1 | #include "Database.h"
|
---|
2 |
|
---|
3 | #include <regex>
|
---|
4 |
|
---|
5 | #include <boost/algorithm/string/join.hpp>
|
---|
6 | #include <boost/tokenizer.hpp>
|
---|
7 | #include <boost/algorithm/string.hpp>
|
---|
8 |
|
---|
9 | #include "tools.h"
|
---|
10 | #include "Time.h"
|
---|
11 | #include "Configuration.h"
|
---|
12 |
|
---|
13 | #include <TROOT.h>
|
---|
14 | #include <TSystem.h>
|
---|
15 | #include <TFile.h>
|
---|
16 | #include <TTree.h>
|
---|
17 |
|
---|
18 | using namespace std;
|
---|
19 |
|
---|
20 | // ------------------------------------------------------------------------
|
---|
21 |
|
---|
22 | void SetupConfiguration(Configuration &conf)
|
---|
23 | {
|
---|
24 | po::options_description control("Database options");
|
---|
25 | control.add_options()
|
---|
26 | ("uri,u", var<string>()->required(), "Database link as in\n\tuser:password@server[:port]/database[?compress=0|1].")
|
---|
27 | ("query,q", var<string>(""), "MySQL query (overwrites --file)")
|
---|
28 | ("file", var<string>("rootify.sql"), "An ASCII file with the MySQL query (overwrites --query)")
|
---|
29 | ("ignore-null,i", po_switch(), "Do not skip rows containing any NULL field")
|
---|
30 | ("display,d", po_switch(), "Displays contents on the screen (most usefull in combination with mysql statements as SHOW or EXPLAIN)")
|
---|
31 | ("write,w", var<string>(""), "Write output to an ascii file")
|
---|
32 | ("delimiter", var<string>(""), "The delimiter used if contents are displayed with --display (default=\\t)")
|
---|
33 | ("explain", po_switch(), "Requests an EXPLAIN from the server (shows the server optimized query)\nsee also https://dev.mysql.com/doc/refman/explain-output.html")
|
---|
34 | ("profiling", po_switch(), "Turn on profiling and print profile")
|
---|
35 | ("var.*", var<string>(), "Predefined SQL user variables (@VAR)")
|
---|
36 | ("env.*", vars<string>(), "Predefined environment for substitutions in the query ($ENV)")
|
---|
37 | ("list.*", var<string>(), "Predefined environment for substitutions in the query ($ENV). The list is read from the given file (one list entry per line)")
|
---|
38 | ("print-connection", po_switch(), "Print database connection information")
|
---|
39 | ("verbose,v", var<uint16_t>(1), "Verbosity (0: quiet, 1: default, 2: more, 3, ...)")
|
---|
40 | ;
|
---|
41 |
|
---|
42 | po::options_description root("Root file options");
|
---|
43 | root.add_options()
|
---|
44 | ("out,o", var<string>("rootify.root"), "Output root file name")
|
---|
45 | ("force,f", po_switch(), "Force overwriting an existing root file ('RECREATE')")
|
---|
46 | ("update", po_switch(), "Update an existing root file with the new tree ('UPDATE')")
|
---|
47 | ("compression,c", var<uint16_t>(1), "zlib compression level for the root file")
|
---|
48 | ("tree,t", var<string>("Result"), "Name of the root tree")
|
---|
49 | ("ignore", vars<string>(), "Ignore the given columns")
|
---|
50 | ("null,n", po_switch(), "Redirect the output file to /dev/null (mainly for debugging purposes, e.g. performance studies)")
|
---|
51 | ("no-fill", po_switch(), "Do not fill events into the root file (mainly for debugging purposes, e.g. performance studies)")
|
---|
52 | ;
|
---|
53 |
|
---|
54 | po::positional_options_description p;
|
---|
55 | p.add("file", 1); // The 1st positional options (n=1)
|
---|
56 | p.add("out", 1); // The 2nd positional options (n=1)
|
---|
57 |
|
---|
58 | conf.AddOptions(control);
|
---|
59 | conf.AddOptions(root);
|
---|
60 | conf.SetArgumentPositions(p);
|
---|
61 | }
|
---|
62 |
|
---|
63 | void PrintUsage()
|
---|
64 | {
|
---|
65 | cout <<
|
---|
66 | "rootifysql - Converts the result of a mysql query into a root file\n"
|
---|
67 | "\n"
|
---|
68 | "For convenience, this documentation uses the extended version of the options, "
|
---|
69 | "refer to the output below to get the abbreviations.\n"
|
---|
70 | "\n"
|
---|
71 | "Writes the result of a mysql query into a root file. For each column, a branch is "
|
---|
72 | "created of type double with the field name as name. This is usually the column name "
|
---|
73 | "if not specified otherwise by the AS mysql directive.\n"
|
---|
74 | "\n"
|
---|
75 | "Columns with CHAR or VARCHAR as field type are ignored. DATETIME, DATE and TIME "
|
---|
76 | "columns are converted to unix time (time_t). Rows containing any file which is "
|
---|
77 | "NULL are skipped if not suppressed by the --ignore-null option. Ideally, the query "
|
---|
78 | "is compiled in a way that no NULL field is returned. With the --display option the "
|
---|
79 | "result of the request is printed on the screen (NULL skipping still in action). "
|
---|
80 | "This can be useful to create an ascii file or to show results as 'SHOW DATABASES' "
|
---|
81 | "or 'EXPLAIN table'. To redirect the contents into an ascii file, the option -v0 "
|
---|
82 | "is useful. To suppress writing to an output file --null can be used.\n"
|
---|
83 | "\n"
|
---|
84 | "The default is to read the query from a file called rootify.sql. Except if a different "
|
---|
85 | "filename is specified by the --file option or a query is given with --query.\n"
|
---|
86 | "\n"
|
---|
87 | "As a trick, the rootify.sql file can be made excutable (chmod u+x rootify.sql). "
|
---|
88 | "If the first line contains '#!rootifysql', the script can be executed directly.\n"
|
---|
89 | "\n"
|
---|
90 | "Columns whose name start with @ are skipped. If you want them in your output file "
|
---|
91 | "give them a name using AS, e.g. 'SELECT @A:=5 AS A'.\n"
|
---|
92 | "\n"
|
---|
93 | "You can use variables in your sql query like @MyVar and define them on the "
|
---|
94 | "command line. In this example with --var.MyVar=5\n"
|
---|
95 | "\n"
|
---|
96 | "You can use environment definitions for substitutions in your SQL query. "
|
---|
97 | "For example --env.TEST=5 would replace $TEST or ${TEST} in your query by 5."
