Debian Codegraph @

www/rowvalue.html

raw
1 <!DOCTYPE html>
2 <html><head>
3 <meta name="viewport" content="width=device-width, initial-scale=1.0">
4 <meta http-equiv="content-type" content="text/html; charset=UTF-8">
5 <link href="sqlite.css" rel="stylesheet">
6 <title>Row Values</title>
7 <!-- path= -->
8 </head>
9 <body>
10 <div class=nosearch>
11 <a href="index.html">
12 <img class="logo" src="images/sqlite370_banner.svg" alt="SQLite" border="0">
13 </a>
14 <div><!-- IE hack to prevent disappearing logo --></div>
15 <div class="tagline desktoponly">
16 Small. Fast. Reliable.<br>Choose any three.
17 </div>
18 <div class="menu mainmenu">
19 <ul>
20 <li><a href="index.html">Home</a>
21 <li class='mobileonly'><a href="javascript:void(0)" onclick='toggle_div("submenu")'>Menu</a>
22 <li class='wideonly'><a href='about.html'>About</a>
23 <li class='desktoponly'><a href="docs.html">Documentation</a>
24 <li class='desktoponly'><a href="download.html">Download</a>
25 <li class='wideonly'><a href='copyright.html'>License</a>
26 <li class='desktoponly'><a href="support.html">Support</a>
27 <li class='desktoponly'><a href="prosupport.html">Purchase</a>
28 <li class='search' id='search_menubutton'>
29 <a href="javascript:void(0)" onclick='toggle_search()'>Search</a>
30 </ul>
31 </div>
32 <div class="menu submenu" id="submenu">
33 <ul>
34 <li><a href='about.html'>About</a>
35 <li><a href='docs.html'>Documentation</a>
36 <li><a href='download.html'>Download</a>
37 <li><a href='support.html'>Support</a>
38 <li><a href='prosupport.html'>Purchase</a>
39 </ul>
40 </div>
41 <div class="searchmenu" id="searchmenu">
42 <form method="GET" action="search">
43 <select name="s" id="searchtype">
44 <option value="d">Search Documentation</option>
45 <option value="c">Search Changelog</option>
46 </select>
47 <input type="text" name="q" id="searchbox" value="">
48 <input type="submit" value="Go">
49 </form>
50 </div>
51 </div>
52 <script>
53 function toggle_div(nm) {
54 var w = document.getElementById(nm);
55 if( w.style.display=="block" ){
56 w.style.display = "none";
57 }else{
58 w.style.display = "block";
59 }
60 }
61 function toggle_search() {
62 var w = document.getElementById("searchmenu");
63 if( w.style.display=="block" ){
64 w.style.display = "none";
65 } else {
66 w.style.display = "block";
67 setTimeout(function(){
68 document.getElementById("searchbox").focus()
69 }, 30);
70 }
71 }
72 function div_off(nm){document.getElementById(nm).style.display="none";}
73 window.onbeforeunload = function(e){div_off("submenu");}
74 /* Disable the Search feature if we are not operating from CGI, since */
75 /* Search is accomplished using CGI and will not work without it. */
76 if( !location.origin || !location.origin.match || !location.origin.match(/http/) ){
77 document.getElementById("search_menubutton").style.display = "none";
78 }
79 /* Used by the Hide/Show button beside syntax diagrams, to toggle the */
80 function hideorshow(btn,obj){
81 var x = document.getElementById(obj);
82 var b = document.getElementById(btn);
83 if( x.style.display!='none' ){
84 x.style.display = 'none';
85 b.innerHTML='show';
86 }else{
87 x.style.display = '';
88 b.innerHTML='hide';
89 }
90 return false;
91 }
92 var antiRobot = 0;
93 function antiRobotGo(){
94 if( antiRobot!=3 ) return;
95 antiRobot = 7;
96 var j = document.getElementById("mtimelink");
97 if(j && j.hasAttribute("data-href")) j.href=j.getAttribute("data-href");
98 }
99 function antiRobotDefense(){
100 document.body.onmousedown=function(){
