src=%s", dst->path + src_len+1, + e_src[i_src].name); + sd = strcmp (dst->path + src_len+1, e_src[i_src].name); + } + else + sd = -1; + } else if (e_src[i_src].name) sd = 1; else - sd = -1; - + break; + logf (LOG_DEBUG, "trav sd=%d", sd); if (sd == 0) { - strcpy (dst +dst_len+1, e_dst[i_dst].name); - strcpy (src +src_len+1, e_src[i_src].name); + strcpy (src + src_len+1, e_src[i_src].name); + sprintf (tmppath, "%s%s", base, src); - /* check type, date, length */ - - stat (dst, &fs_dst); - stat (src, &fs_src); - - switch (fs_dst.st_mode & S_IFMT) + switch (e_src[i_src].kind) { - case S_IFREG: - if (fs_src.st_ctime > fs_dst.st_ctime) + case dirs_file: + if (e_src[i_src].ctime > dst->ctime) { - file_extract ('d', dst, dst); - file_extract ('a', src, dst); - copy_file (dst, src); + if (fileExtract (&dst->sysno, tmppath, rGroup, 0)) + { + logf (LOG_LOG, "dirs_add"); + dirs_add (di, src, dst->sysno, e_src[i_src].ctime); + } } + dst = dirs_read (di); break; - case S_IFDIR: - repository_update_r (cmd, dst, src); + case dirs_dir: + repositoryUpdateR (di, dst, base, src, rGroup); + dst = dirs_last (di); + logf (LOG_DEBUG, "last is %s", dst ? dst->path : "null"); break; + default: + dst = dirs_read (di); } i_src++; - i_dst++; } else if (sd > 0) { - strcpy (dst +dst_len+1, e_src[i_src].name); - strcpy (src +src_len+1, e_src[i_src].name); - - stat (src, &fs_src); - switch (fs_src.st_mode & S_IFMT) + SYSNO sysno = 0; + strcpy (src + src_len+1, e_src[i_src].name); + sprintf (tmppath, "%s%s", base, src); + + switch (e_src[i_src].kind) { - case S_IFREG: - file_extract ('a', src, dst); - copy_file (dst, src); + case dirs_file: + if (fileExtract (&sysno, tmppath, rGroup, 0)) + dirs_add (di, src, sysno, e_src[i_src].ctime); break; - case S_IFDIR: - repository_add_tree (cmd, dst, src); + case dirs_dir: + repositoryUpdateR (di, dst, base, src, rGroup); + if (dst) + dst = dirs_last (di); break; } i_src++; } - else + else /* sd < 0 */ { - strcpy (dst +dst_len+1, e_dst[i_dst].name); - strcpy (src +src_len+1, e_dst[i_dst].name); - - stat (dst, &fs_dst); - switch (fs_dst.st_mode & S_IFMT) + strcpy (src, dst->path); + sprintf (tmppath, "%s%s", base, dst->path); + + switch (dst->kind) { - case S_IFREG: - file_extract ('d', dst, dst); - del_file (dst); - break; - case S_IFDIR: - repository_del_tree (cmd, dst, src); + case dirs_file: + fileExtract (&dst->sysno, tmppath, rGroup, 1); + dirs_del (di, dst->path); + dst = dirs_read (di); break; + case dirs_dir: + repositoryDeleteR (di, dst, base, src, rGroup); + dst = dirs_last (di); } - i_dst++; } } - dir_free (&e_dst); dir_free (&e_src); } -void repository (int cmd, const char *rep, const char *base_path) +static void groupRes (struct recordGroup *rGroup) +{ + char resStr[256]; + char gPrefix[256]; + + if (!rGroup->groupName || !*rGroup->groupName) + *gPrefix = '\0'; + else + sprintf (gPrefix, "%s.", rGroup->groupName); + + sprintf (resStr, "%srecordId", gPrefix); + rGroup->recordId = res_get (common_resource, resStr); +} + +void repositoryUpdate (struct recordGroup *rGroup) { - char rep_tmp1[2048]; - char rep_tmp2[2048]; + char src[256]; - strcpy (rep_tmp1, rep); - if (base_path) + groupRes (rGroup); + if (rGroup->recordId && !strcmp (rGroup->recordId, "file")) { - strcpy (rep_tmp2, base_path); - repository_update_r (cmd, rep_tmp2, rep_tmp1); + Dict dict; + struct dirs_info *di; + + dict = dict_open ("repdict", 40, 1); + + assert (rGroup->path); + di = dirs_open (dict, rGroup->path); + strcpy (src, ""); + repositoryUpdateR (di, dirs_read (di), rGroup->path, src, rGroup); + dirs_free (&di); + dict_close (dict); + } + else + { + strcpy (src, rGroup->path); + if (*src == '\0') + stdinExtractR (0, rGroup); + else + repositoryExtractR (0, src, rGroup); } +} + +void repositoryDelete (struct recordGroup *rGroup) +{ + char src[256]; + + assert (rGroup->path); + groupRes (rGroup); + strcpy (src, rGroup->path); + if (*src == '\0') + stdinExtractR (1, rGroup); else - repository_extract_r (cmd, rep_tmp1); + repositoryExtractR (1, src, rGroup); }