|
---|
98 | "If you specify one environment variable more than once, a list is created. "
|
---|
99 | "For example --env.TEST=1 --env.TEST=2 --env.TEST=3 would substitute "
|
---|
100 | "$TEST or ${TEST} by '1, 2, 3'. This is useful for the SQL `IN` keyword. "
|
---|
101 | "You can also read the values for an enviroment substitution from a file "
|
---|
102 | "(one element per line), e.g. --env.TEST=file.txt. Empty lines and lines "
|
---|
103 | "starting with a # are skipped.\n"
|
---|
104 | "\n"
|
---|
105 | "Comments in the query-file can be placed according to the SQL standard inline "
|
---|
106 | "/*comment*/ or introduced with # (shell script style) or -- (SQL style).\n"
|
---|
107 | "\n"
|
---|
108 | "In case of success, 0 is returned, a value>0 otherwise.\n"
|
---|
109 | "\n"
|
---|
110 | "Usage: rootifysql [rootify.sql [rootify.root]] [-u URI] [-q query|-f file] [-i] [-o out] [-f] [-cN] [-t tree] [-vN]\n"
|
---|
111 | "\n"
|
---|
112 | ;
|
---|
113 | cout << endl;
|
---|
114 | }
|
---|
115 |
|
---|
116 | struct ExplainParser
|
---|
117 | {
|
---|
118 | string sql;
|
---|
119 |
|
---|
120 | vector<string> vec;
|
---|
121 |
|
---|
122 | string substitute(string _str, const regex &expr)
|
---|
123 | {
|
---|
124 | smatch match;
|
---|
125 | while (regex_search(_str, match, expr, regex_constants::format_first_only))
|
---|
126 | {
|
---|
127 | const auto &len = match.length(0);
|
---|
128 | const auto &pos = match.position(0);
|
---|
129 | const auto &str = match.str(0);
|
---|
130 |
|
---|
131 | const auto it = find(vec.cbegin(), vec.cend(), str);
|
---|
132 | const size_t id = it==vec.cend() ? vec.size() : it-vec.cbegin();
|
---|
133 |
|
---|
134 | _str.replace(pos, len, "{"+to_string(id)+"}");
|
---|
135 |
|
---|
136 | if (it==vec.cend())
|
---|
137 | vec.push_back(str);//.substr(1, str.size()-2));
|
---|
138 | }
|
---|
139 |
|
---|
140 | return _str;
|
---|
141 | }
|
---|
142 |
|
---|
143 | string substitute(const string &str, const string &expr)
|
---|
144 | {
|
---|
145 | return substitute(str, regex(expr));
|
---|
146 | }
|
---|
147 |
|
---|
148 | vector<string> queries;
|
---|
149 |
|
---|
150 | string resub(string str)
|
---|
151 | {
|
---|
152 | // search for "KEYWORD expression"
|
---|
153 | regex reg("\\{[0-9]+\\}");
|
---|
154 |
|
---|
155 | //smatch match;
|
---|
156 | smatch match;
|
---|
157 | while (regex_search(str, match, reg, regex_constants::format_first_only))
|
---|
158 | {
|
---|
159 | const auto &len = match.length(0);
|
---|
160 | const auto &pos = match.position(0);
|
---|
161 | const auto &arg = match.str(0); // Argument
|
---|
162 |
|
---|
163 | const auto idx = atoi(arg.c_str()+1);
|
---|
164 |
|
---|
165 | str.replace(pos, len, resub(vec[idx]));
|
---|
166 | }
|
---|
167 |
|
---|
168 | return str;
|
---|
169 | }
|
---|
170 |
|
---|
171 | void expression(string expr, size_t indent=0)
|
---|
172 | {
|
---|
173 | if (expr[0]=='{')
|
---|
174 | {
|
---|
175 | const auto idx = atoi(expr.c_str()+1);
|
---|
176 |
|
---|
177 | // This is a subquery
|
---|
178 | if (vec[idx].substr(0,3)=="(/*")
|
---|
179 | {
|
---|
180 | cout << setw(indent) << ' ' << "(\n";
|
---|
181 | find_tokens(vec[idx], indent+4);
|
---|
182 | cout << setw(indent) << ' ' << ") ";
|
---|
183 | }
|
---|
184 | else
|
---|
185 | // This is just something to substitute back
|
---|
186 | if (vec[idx].substr(0,2)=="({")
|
---|
187 | {
|
---|
188 | cout << setw(indent) << ' ' << "(" << resub(vec[idx]) << ") ";
|
---|
189 | }
|
---|
190 | else
|
---|
191 | {
|
---|
192 | if (indent>0)
|
---|
193 | cout << setw(indent) << ' ';
|
---|
194 | cout << resub(vec[idx]);
|
---|
195 | }
|
---|
196 | }
|
---|
197 | else
|
---|
198 | {
|
---|
199 | if (indent>0)
|
---|
200 | cout << setw(indent) << ' ';
|
---|
201 | cout << resub(expr);
|
---|
202 | }
|
---|
203 | }
|
---|
204 |
|
---|
205 | void find_tokens(string str, size_t indent=0)
|
---|
206 | {
|
---|
207 | // ( COMMENT )?( TOKEN )?(( {NNN} | NNN )( AS|ON ( {NNN}) ))?(,)?)