101 antiRobot |= 2;
102 antiRobotGo();
103 document.body.onmousedown=null;
104 }
105 document.body.onmousemove=function(){
106 antiRobot |= 2;
107 antiRobotGo();
108 document.body.onmousemove=null;
109 }
110 setTimeout(function(){
111 antiRobot |= 1;
112 antiRobotGo();
113 }, 100)
114 antiRobotGo();
115 }
116 antiRobotDefense();
117 </script>
118 <div class=fancy>
119 <div class=nosearch>
120 <div class="fancy_title">
121 Row Values
122 </div>
123 <details class="fancy_toc">
124 <summary>Table Of Contents</summary>
125 <div id="toc_sub"><div class="fancy-toc1"><a href="#definitions">1. Definitions</a></div>
126 <div class="fancy-toc1"><a href="#syntax">2. Syntax</a></div>
127 <div class="fancy-toc2"><a href="#row_value_comparisons">2.1. Row Value Comparisons</a></div>
128 <div class="fancy-toc2"><a href="#row_value_in_operators">2.2. Row Value IN Operators</a></div>
129 <div class="fancy-toc2"><a href="#row_values_in_update_statements">2.3. Row Values In UPDATE Statements</a></div>
130 <div class="fancy-toc1"><a href="#example_uses_of_row_values">3. Example Uses Of Row Values</a></div>
131 <div class="fancy-toc2"><a href="#scrolling_window_queries">3.1. Scrolling Window Queries</a></div>
132 <div class="fancy-toc2"><a href="#comparison_of_dates_stored_as_separate_fields">3.2. Comparison of dates stored as separate fields</a></div>
133 <div class="fancy-toc2"><a href="#search_against_multi_column_keys">3.3. Search against multi-column keys</a></div>
134 <div class="fancy-toc2"><a href="#update_multiple_columns_of_a_table_based_on_a_query">3.4. Update multiple columns of a table based on a query</a></div>
135 <div class="fancy-toc2"><a href="#clarity_of_presentation">3.5. Clarity of presentation</a></div>
136 <div class="fancy-toc1"><a href="#backwards_compatibility">4. Backwards Compatibility</a></div>
137 </div>
138 </details>
139 </div>
140
141
142
143
144
145 <h1 id="definitions"><span>1. </span>Definitions</h1>
146
147 <p>A "value" is a single number, string, BLOB or NULL.
148 Sometimes the qualified name "scalar value" is used to emphasize that
149 only a single quantity is involved.
150
151 </p><p>A "row value" is an ordered list of two or more scalar values.
152 In other words, a "row value" is a vector or tuple.
153
154 </p><p>The "size" of a row value is the number of scalar values the row value contains.
155 The size of a row value is always at least 2.
156 A row value with a single column is just a scalar value.
157 A row value with no columns is a syntax error.
158
159 </p><h1 id="syntax"><span>2. </span>Syntax</h1>
160
161 <p>SQLite allows row values to be expressed in two ways:
162 </p><ol>
163 <li>A parenthesized, comma-separated list of scalar values.
164 </li><li>A subquery expression with two or more result columns.
165 </li></ol>
166
167 <p>SQLite can use row values in two contexts:
168 </p><ol>
169 <li>Two row values of the same size
170 can be compared using operators &lt;, &lt;=, &gt;, &gt;=,
171 =, &lt;&gt;, IS, IS NOT, IN, NOT IN, BETWEEN, or CASE.
172 </li><li>In an <a href="lang_update.html">UPDATE</a> statement, a list of column names can be set to a row value of
173 the same size.
174 </li></ol>
175
176 <p>The syntax for row values and the circumstances in which row values
177 can be used are illustrated in examples below.
178
179 </p><h2 id="row_value_comparisons"><span>2.1. </span>Row Value Comparisons</h2>
180
181 <p>Two row values are compared by looking at the constituent scalar
182 values from left to right.
