src/bin/lttng-relayd/backward-compatibility-group-by.c

   1 /*
   2  * Copyright (C) 2019 Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
   3  *
   4  * SPDX-License-Identifier: GPL-2.0-only
   5  *
   6  */
   7
   8 #include <assert.h>
   9 #include <regex.h>
  10 #include <stdio.h>
  11 #include <stdlib.h>
  12 #include <string.h>
  13
  14 #include <common/common.h>
  15 #include <common/defaults.h>
  16 #include <common/utils.h>
  17
  18 #include "backward-compatibility-group-by.h"
  19
  20 #define DATETIME_STRING_SIZE 16
  21 #define DATETIME_REGEX \
  22         ".*-[1-2][0-9][0-9][0-9][0-1][0-9][0-3][0-9]-[0-2][0-9][0-5][0-9][0-5][0-9]$"
  23
  24 /*
  25  * Provide support for --group-output-by-session for producer >= 2.4 and < 2.11.
  26  * Take the stream path, extract all available information, craft a new path to
  27  * the best of our ability enforcing the group by session.
  28  *
  29  * Return the allocated string containing the new stream path or else NULL.
  30  */
  31 char *backward_compat_group_by_session(
  32                 const char *path, const char *local_session_name)
  33 {
  34         int ret;
  35         size_t len;
  36         char *leftover_ptr;
  37         char *local_copy = NULL;
  38         char *datetime = NULL;
  39         char *partial_base_path = NULL;
  40         char *filepath_per_session = NULL;
  41         const char *second_token_ptr;
  42         const char *leftover_second_token_ptr;
  43         const char *hostname_ptr;
  44         regex_t regex;
  45
  46         assert(path);
  47         assert(local_session_name);
  48         assert(local_session_name[0] != '\0');
  49
  50         DBG("Parsing path \"%s\" of session \"%s\" to create a new path that is grouped by session",
  51                         path, local_session_name);
  52
  53         /* Get a local copy for strtok */
  54         local_copy = strdup(path);
  55         if (!local_copy) {
  56                 PERROR("Failed to parse session path: couldn't copy input path");
  57                 goto error;
  58         }
  59
  60         /*
  61          * The use of strtok with '/' as delimiter is valid since we refuse '/'
  62          * in session name and '/' is not a valid hostname character based on
  63          * RFC-952 [1], RFC-921 [2] and refined in RFC-1123 [3].
  64          * [1] https://tools.ietf.org/html/rfc952
  65          * [2] https://tools.ietf.org/html/rfc921
  66          * [3] https://tools.ietf.org/html/rfc1123#page-13
  67          */
  68
  69         /*
  70          * Get the hostname and possible session_name.
  71          * Note that we can get the hostname and session name from the
  72          * relay_session object we already have. Still, it is easier to
  73          * tokenized the passed path to obtain the start of the path leftover.
  74          */
  75         hostname_ptr = strtok_r(local_copy, "/", &leftover_ptr);
  76         if (!hostname_ptr) {
  77                 ERR("Failed to parse session path \"%s\": couldn't identify hostname",
  78                                 path);
  79                 goto error;
  80         }
  81
  82         second_token_ptr = strtok_r(NULL, "/", &leftover_ptr);
  83         if (!second_token_ptr) {
  84                 ERR("Failed to parse session path \"%s\": couldn't identify session name",
  85                                 path);
  86                 goto error;
  87         }
  88
  89         /*
  90          * Check if the second token is a base path set at url level. This is
  91          * legal in streaming, live and snapshot [1]. Otherwise it is the
  92          * session name with possibly a datetime attached [2]. Note that when
  93          * "adding" snapshot output (lttng snapshot add-output), no session name
  94          * is present in the path by default. The handling for "base path" take
  95          * care of this case as well.
  96          * [1] e.g --set-url net://localhost/my_marvellous_path
  97          * [2] Can be:
  98          *            <session_name>
  99          *                When using --snapshot on session create.
 100          *            <session_name>-<date>-<time>
 101          *            <auto>-<date>-<time>
 102          */
 103         if (strncmp(second_token_ptr, local_session_name,
 104                             strlen(local_session_name)) != 0) {
 105                 /*
 106                  * Token does not start with session name.
 107                  * This mean this is an extra path scenario.
 108                  * Duplicate the current token since it is part of an
 109                  * base_path.
 110                  * Set secDuplicate the current token since it is part of an
 111                  * base_path. The rest is the leftover.