|
---|
208 | //regex reg("(\\/\\*\\ select\\#[0-9]+\\ \\*\\/\\ *)?([a-zA-Z ]+)?((\\{[0-9]+\\}|[0-9]+)(\\ ?([Aa][Ss]|[Oo][Nn])\\ ?(\\{[0-9]+\\}))?(,)?)");
|
---|
209 |
|
---|
210 | const string _com = "\\/\\*\\ select\\#[0-9]+\\ \\*\\/\\ *";
|
---|
211 |
|
---|
212 | const string _tok = "[a-zA-Z_ ]+";
|
---|
213 |
|
---|
214 | const string _nnn = "\\{[0-9]+\\}|[0-9]+";
|
---|
215 |
|
---|
216 | const string _as = "\\ ?([Aa][Ss])\\ ?";
|
---|
217 |
|
---|
218 | // ( _nnn ) ( _as ( _nnn ))?(,)? // can also match noting in between two {NNN}
|
---|
219 | const string _exp = "("+_nnn+")" + "("+_as+"("+_nnn+"))?(,)?";
|
---|
220 |
|
---|
221 | // Matche: ( _com )? ( ( _tok )? ( _exp ) | ( _tok ) )
|
---|
222 | regex reg("("+_com+")?" + "(" + "("+_tok+")?"+"("+_exp+")" + "|" + "("+_tok+")" + ")");
|
---|
223 |
|
---|
224 | smatch match;
|
---|
225 | while (regex_search(str, match, reg, regex_constants::format_first_only))
|
---|
226 | {
|
---|
227 |
|
---|
228 | const auto &com = match.str(1); // comment
|
---|
229 | const auto &tok1 = Tools::Trim(match.str(3)); // token with expression
|
---|
230 | const auto &arg1 = match.str(5); // argument 1
|
---|
231 | const auto &as = match.str(7); // as
|
---|
232 | const auto &arg2 = match.str(8); // argument 2
|
---|
233 | const auto &comma = match.str(9); // comma
|
---|
234 | const auto &tok2 = Tools::Trim(match.str(10)); // token without expression
|
---|
235 |
|
---|
236 | if (!com.empty())
|
---|
237 | cout << setw(indent) << ' ' << "\033[34m" << com << "\033[0m" << '\n';
|
---|
238 |
|
---|
239 | if (!tok1.empty())
|
---|
240 | cout << setw(indent) << ' ' << "\033[32m" << tok1 << "\033[0m" << '\n';
|
---|
241 | if (!tok2.empty())
|
---|
242 | cout << setw(indent) << ' ' << "\033[32m" << tok2 << "\033[0m" << '\n';
|
---|
243 |
|
---|
244 | if (!arg1.empty())
|
---|
245 | {
|
---|
246 | expression(arg1, indent+4);
|
---|
247 |
|
---|
248 | if (!as.empty())
|
---|
249 | cout << " \033[33m" << as << "\033[0m ";
|
---|
250 |
|
---|
251 | if (!arg2.empty())
|
---|
252 | expression(arg2);
|
---|
253 |
|
---|
254 | if (!comma.empty())
|
---|
255 | cout << ',';
|
---|
256 |
|
---|
257 | cout << '\n';
|
---|
258 | }
|
---|
259 |
|
---|
260 | str = str.substr(match.position(0)+match.length(0));
|
---|
261 | }
|
---|
262 | }
|
---|
263 |
|
---|
264 |
|
---|
265 | ExplainParser(const string &_sql) : sql(_sql)
|
---|
266 | {
|
---|
267 | // substitute all strings
|
---|
268 | sql = substitute(sql, "'[^']*'");
|
---|
269 |
|
---|
270 | // substitute all escaped sequences (`something`.`something-else`)
|
---|
271 | sql = substitute(sql, "`[^`]*`(\\.`[^`]*`)*");
|
---|
272 |
|
---|
273 | // substitute all paranthesis
|
---|
274 | sql = substitute(sql, "[a-zA-Z0-9_]*\\([^\\(\\)]*\\)");
|
---|
275 |
|
---|
276 | //cout << sql << "\n\n";
|
---|
277 | find_tokens(sql);
|
---|
278 | cout << endl;
|
---|
279 | }
|
---|
280 | };
|
---|
281 |
|
---|
282 | // Remove queries...
|
---|
283 | void format(string sql)
|
---|
284 | {
|
---|
285 | ExplainParser p(sql);
|
---|
286 |
|
---|
287 | /*
|
---|
288 |
|
---|
289 | SELECT
|
---|
290 | [ALL | DISTINCT | DISTINCTROW ]
|
---|
291 | [HIGH_PRIORITY]
|
---|
292 | [STRAIGHT_JOIN]
|
---|
293 | [SQL_SMALL_RESULT] [SQL_BIG_RESULT] [SQL_BUFFER_RESULT]
|
---|
294 | [SQL_CACHE | SQL_NO_CACHE] [SQL_CALC_FOUND_ROWS]
|
---|
295 | select_expr [, select_expr ...]
|
---|
296 | [FROM table_references
|
---|
297 | [PARTITION partition_list]
|
---|
298 | [WHERE where_condition]
|
---|
299 | [GROUP BY {col_name | expr | position}, ... [WITH ROLLUP]]
|
---|
300 | [HAVING where_condition]
|
---|
301 | [WINDOW window_name AS (window_spec)
|
---|
302 | [, window_name AS (window_spec)] ...]
|
---|
303 | [ORDER BY {col_name | expr | position}
|
---|
304 | [ASC | DESC], ... [WITH ROLLUP]]
|
---|
305 | [LIMIT {[offset,] row_count | row_count OFFSET offset}]
|
---|
306 | [INTO OUTFILE 'file_name'
|
---|
307 | [CHARACTER SET charset_name]
|
---|
308 | export_options
|
---|
309 | | INTO DUMPFILE 'file_name'
|
---|
310 | | INTO var_name [, var_name]]
|
---|
311 | [FOR {UPDATE | SHARE} [OF tbl_name [, tbl_name] ...] [NOWAIT | SKIP LOCKED]
|
---|
312 | | LOCK IN SHARE MODE]]
|
---|
313 | */
|
---|
314 |
|
---|
315 | /*
|
---|
316 | table_references:
|
---|
317 | escaped_table_reference [, escaped_table_reference] ...