183 A NULL means "unknown".
184 The overall result of comparison is NULL if it is possible to make the
185 result either true or false by substituting alternative values in place
186 of the constituent NULLs.
187 The following query demonstrates some row value comparisons:
188
189 </p><div class="codeblock"><pre>SELECT
190 (1,2,3) = (1,2,3), -- 1
191 (1,2,3) = (1,NULL,3), -- NULL
192 (1,2,3) = (1,NULL,4), -- 0
193 (1,2,3) &lt; (2,3,4), -- 1
194 (1,2,3) &lt; (1,2,4), -- 1
195 (1,2,3) &lt; (1,3,NULL), -- 1
196 (1,2,3) &lt; (1,2,NULL), -- NULL
197 (1,3,5) &lt; (1,2,NULL), -- 0
198 (1,2,NULL) IS (1,2,NULL); -- 1
199 </pre></div>
200
201 <p>The result of "(1,2,3)=(1,NULL,3)" is NULL because the result might be
202 true if we replaced NULL&rarr;2 or false if we replaced NULL&rarr;9.
203 The result of "(1,2,3)=(1,NULL,4)" is not NULL because there is no
204 substitution of the constituent NULL that will make the expression true,
205 since 3 will never equal 4 in the third column.
206
207 </p><p>Any of the row values in the previous example could be replaced by a
208 subquery that returns three columns and the same answer would result.
209 For example:
210
211 </p><div class="codeblock"><pre>CREATE TABLE t1(a,b,c);
212 INSERT INTO t1(a,b,c) VALUES(1,2,3);
213 SELECT (1,2,3)=(SELECT * FROM t1); -- 1
214 </pre></div>
215
216 <a name="rvinop"></a>
217
218 <h2 id="row_value_in_operators"><span>2.2. </span>Row Value IN Operators</h2>
219
220 <p>For a row-value <a href="lang_expr.html#in_op">IN operator</a>, the left-hand side (hereafter "LHS") can be either
221 a parenthesized list of values or a subquery with multiple columns. But the
222 right-hand side (hereafter "RHS") must be a subquery expression.
223
224 </p><div class="codeblock"><pre>CREATE TABLE t2(x,y,z);
225 INSERT INTO t2(x,y,z) VALUES(1,2,3),(2,3,4),(1,NULL,5);
226 SELECT
227 (1,2,3) IN (SELECT * FROM t2), -- 1
228 (7,8,9) IN (SELECT * FROM t2), -- 0
229 (1,3,5) IN (SELECT * FROM t2); -- NULL
230 </pre></div>
231
232 <h2 id="row_values_in_update_statements"><span>2.3. </span>Row Values In UPDATE Statements</h2>
233
234 <p>Row values can also be used in the SET clause of an <a href="lang_update.html">UPDATE</a> statement.
235 The LHS must be a list of column names. The RHS can be any row value.
236 For example:
237
238 </p><div class="codeblock"><pre>UPDATE tab3
239 SET (a,b,c) = (SELECT x,y,z
240 FROM tab4
241 WHERE tab4.w=tab3.d)
242 WHERE tab3.e BETWEEN 55 AND 66;
243 </pre></div>
244
245 <h1 id="example_uses_of_row_values"><span>3. </span>Example Uses Of Row Values</h1>
246
247 <h2 id="scrolling_window_queries"><span>3.1. </span>Scrolling Window Queries</h2>
248
249 <p>Suppose an application wants to display a list of contacts
250 in alphabetical order by lastname, firstname, in a scrolling window
251 that can only show 7 contacts at a time. Initializing the scrolling
252 window to the first 7 entries is easy:
253
254 </p><div class="codeblock"><pre>SELECT * FROM contacts
255 ORDER BY lastname, firstname
256 LIMIT 7;
257 </pre></div>
258
259 <p>When the user scrolls down, the application needs to find the
260 second set of 7 entries. One way to do this is to use the OFFSET clause:
261
262 </p><div class="codeblock"><pre>SELECT * FROM contacts
263 ORDER BY lastname, firstname
264 LIMIT 7 OFFSET 7;
265 </pre></div>
266
267 <p>OFFSET gives the correct answer. However, OFFSET requires time
268 proportional to the offset value. What really happens
269 with "LIMIT x OFFSET y" is that SQLite computes the query as
270 "LIMIT x+y" and discards the first y values without returning them
271 to the application. So as the window scrolls down toward
272 the bottom of a long list, and the y value becomes larger and larger,
273 successive offset computations take more and more time.