 112                  * Set second_token_ptr to the local_session_name for further
 113                  * processing.
 114                  */
 115                 partial_base_path = strdup(second_token_ptr);
 116                 if (!partial_base_path) {
 117                         PERROR("Failed to parse session path: couldn't copy partial base path");
 118                         goto error;
 119                 }
 120
 121                 second_token_ptr = local_session_name;
 122         }
 123
 124         /*
 125          * Based on the previous test, we can move inside the token ptr to
 126          * remove the "local_session_name" and inspect the rest of the token.
 127          * We are looking into extracting the creation datetime from either the
 128          * session_name or the token. We need to to all this gymnastic because
 129          * an extra path could decide to append a datetime to its first
 130          * subdirectory.
 131          * Possible scenario:
 132          *     <session_name>
 133          *     <session_name>-<date>-<time>
 134          *     <auto>-<date>-<time>
 135          *     <session_name>_base_path_foo_bar
 136          *     <session_name>-<false date>-<false-time> (via a base path)
 137          *
 138          * We have no way to discern from the basic scenario of:
 139          *     <session_name>-<date>-<time>
 140          * and one done using a base path with the exact format we normally
 141          * expect.
 142          *
 143          * e.g:
 144          *     lttng create my_session -U
 145          *         net://localhost/my_session-19910319-120000/
 146          */
 147         ret = regcomp(&regex, DATETIME_REGEX, 0);
 148         if (ret) {
 149                 ERR("Failed to parse session path: regex compilation failed with code %d", ret);
 150                 goto error;
 151         }
 152
 153         leftover_second_token_ptr =
 154                         second_token_ptr + strlen(local_session_name);
 155         len = strlen(leftover_second_token_ptr);
 156         if (len == 0) {
 157                 /*
 158                  * We are either dealing with an auto session name or only the
 159                  * session_name. If this is a auto session name, we need to
 160                  * fetch the creation datetime.
 161                  */
 162                 ret = regexec(&regex, local_session_name, 0, NULL, 0);
 163                 if (ret == 0) {
 164                         const ssize_t local_session_name_offset =
 165                                         strlen(local_session_name) - DATETIME_STRING_SIZE + 1;
 166
 167                         assert(local_session_name_offset >= 0);
 168                         datetime = strdup(local_session_name +
 169                                         local_session_name_offset);
 170                         if (!datetime) {
 171                                 PERROR("Failed to parse session path: couldn't copy datetime on regex match");
 172                                 goto error_regex;
 173                         }
 174                 }
 175         } else if (len == DATETIME_STRING_SIZE &&
 176                         !regexec(&regex, leftover_second_token_ptr, 0, NULL,
 177                                         0)) {
 178                 /*
 179                  * The leftover from the second token is of format
 180                  * "-<datetime>", use it as the creation time.
 181                  * Ignore leading "-".
 182                  */
 183                 datetime = strdup(&leftover_second_token_ptr[1]);
 184                 if (!datetime) {
 185                         PERROR("Failed to parse session path: couldn't copy datetime on regex match");
 186                         goto error_regex;
 187                 }
 188         } else {
 189                 /*
 190                  * Base path scenario.
 191                  * We cannot try to extract the datetime from the session name
 192                  * since nothing prevent a user to name a session in the
 193                  * "name-<datetime>" format. Using the datetime from such a
 194                  * session would be invalid.
 195                  * */
 196                 assert(partial_base_path == NULL);
 197                 assert(datetime == NULL);
 198
 199                 partial_base_path = strdup(second_token_ptr);
 200                 if (!partial_base_path) {
 201                         PERROR("Failed to parse session path: couldn't copy partial base path");
 202                         goto error_regex;
 203                 }
 204         }
 205
 206         ret = asprintf(&filepath_per_session, "%s/%s%s%s/%s%s%s",
 207                         local_session_name, hostname_ptr, datetime ? "-" : "",
 208                         datetime ? datetime : "",
 209                         partial_base_path ? partial_base_path : "",
 210                         partial_base_path ? "/" : "", leftover_ptr);
 211         if (ret < 0) {
 212                 filepath_per_session = NULL;
 213                 goto error;
 214         }
 215 error_regex:
 216         regfree(&regex);
 217 error:
 218         free(local_copy);
 219         free(partial_base_path);
 220         free(datetime);
 221         return filepath_per_session;
 222 }