|
---|
318 |
|
---|
319 | escaped_table_reference:
|
---|
320 | table_reference
|
---|
321 | | { OJ table_reference }
|
---|
322 |
|
---|
323 | table_reference:
|
---|
324 | table_factor
|
---|
325 | | join_table
|
---|
326 |
|
---|
327 | table_factor:
|
---|
328 | tbl_name [PARTITION (partition_names)]
|
---|
329 | [[AS] alias] [index_hint_list]
|
---|
330 | | table_subquery [AS] alias [(col_list)]
|
---|
331 | | ( table_references )
|
---|
332 |
|
---|
333 | join_table:
|
---|
334 | table_reference [INNER | CROSS] JOIN table_factor [join_condition]
|
---|
335 | | table_reference STRAIGHT_JOIN table_factor
|
---|
336 | | table_reference STRAIGHT_JOIN table_factor ON conditional_expr
|
---|
337 | | table_reference {LEFT|RIGHT} [OUTER] JOIN table_reference join_condition
|
---|
338 | | table_reference NATURAL [INNER | {LEFT|RIGHT} [OUTER]] JOIN table_factor
|
---|
339 |
|
---|
340 | join_condition:
|
---|
341 | ON conditional_expr
|
---|
342 | | USING (column_list)
|
---|
343 |
|
---|
344 | index_hint_list:
|
---|
345 | index_hint [, index_hint] ...
|
---|
346 |
|
---|
347 | index_hint:
|
---|
348 | USE {INDEX|KEY}
|
---|
349 | [FOR {JOIN|ORDER BY|GROUP BY}] ([index_list])
|
---|
350 | | IGNORE {INDEX|KEY}
|
---|
351 | [FOR {JOIN|ORDER BY|GROUP BY}] (index_list)
|
---|
352 | | FORCE {INDEX|KEY}
|
---|
353 | [FOR {JOIN|ORDER BY|GROUP BY}] (index_list)
|
---|
354 |
|
---|
355 | index_list:
|
---|
356 | index_name [, index_name] ...
|
---|
357 | */
|
---|
358 |
|
---|
359 | }
|
---|
360 |
|
---|
361 | int finish(Database &connection, const uint16_t &verbose, const bool &profiling, const bool &print_connection)
|
---|
362 | {
|
---|
363 | if (verbose>0)
|
---|
364 | {
|
---|
365 | try
|
---|
366 | {
|
---|
367 | const auto resw =
|
---|
368 | connection.query("SHOW WARNINGS").store();
|
---|
369 |
|
---|
370 | if (resw.num_rows()>0)
|
---|
371 | cout << "\n" << resw.num_rows() << " Warning(s) issued:\n\n";
|
---|
372 |
|
---|
373 | for (size_t i=0; i<resw.num_rows(); i++)
|
---|
374 | {
|
---|
375 | const mysqlpp::Row &roww = resw[i];
|
---|
376 |
|
---|
377 | cout << roww["Level"] << '[' << roww["Code"] << "]: ";
|
---|
378 | cout << roww["Message"] << '\n';
|
---|
379 | }
|
---|
380 | cout << endl;
|
---|
381 |
|
---|
382 | }
|
---|
383 | catch (const exception &e)
|
---|
384 | {
|
---|
385 | cerr << "\nSHOW WARNINGS\n\n";
|
---|
386 | cerr << "SQL query failed:\n" << e.what() << endl;
|
---|
387 | return 1;
|
---|
388 | }
|
---|
389 | }
|
---|
390 |
|
---|
391 | if (profiling)
|
---|
392 | {
|
---|
393 | try
|
---|
394 | {
|
---|
395 | const auto N =
|
---|
396 | connection.query("SHOW PROFILES").store().num_rows();
|
---|
397 |
|
---|
398 | const auto resp =
|
---|
399 | connection.query("SHOW PROFILE ALL FOR QUERY "+to_string(verbose?N-1:N)).store();
|
---|
400 |
|
---|
401 | cout << '\n';
|
---|
402 | cout << left;
|
---|
403 | cout << setw(26) << "Status" << ' ';
|
---|
404 | cout << right;
|
---|
405 | cout << setw(11) << "Duration" << ' ';
|
---|
406 | cout << setw(11) << "CPU User" << ' ';
|
---|
407 | cout << setw(11) << "CPU System" << '\n';
|
---|
408 | cout << "--------------------------------------------------------------\n";
|
---|
409 | for (size_t i=0; i<resp.num_rows(); i++)
|
---|
410 | {
|
---|
411 | const mysqlpp::Row &rowp = resp[i];
|
---|
412 |
|
---|
413 | cout << left;
|
---|
414 | cout << setw(26) << rowp["Status"] << ' ';
|
---|
415 | cout << right;
|
---|
416 | cout << setw(11) << rowp["Duration"] << ' ';
|
---|
417 | cout << setw(11) << rowp["CPU_user"] << ' ';
|
---|
418 | cout << setw(11) << rowp["CPU_system"] << '\n';
|
---|
419 | }
|
---|
420 | cout << "--------------------------------------------------------------\n";
|
---|
421 | cout << endl;
|
---|
422 | }
|
---|
423 | catch (const exception &e)
|
---|
424 | {
|
---|
425 | cerr << "\nSHOW PROFILE ALL\n\n";
|
---|
426 | cerr << "SQL query failed:\n" << e.what() << '\n' <<endl;
|
---|
427 | return 2;
|
---|
428 | }
|
---|
429 | }
|
---|
430 |
|
---|
431 | if (print_connection)
|
---|
432 | {
|
---|
433 | try
|
---|
434 | {
|
---|
435 | // Exchange _send and _received as it is the view of the server
|
---|
436 | const auto &res1 = connection.query("SHOW STATUS LIKE 'Bytes_%'").store();
|
---|
437 | cout << left << setw(16) << res1[1]["Variable_name"] << ' ' << Tools::Scientific(res1[0]["Value"]) << endl;
|
---|
438 | cout << left << setw(16) << res1[0]["Variable_name"] << ' ' << Tools::Scientific(res1[1]["Value"]) << endl;
|
---|
439 | cout << endl;
|
---|
440 | }
|
---|
441 | catch (const exception &e)
|
---|
442 | {
|
---|
443 | cerr << "\nSHOW STATUS LIKE 'Bytes_%'\n\n";
|
---|
444 | cerr << "SQL query failed:\n" << e.what() << endl;
|
---|
445 | return 3;
|
---|