274
275 </p><p>A more efficient approach is to remember the last entry currently
276 displayed and then use a row value comparison in the WHERE
277 clause:
278
279 </p><div class="codeblock"><pre>SELECT * FROM contacts
280 WHERE (lastname,firstname) &gt; (?1,?2)
281 ORDER BY lastname, firstname
282 LIMIT 7;
283 </pre></div>
284
285 <p>If the lastname and firstname on the bottom row of the previous
286 screen are bound to ?1 and ?2, then the query above computes the next
287 7 rows. And, assuming there is an appropriate index, it does so
288 very efficiently &mdash; much more efficiently than OFFSET.
289
290 </p><h2 id="comparison_of_dates_stored_as_separate_fields"><span>3.2. </span>Comparison of dates stored as separate fields</h2>
291
292 <p>The usual way of storing a date in a database table is as a single
293 field, as either a unix timestamp, a julian day number, or an ISO-8601
294 dates string. But some applications store dates as three separate
295 fields for the year, month, and day.
296
297 </p><div class="codeblock"><pre>CREATE TABLE info(
298 year INT, -- 4 digit year
299 month INT, -- 1 through 12
300 day INT, -- 1 through 31
301 other_stuff BLOB -- blah blah blah
302 );
303 </pre></div>
304
305 <p>When dates are stored this way, row value comparisons provide a
306 convenient way to compare dates:
307
308 </p><div class="codeblock"><pre>SELECT * FROM info
309 WHERE (year,month,day) BETWEEN (2015,9,12) AND (2016,9,12);
310 </pre></div>
311
312 <h2 id="search_against_multi_column_keys"><span>3.3. </span>Search against multi-column keys</h2>
313
314 <p>Suppose we want to know the order number, product number, and quantity
315 for any item in which the product number and quantity match the product
316 number and quantity of any item in order number 365:
317
318 </p><div class="codeblock"><pre>SELECT ordid, prodid, qty
319 FROM item
320 WHERE (prodid, qty) IN (SELECT prodid, qty
321 FROM item
322 WHERE ordid = 365);
323 </pre></div>
324
325 <p>The query above could be rewritten as a join and without the use
326 of row values:
327
328 </p><div class="codeblock"><pre>SELECT t1.ordid, t1.prodid, t1.qty
329 FROM item AS t1, item AS t2
330 WHERE t1.prodid=t2.prodid
331 AND t1.qty=t2.qty
332 AND t2.ordid=365;
333 </pre></div>
334
335 <p>Because the same query could be written without the use of row values,
336 row values do not provide new capabilities. However, many developers say
337 that the row value format is easier to read, write, and debug.
338
339 </p><p>Even in the JOIN form, the query can be made clearer through the use of
340 row values:
341
342 </p><div class="codeblock"><pre>SELECT t1.ordid, t1.prodid, t1.qty
343 FROM item AS t1, item AS t2
344 WHERE (t1.prodid,t1.qty) = (t2.prodid,t2.qty)
345 AND t2.ordid=365;
346 </pre></div>
347
348 <p>This later query generates exactly the same <a href="opcode.html">bytecode</a> as the previous
349 scalar formulation, but using syntax that is cleaner and
350 easier to read.
351
352 </p><h2 id="update_multiple_columns_of_a_table_based_on_a_query"><span>3.4. </span>Update multiple columns of a table based on a query</h2>
353
354 <p>The row-value notation is useful for updating two or more columns
355 of a table from the result of a single query.
