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