446 | }
|
---|
447 | }
|
---|
448 |
|
---|
449 | return 0;
|
---|
450 |
|
---|
451 | }
|
---|
452 |
|
---|
453 | int main(int argc, const char* argv[])
|
---|
454 | {
|
---|
455 | Time start;
|
---|
456 |
|
---|
457 | gROOT->SetBatch();
|
---|
458 |
|
---|
459 | Configuration conf(argv[0]);
|
---|
460 | conf.SetPrintUsage(PrintUsage);
|
---|
461 | SetupConfiguration(conf);
|
---|
462 |
|
---|
463 | if (!conf.DoParse(argc, argv))
|
---|
464 | return 127;
|
---|
465 |
|
---|
466 | // ----------------------------- Evaluate options --------------------------
|
---|
467 | const string uri = conf.Get<string>("uri");
|
---|
468 | const string out = conf.Get<string>("out");
|
---|
469 | const string file = conf.Get<string>("file");
|
---|
470 | const string tree = conf.Get<string>("tree");
|
---|
471 | const bool force = conf.Get<bool>("force");
|
---|
472 | const bool ignorenull = conf.Get<bool>("ignore-null");
|
---|
473 | const bool update = conf.Get<bool>("update");
|
---|
474 | const bool display = conf.Get<bool>("display");
|
---|
475 | const string write = conf.Get<string>("write");
|
---|
476 | const bool noout = conf.Get<bool>("null");
|
---|
477 | const bool nofill = conf.Get<bool>("no-fill");
|
---|
478 | const bool explain = conf.Get<bool>("explain");
|
---|
479 | const bool profiling = conf.Get<bool>("profiling");
|
---|
480 | const uint16_t verbose = conf.Get<uint16_t>("verbose");
|
---|
481 | const uint16_t compression = conf.Get<uint16_t>("compression");
|
---|
482 | const string delimiter = conf.Get<string>("delimiter");
|
---|
483 | const vector<string> _ignore = conf.Vec<string>("ignore");
|
---|
484 | const bool print_connection = conf.Get<bool>("print-connection");
|
---|
485 | //const vector<Map> mymap = conf.Vec<Map>("map");
|
---|
486 |
|
---|
487 | // -------------------------------------------------------------------------
|
---|
488 |
|
---|
489 | const auto vars = conf.GetWildcardOptions("var.*");
|
---|
490 |
|
---|
491 | vector<string> variables;
|
---|
492 | for (const auto &var : vars)
|
---|
493 | variables.emplace_back('@'+var.substr(4)+":="+Tools::Trim(conf.Get<string>(var)));
|
---|
494 |
|
---|
495 | // -------------------------------------------------------------------------
|
---|
496 |
|
---|
497 | if (verbose>0)
|
---|
498 | cout << "\n------------------------ Rootify SQL -------------------------" << endl;
|
---|
499 |
|
---|
500 | string query = conf.Get<string>("query");
|
---|
501 | if (query.empty())
|
---|
502 | {
|
---|
503 | if (verbose>0)
|
---|
504 | cout << "Reading query from file '" << file << "'." << endl;
|
---|
505 |
|
---|
506 | ifstream fin(file);
|
---|
507 | if (!fin)
|
---|
508 | {
|
---|
509 | cerr << "Could not open query in '" << file << "': " << strerror(errno) << endl;
|
---|
510 | return 4;
|
---|
511 | }
|
---|
512 | getline(fin, query, (char)fin.eof());
|
---|
513 | }
|
---|
514 |
|
---|
515 | if (query.empty())
|
---|
516 | {
|
---|
517 | cerr << "No query specified." << endl;
|
---|
518 | return 5;
|
---|
519 | }
|
---|
520 |
|
---|
521 | // -------------------------------------------------------------------------
|
---|
522 |
|
---|
523 | map<string, vector<string>> envs;
|
---|
524 |
|
---|
525 | for (const auto &env : conf.GetWildcardOptions("env.*"))
|
---|
526 | envs[env.substr(4)] = conf.Vec<string>(env);
|
---|
527 |
|
---|
528 | for (const auto &env : conf.GetWildcardOptions("list.*"))
|
---|
529 | {
|
---|
530 | const string fname = conf.Get<string>(env);
|
---|
531 | const string &ident = env.substr(5);
|
---|
532 |
|
---|
533 | ifstream fin(fname);
|
---|
534 | if (!fin)
|
---|
535 | {
|
---|
536 | cerr << "Could not open environment in '" << fname << "' for ${" << ident << "}: " << strerror(errno) << endl;
|
---|
537 | return 6;
|
---|
538 | }
|
---|
539 |
|
---|
540 | for (string line; getline(fin, line); )
|
---|
541 | {
|
---|
542 | const auto &l = Tools::Trim(line);
|
---|
543 | if (!l.empty() && l[0]!='#')
|
---|
544 | envs[ident].push_back(line);
|
---|
545 | }
|
---|
546 |
|
---|
547 | if (verbose>0)
|
---|
548 | cout << "Found " << envs[ident].size() << " list element(s) for ${" << ident << "}" << endl;
|
---|
549 | }
|
---|
550 |
|
---|
551 | for (const auto &env : envs)
|
---|
552 | {
|
---|
553 | regex rexpr("\\$(\\{"+env.first+"\\}|"+env.first+"\\b)");
|
---|
554 | query = regex_replace(query, rexpr, boost::join(env.second, ", "));
|
---|
555 | }
|
---|
556 |
|
---|
557 | // -------------------------- Check for file permssion ---------------------
|
---|
558 | // Strictly speaking, checking for write permission and existance is not necessary,
|
---|
559 | // but it is convenient that the user does not find out that it failed after
|
---|
560 | // waiting long for the query result