356 An example of this is in the full-text search feature of the
357 <a href="https://www.fossil-scm.org/">Fossil version control system</a>.
358
359 </p><p>In the Fossil full-text search system,
360 documents that participate in the full-text search (wiki pages, tickets,
361 check-ins, documentation files, etc) are tracked by a table called
362 "ftsdocs" (<u>f</u>ull <u>t</u>ext <u>s</u>earch <u>doc</u>ument<u>s</u>).
363 As new documents are added to the repository, they are not indexed right
364 away. Indexing is deferred until there is a search request. The
365 ftsdocs table contains an "idxed" field which is true if the document
366 has been indexed and false if not.
367
368 </p><p>When a search request occurs and pending documents are indexed for the
369 first time, the ftsdocs table must be updated by setting the idxed column
370 to true and also filling in several other columns with information pertinent
371 to the search. That other information is obtained from a join. The
372 query is this:
373
374 </p><div class="codeblock"><pre>UPDATE ftsdocs SET
375 idxed=1,
376 name=NULL,
377 (label,url,mtime) =
378 (SELECT printf('Check-in &#91;%%.16s&#93; on %%s',blob.uuid,
379 datetime(event.mtime)),
380 printf('/timeline?y=ci&amp;c=%%.20s',blob.uuid),
381 event.mtime
382 FROM event, blob
383 WHERE event.objid=ftsdocs.rid
384 AND blob.rid=ftsdocs.rid)
385 WHERE ftsdocs.type='c' AND NOT ftsdocs.idxed
386 </pre></div>
387
388 <p>(See the
389 <a href="https://www.fossil-scm.org/fossil/artifact/e5d6a82d?ln=1594-1605">source code</a>
390 for further detail. Other examples
391 <a href="https://www.fossil-scm.org/fossil/artifact/e5d6a82d?ln=1618-1628">here</a> and
392 <a href="https://www.fossil-scm.org/fossil/artifact/e5d6a82d?ln=1641-1650">here</a>.)
393
394 </p><p>Five out of nine columns in the ftsdocs table are updated. Two of
395 the modified columns, "idxed" and "name", can be updated independently of
396 the query. But the three columns "label", "url", and "mtime" all require
397 a join query against the "event" and "blob" tables. Without row values,
398 the equivalent UPDATE would require that the join be repeated three times,
399 once for each column to be updated.
400
401 </p><h2 id="clarity_of_presentation"><span>3.5. </span>Clarity of presentation</h2>
402
403 <p>Sometimes the use of row values just makes the SQL easier to read
404 and write. Consider the following two UPDATE statements:
405
406 </p><div class="codeblock"><pre>UPDATE tab1 SET (a,b)=(b,a);
407 UPDATE tab1 SET a=b, b=a;
408 </pre></div>
409
410 <p>Both UPDATE statements do exactly the same thing. (They generate
411 identical <a href="opcode.html">bytecode</a>.) But the first form, the row value form, seems
412 to make it clearer that the intent of the statement is to swap the
413 values in columns A and B.
414
415 </p><p>Or consider these identical queries:
416
417 </p><div class="codeblock"><pre>SELECT * FROM tab1 WHERE a=?1 AND b=?2;
418 SELECT * FROM tab1 WHERE (a,b)=(?1,?2);
419 </pre></div>
420
421 <p>Once again, the SQL statements generate identical bytecode and thus
422 do exactly the same job in exactly the same way. But the second form
423 is made easier for humans to read by grouping the query parameters together
424 into a single row value rather than scattering them across the WHERE
425 clause.
426
427 </p><h1 id="backwards_compatibility"><span>4. </span>Backwards Compatibility</h1>
428
429 <p>Row values were added to SQLite
430 <a href="releaselog/3_15_0.html">version 3.15.0</a> (2016-10-14). Attempts to use row values in
431 prior versions of SQLite will generate syntax errors.
432 </p><p align="center"><small><i>This page was last updated on 2025-05-31 13:08:22Z </small></i></p>
433