|
---|
561 | //
|
---|
562 | // I am using root here instead of boost to be
|
---|
563 | // consistent with the access pattern by TFile
|
---|
564 | TString path(noout?"/dev/null":out.c_str());
|
---|
565 | gSystem->ExpandPathName(path);
|
---|
566 |
|
---|
567 | if (!noout)
|
---|
568 | {
|
---|
569 | FileStat_t stat;
|
---|
570 | const Int_t exist = !gSystem->GetPathInfo(path, stat);
|
---|
571 | const Bool_t _write = !gSystem->AccessPathName(path, kWritePermission) && R_ISREG(stat.fMode);
|
---|
572 |
|
---|
573 | if ((update && !exist) || (update && exist && !_write) || (force && exist && !_write))
|
---|
574 | {
|
---|
575 | cerr << "File '" << path << "' is not writable." << endl;
|
---|
576 | return 7;
|
---|
577 | }
|
---|
578 |
|
---|
579 | if (!update && !force && exist)
|
---|
580 | {
|
---|
581 | cerr << "File '" << path << "' already exists." << endl;
|
---|
582 | return 8;
|
---|
583 | }
|
---|
584 | }
|
---|
585 |
|
---|
586 | Time start2;
|
---|
587 |
|
---|
588 | // --------------------------- Connect to database -------------------------------------------------
|
---|
589 |
|
---|
590 | if (query.back()!='\n')
|
---|
591 | query += '\n';
|
---|
592 |
|
---|
593 | if (verbose>0)
|
---|
594 | cout << "Connecting to database..." << endl;
|
---|
595 |
|
---|
596 |
|
---|
597 | Database connection(uri); // Keep alive while fetching rows
|
---|
598 |
|
---|
599 | if (verbose>0)
|
---|
600 | {
|
---|
601 | cout << "Client Version: " << connection.client_version() << '\n';
|
---|
602 | cout << "Server Version: " << connection.server_version() << endl;
|
---|
603 | }
|
---|
604 |
|
---|
605 | if (print_connection)
|
---|
606 | {
|
---|
607 | try
|
---|
608 | {
|
---|
609 | const auto &res1 = connection.query("SHOW STATUS LIKE 'Compression'").store();
|
---|
610 | cout << "Compression of database connection is " << string(res1[0][1]) << endl;
|
---|
611 |
|
---|
612 | const auto &res2 = connection.query("SHOW STATUS LIKE 'Ssl_cipher'").store();
|
---|
613 | cout << "Connection to databases is " << (string(res2[0][1]).empty()?"UNENCRYPTED":"ENCRYPTED ("+string(res2[0][1])+")") << endl;
|
---|
614 | }
|
---|
615 | catch (const exception &e)
|
---|
616 | {
|
---|
617 | cerr << "\nSHOW STATUS LIKE 'Compression'\n\n";
|
---|
618 | cerr << "SQL query failed:\n" << e.what() << endl;
|
---|
619 | return 9;
|
---|
620 | }
|
---|
621 | }
|
---|
622 |
|
---|
623 | try
|
---|
624 | {
|
---|
625 | if (profiling)
|
---|
626 | connection.query("SET PROFILING=1").execute();
|
---|
627 | }
|
---|
628 | catch (const exception &e)
|
---|
629 | {
|
---|
630 | cerr << "\nSET profiling=1\n\n";
|
---|
631 | cerr << "SQL query failed:\n" << e.what() << endl;
|
---|
632 | return 10;
|
---|
633 | }
|
---|
634 |
|
---|
635 | // -------------------------- Set user defined variables -------------------
|
---|
636 | if (variables.size()>0)
|
---|
637 | {
|
---|
638 | if (verbose>0)
|
---|
639 | cout << "Setting user defined variables..." << endl;
|
---|
640 |
|
---|
641 | const string varset =
|
---|
642 | "SET\n "+boost::algorithm::join(variables, ",\n ");
|
---|
643 |
|
---|
644 | try
|
---|
645 | {
|
---|
646 | connection.query(varset).execute();
|
---|
647 | }
|
---|
648 | catch (const exception &e)
|
---|
649 | {
|
---|
650 | cerr << '\n' << varset << "\n\n";
|
---|
651 | cerr << "SQL query failed:\n" << e.what() << endl;
|
---|
652 | return 11;
|
---|
653 | }
|
---|
654 |
|
---|
655 | if (verbose>2)
|
---|
656 | cout << '\n' << varset << '\n' << endl;
|
---|
657 | }
|
---|
658 |
|
---|
659 | // ------------------------- Explain query if requested --------------------
|
---|
660 |
|
---|
661 | if (explain)
|
---|
662 | {
|
---|
663 | try
|
---|
664 | {
|
---|
665 | const auto res0 =
|
---|
666 | connection.query("EXPLAIN FORMAT=JSON "+query).store();
|
---|
667 |
|
---|
668 | cout << res0[0][0] << endl;
|
---|
669 | cout << endl;
|
---|
670 |
|
---|
671 | const mysqlpp::StoreQueryResult res1 =
|
---|
672 | connection.query("EXPLAIN "+query).store();
|
---|
673 |
|
---|
674 | for (size_t i=0; i<res1.num_rows(); i++)
|
---|
675 | {
|
---|
676 | const mysqlpp::Row &row = res1[i];
|
---|
677 |
|
---|
678 | cout << "\nid : " << row["id"];
|
---|
679 | cout << "\nselect type : " << row["select_type"];
|
---|
680 |
|
---|
681 | if (!row["table"].is_null())
|
---|
682 | cout << "\ntable : " << row["table"];
|
---|
683 |
|
---|
684 | if (!row["partitions"].is_null())
|
---|
685 | cout << "\npartitions : " << row["partitions"];
|
---|
686 |
|
---|
687 | if (!row["key"].is_null())
|
---|
688 | cout << "\nselected key : " << row["key"] << " [len=" << row["key_len"] << "] out of (" << row["possible_keys"] << ")";
|
---|
689 |
|
---|
690 | if (!row["type"].is_null())
|
---|
691 | cout << "\njoin type : " << row["type"];
|
---|
692 |
|
---|
693 | //if (!row["possible_keys"].is_null())
|
---|
694 | // cout << "\npossible_keys: " << row["possible_keys"];
|
---|
695 |
|
---|
696 | //if (!row["key_len"].is_null())
|
---|
697 | // cout << "\nkey_len : " << row["key_len"];
|
---|
698 |
|
---|
699 | if (!row["ref"].is_null())
|
---|
700 | cout << "\nref : (" << row["ref"] << ") compared to the index";
|
---|
701 |
|
---|
702 | if (!row["rows"].is_null())
|
---|
703 | cout << "\nrows : " << row["rows"];
|
---|
704 |
|
---|
705 | if (!row["filtered"].is_null())
|
---|
706 | cout << "\nfiltered : " << row["filtered"];
|
---|
707 |
|
---|
708 | if (!row["extra"].is_null())
|
---|
709 | cout << "\nExtra : " << row["extra"];
|
---|
710 |
|
---|
711 | cout << endl;
|
---|
712 | }
|
---|
713 |
|
---|
714 | cout << endl;
|
---|
715 |
|
---|
716 | const mysqlpp::StoreQueryResult res2 =
|
---|
717 | connection.query("SHOW WARNINGS").store();
|
---|
718 |
|
---|
719 | for (size_t i=0; i<res2.num_rows(); i++)
|
---|
720 | {
|
---|
721 | const mysqlpp::Row &row = res2[i];
|
---|
722 |
|
---|
723 | // 1003 //
|
---|
724 | cout << row["Level"] << '[' << row["Code"] << "]:\n";
|
---|
725 | if (uint32_t(row["Code"])==1003)
|
---|
726 | format(row["Message"].c_str());
|
---|
727 | else
|
---|
728 | cout << row["Message"] << '\n' << endl;
|
---|
729 |
|
---|
730 | }
|
---|
731 |
|
---|
732 | }
|
---|
733 | catch (const exception &e)
|
---|
734 | {
|
---|
735 | cerr << '\n' << query << "\n\n";
|
---|
736 | cerr << "SQL query failed:\n" << e.what() << endl;
|
---|
737 | return 12;
|
---|
738 | }
|
---|
739 |
|
---|
740 | return 0;
|
---|
741 | }
|
---|
742 |
|
---|
743 | // -------------------------- Request data from database -------------------
|
---|
744 | if (verbose>0)
|
---|
745 | cout << "Requesting data..." << endl;
|
---|
746 |
|
---|
747 | if (verbose>2)
|
---|
748 | cout << '\n' << query << endl;
|
---|
749 |
|
---|
750 | const mysqlpp::UseQueryResult res =
|
---|
751 | connection.query(query).use();
|
---|
752 |
|
---|
753 | // -------------------------------------------------------------------------
|
---|
754 |
|
---|
755 | if (verbose>0)
|
---|
756 | {
|
---|
757 | cout << "Opening file '" << path << "' [compression=" << compression << "]...\n";
|
---|
758 | cout << "Writing data to tree '" << tree << "'" << endl;
|
---|
759 | }
|
---|
760 |
|
---|
761 | // ----------------------------- Open output file --------------------------
|
---|
762 | TFile tfile(path, update?"UPDATE":(force?"RECREATE":"CREATE"), "Rootify SQL", compression);
|
---|
763 | if (tfile.IsZombie())
|
---|
764 | return 13;
|
---|
765 |
|
---|
766 | // -------------------------------------------------------------------------
|
---|
767 |
|
---|
768 | // get the first row to get the field description
|
---|
769 | mysqlpp::Row row = res.fetch_row();
|
---|
770 | if (!row)
|
---|
771 | {
|
---|
772 | cerr << "Empty set returned... nothing to write." << endl;
|
---|
773 | finish(connection, verbose, profiling, print_connection);
|
---|
774 | return 14;
|
---|
775 | }
|
---|
776 |
|
---|
777 | if (verbose>0)
|
---|
778 | cout << "Trying to setup " << row.size() << " branches..." << endl;
|
---|
779 |
|
---|
780 | if (verbose>1)
|
---|
781 | cout << endl;
|
---|
782 |
|
---|
783 | const mysqlpp::FieldNames &l = *row.field_list().list;
|
---|
784 |
|
---|
785 | vector<double> buf(l.size());
|
---|
786 | vector<uint8_t> typ(l.size(),'n'); // n=number [double], d is used for DateTime
|
---|
787 |
|
---|
788 | UInt_t cols = 0;
|
---|
789 |
|
---|
790 |
|
---|
791 | // -------------------- Configure branches of TTree ------------------------
|
---|
792 | TTree *ttree = new TTree(tree.c_str(), query.c_str());
|
---|
793 |
|
---|
794 | size_t skipat = 0;
|
---|
795 | size_t skipreg = 0;
|
---|
796 | for (size_t i=0; i<l.size(); i++)
|
---|
797 | {
|
---|
798 | const string t = row[i].type().sql_name();
|
---|
799 |
|
---|
800 | if (t.find("DATETIME")!=string::npos)
|
---|
801 | typ[i] = 'd';
|
---|
802 | else
|
---|
803 | if (t.find("DATE")!=string::npos)
|
---|
804 | typ[i] = 'D';
|
---|
805 | else
|
---|
806 | if (t.find("TIME")!=string::npos)
|
---|
807 | typ[i] = 'T';
|
---|
808 | else
|
---|
809 | if (t.find("VARCHAR")!=string::npos)
|
---|
810 | typ[i] = 'V';
|
---|
811 | else
|
---|
812 | if (t.find("CHAR")!=string::npos)
|
---|
813 | typ[i] = 'C';
|
---|
814 |
|
---|
815 | bool found = false;
|
---|
816 | for (const auto &pattern: _ignore)
|
---|
817 | {
|
---|
818 | if (regex_match(l[i], regex(pattern)))
|
---|
819 | {
|
---|
820 | found = true;
|
---|
821 | typ[i] = '-';
|
---|
822 | skipreg++;
|
---|
823 | break;
|
---|
824 | }
|
---|
825 | }
|
---|
826 |
|
---|
827 | if (l[i][0]=='@')
|
---|
828 | {
|
---|
829 | typ[i] = '@';
|
---|
830 | skipat++;
|
---|
831 | }
|
---|
832 |
|
---|
833 | const bool use = l[i][0]!='@' && typ[i]!='V' && typ[i]!='C' && !found;
|
---|
834 |
|
---|
835 | if (verbose>1)
|
---|
836 | cout << (use?" + ":" - ") << l[i].c_str() << " [" << t << "] {" << typ[i] << "}\n";
|
---|
837 |
|
---|
838 | if (use)
|
---|
839 | {
|
---|
840 | // string name = l[i];
|
---|
841 | // for (const auto &m: mymap)
|
---|
842 | // name = boost::regex_replace(l[i], boost::regex(m.first), m.second);
|
---|
843 |
|
---|
844 | ttree->Branch(l[i].c_str(), buf.data()+i);
|
---|
845 | cols++;
|
---|
846 | }
|
---|
847 | }
|
---|
848 | // -------------------------------------------------------------------------
|
---|
849 |
|
---|
850 | if (verbose>1)
|
---|
851 | cout << endl;
|
---|
852 | if (verbose>0)
|
---|
853 | {
|
---|
854 | if (skipreg)
|
---|
855 | cout << skipreg << " branches skipped due to ignore list." << endl;
|
---|
856 | if (skipat)
|
---|
857 | cout << skipat << " branches skipped due to name starting with @." << endl;
|
---|
858 | cout << "Configured " << cols << " branches.\nFilling branches..." << endl;
|
---|
859 | }
|
---|
860 |
|
---|
861 | ofstream fout(write);
|
---|
862 | if (!write.empty() && !fout)
|
---|
863 | cout << "WARNING: Writing to '" << write << "' failed: " << strerror(errno) << endl;
|
---|
864 |
|
---|
865 | if (display)
|
---|
866 | {
|
---|
867 | cout << endl;
|
---|
868 | cout << "#";
|
---|
869 | for (size_t i=0; i<l.size(); i++)
|
---|
870 | cout << ' ' << l[i].c_str();
|
---|
871 | cout << endl;
|
---|
872 | }
|
---|
873 |
|
---|
874 | if (!write.empty())
|
---|
875 | {
|
---|
876 | fout << "#";
|
---|
877 | for (size_t i=0; i<l.size(); i++)
|
---|
878 | fout << ' ' << l[i].c_str();
|
---|
879 | fout << endl;
|
---|
880 | }
|
---|
881 |
|
---|
882 | // ---------------------- Fill TTree with DB data --------------------------
|
---|
883 | size_t count = 0;
|
---|
884 | size_t skip = 0;
|
---|
885 | do
|
---|
886 | {
|
---|
887 | count++;
|
---|
888 |
|
---|
889 | ostringstream sout;
|
---|
890 |
|
---|
891 | size_t idx=0;
|
---|
892 | for (auto col=row.begin(); col!=row.end(); col++, idx++)
|
---|
893 | {
|
---|
894 | if (display || !write.empty())
|
---|
895 | {
|
---|
896 | if (idx>0)
|
---|
897 | sout << (delimiter.empty()?"\t":delimiter);
|
---|
898 | sout << col->c_str();
|
---|
899 | }
|
---|
900 |
|
---|
901 | if (!ignorenull && col->is_null())
|
---|
902 | {
|
---|
903 | skip++;
|
---|
904 | break;
|
---|
905 | }
|
---|
906 |
|
---|
907 | switch (typ[idx])
|
---|
908 | {
|
---|
909 | case 'd':
|
---|
910 | buf[idx] = time_t((mysqlpp::DateTime)(*col));
|
---|
911 | break;
|
---|
912 |
|
---|
913 | case 'D':
|
---|
914 | buf[idx] = time_t((mysqlpp::Date)(*col));
|
---|
915 | break;
|
---|
916 |
|
---|
917 | case 'T':
|
---|
918 | buf[idx] = time_t((mysqlpp::Time)(*col));
|
---|
919 | break;
|
---|
920 |
|
---|
921 | case 'V':
|
---|
922 | case 'C':
|
---|
923 | case '-':
|
---|
924 | case '@':
|
---|
925 | break;
|
---|
926 |
|
---|
927 | default:
|
---|
928 | buf[idx] = atof(col->c_str());
|
---|
929 | }
|
---|
930 | }
|
---|
931 |
|
---|
932 | if (idx==row.size())
|
---|
933 | {
|
---|
934 | if (!nofill)
|
---|
935 | ttree->Fill();
|
---|
936 |
|
---|
937 | if (display)
|
---|
938 | cout << sout.str() << endl;
|
---|
939 | if (!write.empty())
|
---|
940 | fout << sout.str() << '\n';
|
---|
941 | }
|
---|
942 |
|
---|
943 | row = res.fetch_row();
|
---|
944 |
|
---|
945 |
|
---|
946 | } while (row);
|
---|
947 |
|
---|
948 | // -------------------------------------------------------------------------
|
---|
949 |
|
---|
950 | if (display)
|
---|
951 | cout << '\n' << endl;
|
---|
952 |
|
---|
953 | if (verbose>0)
|
---|
954 | {
|
---|
955 | cout << count << " rows fetched." << endl;
|
---|
956 | if (skip>0)
|
---|
957 | cout << skip << " rows skipped due to NULL field." << endl;
|
---|
958 |
|
---|
959 | cout << ttree->GetEntries() << " rows filled into tree." << endl;
|
---|
960 | }
|
---|
961 |
|
---|
962 | ttree->Write();
|
---|
963 | tfile.Close();
|
---|
964 |
|
---|
965 | if (verbose>0)
|
---|
966 | {
|
---|
967 | const auto sec = Time().UnixTime()-start.UnixTime();
|
---|
968 |
|
---|
969 | cout << Tools::Scientific(tfile.GetSize()) << "B written to disk.\n";
|
---|
970 | cout << "File closed.\n";
|
---|
971 | cout << "Execution time: " << sec << "s ";
|
---|
972 | cout << "(" << Tools::Fractional(sec/count) << "s/row)\n";
|
---|
973 | cout << "--------------------------------------------------------------" << endl;
|
---|
974 | }
|
---|
975 |
|
---|
976 | return finish(connection, verbose, profiling, print_connection);
|
---|
977 | }
|
---|